" + Encode(l.Title) + "
"); + sb.AppendLine("" + Encode(summary) + "
"); + sb.AppendLine(" "); + sb.AppendLine("` | Force the language for FCE language-gated rules (code or English name, e.g. `es` / `Spanish`); default: auto-detect from content |
| `--fix-rtl-via-unicode-chars` | Fix RTL via Unicode characters |
@@ -532,7 +632,7 @@ In most batch pipelines `--apply-min-gap` is the better choice; reach for the FC
### FixCommonErrors rule selection
-`--fix-common-errors` (no value) runs all 39 rules. Pass `--fix-common-errors-rules:` to pick a subset — supplying that option implies `--fix-common-errors`.
+`--fix-common-errors` (no value) runs all 40 rules (39 fixes plus the `FixCommonOcrErrors` pass). Pass `--fix-common-errors-rules:` to pick a subset — supplying that option implies `--fix-common-errors`.
```bash
seconv movie.srt subrip --fix-common-errors # all rules
@@ -645,18 +745,19 @@ This mirrors the desktop app, where batch convert's *Remove formatting* function
| `unipac`, `pacunicode` | PAC Unicode |
| `ebu`, `ebustl`, `stl` | EBU STL — binary |
| `cavena`, `cavena890` | Cavena 890 — binary |
-| `cheetahcaption` | CheetahCaption — binary |
-| `capmakerplus` | CapMakerPlus — binary |
+| `cheetah`, `cheetahcaption` | CheetahCaption — binary |
+| `capmaker`, `capmakerplus` | CapMakerPlus — binary |
| `ayato` | Ayato — binary |
-| `bluraysup`, `sup` | Blu-Ray sup — image |
+| `bluraysup`, `blurayup`, `sup` | Blu-Ray sup — image |
| `vobsub` | VobSub — image |
| `bdnxml`, `bdn-xml` | BDN-XML — image (folder of PNGs + index.xml) |
| `bdnxml8bit`, `bdn-xml8-bit` | BDN-XML with 8-bit palette-indexed PNGs — image |
| `dost`, `dostimage` | DOST/image |
| `fcpimage`, `fcp` | FCP/image |
-| `dcinemainterop` | D-Cinema interop/png |
-| `dcinemasmpte2014` | D-Cinema SMPTE 2014/png |
-| `imageswithtimecode` | Images with time codes in file name |
+| `dcinemainterop`, `dcinema-interop` | D-Cinema interop/png |
+| `dcinemasmpte2014`, `dcinema-smpte` | D-Cinema SMPTE 2014/png |
+| `imageswithtimecode`, `imagesintc` | Images with time codes in file name |
+| `webvttthumbnail`, `webvtt-thumbnail`, `vttthumb` | WebVTT Thumbnail — image (sprite sheet + `.vtt`) |
| `plaintext`, `text`, `txt` | Plain text (HTML stripped) |
| `customtext`, `customtextformat` | Custom-templated text (requires `--custom-format`) |
diff --git a/docs/reference/keyboard-shortcuts.md b/docs/reference/keyboard-shortcuts.md
index 9452061b34d..e8345faaf89 100644
--- a/docs/reference/keyboard-shortcuts.md
+++ b/docs/reference/keyboard-shortcuts.md
@@ -80,6 +80,8 @@ Inside the [source view](../features/source-view.md) window (F2), which edits th
| Shortcut | Action |
|----------|--------|
+| Ctrl+Z | Undo |
+| Ctrl+Shift+Z / Ctrl+Y | Redo |
| Ctrl+F | Find in the source |
| Ctrl+H | Replace in the source (Cmd+Alt+F also works on macOS) |
| F3 / Shift+F3 | Find next / previous |
diff --git a/docs/reference/mouse-controls.md b/docs/reference/mouse-controls.md
index bc2a2c131f6..5aac4e09155 100644
--- a/docs/reference/mouse-controls.md
+++ b/docs/reference/mouse-controls.md
@@ -9,7 +9,9 @@ See also: [Audio Visualizer](../features/audio-visualizer.md) | [Video Player](.
| Action | Effect |
|--------|--------|
| Click on video | Play/Pause toggle |
-| Double-click on video | Toggle fullscreen |
+| Scroll wheel over video | Seek ±0.5 s per notch (direction follows the waveform's "Invert mouse-wheel" setting) |
+
+Full screen is toggled with Alt+Enter — see [Keyboard Shortcuts](keyboard-shortcuts.md).
## Audio Visualizer / Waveform
@@ -18,19 +20,23 @@ See also: [Audio Visualizer](../features/audio-visualizer.md) | [Video Player](.
| Left-click | Set current position |
| Left-click + drag on subtitle edge | Adjust start/end time of subtitle |
| Left-click + drag in empty area | Create a new selection |
-| Right-click | Open context menu |
-| Scroll wheel | Scroll waveform horizontally |
+| Shift + Click | Set the selected subtitle's start time to the clicked position |
+| Ctrl (or Cmd) + Click | Set the selected subtitle's end time to the clicked position |
+| Ctrl + Shift + Click | Set the selected subtitle's start time and offset all following subtitles by the same amount |
+| Alt + Click | Move the selected subtitle's start to the clicked position, keeping its duration |
+| Right-click | Select the subtitle under the cursor ("Select subtitle on right click", on by default), move the video position there, and open the context menu. On macOS, Ctrl + left-click opens the same menu |
+| Scroll wheel | Scroll waveform horizontally — or seek the video by one step per notch when "Mouse-wheel sets video position" is on; "Invert mouse-wheel" flips the direction (Options → Settings → Waveform/spectrogram) |
| Alt + Scroll wheel | Horizontal zoom in/out |
| Shift + Scroll wheel | Vertical zoom in/out |
| Ctrl (or Cmd) + Scroll wheel | Scroll waveform and seek video to cursor |
-| Double-click on subtitle | Select the subtitle in the grid |
+| Double-click | Nothing by default — set "Waveform double-click action" in Options → Settings → Waveform/spectrogram to select the subtitle, center on it, pause, or play |
## Subtitle Grid
| Action | Effect |
|--------|--------|
| Click on row | Select subtitle line |
-| Double-click on row | Edit subtitle line in text editor |
+| Double-click on row | Go to the subtitle in the video and pause (default; "Subtitle grid double-click action" in Options → Settings → General offers other actions, e.g. play, or also focus the text box) |
| Ctrl + Click | Toggle selection of individual rows |
| Shift + Click | Select range of rows |
| Right-click | Open context menu |
diff --git a/docs/reference/supported-formats.md b/docs/reference/supported-formats.md
index 36756692360..e852529d90d 100644
--- a/docs/reference/supported-formats.md
+++ b/docs/reference/supported-formats.md
@@ -17,11 +17,12 @@ Subtitle Edit supports a wide range of subtitle formats for reading and writing.
| EBU STL | .stl | |
| Spruce STL | .stl | |
| Scenarist Closed Captions (SCC) | .scc | |
-| DVD Studio Pro | .txt | |
+| DVD Studio Pro | .stl | |
| Cavena 890 | .890 | |
| PAC | .pac | |
| Cheetah | .cap | |
-| Avid DS | .txt | |
+| Avid DVD, Avid Caption, Avid Caption Drop Frame, Avid Loc Markers | .txt | |
+| Avid STL | .stl | |
| JSON (various) | .json | |
| LRC (Lyrics) | .lrc | |
| and many more... | | |
@@ -47,6 +48,16 @@ Formats used to move captions and markers between Subtitle Edit and video editor
| Adobe Premiere PrProj Xml | .xml |
| Adobe Encore (tabs) | .txt |
+## Spreadsheets
+
+Import only — a spreadsheet with a header row naming start/end/text columns is read directly, and any layout can be mapped by hand in *File → Import → CSV/XLSX/ODS with custom columns*. See [Import Spreadsheets](../features/import-csv-xlsx.md).
+
+| Format | Extension(s) |
+|--------|--------------|
+| Excel workbook (Office Open XML) | .xlsx |
+| OpenDocument spreadsheet | .ods |
+| Delimited text (comma, semicolon, tab, pipe) | .csv, .tsv, .txt |
+
## Image-Based Formats
| Format | Extension(s) |
@@ -55,16 +66,25 @@ Formats used to move captions and markers between Subtitle Edit and video editor
| VobSub (DVD) | .sub/.idx |
| BDN XML | .xml |
| Timed Text Base64 Image (SMPTE-TT bitmap) | .xml |
+| Timed Text Image | .xml |
+| BDN XML 8-bit (palette-indexed PNGs, export) | .xml |
+| DOST | .dost |
+| Final Cut Pro Image (FCP/image) | .xml |
+| SPU Image | .xml |
+| D-Cinema interop/png, D-Cinema SMPTE 2014/png (export) | .xml |
+| Images with time codes in file name (export) | image files |
+| WebVTT Thumbnail (sprite sheet) | .vtt |
## Container Formats (with embedded subtitles)
| Format | Extension(s) |
|--------|--------------|
-| Matroska (MKV/MKS) | .mkv, .mks |
-| MP4 / MOV (text tracks, including fragmented MP4/DASH — wvtt, stpp, tx3g) | .mp4, .m4v, .m4s, .3gp |
+| Matroska (MKV/MKS/WebM) | .mkv, .mks, .webm |
+| MP4 / MOV (text tracks, including fragmented MP4/DASH — wvtt, stpp, tx3g) | .mp4, .m4v, .m4s, .3gp, .mov |
| Transport Stream (teletext, DVB-sub) | .ts, .m2ts, .mts |
-| AVI (XSUB) | .avi |
+| AVI (XSUB) | .avi, .divx |
| MacCaption | .mcc |
+| MXF (timed-text essences) | .mxf |
## Video Formats (for loading video)
diff --git a/docs/screenshots/README.md b/docs/screenshots/README.md
index b016846a4e1..0a3a2d47bae 100644
--- a/docs/screenshots/README.md
+++ b/docs/screenshots/README.md
@@ -37,6 +37,7 @@ All listed files exist on disk. Filenames are stable and may be linked from `../
- `multiple-replace.png` — Multiple replace window
- `modify-selection.png` — Modify selection window
- `show-history.png` — Show history window
+- `ai-assistant.png` — AI assistant window
## Tools
@@ -87,6 +88,8 @@ All listed files exist on disk. Filenames are stable and may be linked from `../
- `blank-video.png` — Blank video window
- `cut-video.png` — Cut video window
- `re-encode-video.png` — Re-encode video window
+- `video-ocr.png` — Video OCR window
+- `chapters.png` — Chapters window
## Translation
@@ -157,6 +160,10 @@ Used by `../reference/assa-override-tags.md` to illustrate individual override t
## Counts
-- Top-level screenshots: **92**
+- Top-level screenshots: **95**
- ASSA override-tag examples (`assa/`): **35**
-- Total: **127**
+- Total: **130**
+
+## How the dialog screenshots are made
+
+Most dialog screenshots are headless renders of the real windows (dark theme, English, sample subtitle loaded), captured with Avalonia's headless platform and real Skia drawing — the same setup the UI tests use (`UseHeadless(new AvaloniaHeadlessPlatformOptions { UseHeadlessDrawing = false }).UseSkia()`, see `tests/UI/Logic/Accessibility/EditBoxAccessibilityNameTests.cs`). A temporary `[AvaloniaFact]` resolves the view model from `AddSubtitleEditServices()`, calls its `Initialize(...)` with sample lines, constructs the window, runs `Dispatcher.UIThread.RunJobs()` and saves `window.CaptureRenderedFrame()`. Such renders have no OS title bar, and video/waveform areas stay black because libmpv does not run headlessly, so windows that are mostly video (main window, video player) are still real screenshots. PNGs are quantized to a 256-color palette before committing.
diff --git a/docs/screenshots/ai-assistant.png b/docs/screenshots/ai-assistant.png
new file mode 100644
index 00000000000..b9acd45e219
Binary files /dev/null and b/docs/screenshots/ai-assistant.png differ
diff --git a/docs/screenshots/ai-review.png b/docs/screenshots/ai-review.png
index f5d126baa74..a9f19a58f8d 100644
Binary files a/docs/screenshots/ai-review.png and b/docs/screenshots/ai-review.png differ
diff --git a/docs/screenshots/apply-duration-limits.png b/docs/screenshots/apply-duration-limits.png
index dc448aca621..567011795b3 100644
Binary files a/docs/screenshots/apply-duration-limits.png and b/docs/screenshots/apply-duration-limits.png differ
diff --git a/docs/screenshots/assa-attachments.png b/docs/screenshots/assa-attachments.png
index fbc65d9cce8..d0411b15a61 100644
Binary files a/docs/screenshots/assa-attachments.png and b/docs/screenshots/assa-attachments.png differ
diff --git a/docs/screenshots/assa-set-position.png b/docs/screenshots/assa-set-position.png
index 43db8d2d039..98e40406a37 100644
Binary files a/docs/screenshots/assa-set-position.png and b/docs/screenshots/assa-set-position.png differ
diff --git a/docs/screenshots/assa-styles.png b/docs/screenshots/assa-styles.png
index a7fb4b9d2ea..e26295eaa93 100644
Binary files a/docs/screenshots/assa-styles.png and b/docs/screenshots/assa-styles.png differ
diff --git a/docs/screenshots/auto-translate.png b/docs/screenshots/auto-translate.png
index f6fe3b32841..d1aff685676 100644
Binary files a/docs/screenshots/auto-translate.png and b/docs/screenshots/auto-translate.png differ
diff --git a/docs/screenshots/batch-convert.png b/docs/screenshots/batch-convert.png
index 48a7c466578..d5fab211087 100644
Binary files a/docs/screenshots/batch-convert.png and b/docs/screenshots/batch-convert.png differ
diff --git a/docs/screenshots/beautify-time-codes.png b/docs/screenshots/beautify-time-codes.png
index 1fd6717eaf0..19e9fa50de0 100644
Binary files a/docs/screenshots/beautify-time-codes.png and b/docs/screenshots/beautify-time-codes.png differ
diff --git a/docs/screenshots/burn-in.png b/docs/screenshots/burn-in.png
index 1fc9171ffa9..0ebe4fc9f7d 100644
Binary files a/docs/screenshots/burn-in.png and b/docs/screenshots/burn-in.png differ
diff --git a/docs/screenshots/change-speed.png b/docs/screenshots/change-speed.png
index 46ed6845ae8..bf4a9fb47c4 100644
Binary files a/docs/screenshots/change-speed.png and b/docs/screenshots/change-speed.png differ
diff --git a/docs/screenshots/chapters.png b/docs/screenshots/chapters.png
new file mode 100644
index 00000000000..86a162bd279
Binary files /dev/null and b/docs/screenshots/chapters.png differ
diff --git a/docs/screenshots/compare.png b/docs/screenshots/compare.png
index bb4409cb799..2e76701d64b 100644
Binary files a/docs/screenshots/compare.png and b/docs/screenshots/compare.png differ
diff --git a/docs/screenshots/convert-actors.png b/docs/screenshots/convert-actors.png
index 519f776ebe2..665bb3a580c 100644
Binary files a/docs/screenshots/convert-actors.png and b/docs/screenshots/convert-actors.png differ
diff --git a/docs/screenshots/export-image-based.png b/docs/screenshots/export-image-based.png
index f8cbd94301b..fe3c1d45126 100644
Binary files a/docs/screenshots/export-image-based.png and b/docs/screenshots/export-image-based.png differ
diff --git a/docs/screenshots/find.png b/docs/screenshots/find.png
index f0c229fc8e6..6ba737dfee6 100644
Binary files a/docs/screenshots/find.png and b/docs/screenshots/find.png differ
diff --git a/docs/screenshots/fix-names.png b/docs/screenshots/fix-names.png
index a592999727b..46044600b7c 100644
Binary files a/docs/screenshots/fix-names.png and b/docs/screenshots/fix-names.png differ
diff --git a/docs/screenshots/import-plain-text.png b/docs/screenshots/import-plain-text.png
index 150b6cfff41..ac4ac847728 100644
Binary files a/docs/screenshots/import-plain-text.png and b/docs/screenshots/import-plain-text.png differ
diff --git a/docs/screenshots/modify-selection.png b/docs/screenshots/modify-selection.png
index 84822cc4e3d..9720b558b29 100644
Binary files a/docs/screenshots/modify-selection.png and b/docs/screenshots/modify-selection.png differ
diff --git a/docs/screenshots/netflix-errors.png b/docs/screenshots/netflix-errors.png
index e742b48a92a..a2e6b76953e 100644
Binary files a/docs/screenshots/netflix-errors.png and b/docs/screenshots/netflix-errors.png differ
diff --git a/docs/screenshots/point-sync-via-other.png b/docs/screenshots/point-sync-via-other.png
index 5e69ee9ec1c..1676429e289 100644
Binary files a/docs/screenshots/point-sync-via-other.png and b/docs/screenshots/point-sync-via-other.png differ
diff --git a/docs/screenshots/point-sync.png b/docs/screenshots/point-sync.png
index 4b0f48496cf..e785c62fdc3 100644
Binary files a/docs/screenshots/point-sync.png and b/docs/screenshots/point-sync.png differ
diff --git a/docs/screenshots/remove-text-hi.png b/docs/screenshots/remove-text-hi.png
index 2bcb11230b4..9b88cafb781 100644
Binary files a/docs/screenshots/remove-text-hi.png and b/docs/screenshots/remove-text-hi.png differ
diff --git a/docs/screenshots/settings.png b/docs/screenshots/settings.png
index 174484b27b8..ab5f842e90f 100644
Binary files a/docs/screenshots/settings.png and b/docs/screenshots/settings.png differ
diff --git a/docs/screenshots/speech-to-text.png b/docs/screenshots/speech-to-text.png
index d804a7185eb..692fe42ed11 100644
Binary files a/docs/screenshots/speech-to-text.png and b/docs/screenshots/speech-to-text.png differ
diff --git a/docs/screenshots/spell-check.png b/docs/screenshots/spell-check.png
index 411339325bc..88102532a84 100644
Binary files a/docs/screenshots/spell-check.png and b/docs/screenshots/spell-check.png differ
diff --git a/docs/screenshots/split-break-long-lines.png b/docs/screenshots/split-break-long-lines.png
index a2bb918b0ee..31dbe667710 100644
Binary files a/docs/screenshots/split-break-long-lines.png and b/docs/screenshots/split-break-long-lines.png differ
diff --git a/docs/screenshots/statistics.png b/docs/screenshots/statistics.png
index 9040f8b1ce1..c069e9db424 100644
Binary files a/docs/screenshots/statistics.png and b/docs/screenshots/statistics.png differ
diff --git a/docs/screenshots/text-to-speech.png b/docs/screenshots/text-to-speech.png
index 235c4ebae66..fec652a383c 100644
Binary files a/docs/screenshots/text-to-speech.png and b/docs/screenshots/text-to-speech.png differ
diff --git a/docs/screenshots/transparent-subtitles.png b/docs/screenshots/transparent-subtitles.png
index fb87bc76bda..e1d85b6c575 100644
Binary files a/docs/screenshots/transparent-subtitles.png and b/docs/screenshots/transparent-subtitles.png differ
diff --git a/docs/screenshots/video-ocr.png b/docs/screenshots/video-ocr.png
new file mode 100644
index 00000000000..bd9dd6d2f5d
Binary files /dev/null and b/docs/screenshots/video-ocr.png differ
diff --git a/docs/screenshots/visual-sync.png b/docs/screenshots/visual-sync.png
index c4c06c8ea8c..aef6d1395d9 100644
Binary files a/docs/screenshots/visual-sync.png and b/docs/screenshots/visual-sync.png differ
diff --git a/docs/screenshots/word-lists.png b/docs/screenshots/word-lists.png
index 02feda4e648..15e844367b3 100644
Binary files a/docs/screenshots/word-lists.png and b/docs/screenshots/word-lists.png differ
diff --git a/docs/third-party-components.md b/docs/third-party-components.md
index ce0d8d5b564..4bf6c739421 100644
--- a/docs/third-party-components.md
+++ b/docs/third-party-components.md
@@ -36,8 +36,7 @@ Subtitle Edit stores these components in its **Data Folder**.
| **Purfview Faster-Whisper XXL** | `faster-whisper-xxl.exe`, `_models/` folder | `[Data Folder]/SpeechToText/Purfview-Faster-Whisper-XXL` |
| **Crisp ASR** | `crispasr.exe`, `models/` folder | `[Data Folder]/CrispASR` |
| **Qwen3 ASR CPP** | `qwen3-asr-cli.exe`, `models/` folder | `[Data Folder]/Qwen3ASR` |
-| **Parakeet.cpp** | `parakeet.exe`, model folders | `[Data Folder]/parakeet.cpp` |
-| **PaddleOCR** | `paddleocr.exe`, `models/` folder | `[Data Folder]/OCR/PaddleOCR3-4` |
+| **PaddleOCR** | `paddleocr.exe`, `models/` folder | `[Data Folder]/OCR/PaddleOCR3-7` |
| **Qwen3 TTS (CrispASR)** | shares `crispasr.exe` + `models/` from `[Data Folder]/CrispASR`; reference voices in `voices/` | `[Data Folder]/TextToSpeech/Qwen3TtsCrispAsr` (voices only) |
| **Chatterbox TTS (CrispASR)** | shares `crispasr.exe` + `models/` from `[Data Folder]/CrispASR`; reference voices in `voices/` | `[Data Folder]/TextToSpeech/Chatterbox` (voices only) |
| **OmniVoice TTS** | `omnivoice-tts.exe`, `omnivoice-codec.exe`, `models/`, `voices/` | `[Data Folder]/TextToSpeech/OmniVoice` |
@@ -57,7 +56,9 @@ Used for reading media info, extracting audio, and generating waveforms.
* **Alternative: Custom Path**
* You can point to an existing FFmpeg installation in `Settings.json` (located in the Data Folder):
```json
- "FfmpegPath": "C:\\path\\to\\your\\ffmpeg.exe"
+ "General": {
+ "FfmpegPath": "C:\\path\\to\\your\\ffmpeg.exe"
+ }
```
* Use double backslashes (`\\`) for Windows paths in JSON.
@@ -110,30 +111,33 @@ Used for GPU-accelerated AI-based speech recognition.
### SE5 Speech-to-Text Engines
Subtitle Edit 5 can download additional ASR engines directly from the **Speech to text** window.
-* **Crisp ASR:** Stored in `[Data Folder]/CrispASR`. Models go into its `models` folder. Crisp ASR backends include Parakeet, Canary, Cohere, Fire Red, GLM, Granite, Qwen3, Mega, Omni, and Kyutai.
+* **Crisp ASR:** Stored in `[Data Folder]/CrispASR`. Models go into its `models` folder. Crisp ASR backends include Parakeet, Canary, Cohere, Fire Red, Fun-ASR Nano, GigaAM, GLM, Granite, Qwen3, Mega, MOSS Diarize, Omni, Kyutai, SenseVoice, ARK, and Voxtral.
* The speech-to-text dialog also offers a **Forced aligner** combo for word-level timestamps. Built-in (where the backend supports it), Canary CTC, Qwen3, and 12 language-specific wav2vec2 aligners (the WhisperX aligner zoo): `en`, `de`, `fr`, `es`, `it`, `ja`, `zh`, `nl`, `pt`, `ar`, `uk`, `cs`. The default is the built-in aligner when the backend supports it, otherwise Qwen3 or Canary CTC depending on the backend; pick a wav2vec2 entry manually to use one of those.
-* **Qwen3 ASR CPP:** Stored in `[Data Folder]/Qwen3ASR`. Models go into `[Data Folder]/Qwen3ASR/models`.
-* **Parakeet.cpp:** Stored in `[Data Folder]/parakeet.cpp`. Each model has its own folder because the model weights and `vocab.txt` must stay together.
+* **Qwen3 ASR CPP:** Stored in `[Data Folder]/Qwen3ASR`. Models go into `[Data Folder]/Qwen3ASR/models`. (Parakeet is no longer a standalone engine; it is a Crisp ASR backend.)
Use [Speech to Text](features/speech-to-text.md) for the current engine list and workflow.
### PaddleOCR
Used for OCR of image-based subtitles.
-* **Destination:** `[Data Folder]/OCR/PaddleOCR3-4`
-* **Models:** `[Data Folder]/OCR/PaddleOCR3-4/models`
+* **Destination:** `[Data Folder]/OCR/PaddleOCR3-7`
+* **Models:** `[Data Folder]/OCR/PaddleOCR3-7/models`
* **Builds:** Subtitle Edit can download CPU, CUDA 11.8, or CUDA 12.9 builds, on both Windows and Linux (Linux x64 only; the Linux builds need glibc 2.35 or newer).
-* **Version:** The folder name follows the PaddleOCR release the standalone engine is built from, so upgrading installs into a new folder instead of mixing files. The old `[Data Folder]/OCR/PaddleOCR3-1` folder is deleted automatically after the new one is installed.
+* **Recognition models:** PP-OCRv6 (PaddleOCR 3.7) recognizes Chinese, English, Japanese and the Latin languages with one unified model; Arabic, Cyrillic, E-Slavic, Devanagari, Korean, Greek, Tamil, Telugu, Thai and Pali still use their PP-OCRv5 models, and Georgian its PP-OCRv3 one. All of them ship in the same bundle.
+* **Version:** The folder name follows the PaddleOCR release the standalone engine is built from, so upgrading installs into a new folder instead of mixing files. The old `[Data Folder]/OCR/PaddleOCR3-1` and `[Data Folder]/OCR/PaddleOCR3-4` folders are deleted automatically after the new one is installed.
+* **Paddle OCR Python:** The pip-installed engine uses the same downloaded models, so it needs `paddleocr` 3.7 or newer - older versions do not know the PP-OCRv6 model names.
### Local Text-to-Speech Engines
Subtitle Edit 5 can download local TTS servers and models from the **Text to speech** window.
-* **Qwen3 TTS (CrispASR):** Reference voices are stored in `[Data Folder]/TextToSpeech/Qwen3TtsCrispAsr/voices`. The talker GGUFs (VoiceDesign 1.7B or CustomVoice 1.7B) and the 12 Hz codec are downloaded into the shared `[Data Folder]/CrispASR/models` cache alongside the Crisp ASR speech-to-text models, not under `TextToSpeech/Qwen3TtsCrispAsr/models` — installing Crisp ASR first is therefore recommended. Older installs that still have model files under the legacy `TextToSpeech/Qwen3TtsCrispAsr/models` folder are migrated automatically the first time the engine is used.
+* **Qwen3 TTS (CrispASR):** Reference voices are stored in `[Data Folder]/TextToSpeech/Qwen3TtsCrispAsr/voices`. The talker GGUFs (VoiceDesign 1.7B, CustomVoice 1.7B or Voice clone Base 1.7B) and the 12 Hz codec are downloaded into the shared `[Data Folder]/CrispASR/models` cache alongside the Crisp ASR speech-to-text models, not under `TextToSpeech/Qwen3TtsCrispAsr/models` — installing Crisp ASR first is therefore recommended. Older installs that still have model files under the legacy `TextToSpeech/Qwen3TtsCrispAsr/models` folder are migrated automatically the first time the engine is used.
* **Chatterbox TTS (CrispASR):** Reference voices are stored in `[Data Folder]/TextToSpeech/Chatterbox/voices`. The Base / Turbo model GGUFs (T3 + S3Gen) are downloaded into the shared `[Data Folder]/CrispASR/models` cache alongside the Crisp ASR speech-to-text models, not under `TextToSpeech/Chatterbox/models` — installing Crisp ASR first is therefore recommended. Older installs that still have model files under the legacy `TextToSpeech/Chatterbox/models` folder are migrated automatically the first time the engine is used.
* **OmniVoice TTS:** Stored in `[Data Folder]/TextToSpeech/OmniVoice`. Brings its own `omnivoice-tts` and `omnivoice-codec` binaries. Supports 646 languages and voice cloning on CPU. `models/` and `voices/` subfolders.
* **Kokoro TTS:** Stored in `[Data Folder]/TextToSpeech/KokoroTtsCpp`. Models go into the `models` folder.
-Use [Text to Speech](features/text-to-speech.md) for engine-specific options.
+These are examples, not the full set — many more local engines are downloadable from the Text to speech window (IndexTTS, CosyVoice3, dots.tts, VoxCPM2, MOSS-TTS, Zonos, VibeVoice, Confucius4-TTS, Pocket TTS, Higgs Audio, Fish Audio, and more), following the same layout: CrispASR-based engines share the `[Data Folder]/CrispASR` cache, and the rest live under `[Data Folder]/TextToSpeech/`.
+
+Use [Text to Speech](features/text-to-speech.md) for the full engine list and engine-specific options.
---
@@ -147,7 +151,9 @@ Used for reading media info, extracting audio, and generating waveforms.
* **Alternative:** Place the `ffmpeg` binary in `[Data Folder]/ffmpeg`.
* **Custom Path:** You can specify a custom path in `Settings.json` (located in the Data Folder):
```json
- "FfmpegPath": "/path/to/your/ffmpeg"
+ "General": {
+ "FfmpegPath": "/path/to/your/ffmpeg"
+ }
```
### MPV Media Player (libmpv)
@@ -190,7 +196,7 @@ Used for GPU-accelerated AI-based speech recognition.
### SE5 Speech-to-Text, OCR, and TTS Engines
-The same data-folder layout is used on Linux. Prefer the in-app downloaders for Crisp ASR, Qwen3 ASR, Parakeet.cpp, PaddleOCR, Qwen3 TTS (CrispASR), Chatterbox TTS (CrispASR), OmniVoice TTS, and Kokoro TTS because the required files differ by build and model.
+The same data-folder layout is used on Linux. Prefer the in-app downloaders for Crisp ASR, Qwen3 ASR, PaddleOCR, and the local TTS engines because the required files differ by build and model.
---
@@ -204,7 +210,9 @@ Used for reading media info, extracting audio, and generating waveforms.
* **Alternative:** Place the `ffmpeg` binary in `[Data Folder]/ffmpeg`.
* **Custom Path:** You can specify a custom path in `Settings.json` (located in the Data Folder):
```json
- "FfmpegPath": "/path/to/your/ffmpeg"
+ "General": {
+ "FfmpegPath": "/path/to/your/ffmpeg"
+ }
```
### MPV Media Player (libmpv)
@@ -237,15 +245,6 @@ Used for AI-based speech recognition.
* **Files:** Download or build the binary and ensure it is named `whisper-cli`.
* **Models:** Models (`.bin` files) go into a `Models` subfolder: `[Data Folder]/SpeechToText/Cpp/Models`.
-### MLX Whisper (Speech-to-Text, Apple Silicon)
-Runs Apple's `mlx-whisper` Python package on the GPU / Neural Engine. **Subtitle Edit does not download this engine** — you install the Python package yourself, and there is no file in the Data Folder.
-
-* **Requirements:** An Apple Silicon Mac (M1 or newer) and Python 3.
-* **Install:** `pip3 install mlx-whisper` — or, for an isolated install, `pipx install mlx-whisper`.
-* **Models:** MLX-format Whisper weights (`tiny` … `large-v3-turbo`) are downloaded from Hugging Face (the `mlx-community` org) into the Hugging Face cache on first use.
-* **How detection works:** Subtitle Edit does not look for a binary; it looks for a Python interpreter that can `import mlx_whisper`. It probes Homebrew (`/opt/homebrew/bin`, `/usr/local/bin`), python.org framework builds, pyenv, and the system Python. Because **pipx, virtual environments, and conda install the package into an isolated environment** that those shared interpreters cannot import, Subtitle Edit also reads the shebang of the installed `mlx_whisper` command — found on your `PATH` or at `~/.local/bin/mlx_whisper` (where pipx places it) — to locate the exact matching interpreter.
-* **If it reports "not found" after installing:** confirm the command resolves with `which mlx_whisper`. If it does not, add its directory to your `PATH`, or symlink the command into `~/.local/bin`. (Symlinking or copying only the *model files* does nothing — detection is by the Python package, not a binary.)
-
### SE5 Speech-to-Text, OCR, and TTS Engines
Some newer local engines are platform-specific or model-specific. Use the in-app downloaders where available, and check [Speech to Text](features/speech-to-text.md), [Text to Speech](features/text-to-speech.md), and [OCR](features/ocr.md) for current engine notes.
diff --git a/docs/translating.md b/docs/translating.md
index d82ecda0e1d..9d9a47cdcbc 100644
--- a/docs/translating.md
+++ b/docs/translating.md
@@ -58,7 +58,7 @@ Near the top of the JSON file, update these fields:
```json
{
"title": "Subtitle Edit",
- "version": null,
+ "version": "v5.2.0-beta32",
"translatedBy": "Your Name (or email / homepage)",
"cultureName": "de-DE",
...
@@ -76,7 +76,7 @@ The `cultureName` value **must** be a valid .NET culture identifier.
## Step 4 — Save and Install the File
-Save your translated file as **`{CultureName}.json`** (e.g., `de-DE.json`) and place it in the **`Languages`** subfolder inside Subtitle Edit's data folder:
+Save your translated file as **`{English language name}.json`** (e.g., `German.json` — the file name is what the language list shows, and the shipped translations are named the same way) and place it in the **`Languages`** subfolder inside Subtitle Edit's data folder:
| Platform | Languages folder path |
|----------|-----------------------|
diff --git a/installer/WindowsInno/Subtitle_Edit_Localization.iss b/installer/WindowsInno/Subtitle_Edit_Localization.iss
index 55be25bca4d..74d5da390a0 100644
--- a/installer/WindowsInno/Subtitle_Edit_Localization.iss
+++ b/installer/WindowsInno/Subtitle_Edit_Localization.iss
@@ -1,4 +1,4 @@
-;* Subtitle Edit - installer localization script
+;* Subtitle Edit - installer localization script
; English
[CustomMessages]
@@ -242,7 +242,7 @@ hu.types_default=Default installation
; Italian
[CustomMessages]
it.sm_com_Changelog=Elenco novità di Subtitle Edit
-it.run_ViewChangelog=Visualizza le novità
+it.run_ViewChangelog=Visualizza elenco novità di Subtitle Edit
it.msg_DeleteSettings=Vuoi eliminare le impostazioni personali di Subtitle Edit?
it.tsk_AllUsers=Per tutti gli utenti
it.tsk_CurrentUser=Solo per l'utente attuale
@@ -254,6 +254,9 @@ it.types_custom=Installazione personalizzata
it.types_default=Installazione predefinita
it.NameAndVersion=%1 %2
it.LaunchProgram=Esegui %1
+it.AdditionalIcons=Collegamenti:
+it.CreateDesktopIcon=Crea collegamento programma sul &desktop
+it.CreateQuickLaunchIcon=Crea collegamento programma nella &barra 'Avvio veloce'
; Japanese
[CustomMessages]
diff --git a/installer/macBundle/SubtitleEdit.app/Contents/Info.plist b/installer/macBundle/SubtitleEdit.app/Contents/Info.plist
index 120286a6157..5152ef06876 100644
--- a/installer/macBundle/SubtitleEdit.app/Contents/Info.plist
+++ b/installer/macBundle/SubtitleEdit.app/Contents/Info.plist
@@ -20,8 +20,19 @@
SBED
CFBundleIconFile
SE.icns
+
LSMinimumSystemVersion
- 10.15
+ 12.0
NSHighResolutionCapable
NSSupportsAutomaticGraphicsSwitching
@@ -41,7 +52,16 @@
vtt
sbv
stl
- xml
+ ttml
+ dfxp
+ itt
+ smi
+ scc
+ mcc
+ cap
+ pac
+ lrc
+ idx
CFBundleTypeName
Subtitle Files
@@ -50,6 +70,22 @@
LSHandlerRank
Owner
+
+
+ CFBundleTypeExtensions
+
+ xml
+
+ CFBundleTypeName
+ XML Subtitle Files
+ CFBundleTypeRole
+ Editor
+ LSHandlerRank
+ Alternate
+
CFBundleTypeExtensions
diff --git a/src/libse/BluRaySup/BluRaySupFadeStep.cs b/src/libse/BluRaySup/BluRaySupFadeStep.cs
new file mode 100644
index 00000000000..aa4eb60a0d5
--- /dev/null
+++ b/src/libse/BluRaySup/BluRaySupFadeStep.cs
@@ -0,0 +1,38 @@
+using System;
+
+namespace Nikse.SubtitleEdit.Core.BluRaySup
+{
+ ///
+ /// One alpha level of a Blu-ray fade. The object is sent once, at the start of the epoch, and
+ /// each step after that is written as a "palette update display set" - a PCS with
+ /// palette_update_flag set plus a PDS holding the same palette with every entry's alpha
+ /// scaled by
+ public class BluRaySupFadeStep
+ {
+ ///
+ /// When the alpha level takes effect, in milliseconds - the same time base as
+ ///
+ public long TimeMs { get; set; }
+
+ ///
+ /// Alpha of the palette at this step, 0 (invisible) to 100 (fully opaque).
+ ///
+ public int AlphaPercent { get; set; }
+
+ public long TimeForWrite => (long)Math.Round(TimeMs * 90.0, MidpointRounding.AwayFromZero);
+
+ public BluRaySupFadeStep()
+ {
+ }
+
+ public BluRaySupFadeStep(long timeMs, int alphaPercent)
+ {
+ TimeMs = timeMs;
+ AlphaPercent = alphaPercent;
+ }
+ }
+}
diff --git a/src/libse/BluRaySup/BluRaySupPalette.cs b/src/libse/BluRaySup/BluRaySupPalette.cs
index 3fcadd61bb5..1b5b00b92ff 100644
--- a/src/libse/BluRaySup/BluRaySupPalette.cs
+++ b/src/libse/BluRaySup/BluRaySupPalette.cs
@@ -50,6 +50,17 @@ public class BluRaySupPalette
* @return Integer array with red, blue, green component (in this order)
*/
public static int[] YCbCr2Rgb(int y, int cb, int cr, bool useBt601)
+ {
+ YCbCr2Rgb(y, cb, cr, useBt601, out var ir, out var ig, out var ib);
+ return new[] { ir, ig, ib };
+ }
+
+ ///
+ /// Same conversion as
+ public static void YCbCr2Rgb(int y, int cb, int cr, bool useBt601, out int r8, out int g8, out int b8)
{
// Studio range → center/offset removal
y -= 16;
@@ -82,7 +93,7 @@ public static int[] YCbCr2Rgb(int y, int cb, int cr, bool useBt601)
if (r > 254.0) r += 0.35;
if (g > 254.0) g += 0.35;
if (b > 254.0) b += 0.35;
-
+
int ir = (int)Math.Round(r, MidpointRounding.AwayFromZero);
int ig = (int)Math.Round(g, MidpointRounding.AwayFromZero);
int ib = (int)Math.Round(b, MidpointRounding.AwayFromZero);
@@ -91,7 +102,9 @@ public static int[] YCbCr2Rgb(int y, int cb, int cr, bool useBt601)
if (ig < 0) ig = 0; else if (ig > 255) ig = 255;
if (ib < 0) ib = 0; else if (ib > 255) ib = 255;
- return new[] { ir, ig, ib };
+ r8 = ir;
+ g8 = ig;
+ b8 = ib;
}
/**
@@ -300,10 +313,10 @@ public void SetYCbCr(int index, int yn, int cbn, int crn)
_cb[index] = (byte)cbn;
_cr[index] = (byte)crn;
// create RGB
- var rgb = YCbCr2Rgb(yn, cbn, crn, _useBt601);
- _r[index] = (byte)rgb[0];
- _g[index] = (byte)rgb[1];
- _b[index] = (byte)rgb[2];
+ YCbCr2Rgb(yn, cbn, crn, _useBt601, out var r8, out var g8, out var b8);
+ _r[index] = (byte)r8;
+ _g[index] = (byte)g8;
+ _b[index] = (byte)b8;
}
/**
diff --git a/src/libse/BluRaySup/BluRaySupParser.cs b/src/libse/BluRaySup/BluRaySupParser.cs
index 7ca7677c1c4..8258e5e87e0 100644
--- a/src/libse/BluRaySup/BluRaySupParser.cs
+++ b/src/libse/BluRaySup/BluRaySupParser.cs
@@ -557,7 +557,10 @@ public static List ParseBluRaySupFromMatroska(MatroskaTrackInfo matrosk
subtitles.Add(sup);
// fix overlapping
- if (subtitles.Count > 1 && sub[subtitles.Count - 2].End > sub[subtitles.Count - 1].Start)
+ // Index "subtitles", not "sub": the two lists are not parallel (one
+ // container block can yield several display sets), so this compared
+ // unrelated blocks and threw once subtitles outgrew sub.
+ if (subtitles.Count > 1 && subtitles[subtitles.Count - 2].EndTime > subtitles[subtitles.Count - 1].StartTime)
{
subtitles[subtitles.Count - 2].EndTime = subtitles[subtitles.Count - 1].StartTime - 1;
}
diff --git a/src/libse/BluRaySup/BluRaySupPicture.cs b/src/libse/BluRaySup/BluRaySupPicture.cs
index 00ddd320639..498f3d90c0d 100644
--- a/src/libse/BluRaySup/BluRaySupPicture.cs
+++ b/src/libse/BluRaySup/BluRaySupPicture.cs
@@ -103,6 +103,13 @@ public class BluRaySupPicture
///
public List> Palettes { get; set; } = new List>();
+ ///
+ /// Alpha levels for a fade in/out, in presentation order. A step at
+ public List FadeSteps { get; set; } = new List();
+
///
/// Create RLE buffer from bitmap
///
@@ -121,6 +128,11 @@ private static byte[] EncodeImage(SKBitmap bm, List palette)
}
}
+ // Cap on the approximate-match memo below. A caption repeats its anti-aliasing
+ // colours on every row, but a photographic source can hold millions of distinct
+ // ones - past the cap the linear scan simply runs again, as it always did.
+ const int maxCachedColors = 1 << 16;
+
var transparentColor = (byte)palette[palette.Count - 1];
var bytes = new List(bm.Width * 2);
var reader = new BitmapRowReader(bm);
@@ -145,7 +157,15 @@ private static byte[] EncodeImage(SKBitmap bm, List palette)
}
else
{
+ // FindBestMatch is a linear scan over the (up to 255 entry) palette and
+ // the palette does not change while encoding, so the answer for a color
+ // is fixed. Anti-aliased edges hit this for the same handful of colors
+ // on every row of the caption; remember them in the same lookup.
color = FindBestMatch(c, palette);
+ if (lookup.Count < maxCachedColors)
+ {
+ lookup[c] = color;
+ }
}
for (len = 1; x + len < bm.Width; len++)
@@ -361,6 +381,15 @@ private static List GetBitmapPalette(SKBitmap bitmap, SKColor fontColor
lookup = new HashSet();
pal.Add(fontColor);
lookup.Add(fontColor);
+
+ // Colors already known to have a close palette entry. HasCloseColor is a linear scan
+ // over the palette, and this pass reached it for every pixel of every anti-aliased
+ // edge - a 1920x200 caption ran it hundreds of thousands of times over a palette
+ // growing towards 254 entries. A rejection can never be undone: the palette only
+ // grows, and the tolerance only widens as it does (1 -> 5 -> 25), so a color that
+ // had a close entry once still has one later. Rejections take no palette slot, so
+ // caching one can never skip an insertion.
+ var rejected = new HashSet();
for (var y = 0; y < bitmap.Height; y++)
{
reader.ReadRow(y, row);
@@ -373,6 +402,10 @@ private static List GetBitmapPalette(SKBitmap bitmap, SKColor fontColor
{
// exact color already exists
}
+ else if (rejected.Contains(c))
+ {
+ // a close enough color already exists
+ }
else if (pal.Count < 100)
{
if (!HasCloseColor(c, pal, 1))
@@ -380,6 +413,10 @@ private static List GetBitmapPalette(SKBitmap bitmap, SKColor fontColor
pal.Add(c);
lookup.Add(c);
}
+ else
+ {
+ rejected.Add(c);
+ }
}
else if (pal.Count < 240)
{
@@ -388,14 +425,32 @@ private static List GetBitmapPalette(SKBitmap bitmap, SKColor fontColor
pal.Add(c);
lookup.Add(c);
}
+ else
+ {
+ rejected.Add(c);
+ }
}
- else if (pal.Count < 254 && !HasCloseColor(c, pal, 25))
+ else if (pal.Count < 254)
{
- pal.Add(c);
- lookup.Add(c);
+ if (!HasCloseColor(c, pal, 25))
+ {
+ pal.Add(c);
+ lookup.Add(c);
+ }
+ else
+ {
+ rejected.Add(c);
+ }
}
}
}
+
+ // Every branch above requires pal.Count < 254, so a full palette freezes the
+ // result - the rest of the image only cost scans that could not change it.
+ if (pal.Count >= 254)
+ {
+ break;
+ }
}
pal.Add(SKColors.Transparent); // last entry must be transparent
@@ -444,6 +499,78 @@ private static int GetFpsId(double fps)
private static long _lastEndTimeForWrite = -1000;
+ ///
+ /// Splits
+ private static List GetFadeSteps(BluRaySupPicture pic, out int startAlphaPercent)
+ {
+ startAlphaPercent = 100;
+ var updates = new List();
+ if (pic.FadeSteps == null || pic.FadeSteps.Count == 0)
+ {
+ return updates;
+ }
+
+ var sorted = new List(pic.FadeSteps);
+ sorted.Sort((a, b) => a.TimeMs.CompareTo(b.TimeMs));
+
+ var lastAlpha = 100;
+ var lastTime = long.MinValue;
+ foreach (var step in sorted)
+ {
+ var alpha = Math.Min(100, Math.Max(0, step.AlphaPercent));
+ if (step.TimeMs <= pic.StartTime)
+ {
+ startAlphaPercent = alpha;
+ lastAlpha = alpha;
+ continue;
+ }
+
+ if (step.TimeMs >= pic.EndTime || alpha == lastAlpha || step.TimeMs == lastTime)
+ {
+ continue;
+ }
+
+ updates.Add(new BluRaySupFadeStep(step.TimeMs, alpha));
+ lastAlpha = alpha;
+ lastTime = step.TimeMs;
+ }
+
+ return updates;
+ }
+
+ ///
+ /// Writes a Palette Definition Segment with every alpha scaled by
+ ///
+ private static int WritePds(byte[] buf, int index, byte[] packetHeader, BluRaySupPalette pal, int palSize, int paletteVersion, int alphaPercent)
+ {
+ packetHeader[10] = 0x14; // ID (keep PTS & DTS)
+ ToolBox.SetWord(packetHeader, 11, 2 + palSize * 5); // size
+ for (var i = 0; i < packetHeader.Length; i++)
+ {
+ buf[index++] = packetHeader[i];
+ }
+
+ buf[index++] = 0; // palette_id
+ buf[index++] = (byte)paletteVersion; // palette_version_number
+ var alpha = pal.GetAlpha();
+ for (var i = 0; i < palSize; i++)
+ {
+ buf[index++] = (byte)i; // index
+ buf[index++] = pal.GetY()[i]; // Y
+ buf[index++] = pal.GetCr()[i]; // Cr
+ buf[index++] = pal.GetCb()[i]; // Cb
+ buf[index++] = (byte)(alpha[i] * alphaPercent / 100); // Alpha
+ }
+
+ return index;
+ }
+
///
/// Create the binary stream representation of one caption
///
@@ -482,7 +609,8 @@ public static byte[] CreateSupFrame(BluRaySupPicture pic, SKBitmap bmp, SKColor
}
else
{
- numAddPackets = 1 + (rleBuf.Length - 0xffe4) / 0xffeb;
+ // round up, but without an extra empty packet when the rest divides evenly
+ numAddPackets = (rleBuf.Length - 0xffe4 + 0xffeb - 1) / 0xffeb;
}
// a typical frame consists of 8 packets. It can be elongated by additional object frames
@@ -542,14 +670,20 @@ public static byte[] CreateSupFrame(BluRaySupPicture pic, SKBitmap bmp, SKColor
{
0x00, 0x00, // 0: object_id
0x00, // 2: object_version_number
- 0x40 // 3: first_in_sequence (0x80), last_in_sequence (0x40), 6bits reserved
+ 0x00 // 3: first_in_sequence (0x80), last_in_sequence (0x40), 6bits reserved
+ // set per packet below - only the final one is last_in_sequence
};
- var size = packetHeader.Length * (8 + numAddPackets);
+ // Fade steps ride along as palette update display sets (PCS + PDS + END) between the
+ // caption and the screen clear - same object, only the palette alpha changes.
+ var fadeSteps = GetFadeSteps(pic, out var startAlphaPercent);
+
+ var size = packetHeader.Length * (8 + numAddPackets + fadeSteps.Count * 3);
size += headerPcsStart.Length + headerPcsEnd.Length;
size += 2 * headerWds.Length + headerOdsFirst.Length;
size += numAddPackets * headerOdsNext.Length;
size += (2 + palSize * 5) /* PDS */;
+ size += fadeSteps.Count * (headerPcsStart.Length + 2 + palSize * 5);
size += rleBuf.Length;
switch (alignment)
@@ -679,23 +813,7 @@ public static byte[] CreateSupFrame(BluRaySupPicture pic, SKBitmap bmp, SKColor
}
// write PDS - Palette Definition Segment
- packetHeader[10] = 0x14; // ID (keep PTS & DTS)
- ToolBox.SetWord(packetHeader, 11, 2 + palSize * 5); // size
- for (var i = 0; i < packetHeader.Length; i++)
- {
- buf[index++] = packetHeader[i];
- }
-
- buf[index++] = 0;
- buf[index++] = 0;
- for (var i = 0; i < palSize; i++)
- {
- buf[index++] = (byte)i; // index
- buf[index++] = pal.GetY()[i]; // Y
- buf[index++] = pal.GetCr()[i]; // Cr
- buf[index++] = pal.GetCb()[i]; // Cb
- buf[index++] = pal.GetAlpha()[i]; // Alpha
- }
+ index = WritePds(buf, index, packetHeader, pal, palSize, 0, startAlphaPercent);
// write first OBJ
var bufSize = rleBuf.Length;
@@ -742,6 +860,8 @@ public static byte[] CreateSupFrame(BluRaySupPicture pic, SKBitmap bmp, SKColor
buf[index++] = packetHeader[i];
}
+ // only the final fragment carries last_in_sequence - middle ones must be 0x00
+ headerOdsNext[3] = (byte)(p == numAddPackets - 1 ? 0x40 : 0x00);
for (var i = 0; i < headerOdsNext.Length; i++)
{
buf[index++] = headerOdsNext[i];
@@ -763,6 +883,47 @@ public static byte[] CreateSupFrame(BluRaySupPicture pic, SKBitmap bmp, SKColor
buf[index++] = packetHeader[i];
}
+ // write the fade steps - one palette update display set each (PCS + PDS + END). The
+ // PCS repeats the composition of the epoch start with palette_update_flag set, which
+ // tells the decoder to keep the object it already has and only take the new palette.
+ var endPts = pic.EndTimeForWrite;
+ var compositionNumber = pic.CompositionNumber;
+ headerPcsStart[7] = 0x00; // composition_state: normal case
+ headerPcsStart[8] = 0x80; // palette_update_flag
+ for (var step = 0; step < fadeSteps.Count; step++)
+ {
+ // A step may only be scheduled after the caption is up and before it is taken
+ // down; the start PTS can have been nudged by the small gap removal above.
+ var stepPts = Math.Min(Math.Max(fadeSteps[step].TimeForWrite, pts + 1), endPts - 1);
+
+ compositionNumber++;
+ packetHeader[10] = 0x16; // ID
+ ToolBox.SetDWord(packetHeader, 2, (uint)stepPts); // PTS
+ ToolBox.SetDWord(packetHeader, 6, 0); // DTS (0 = unset)
+ ToolBox.SetWord(packetHeader, 11, headerPcsStart.Length); // size
+ for (var i = 0; i < packetHeader.Length; i++)
+ {
+ buf[index++] = packetHeader[i];
+ }
+
+ ToolBox.SetWord(headerPcsStart, 5, compositionNumber);
+ for (var i = 0; i < headerPcsStart.Length; i++)
+ {
+ buf[index++] = headerPcsStart[i];
+ }
+
+ // The palette version has to move for the decoder to take the update; it is a
+ // byte, so it wraps on captions with more than 255 steps.
+ index = WritePds(buf, index, packetHeader, pal, palSize, (step + 1) & 0xff, fadeSteps[step].AlphaPercent);
+
+ packetHeader[10] = 0x80; // END (keep PTS & DTS)
+ ToolBox.SetWord(packetHeader, 11, 0); // size
+ for (var i = 0; i < packetHeader.Length; i++)
+ {
+ buf[index++] = packetHeader[i];
+ }
+ }
+
// write PCS end
packetHeader[10] = 0x16; // ID
ToolBox.SetDWord(packetHeader, 2, (uint)pic.EndTimeForWrite); // PTS
@@ -776,7 +937,7 @@ public static byte[] CreateSupFrame(BluRaySupPicture pic, SKBitmap bmp, SKColor
ToolBox.SetWord(headerPcsEnd, 0, pic.Width);
ToolBox.SetWord(headerPcsEnd, 2, h); // cropped height
ToolBox.SetByte(headerPcsEnd, 4, fpsId);
- ToolBox.SetWord(headerPcsEnd, 5, pic.CompositionNumber + 1);
+ ToolBox.SetWord(headerPcsEnd, 5, compositionNumber + 1);
for (var i = 0; i < headerPcsEnd.Length; i++)
{
buf[index++] = headerPcsEnd[i];
diff --git a/src/libse/BluRaySup/SkiaExt.cs b/src/libse/BluRaySup/SkiaExt.cs
index 9825c38f9ce..504965c1e95 100644
--- a/src/libse/BluRaySup/SkiaExt.cs
+++ b/src/libse/BluRaySup/SkiaExt.cs
@@ -111,13 +111,23 @@ private static bool ByteArraysEqual(ReadOnlySpan a1, ReadOnlySpan a2
return a1.SequenceEqual(a2);
}
- public class TrimResult
+ ///
+ ///
+ public class TrimResult : IDisposable
{
public SKBitmap TrimmedBitmap { get; set; }
public int Top { get; set; }
public int Left { get; set; }
public int Right { get; set; }
public int Bottom { get; set; }
+
+ public void Dispose()
+ {
+ TrimmedBitmap?.Dispose();
+ TrimmedBitmap = null;
+ }
}
///
diff --git a/src/libse/BluRaySup/ToolBox.cs b/src/libse/BluRaySup/ToolBox.cs
index 1ffa4a1bea8..9d3970d70a1 100644
--- a/src/libse/BluRaySup/ToolBox.cs
+++ b/src/libse/BluRaySup/ToolBox.cs
@@ -29,15 +29,17 @@ public static class ToolBox
*/
public static long[] MillisecondsToTime(double ms)
{
+ // Each component must be truncated, not rounded: rounding one up makes the remainder
+ // negative and cascades, so 3500 ms came out as 0:0:4:-500 and 40 minutes as 1:-20:0:0.
var time = new long[4];
// time[0] = hours
- time[0] = (long)Math.Round(ms / (60 * 60 * 1000), MidpointRounding.AwayFromZero);
+ time[0] = (long)(ms / (60.0 * 60.0 * 1000.0));
ms -= time[0] * 60.0 * 60.0 * 1000.0;
// time[1] = minutes
- time[1] = (long)Math.Round(ms / (60.0 * 1000.0), MidpointRounding.AwayFromZero);
+ time[1] = (long)(ms / (60.0 * 1000.0));
ms -= time[1] * 60 * 1000;
// time[2] = seconds
- time[2] = (long)Math.Round(ms / 1000.0, MidpointRounding.AwayFromZero);
+ time[2] = (long)(ms / 1000.0);
ms -= time[2] * 1000.0;
time[3] = (long)Math.Round(ms, MidpointRounding.AwayFromZero);
return time;
diff --git a/src/libse/CDG/CdgGraphics.cs b/src/libse/CDG/CdgGraphics.cs
index c60cde3c67d..ebe438b87cc 100644
--- a/src/libse/CDG/CdgGraphics.cs
+++ b/src/libse/CDG/CdgGraphics.cs
@@ -79,7 +79,10 @@ public SKBitmap ToBitmap(int packetNumber) // long timeInMilliseconds)
}
var graphicData = GetGraphicData();
- var bitmap = new SKBitmap(FullWidth, FullHeight, SKColorType.Rgba8888, SKAlphaType.Premul);
+ // Bgra8888, not Rgba8888: the loop below packs (a<<24)|(r<<16)|(g<<8)|b, which on
+ // a little-endian machine lands in memory as B,G,R,A. Declaring Rgba8888 made
+ // Skia read that back as R,G,B,A - i.e. red and blue swapped in every frame.
+ var bitmap = new SKBitmap(FullWidth, FullHeight, SKColorType.Bgra8888, SKAlphaType.Premul);
// Copy pixel data into the bitmap
var pixelData = bitmap.GetPixels();
diff --git a/src/libse/Cea608/CaptionScreen.cs b/src/libse/Cea608/CaptionScreen.cs
index ce0a7514e04..a4a0550358c 100644
--- a/src/libse/Cea608/CaptionScreen.cs
+++ b/src/libse/Cea608/CaptionScreen.cs
@@ -184,8 +184,19 @@ public void RollUp()
return;
}
+ // Reuse the row that scrolled off the top, cleared - a plain `new CcRow()` is NOT
+ // empty: its chars start with a null foreground/background pen state, so IsEmpty()
+ // (and with it CaptionScreen.IsEmpty and Serialize) reported a blank row as content
+ // for the rest of the file once a roll-up had happened.
+ var topRow = rows[removeIndex];
rows.RemoveAt(removeIndex);
- rows.Add(new CcRow());
+ topRow.Clear();
+
+ // It goes back at the roll-up base row, not at the bottom of the screen. Appending
+ // only happens to be right when the base row IS the bottom row - with a roll-up
+ // window placed higher up (a PAC can put it anywhere) appending scrolled every row
+ // below the window along with it.
+ rows.Insert(Math.Min(CurrentRow, rows.Count), topRow);
Rows = rows.ToArray();
}
}
diff --git a/src/libse/Cea708/VancDataWriter.cs b/src/libse/Cea708/VancDataWriter.cs
index 24bddaed78f..b31cbdcaf9b 100644
--- a/src/libse/Cea708/VancDataWriter.cs
+++ b/src/libse/Cea708/VancDataWriter.cs
@@ -76,12 +76,49 @@ public static string[] GenerateLinesFromText(string input, int counter)
}
bytes.AddRange(c1.GetBytes());
- var c2 = new SetText(line);
- if (c2.GetBytes().Length + bytes.Count > 32)
+ // Split the line across packets instead of appending it whole: a cc_data
+ // section holds at most 16 byte-pairs, so a single SetText longer than the
+ // remaining room made CcDataSection throw "Too many bytes for CCData!" and the
+ // whole save failed on any line of ~34 characters or more.
+ var remaining = line;
+ while (remaining.Length > 0)
{
- counter = FlushCommands(counter, bytes, results);
+ var room = 32 - bytes.Count;
+ if (room <= 0)
+ {
+ counter = FlushCommands(counter, bytes, results);
+ room = 32 - bytes.Count;
+ }
+
+ // Count whole characters: a G2 character encodes as EXT1 + code, and
+ // splitting that pair apart would emit a stray escape.
+ var take = 0;
+ var used = 0;
+ while (take < remaining.Length)
+ {
+ var charLength = new SetText(remaining[take].ToString()).GetBytes().Length;
+ if (used + charLength > room)
+ {
+ break;
+ }
+
+ used += charLength;
+ take++;
+ }
+
+ if (take == 0)
+ {
+ counter = FlushCommands(counter, bytes, results);
+ continue;
+ }
+
+ bytes.AddRange(new SetText(remaining.Substring(0, take)).GetBytes());
+ remaining = remaining.Substring(take);
+ if (remaining.Length > 0)
+ {
+ counter = FlushCommands(counter, bytes, results);
+ }
}
- bytes.AddRange(c2.GetBytes());
}
FlushCommands(counter, bytes, results);
diff --git a/src/libse/Common/ActorConverter.cs b/src/libse/Common/ActorConverter.cs
index 1fb76b36c7f..861bbe7c888 100644
--- a/src/libse/Common/ActorConverter.cs
+++ b/src/libse/Common/ActorConverter.cs
@@ -4,7 +4,6 @@
using System;
using System.Collections.Generic;
using System.Linq;
-using System.Text;
namespace Nikse.SubtitleEdit.Core.Common
{
@@ -35,8 +34,19 @@ public ActorConverter(SubtitleFormat subtitleFormat, string languageCode)
_nameListInclMulti = _namesList.GetAllNames();
}
- public string FixActorsFromActor(Paragraph p, int? changeCasing, SKColor? color)
+ ///
+ /// Moves the actor column into the text. The converted paragraph is returned in the result -
+ /// the actor column is cleared, as the name now lives in the text and would otherwise be
+ /// written twice (#14077).
+ ///
+ public ActorConverterResult FixActorsFromActor(Paragraph paragraph, int? changeCasing, SKColor? color)
{
+ var p = new Paragraph(paragraph, false);
+ if (ToActor)
+ {
+ return new ActorConverterResult { Paragraph = p, Selected = true };
+ }
+
var actor = p.Actor;
if (changeCasing.HasValue)
{
@@ -55,85 +65,161 @@ public string FixActorsFromActor(Paragraph p, int? changeCasing, SKColor? color)
{
actor = actor + ":";
}
- else if (ToActor)
- {
- return p.Text;
- }
- if (color.HasValue && !ToActor)
+ if (color.HasValue)
{
actor = SetColor(_subtitleFormat, color.Value, actor);
}
p.Text = actor + " " + p.Text.TrimStart(' ');
+ p.Actor = string.Empty;
- return p.Text;
+ return new ActorConverterResult { Paragraph = p, Selected = true };
}
- public string FixActorsFromBeforeColon(Paragraph p, char ch, int? changeCasing, SKColor? color)
+ ///
+ /// Converts "Actor: text" lines. The converted paragraph is returned in the result: converting
+ /// to the actor column writes
+ public ActorConverterResult FixActorsFromBeforeColon(Paragraph paragraph, char ch, int? changeCasing, SKColor? color)
{
- var sb = new StringBuilder();
- foreach (var line in p.Text.SplitToLines())
+ var p = new Paragraph(paragraph, false);
+ var lines = p.Text.SplitToLines();
+
+ // Only one extra paragraph can be split off, so a third speaker would be lost.
+ if (ToActor && lines.Count(line => HasActor(line, ch)) > 2)
{
+ return new ActorConverterResult { Paragraph = paragraph, Skip = true };
+ }
+
+ Paragraph nextParagraph = null;
+ var selectFix = true;
+ var actorAssigned = false;
+ var textLines = new List();
+ foreach (var line in lines)
+ {
+ // index into the trimmed line - leading whitespace goes with the actor it precedes
var s = line.Trim();
- var startIdx = line.IndexOf(ch);
- if (startIdx > 0)
+ var startIdx = s.IndexOf(ch);
+ if (startIdx <= 0)
{
- var actor = s.Substring(0, startIdx).Trim(' ', '-', '"');
- if (changeCasing.HasValue)
- {
- actor = SetCasing(_subtitleFormat, changeCasing, actor);
- }
-
- if (ToSquare)
- {
- actor = "[" + actor + "]";
- }
- else if (ToParentheses)
- {
- actor = "(" + actor + ")";
- }
- else if (ToColon)
+ // A line without an actor belongs to the paragraph the previous line went to.
+ if (nextParagraph != null)
{
- actor = actor + ":";
+ nextParagraph.Text = (nextParagraph.Text + Environment.NewLine + s).Trim();
}
- else if (ToActor)
+ else
{
+ textLines.Add(s);
}
- if (color.HasValue && !ToActor)
- {
- SetColor(_subtitleFormat, color.Value, actor);
- }
+ continue;
+ }
- if (ToSquare)
- {
- s = actor + " " + s.Substring(startIdx + 1).TrimStart(' ');
- }
- else if (ToParentheses)
- {
- s = actor + " " + s.Substring(startIdx + 1).TrimStart(' ');
- }
- else if (ToColon)
- {
- s = actor + " " + s.Substring(startIdx + 1).TrimStart(' ');
- }
- else if (ToActor)
- {
- s = s.Substring(startIdx + 1);
- }
+ var actor = s.Substring(0, startIdx).Trim(' ', '-', '"');
+ // AND, not assignment: selectFix is declared outside the per-line loop, so a later line's valid speaker used to mask an earlier line whose "actor" was really a clock ("It's 12:30 now.") - pre-checking a fix that destroys text.
+ selectFix &= IsActor(actor);
+ if (changeCasing.HasValue)
+ {
+ actor = SetCasing(_subtitleFormat, changeCasing, actor);
}
- sb.AppendLine(s);
+ if (ToSquare)
+ {
+ actor = "[" + actor + "]";
+ }
+ else if (ToParentheses)
+ {
+ actor = "(" + actor + ")";
+ }
+ else if (ToColon)
+ {
+ actor = actor + ":";
+ }
+
+ if (color.HasValue && !ToActor)
+ {
+ actor = SetColor(_subtitleFormat, color.Value, actor);
+ }
+
+ var text = s.Substring(startIdx + 1).TrimStart(' ');
+ if (!ToActor)
+ {
+ textLines.Add(actor + " " + text);
+ }
+ else if (!actorAssigned)
+ {
+ // The first name found goes in the actor column...
+ p.Actor = actor;
+ actorAssigned = true;
+ textLines.Add(text);
+ }
+ else
+ {
+ // ...a second one needs a paragraph of its own, as the column holds one name.
+ nextParagraph = new Paragraph(p) { Text = text, Actor = actor };
+ }
}
- return sb.ToString().Trim();
+ p.Text = string.Join(Environment.NewLine, textLines).Trim();
+
+ return new ActorConverterResult
+ {
+ Paragraph = p,
+ NextParagraph = nextParagraph,
+ Selected = selectFix,
+ };
+ }
+
+ private static bool HasActor(string line, char ch)
+ {
+ return line.Trim().IndexOf(ch) > 0;
+ }
+
+ ///
+ /// A line splitter can break a tag in half - "Princess Peach on (Speaker\n2) Super Smash
+ /// Bros." - and the per-line scan in
+ private static string JoinTagBrokenOverLineBreak(string text, char start, char end)
+ {
+ const int maxTagLength = 50;
+
+ var startIdx = text.IndexOf(start);
+ while (startIdx >= 0)
+ {
+ var endIdx = text.IndexOf(end, startIdx + 1);
+ if (endIdx < 0)
+ {
+ break;
+ }
+
+ var inner = text.Substring(startIdx + 1, endIdx - startIdx - 1);
+ if (inner.Length <= maxTagLength &&
+ inner.IndexOf(start) < 0 &&
+ (inner.Contains('\n') || inner.Contains('\r')))
+ {
+ var joined = string.Join(" ", inner.Split((char[])null, StringSplitOptions.RemoveEmptyEntries));
+ text = text.Substring(0, startIdx + 1) + joined + text.Substring(endIdx);
+ endIdx = startIdx + 1 + joined.Length;
+ }
+
+ startIdx = text.IndexOf(start, endIdx + 1);
+ }
+
+ return text;
}
public ActorConverterResult FixActors(Paragraph paragraph, char start, char end, int? changeCasing, SKColor? color)
{
var p = new Paragraph(paragraph, false);
Paragraph nextParagraph = null;
+ p.Text = JoinTagBrokenOverLineBreak(p.Text, start, end);
var lines = p.Text.SplitToLines();
if (lines.Count > 2)
{
@@ -148,15 +234,15 @@ public ActorConverterResult FixActors(Paragraph paragraph, char start, char end,
var s = line;
var startIdx = s.IndexOf(start);
var endIdx = s.IndexOf(end);
- if (startIdx != -1 && endIdx != -1)
- {
- if (endIdx < startIdx)
- {
- break;
- }
+ // A closing bracket before the opening one is not an actor - the line is kept as it
+ // is. Giving up on the whole paragraph here dropped this line and every line after
+ // it from the text.
+ if (startIdx != -1 && endIdx > startIdx)
+ {
var actor = s.Substring(startIdx + 1, endIdx - startIdx - 1).Trim(' ', '-', '"');
- selectFix = IsActor(actor);
+ // AND, not assignment: selectFix is declared outside the per-line loop, so a later line's valid speaker used to mask an earlier line whose "actor" was really a clock ("It's 12:30 now.") - pre-checking a fix that destroys text.
+ selectFix &= IsActor(actor);
if (changeCasing.HasValue)
{
actor = SetCasing(_subtitleFormat, changeCasing, actor);
@@ -208,9 +294,20 @@ public ActorConverterResult FixActors(Paragraph paragraph, char start, char end,
}
else if (lineIdx == 1 && ToActor)
{
- nextParagraph = new Paragraph(p);
- nextParagraph.Text = s.Trim();
- nextParagraph.Actor = actor;
+ if (string.IsNullOrEmpty(p.Actor))
+ {
+ // Only the second line names a speaker, so it belongs to this paragraph -
+ // splitting off a paragraph with no actor at all would leave the name
+ // nowhere (#14077).
+ p.Actor = actor;
+ p.Text = (p.Text + Environment.NewLine + s.Trim()).Trim();
+ }
+ else
+ {
+ nextParagraph = new Paragraph(p);
+ nextParagraph.Text = s.Trim();
+ nextParagraph.Actor = actor;
+ }
}
else if (lineIdx == 1)
{
@@ -269,6 +366,8 @@ private static string SetColor(SubtitleFormat format, SKColor color, string acto
return actor;
}
+ private static readonly string[] CommonTitles = { "Mr.", "Mrs.", "Dr." };
+
private bool IsActor(string s)
{
if (string.IsNullOrWhiteSpace(s))
@@ -294,15 +393,14 @@ private bool IsActor(string s)
return false;
}
- if (word.Any(c => char.IsDigit(c) || (!char.IsLetter(c) && c != '-' && c != '\'')))
+ if (CommonTitles.Contains(word))
{
- return false;
+ continue;
}
- var commonTitles = new[] { "Mr.", "Mrs.", "Dr.", };
- if (commonTitles.Contains(word))
+ if (word.Any(c => char.IsDigit(c) || (!char.IsLetter(c) && c != '-' && c != '\'')))
{
- continue;
+ return false;
}
if (!_nameListInclMulti.Contains(word, StringComparer.OrdinalIgnoreCase))
diff --git a/src/libse/Common/AssaResampler.cs b/src/libse/Common/AssaResampler.cs
index 92b5c869930..84c3d969f02 100644
--- a/src/libse/Common/AssaResampler.cs
+++ b/src/libse/Common/AssaResampler.cs
@@ -183,6 +183,20 @@ private static string FixDrawing(decimal sourceWidth, decimal targetWidth, decim
return sb.ToString().TrimEnd() + s;
}
+ // The tag regexes below allow whitespace ("\pos (10,11)"), so the parameters cannot be cut
+ // at a fixed offset from the tag name - take everything between the parentheses instead.
+ private static string GetTagParameters(string matchValue)
+ {
+ var open = matchValue.IndexOf('(');
+ var close = matchValue.LastIndexOf(')');
+ if (open < 0 || close <= open)
+ {
+ return string.Empty;
+ }
+
+ return matchValue.Substring(open + 1, close - open - 1).RemoveChar(' ');
+ }
+
private static string FixMethodFourParameters(decimal sourceWidth, decimal targetWidth, decimal sourceHeight, decimal targetHeight, string input, string tag)
{
var regex = GetCachedRegex("\\\\" + tag + "\\s*\\(\\s*[-+]?\\d+[\\.\\d+]*\\s*,\\s*[-+]?\\d+[\\.\\d+]*\\s*,\\s*[-+]?\\d+[\\.\\d+]*\\s*,\\s*[-+]?\\d+[\\.\\d+]*\\s*\\)");
@@ -190,7 +204,7 @@ private static string FixMethodFourParameters(decimal sourceWidth, decimal targe
var match = regex.Match(s);
while (match.Success)
{
- var value = match.Value.Substring(tag.Length + 2, match.Value.Length - tag.Length - 3).RemoveChar(' ');
+ var value = GetTagParameters(match.Value);
var arr = value.Split(',');
if (arr.Length == 4 &&
decimal.TryParse(arr[0], NumberStyles.AllowDecimalPoint | NumberStyles.AllowLeadingSign, CultureInfo.InvariantCulture, out var x1) &&
@@ -212,7 +226,8 @@ private static string FixMethodFourParameters(decimal sourceWidth, decimal targe
}
else
{
- break;
+ // Skip this tag rather than abandoning every remaining tag on the line.
+ match = regex.Match(s, match.Index + match.Value.Length);
}
}
@@ -226,7 +241,7 @@ private static string FixMethodSixParametersFourActive(decimal sourceWidth, deci
var match = regex.Match(s);
while (match.Success)
{
- var value = match.Value.Substring(tag.Length + 2, match.Value.Length - tag.Length - 3).RemoveChar(' ');
+ var value = GetTagParameters(match.Value);
var arr = value.Split(',');
if (arr.Length == 6 &&
decimal.TryParse(arr[0], NumberStyles.AllowDecimalPoint | NumberStyles.AllowLeadingSign, CultureInfo.InvariantCulture, out var x1) &&
@@ -252,7 +267,8 @@ private static string FixMethodSixParametersFourActive(decimal sourceWidth, deci
}
else
{
- break;
+ // Skip this tag rather than abandoning every remaining tag on the line.
+ match = regex.Match(s, match.Index + match.Value.Length);
}
}
@@ -266,7 +282,7 @@ private static string FixMethodTwoParameters(decimal sourceWidth, decimal target
var match = regex.Match(s);
while (match.Success)
{
- var value = match.Value.Substring(tag.Length + 2, match.Value.Length - tag.Length - 3).RemoveChar(' ');
+ var value = GetTagParameters(match.Value);
var arr = value.Split(',');
if (arr.Length == 2 &&
decimal.TryParse(arr[0], NumberStyles.AllowDecimalPoint | NumberStyles.AllowLeadingSign, CultureInfo.InvariantCulture, out var x) &&
@@ -282,7 +298,8 @@ private static string FixMethodTwoParameters(decimal sourceWidth, decimal target
}
else
{
- break;
+ // Skip this tag rather than abandoning every remaining tag on the line.
+ match = regex.Match(s, match.Index + match.Value.Length);
}
}
@@ -306,7 +323,8 @@ private static string FixTagWithNumber(decimal sourceHeight, decimal targetHeigh
}
else
{
- break;
+ // Skip this tag rather than abandoning every remaining tag on the line.
+ match = regex.Match(s, match.Index + match.Value.Length);
}
}
diff --git a/src/libse/Common/CharUtils.cs b/src/libse/Common/CharUtils.cs
index 6ea30179028..a9b071e4035 100644
--- a/src/libse/Common/CharUtils.cs
+++ b/src/libse/Common/CharUtils.cs
@@ -6,7 +6,7 @@ public static class CharUtils
/// Checks if character matches [0-9]
///
///
- public static bool IsDigit(char ch) => ch >= '0' && ch <= '9';
+ public static bool IsAsciiDigit(char ch) => ch >= '0' && ch <= '9';
///
/// Checks if given character is hexadecimal
@@ -17,6 +17,7 @@ public static class CharUtils
///
/// Checks if character is between A-Z or a-z
///
- public static bool IsEnglishAlphabet(char ch) => ch >= 'A' && ch <= 'z' && (ch <= 'Z' || ch >= 'a');
+ /// char.IsAsciiLetter (.NET 7+) is not available in netstandard2.1, which this library also targets.
+ public static bool IsAsciiLetter(char ch) => ch >= 'A' && ch <= 'z' && (ch <= 'Z' || ch >= 'a');
}
}
\ No newline at end of file
diff --git a/src/libse/Common/ContinuationUtilities.cs b/src/libse/Common/ContinuationUtilities.cs
index d2321434bb5..ccc26e51cd5 100644
--- a/src/libse/Common/ContinuationUtilities.cs
+++ b/src/libse/Common/ContinuationUtilities.cs
@@ -708,7 +708,9 @@ public static string RemovePrefix(string originalText, ContinuationProfile profi
foreach (var prefix in Prefixes)
{
- if (newFirstWord.StartsWith(prefix, StringComparison.Ordinal) && !newFirstWord.EndsWith(prefix + Environment.NewLine, StringComparison.Ordinal))
+ // EndsWith cannot express "the prefix is followed by a line break" - the prefix was
+ // just matched at the START. HasPrefix below is the correct twin of this test.
+ if (newFirstWord.StartsWith(prefix, StringComparison.Ordinal) && !newFirstWord.StartsWith(prefix + Environment.NewLine, StringComparison.Ordinal))
{
newFirstWord = newFirstWord.Substring(prefix.Length);
}
@@ -911,7 +913,9 @@ public static bool IsBold(string input)
while (input.IndexOf("", StringComparison.Ordinal) >= 0)
{
var startIndex = input.IndexOf("", StringComparison.Ordinal);
- var endIndex = input.IndexOf("", StringComparison.Ordinal);
+ // From startIndex, as IsItalic does: searching from 0 found a stray leading
+ // "" and produced a negative Substring length, throwing on ordinary text.
+ var endIndex = input.IndexOf("", startIndex, StringComparison.Ordinal);
var textToRemove = endIndex >= 0 ? input.Substring(startIndex, (endIndex + 4) - startIndex) : input.Substring(startIndex);
input = input.Replace(textToRemove, string.Empty);
}
@@ -960,15 +964,15 @@ public static bool IsFullLineTag(string input, int position)
var endIndex = input.LastIndexOf("= 0)
{
+ // A truncated closing tag ("') made IndexOf return -1, and the +1
+ // turned that into 0 - a negative Substring length below, which threw.
+ var closeIndex = input.IndexOf('>', endIndex);
if (startIndex == endIndex)
{
startIndex = 0;
- endIndex = input.IndexOf('>', endIndex) + 1;
- }
- else
- {
- endIndex = input.IndexOf('>', endIndex) + 1;
}
+
+ endIndex = closeIndex >= 0 ? closeIndex + 1 : input.Length;
}
else
{
@@ -1002,8 +1006,16 @@ public static bool IsFullLineQuote(string originalInput, int position, char quot
return false;
}
- // Shift index if needed after deleting { } tags
+ // Shift index if needed after deleting { } tags. IndexOf only finds the shift when
+ // the removed block sat at the very start - with a tag in the middle, input is not a
+ // contiguous substring of originalInput and no shift happened at all, leaving
+ // position past the end of input.
position -= Math.Max(0, originalInput.IndexOf(input, StringComparison.Ordinal));
+ position = Math.Min(position, input.Length - 1);
+ if (position < 0)
+ {
+ return false;
+ }
var startIndex = position;
var endIndex = position;
@@ -1152,7 +1164,7 @@ public static bool HasSuffix(string input, ContinuationProfile profile)
return true;
}
- if (profile.UseDifferentStyleGap && profile.GapSuffix.Length > 0 && input.EndsWith(profile.GapSuffix) && !input.EndsWith(Environment.NewLine + profile.GapSuffix, StringComparison.Ordinal))
+ if (profile.UseDifferentStyleGap && profile.GapSuffix.Length > 0 && input.EndsWith(profile.GapSuffix, StringComparison.Ordinal) && !input.EndsWith(Environment.NewLine + profile.GapSuffix, StringComparison.Ordinal))
{
return true;
}
diff --git a/src/libse/Common/ConvertColorsToDialogUtils.cs b/src/libse/Common/ConvertColorsToDialogUtils.cs
index 9c9a3c5e2d5..45d7abb6889 100644
--- a/src/libse/Common/ConvertColorsToDialogUtils.cs
+++ b/src/libse/Common/ConvertColorsToDialogUtils.cs
@@ -93,7 +93,16 @@ public static void ConvertColorsToDialogInSubtitle(Subtitle subtitle, bool remov
currentColor = newColor;
}
- index = p.Text.IndexOf(">", index) + 1;
+ // An unterminated tag has no '>': IndexOf returns -1 and "+ 1" put the
+ // scan back at 0, so the same tag matched again forever and the whole
+ // app froze. Nothing after an unterminated tag is parseable - stop.
+ var tagEnd = p.Text.IndexOf(">", index, StringComparison.Ordinal);
+ if (tagEnd < 0)
+ {
+ break;
+ }
+
+ index = tagEnd + 1;
endOfColor = false;
}
@@ -113,8 +122,15 @@ public static void ConvertColorsToDialogInSubtitle(Subtitle subtitle, bool remov
}
else if (index + "{".Length <= p.Text.Length && p.Text.SafeSubstring(index, "{".Length) == "{")
{
- // ASS tag, jump over
- index = p.Text.IndexOf("}", index) + 1;
+ // ASS tag, jump over. Same trap as the '>' search above: an unclosed
+ // '{' has no '}', and "-1 + 1" restarted the scan at 0 forever.
+ var assaTagEnd = p.Text.IndexOf("}", index, StringComparison.Ordinal);
+ if (assaTagEnd < 0)
+ {
+ break;
+ }
+
+ index = assaTagEnd + 1;
}
else if (index + 1 <= p.Text.Length && p.Text.SafeSubstring(index, 1) == " " || p.Text.SafeSubstring(index, 1) == "\r" || p.Text.SafeSubstring(index, 1) == "\n")
{
diff --git a/src/libse/Common/CsvUtil.cs b/src/libse/Common/CsvUtil.cs
index cb0495d5839..988b8b206b7 100644
--- a/src/libse/Common/CsvUtil.cs
+++ b/src/libse/Common/CsvUtil.cs
@@ -66,8 +66,15 @@ public static List> CsvSplitLines(List lines, char separato
public static string[] CsvSplit(string line, bool quoteOn, out bool continuation, char separator = ',')
{
var lines = new List();
- var stringOn = false;
+ // A continuation line resumes inside the field the previous line left open, so it
+ // starts "in a string" too. Starting at false handed the first character of such a
+ // line to the not-in-a-string branches below: a line beginning with the closing
+ // quote appended it as literal text and never closed the field (swallowing the rest
+ // of the file), and a line beginning with the separator split a field that is
+ // inside quotes.
+ var stringOn = quoteOn;
var item = new StringBuilder();
+ var sawSeparator = false;
var index = 0;
while (index < line.Length)
{
@@ -82,21 +89,13 @@ public static string[] CsvSplit(string line, bool quoteOn, out bool continuation
{
switch (ch)
{
- case ',' when !quoteOn && separator == ',':
- lines.Add(item.ToString());
- item.Clear();
- index++;
- stringOn = false;
- continue;
- case ';' when !quoteOn && separator == ';':
- lines.Add(item.ToString());
- item.Clear();
- index++;
- stringOn = false;
- continue;
- case '\t' when !quoteOn && separator == '\t':
+ // Any separator, not just comma/semicolon/tab: the import window's
+ // detection also picks '|', and an unhandled separator left the whole
+ // row in the first field with every other column empty.
+ case var _ when !quoteOn && ch == separator:
lines.Add(item.ToString());
item.Clear();
+ sawSeparator = true;
index++;
stringOn = false;
continue;
@@ -135,6 +134,7 @@ public static string[] CsvSplit(string line, bool quoteOn, out bool continuation
{
lines.Add(item.ToString());
item.Clear();
+ sawSeparator = true;
index++;
continue;
}
@@ -149,7 +149,10 @@ public static string[] CsvSplit(string line, bool quoteOn, out bool continuation
index++;
}
- if (item.Length > 0)
+ // "sawSeparator" keeps a trailing empty field: "a," is two columns, the second one
+ // empty, and dropping it made every row that ends on an empty column one field short.
+ // A line with no separator and no content stays zero fields, as before.
+ if (item.Length > 0 || sawSeparator)
{
lines.Add(item.ToString());
}
diff --git a/src/libse/Common/FastBitmap.cs b/src/libse/Common/FastBitmap.cs
index 244eb1e733a..008d70bc2f0 100644
--- a/src/libse/Common/FastBitmap.cs
+++ b/src/libse/Common/FastBitmap.cs
@@ -103,15 +103,53 @@ public void SetPixel(int x, int y, PixelData color)
public void SetPixel(int x, int y, SKColor color, int length)
{
- var data = (PixelData*)(_pBase + y * _width + x * sizeof(PixelData));
- for (int i = 0; i < length; i++)
+ if (length <= 0)
+ {
+ return;
+ }
+
+ // Bgra8888 in memory is B,G,R,A at increasing addresses; as a little-endian
+ // uint that is B | G<<8 | R<<16 | A<<24. A span Fill vectorizes the run.
+ var value = color.Blue | ((uint)color.Green << 8) | ((uint)color.Red << 16) | ((uint)color.Alpha << 24);
+ var data = (uint*)(_pBase + y * _width + x * sizeof(PixelData));
+ new Span(data, length).Fill(value);
+ }
+
+ ///
+ /// True when every pixel in row
+ public bool IsRowTransparent(int y, byte alphaLimit)
+ {
+ var p = _pBase + y * _width + 3;
+ for (var x = 0; x < Width; x++, p += sizeof(PixelData))
{
- data->Alpha = color.Alpha;
- data->Red = color.Red;
- data->Green = color.Green;
- data->Blue = color.Blue;
- data++;
+ if (*p >= alphaLimit)
+ {
+ return false;
+ }
}
+
+ return true;
+ }
+
+ ///
+ /// True when every pixel in column
+ public bool IsColumnTransparent(int x, int startY, byte alphaLimit)
+ {
+ var p = _pBase + startY * _width + x * sizeof(PixelData) + 3;
+ for (var y = startY; y < Height; y++, p += _width)
+ {
+ if (*p >= alphaLimit)
+ {
+ return false;
+ }
+ }
+
+ return true;
}
public SKBitmap GetBitmap()
diff --git a/src/libse/Common/FileUtil.cs b/src/libse/Common/FileUtil.cs
index 46cbca06c61..a23d3e51bdf 100644
--- a/src/libse/Common/FileUtil.cs
+++ b/src/libse/Common/FileUtil.cs
@@ -722,7 +722,7 @@ public static bool IsPlainText(string fileName)
{
return false;
}
- else if (CharUtils.IsDigit(ch))
+ else if (CharUtils.IsAsciiDigit(ch))
{
numberCount++;
}
diff --git a/src/libse/Common/FixCasing.cs b/src/libse/Common/FixCasing.cs
index 7ba5b75c677..f2d7943db3b 100644
--- a/src/libse/Common/FixCasing.cs
+++ b/src/libse/Common/FixCasing.cs
@@ -95,7 +95,7 @@ private static string FixEnglishAloneILowerToUpper(string input)
if (text.StartsWith("I-if ", StringComparison.Ordinal))
{
- text = text.Remove(0, 4).Insert(0, "I-If ");
+ text = text.Remove(0, 4).Insert(0, "I-If");
}
for (var indexOfI = text.IndexOf('i'); indexOfI >= 0; indexOfI = text.IndexOf('i', indexOfI + 1))
@@ -175,7 +175,10 @@ private string FixCasingAfterTitles(string input)
if (start.StartsWith(title.AsSpan(), StringComparison.OrdinalIgnoreCase))
{
var idx = i + title.Length;
- if (idx < text.Length - 2 && text[idx] == ' ')
+ // The body reads text[idx] after the increment below, so the bound only
+ // needs to leave one character - "- 2" skipped a one-letter word at the
+ // end of the line ("Mr. t" was left alone while "Mr. to" was fixed).
+ if (idx < text.Length - 1 && text[idx] == ' ')
{
idx++;
var words = text.Substring(idx).Split(' ', '\r', '\n', ',', '"', '?', '!', '.', '\'');
@@ -220,7 +223,7 @@ public static string FixStutter(string text)
}
else if (!tagOn && char.IsLetter(ch))
{
- if (firstLetter && index < text.Length - 6 && char.IsUpper(text[index]) &&
+ if (firstLetter && index < text.Length - 5 && char.IsUpper(text[index]) &&
text[index + 1] == '-' && char.IsLower(text[index + 2]) && text[index] == char.ToUpperInvariant(text[index + 2]) &&
text[index + 3] == '-' && char.IsLower(text[index + 4]) && text[index] == char.ToUpperInvariant(text[index + 4]) &&
text[index + 5] != '-')
@@ -231,7 +234,7 @@ public static string FixStutter(string text)
sb.Append('-');
index += 4;
}
- else if (firstLetter && index < text.Length - 4 && char.IsUpper(text[index]) &&
+ else if (firstLetter && index < text.Length - 3 && char.IsUpper(text[index]) &&
text[index + 1] == '-' && char.IsLower(text[index + 2]) && text[index] == char.ToUpperInvariant(text[index + 2]) &&
text[index + 3] != '-')
{
@@ -281,12 +284,12 @@ private string Fix(string original, string lastLine, List nameList, Cult
else if (FixMakeUppercase)
{
var st = new StrippableText(text);
- text = st.Pre + MakeUpperCaseExceptTags(st.StrippedText) + st.Post;
+ text = st.Pre + MakeUpperCaseExceptTags(st.StrippedText, subtitleCulture) + st.Post;
text = HtmlUtil.FixUpperTags(text); // tags inside text
}
else if (FixMakeLowercase)
{
- text = MakeLowerCaseExceptTags(text);
+ text = MakeLowerCaseExceptTags(text, subtitleCulture);
}
else if (FixMakeProperCase)
{
@@ -312,15 +315,19 @@ private string Fix(string original, string lastLine, List nameList, Cult
return text;
}
- private string MakeUpperCaseExceptTags(string text)
+ // The case conversion follows the SUBTITLE's language, not the machine's: char.ToUpper /
+ // char.ToLower without a culture use CurrentCulture, so on a Turkish system every "i" in
+ // an English subtitle became "I-with-dot" and every "I" became a dotless "i".
+ private string MakeUpperCaseExceptTags(string text, CultureInfo culture)
{
if (string.IsNullOrEmpty(text))
{
return text;
}
- return string.Create(text.Length, text, (span, src) =>
+ return string.Create(text.Length, (text, culture), (span, state) =>
{
+ var (src, ci) = state;
var insideAngle = false;
var insideCurly = false;
@@ -344,20 +351,21 @@ private string MakeUpperCaseExceptTags(string text)
insideCurly = false;
}
- span[i] = insideAngle || insideCurly ? c : char.ToUpper(c);
+ span[i] = insideAngle || insideCurly ? c : char.ToUpper(c, ci);
}
});
}
- private string MakeLowerCaseExceptTags(string text)
+ private string MakeLowerCaseExceptTags(string text, CultureInfo culture)
{
if (string.IsNullOrEmpty(text))
{
return text;
}
- return string.Create(text.Length, text, (span, src) =>
+ return string.Create(text.Length, (text, culture), (span, state) =>
{
+ var (src, ci) = state;
var insideAngle = false;
var insideCurly = false;
@@ -381,7 +389,7 @@ private string MakeLowerCaseExceptTags(string text)
insideCurly = false;
}
- span[i] = insideAngle || insideCurly ? c : char.ToLower(c);
+ span[i] = insideAngle || insideCurly ? c : char.ToLower(c, ci);
}
});
}
diff --git a/src/libse/Common/HtmlUtil.cs b/src/libse/Common/HtmlUtil.cs
index a251c2e7a03..2bcd8280420 100644
--- a/src/libse/Common/HtmlUtil.cs
+++ b/src/libse/Common/HtmlUtil.cs
@@ -879,6 +879,14 @@ public static bool IsTextFormattable(in string text)
/// A string with corrected italic tags.
public static string FixInvalidItalicTags(string input)
{
+ // Every transformation below requires a '<' somewhere in the text (the "{\...}"
+ // prefix is only sliced off to be re-prepended verbatim), and most subtitle lines
+ // carry no tags at all - so skip the ~50 full-string Replace scans for those.
+ if (string.IsNullOrEmpty(input) || input.IndexOf('<') < 0)
+ {
+ return input;
+ }
+
var text = input;
var preTags = string.Empty;
@@ -1099,7 +1107,9 @@ int CalculateEarlyInsertIndex(string s)
text = beginTag + text + endTag;
}
- if (italicBeginTagCount == 0 && italicEndTagCount == 2)
+ // "else if": the counts above are stale after the block just ran, so this used to undo
+ // it - rewriting the closing tag it had just added into a second opening tag.
+ else if (italicBeginTagCount == 0 && italicEndTagCount == 2)
{
var firstIndex = text.IndexOf(endTag, StringComparison.Ordinal);
text = text.Remove(firstIndex, endTag.Length).Insert(firstIndex, beginTag);
@@ -1176,7 +1186,11 @@ int CalculateEarlyInsertIndex(string s)
//- You think they're they gone?
//- That can't be.
- if (italicBeginTagCount == 3 && italicEndTagCount == 1 && noOfLines == 2)
+ // GetNumberOfLines counts '\n', so noOfLines is 2 for a text broken with a bare "\n"
+ // too - and on Windows IndexOf(Environment.NewLine) then returns -1 and Substring
+ // threw. The sibling branches in this method all guard the index the same way.
+ if (italicBeginTagCount == 3 && italicEndTagCount == 1 && noOfLines == 2 &&
+ text.IndexOf(Environment.NewLine, StringComparison.Ordinal) > 0)
{
var newLineIdx = text.IndexOf(Environment.NewLine, StringComparison.Ordinal);
var firstLine = text.Substring(0, newLineIdx).Trim();
@@ -1592,14 +1606,14 @@ public static string RemoveColorTags(string input)
endIndex = s.IndexOf("< /font>", match.Index - 5, StringComparison.OrdinalIgnoreCase);
if (endIndex >= 0)
{
- s = s.Remove(endIndex, 7);
+ s = s.Remove(endIndex, 8);
}
else
{
endIndex = s.IndexOf("", match.Index - 5, StringComparison.OrdinalIgnoreCase);
if (endIndex >= 0)
{
- s = s.Remove(endIndex, 7);
+ s = s.Remove(endIndex, 8);
}
}
}
diff --git a/src/libse/Common/ListExtensions.cs b/src/libse/Common/ListExtensions.cs
index 9eb5f54ea7e..508bd3526a3 100644
--- a/src/libse/Common/ListExtensions.cs
+++ b/src/libse/Common/ListExtensions.cs
@@ -45,11 +45,17 @@ public static T FirstOnOrBefore(this List orderedList, T target, T default
public static double FirstOnOrAfter(this List orderedList, double target, double maxDifference, double defaultValue)
{
int index = orderedList.BinarySearch(target);
- if (index < 0)
+ if (index >= 0)
{
- index = ~index; // Get the bitwise complement to find the index of the first element greater than the target
+ // An exact hit is always the answer. The tolerance probe below used to run for it
+ // too and could return the PREVIOUS element - so with shot changes [9.98, 10.0]
+ // and a cue already snapped to 10.0, "the next shot change" came back as 9.98,
+ // 20 ms earlier. The generic overloads above only rewrite index when it is < 0.
+ return orderedList[index];
}
+ index = ~index; // Get the bitwise complement to find the index of the first element greater than the target
+
if ((index - 1) >= 0 && target - orderedList[index - 1] <= maxDifference)
{
return orderedList[index - 1];
@@ -66,11 +72,15 @@ public static double FirstOnOrAfter(this List orderedList, double target
public static double FirstOnOrBefore(this List orderedList, double target, double maxDifference, double defaultValue)
{
int index = orderedList.BinarySearch(target);
- if (index < 0)
+ if (index >= 0)
{
- index = ~index - 1; // Get the bitwise complement to find the index of the last element smaller than the target
+ // Exact hit wins - see the note in FirstOnOrAfter. Here the probe returned the
+ // element AFTER the target instead.
+ return orderedList[index];
}
+ index = ~index - 1; // Get the bitwise complement to find the index of the last element smaller than the target
+
if (index + 1 < orderedList.Count && orderedList[index + 1] - target <= maxDifference)
{
return orderedList[index + 1];
diff --git a/src/libse/Common/NetflixImsc11JapaneseToAss.cs b/src/libse/Common/NetflixImsc11JapaneseToAss.cs
index be29f4f9413..ace31eadd3a 100644
--- a/src/libse/Common/NetflixImsc11JapaneseToAss.cs
+++ b/src/libse/Common/NetflixImsc11JapaneseToAss.cs
@@ -97,6 +97,34 @@ private static bool IsFullWidth(char ch)
ch >= 0xFFE0 && ch <= 0xFFE6);
}
+ ///
+ /// True when the subtitle carries the Japanese profile markup libass cannot draw by itself.
+ /// The format alone does not say: "Lambda Cap" decodes furigana, emphasis marks and
+ /// tate-chu-yoko to the same markup, and its control codes would otherwise end up on screen
+ /// as literal text (issue #14165).
+ ///
+ public static bool HasJapaneseMarkup(Subtitle subtitle)
+ {
+ if (subtitle == null)
+ {
+ return false;
+ }
+
+ foreach (var p in subtitle.Paragraphs)
+ {
+ var text = p.Text;
+ if (!string.IsNullOrEmpty(text) && text.Contains('<') &&
+ (text.Contains("", StringComparison.Ordinal)))
+ {
+ return true;
+ }
+ }
+
+ return false;
+ }
+
public static List SplitToAssRenderLines(Paragraph p, int width, int height)
{
var metrics = new Metrics(height);
@@ -183,6 +211,16 @@ private static List MakeHorizontalParagraphs(Paragraph p, int width,
i += 4;
italicOn = false;
}
+ else if (line.Substring(i).StartsWith("", StringComparison.Ordinal))
+ {
+ // Tate-chu-yoko only means something in a vertical column - horizontal digits
+ // are already horizontal, so just drop the tags (issue #14165).
+ i += "".Length;
+ }
+ else if (line.Substring(i).StartsWith(" ", StringComparison.Ordinal))
+ {
+ i += " ".Length;
+ }
else if (line.Substring(i).StartsWith("', i);
diff --git a/src/libse/Common/NikseBitmap.cs b/src/libse/Common/NikseBitmap.cs
index 15ac4246a4e..6868be6fcc5 100644
--- a/src/libse/Common/NikseBitmap.cs
+++ b/src/libse/Common/NikseBitmap.cs
@@ -304,28 +304,51 @@ public SKColor ConvertToFourColors(SKColor background, SKColor pattern, SKColor
emphasis2Buffer[3] = emphasis2.Alpha;
}
- for (int i = 0; i < _bitmapData.Length; i += 4)
+ // One 32-bit read and one 32-bit write per pixel. The loop used to index the byte
+ // array eight times per candidate colour and then call Buffer.BlockCopy to store the
+ // four result bytes - a full call per pixel of every exported VobSub/DVD image.
+ var packedBackground = PackBgra(backgroundBuffer[0], backgroundBuffer[1], backgroundBuffer[2], backgroundBuffer[3]);
+ var packedPattern = PackBgra(patternBuffer[0], patternBuffer[1], patternBuffer[2], patternBuffer[3]);
+ var packedEmphasis1 = PackBgra(emphasis1Buffer[0], emphasis1Buffer[1], emphasis1Buffer[2], emphasis1Buffer[3]);
+ var packedEmphasis2 = PackBgra(emphasis2Buffer[0], emphasis2Buffer[1], emphasis2Buffer[2], emphasis2Buffer[3]);
+ var backgroundIsTransparent = backgroundBuffer[3] == 0;
+
+ // The four reference colours as plain locals: read out of the byte arrays they were
+ // indexed out of twelve times per pixel, each with its own bounds check.
+ int patternB = patternBuffer[0], patternG = patternBuffer[1], patternR = patternBuffer[2], patternA = patternBuffer[3];
+ int e1B = emphasis1Buffer[0], e1G = emphasis1Buffer[1], e1R = emphasis1Buffer[2], e1A = emphasis1Buffer[3];
+ int e2B = emphasis2Buffer[0], e2G = emphasis2Buffer[1], e2R = emphasis2Buffer[2], e2A = emphasis2Buffer[3];
+
+ var pixels = MemoryMarshal.Cast(_bitmapData.AsSpan());
+
+ for (var p = 0; p < pixels.Length; p++)
{
- int smallestDiff = 10000;
- byte[] buffer = backgroundBuffer;
- if (backgroundBuffer[3] == 0 && _bitmapData[i + 3] < 10) // transparent
+ var pixel = pixels[p];
+ int blue = (byte)pixel;
+ int green = (byte)(pixel >> 8);
+ int red = (byte)(pixel >> 16);
+ int alpha = (byte)(pixel >> 24);
+
+ var smallestDiff = 10000;
+ var result = packedBackground;
+ if (backgroundIsTransparent && alpha < 10) // transparent
{
}
else
{
- int patternDiff = Math.Abs(patternBuffer[0] - _bitmapData[i]) + Math.Abs(patternBuffer[1] - _bitmapData[i + 1]) + Math.Abs(patternBuffer[2] - _bitmapData[i + 2]) + Math.Abs(patternBuffer[3] - _bitmapData[i + 3]);
+ var patternDiff = Math.Abs(patternB - blue) + Math.Abs(patternG - green) + Math.Abs(patternR - red) + Math.Abs(patternA - alpha);
if (patternDiff < smallestDiff)
{
smallestDiff = patternDiff;
- buffer = patternBuffer;
+ result = packedPattern;
}
- int emphasis1Diff = Math.Abs(emphasis1Buffer[0] - _bitmapData[i]) + Math.Abs(emphasis1Buffer[1] - _bitmapData[i + 1]) + Math.Abs(emphasis1Buffer[2] - _bitmapData[i + 2]) + Math.Abs(emphasis1Buffer[3] - _bitmapData[i + 3]);
+ var emphasis1Diff = Math.Abs(e1B - blue) + Math.Abs(e1G - green) + Math.Abs(e1R - red) + Math.Abs(e1A - alpha);
if (useInnerAntialize)
{
if (emphasis1Diff - 20 < smallestDiff)
{
- buffer = emphasis1Buffer;
+ result = packedEmphasis1;
}
}
else
@@ -333,21 +356,22 @@ public SKColor ConvertToFourColors(SKColor background, SKColor pattern, SKColor
if (emphasis1Diff < smallestDiff)
{
smallestDiff = emphasis1Diff;
- buffer = emphasis1Buffer;
+ result = packedEmphasis1;
}
- int emphasis2Diff = Math.Abs(emphasis2Buffer[0] - _bitmapData[i]) + Math.Abs(emphasis2Buffer[1] - _bitmapData[i + 1]) + Math.Abs(emphasis2Buffer[2] - _bitmapData[i + 2]) + Math.Abs(emphasis2Buffer[3] - _bitmapData[i + 3]);
+ var emphasis2Diff = Math.Abs(e2B - blue) + Math.Abs(e2G - green) + Math.Abs(e2R - red) + Math.Abs(e2A - alpha);
if (emphasis2Diff < smallestDiff)
{
- buffer = emphasis2Buffer;
+ result = packedEmphasis2;
}
- else if (_bitmapData[i + 3] >= 10 && _bitmapData[i + 3] < 90) // anti-alias
+ else if (alpha >= 10 && alpha < 90) // anti-alias
{
- buffer = emphasis2Buffer;
+ result = packedEmphasis2;
}
}
}
- Buffer.BlockCopy(buffer, 0, _bitmapData, i, 4);
+
+ pixels[p] = result;
}
if (useInnerAntialize)
diff --git a/src/libse/Common/Paragraph.cs b/src/libse/Common/Paragraph.cs
index d251bd8740b..4823d101e98 100644
--- a/src/libse/Common/Paragraph.cs
+++ b/src/libse/Common/Paragraph.cs
@@ -131,6 +131,14 @@ public double WordsPerMinute
return 0;
}
+ // Same guard as GetCharactersPerSecond: a zero or negative duration would give
+ // infinity/a negative rate, which poisons the min/max/average of every statistics
+ // report that sums this up.
+ if (DurationTotalMilliseconds < 1)
+ {
+ return 999;
+ }
+
return 60.0 / DurationTotalSeconds * Text.CountWords();
}
}
diff --git a/src/libse/Common/RegexUtils.cs b/src/libse/Common/RegexUtils.cs
index 26ab1f537b9..09d6cd7f2c7 100644
--- a/src/libse/Common/RegexUtils.cs
+++ b/src/libse/Common/RegexUtils.cs
@@ -12,7 +12,13 @@ public static partial class RegexUtils
/// matches on the UI thread that means a frozen program with no way out. SE 4 used the same five
/// seconds in its find/replace helper; callers treat the timeout as "no match".
///
- public static readonly TimeSpan UserPatternMatchTimeout = TimeSpan.FromSeconds(5);
+ ///
+ /// Settable only so the tests that prove the timeout works do not have to wait it out - fourteen
+ /// of them sat on the full five seconds, close to half the whole UI suite. Nothing in the program
+ /// writes it; callers read it when they build the Regex, so a test sets it before the pattern is
+ /// compiled and puts it back afterwards.
+ ///
+ public static TimeSpan UserPatternMatchTimeout { get; set; } = TimeSpan.FromSeconds(5);
// Others classes may want to use this regex.
#if NET7_0_OR_GREATER
@@ -56,7 +62,7 @@ static DanishLetterI()
RegexFactory(@"\bburde i (?:gøre|ikke|købe|løbe|se|sig|tage)\b"),
RegexFactory(@"\bfanden vil i\?"),
RegexFactory(@"\b[Ff]or ser i\b"),
- RegexFactory(@"\b[Db]a i (?:ankom|forlod|fik|gik|kom|dræbte|ikke|gjorde|har|havde|så)\b"),
+ RegexFactory(@"\b[Dd]a i (?:ankom|forlod|fik|gik|kom|dræbte|ikke|gjorde|har|havde|så)\b"),
RegexFactory(@"\b[Dd]et (?:får i|har i|må i|gør i|gjorde i)\b"),
RegexFactory(@"\b[Dd]et må i (?:fandme|ikke|sgu|gerne|selv|da|faktisk)\b"),
RegexFactory(@"\b[Dd]et Det kan i sgu"),
@@ -144,7 +150,7 @@ static DanishLetterI()
RegexFactory(@"\b[Tt]og i (?:bilen|liften|toget)\b"),
RegexFactory(@"\b[Tt]ræder i frem\b"),
RegexFactory(@"\b[Tt]ror i (?:at|det|jeg|på|virkelig)\b"),
- RegexFactory(@"\b[Tr]ror i(?:, | på\b)"),
+ RegexFactory(@"\b[Tt]ror i(?:, | på\b)"),
RegexFactory(@"\b[Vv]ar i blevet\b"),
RegexFactory(@"\b[Vv]ed i (?:alle|allesammen|er|ikke|hvad|hvem|hvor|hvorfor|hvordan|var|ville|har|havde|hvem|hvad|hvor|mente|tror)\b"),
RegexFactory(@"\b[Vv]enter i på\b"),
@@ -291,7 +297,7 @@ public static string ReplaceNewLineSafe(Regex regularExpression, string text, st
/// \n left behind by the match normalization above was written as a physical newline
/// inside the Dialogue line, corrupting the saved file on Windows (#12620).
///
- private static string RestorePlatformNewLines(string text)
+ public static string RestorePlatformNewLines(string text)
{
if (Environment.NewLine == "\n" || text.IndexOf('\n') < 0)
{
diff --git a/src/libse/Common/RichTextToPlainText.cs b/src/libse/Common/RichTextToPlainText.cs
index 481def99c2e..876568a465b 100644
--- a/src/libse/Common/RichTextToPlainText.cs
+++ b/src/libse/Common/RichTextToPlainText.cs
@@ -146,13 +146,18 @@ public static string ConvertToText(string inputRtf)
// Push state
stack.Push(new StackEntry(ucskip, ignorable));
}
- else
+ else if (stack.Count > 0)
{
// Pop state
StackEntry entry = stack.Pop();
ucskip = entry.NumberOfCharactersToSkip;
ignorable = entry.Ignorable;
}
+
+ // else: a '}' with no matching '{'. Popping an empty stack threw
+ // InvalidOperationException, and because ~20 readers run their input through
+ // FromRtf(), a single unbalanced brace anywhere aborted format detection for
+ // the whole file instead of reporting an unknown format. Ignore the brace.
}
else if (character.Length > 0) // \x (not a letter)
{
diff --git a/src/libse/Common/SeJsonParser.cs b/src/libse/Common/SeJsonParser.cs
index 269997dc0af..11cf94b99c1 100644
--- a/src/libse/Common/SeJsonParser.cs
+++ b/src/libse/Common/SeJsonParser.cs
@@ -86,6 +86,15 @@ public List GetAllTagsByNameAsStrings(string content, string name)
{
i++;
var end = content.IndexOf('"', i);
+ if (end < 0)
+ {
+ // Unterminated name - the file is truncated or damaged. Bail out
+ // instead of letting Substring throw out of the reader (and out of
+ // IsMine, which format detection calls unguarded).
+ Errors.Add($"Fatal - unterminated string at position {i}");
+ return list;
+ }
+
objectName = content.Substring(i, end - i).Trim();
var colon = content.IndexOf(':', end);
if (colon < 0)
@@ -151,7 +160,9 @@ public List GetAllTagsByNameAsStrings(string content, string name)
Errors.Add($"Fatal - expected char \" after position {endSeek}");
return list;
}
- skip = content[end - 1] == '\\';
+ // end can be 0 when the value's opening quote is the very first
+ // character (a truncated file), and content[-1] threw.
+ skip = end > 0 && content[end - 1] == '\\';
if (skip)
{
endSeek = end + 1;
@@ -247,7 +258,10 @@ public List GetAllTagsByNameAsStrings(string content, string name)
else if ("+-0123456789".IndexOf(ch) >= 0)
{
sb.Clear();
- while (IsNumberChar(content[i]) && i < max)
+ // Bounds check FIRST: this indexed content[i] before testing i < max,
+ // so a file whose last characters are a number (a truncated one)
+ // threw IndexOutOfRangeException straight out of the reader.
+ while (i < max && IsNumberChar(content[i]))
{
sb.Append(content[i]);
i++;
@@ -378,6 +392,15 @@ public List GetArrayElementsByName(string content, string name)
{
i++;
var end = content.IndexOf('"', i);
+ if (end < 0)
+ {
+ // Unterminated name - the file is truncated or damaged. Bail out
+ // instead of letting Substring throw out of the reader (and out of
+ // IsMine, which format detection calls unguarded).
+ Errors.Add($"Fatal - unterminated string at position {i}");
+ return list;
+ }
+
objectName = content.Substring(i, end - i).Trim();
var colon = content.IndexOf(':', end);
if (colon < 0)
@@ -443,7 +466,9 @@ public List GetArrayElementsByName(string content, string name)
Errors.Add($"Fatal - expected char \" after position {endSeek}");
return list;
}
- skip = content[end - 1] == '\\';
+ // end can be 0 when the value's opening quote is the very first
+ // character (a truncated file), and content[-1] threw.
+ skip = end > 0 && content[end - 1] == '\\';
if (skip)
{
endSeek = end + 1;
@@ -684,6 +709,15 @@ public List GetArrayElements(string content)
{
i++;
var end = content.IndexOf('"', i);
+ if (end < 0)
+ {
+ // Unterminated name - the file is truncated or damaged. Bail out
+ // instead of letting Substring throw out of the reader (and out of
+ // IsMine, which format detection calls unguarded).
+ Errors.Add($"Fatal - unterminated string at position {i}");
+ return list;
+ }
+
objectName = content.Substring(i, end - i).Trim();
var colon = content.IndexOf(':', end);
if (colon < 0)
@@ -749,7 +783,9 @@ public List GetArrayElements(string content)
Errors.Add($"Fatal - expected char \" after position {endSeek}");
return list;
}
- skip = content[end - 1] == '\\';
+ // end can be 0 when the value's opening quote is the very first
+ // character (a truncated file), and content[-1] threw.
+ skip = end > 0 && content[end - 1] == '\\';
if (skip)
{
endSeek = end + 1;
@@ -826,7 +862,10 @@ public List GetArrayElements(string content)
}
else if ("+-0123456789".IndexOf(ch) >= 0)
{
- while (IsNumberChar(content[i]) && i < max)
+ // Bounds check FIRST: this indexed content[i] before testing i < max,
+ // so a file whose last characters are a number (a truncated one)
+ // threw IndexOutOfRangeException straight out of the reader.
+ while (i < max && IsNumberChar(content[i]))
{
i++;
}
@@ -978,6 +1017,13 @@ public string GetFirstObject(string content, string name)
{
i++;
var end = content.IndexOf('"', i);
+ if (end < 0)
+ {
+ // Unterminated name - see the list-returning overloads above.
+ Errors.Add($"Fatal - unterminated string at position {i}");
+ return string.Empty;
+ }
+
objectName = content.Substring(i, end - i).Trim();
var colon = content.IndexOf(':', end);
if (colon < 0)
@@ -1052,7 +1098,9 @@ public string GetFirstObject(string content, string name)
Errors.Add($"Fatal - expected char \" after position {endSeek}");
return string.Empty;
}
- skip = content[end - 1] == '\\';
+ // end can be 0 when the value's opening quote is the very first
+ // character (a truncated file), and content[-1] threw.
+ skip = end > 0 && content[end - 1] == '\\';
if (skip)
{
endSeek = end + 1;
@@ -1137,7 +1185,10 @@ public string GetFirstObject(string content, string name)
else if ("+-0123456789".IndexOf(ch) >= 0)
{
sb.Clear();
- while (IsNumberChar(content[i]) && i < max)
+ // Bounds check FIRST: this indexed content[i] before testing i < max,
+ // so a file whose last characters are a number (a truncated one)
+ // threw IndexOutOfRangeException straight out of the reader.
+ while (i < max && IsNumberChar(content[i]))
{
sb.Append(content[i]);
i++;
@@ -1280,6 +1331,15 @@ public List GetRootElements(string content)
{
i++;
var end = content.IndexOf('"', i);
+ if (end < 0)
+ {
+ // Unterminated name - the file is truncated or damaged. Bail out
+ // instead of letting Substring throw out of the reader (and out of
+ // IsMine, which format detection calls unguarded).
+ Errors.Add($"Fatal - unterminated string at position {i}");
+ return list;
+ }
+
objectName = content.Substring(i, end - i).Trim();
var colon = content.IndexOf(':', end);
if (colon < 0)
@@ -1364,7 +1424,9 @@ public List GetRootElements(string content)
Errors.Add($"Fatal - expected char \" after position {endSeek}");
return list;
}
- skip = content[end - 1] == '\\';
+ // end can be 0 when the value's opening quote is the very first
+ // character (a truncated file), and content[-1] threw.
+ skip = end > 0 && content[end - 1] == '\\';
if (skip)
{
endSeek = end + 1;
diff --git a/src/libse/Common/ShotChangeHelper.cs b/src/libse/Common/ShotChangeHelper.cs
index 2fb452af3b1..2037d907cfc 100644
--- a/src/libse/Common/ShotChangeHelper.cs
+++ b/src/libse/Common/ShotChangeHelper.cs
@@ -123,11 +123,29 @@ public static void SaveShotChanges(string videoFileName, string shotChangeDirect
/// Directory where shot change files are stored
public static void DeleteShotChanges(string videoFileName, string shotChangeDirectory)
{
+ // Delete every file FromDisk could load, not just the canonical name: the hash is
+ // computed from the video's CONTENT, so a renamed video still finds its old
+ // "{hash}_{oldName}.shotchanges" through the wildcard in FindShotChangesFileName -
+ // while this method used to build "{hash}_{newName}.shotchanges", find nothing, and
+ // leave the shot changes to come straight back.
var shotChangesFileName = GetShotChangesFileName(videoFileName, shotChangeDirectory);
if (File.Exists(shotChangesFileName))
{
File.Delete(shotChangesFileName);
}
+
+ var hash = MovieHasher.GenerateHash(videoFileName);
+ foreach (var fileName in Directory.GetFiles(shotChangeDirectory, $"{hash}*.shotchanges"))
+ {
+ try
+ {
+ File.Delete(fileName);
+ }
+ catch
+ {
+ // a file we cannot delete must not abort the rest
+ }
+ }
}
diff --git a/src/libse/Common/SsaStyle.cs b/src/libse/Common/SsaStyle.cs
index 77eecc7ff28..836d67f9c60 100644
--- a/src/libse/Common/SsaStyle.cs
+++ b/src/libse/Common/SsaStyle.cs
@@ -291,10 +291,6 @@ public string ToRawAss(string styleFormat = DefaultAssStyleFormat)
{
sb.Append('1');
}
- else if (f == "strikeout")
- {
- sb.Append('0');
- }
else if (f == "scalex")
{
sb.Append(ScaleX.ToString(CultureInfo.InvariantCulture));
diff --git a/src/libse/Common/StringExtensions.cs b/src/libse/Common/StringExtensions.cs
index 0cb2324386d..eea2d9adcf9 100644
--- a/src/libse/Common/StringExtensions.cs
+++ b/src/libse/Common/StringExtensions.cs
@@ -35,7 +35,7 @@ public static bool LineEndsWithHtmlTag(this string text, bool threeLengthTag, bo
}
var len = text.Length;
- if (len < 6 || text[len - 1] != '>')
+ if (len < 5 || text[len - 1] != '>')
{
return false;
}
@@ -46,7 +46,7 @@ public static bool LineEndsWithHtmlTag(this string text, bool threeLengthTag, bo
return true;
}
- if (includeFont && len > 8 && text[len - 7] == '<' && text[len - 6] == '/')
+ if (includeFont && len >= 8 && text[len - 7] == '<' && text[len - 6] == '/')
{
return true;
}
@@ -492,7 +492,7 @@ public static bool ContainsNumber(this string s)
var len = s.Length;
for (var i = 0; i < len; i++)
{
- if (CharUtils.IsDigit(s[i]))
+ if (CharUtils.IsAsciiDigit(s[i]))
{
return true;
}
@@ -795,7 +795,11 @@ private static List> RemoveAndSaveTags(string input, S
&& index + 1 < input.Length
&& (input[index + 1] == 'N' || input[index + 1] == 'n' || input[index + 1] == 'h'))
{
- tags.Add(new KeyValuePair(index, input.Substring(index, 2)));
+ // sb.Length, not index: every other tag is recorded in tag-stripped
+ // coordinates, which is what RestoreSavedAndRemovedTags re-inserts into. Using
+ // the input offset put "\N" back in the wrong place whenever a tag preceded it
+ // ("{\an8}One\Ntwo" toggled to "{\an8}ONETWO\N").
+ tags.Add(new KeyValuePair(sb.Length, input.Substring(index, 2)));
skipNext = true;
continue;
}
diff --git a/src/libse/Common/StrippableText.cs b/src/libse/Common/StrippableText.cs
index add2bd48af9..6d41ad61a25 100644
--- a/src/libse/Common/StrippableText.cs
+++ b/src/libse/Common/StrippableText.cs
@@ -357,7 +357,12 @@ public void FixCasing(IEnumerable nameList, bool changeNameCases, bool m
var idx = sb.ToString().IndexOf('[');
if (s == ']' && idx > 1)
{ // I [Motor roaring] love you!
- var temp = sb.ToString(0, idx - 1).Trim();
+ // The text before '[' is [0, idx), so the length is idx - the
+ // "- 1" dropped the last character before the bracket, which is
+ // exactly the one this test looks at. "Yes.[Motor roaring] hello."
+ // saw "Yes" (a letter) instead of "Yes." and left "hello"
+ // lowercase. The documented "I [Motor roaring]" case is unchanged.
+ var temp = sb.ToString(0, idx).Trim();
if (temp.Length > 0 && !char.IsLetterOrDigit(temp[temp.Length - 1]))
{
lastWasBreak = true;
diff --git a/src/libse/Common/Subtitle.cs b/src/libse/Common/Subtitle.cs
index 30311410bb2..c25a18a7ce3 100644
--- a/src/libse/Common/Subtitle.cs
+++ b/src/libse/Common/Subtitle.cs
@@ -291,7 +291,7 @@ public SubtitleFormat LoadSubtitle(string fileName, out Encoding encoding, Encod
var ext = Path.GetExtension(fileName).ToLowerInvariant();
foreach (var subtitleFormat in SubtitleFormat.AllSubtitleFormats.Where(p => p.Extension.Equals(ext, StringComparison.OrdinalIgnoreCase) && !p.Name.StartsWith("Unknown", StringComparison.Ordinal)))
{
- if (subtitleFormat.IsMine(lines, fileName))
+ if (IsFormatMine(subtitleFormat, lines, fileName))
{
return FinalizeFormat(fileName, batchMode, sourceFrameRate, lines, subtitleFormat, loadSubtitle);
}
@@ -299,7 +299,7 @@ public SubtitleFormat LoadSubtitle(string fileName, out Encoding encoding, Encod
foreach (var subtitleFormat in SubtitleFormat.AllSubtitleFormats.Where(p => !p.Extension.Equals(ext, StringComparison.OrdinalIgnoreCase) || p.Name.StartsWith("Unknown", StringComparison.Ordinal)))
{
- if (subtitleFormat.IsMine(lines, fileName))
+ if (IsFormatMine(subtitleFormat, lines, fileName))
{
return FinalizeFormat(fileName, batchMode, sourceFrameRate, lines, subtitleFormat, loadSubtitle);
}
@@ -313,6 +313,25 @@ public SubtitleFormat LoadSubtitle(string fileName, out Encoding encoding, Encod
return null;
}
+ ///
+ /// Asks one format whether the lines are its own. A damaged or truncated file can make a
+ /// reader throw (an unclosed xml element, a json string with no end quote, ...), and this
+ /// runs for EVERY format when a file is opened - so a throw here used to take down the
+ /// whole open instead of moving on to the next format.
+ ///
+ private static bool IsFormatMine(SubtitleFormat subtitleFormat, List lines, string fileName)
+ {
+ try
+ {
+ return subtitleFormat.IsMine(lines, fileName);
+ }
+ catch (Exception exception)
+ {
+ System.Diagnostics.Debug.WriteLine($"{subtitleFormat.Name}.IsMine failed: {exception.Message}");
+ return false;
+ }
+ }
+
private static List ReadLinesFromFile(string fileName, Encoding useThisEncoding, out Encoding encoding)
{
StreamReader sr;
@@ -383,6 +402,43 @@ public void ChangeFrameRate(double oldFrameRate, double newFrameRate)
}
}
+ ///
+ /// Frame-rate conversion that lands on whole milliseconds. Scales the start and the
+ /// duration rather than the start and the end independently, so two cues of equal length
+ /// keep equal lengths afterwards - scaling both ends separately rounds them apart (#14056).
+ ///
+ public void ChangeFrameRateWholeMilliseconds(double oldFrameRate, double newFrameRate)
+ {
+ var factor = SubtitleFormat.GetFrameForCalculation(oldFrameRate) / SubtitleFormat.GetFrameForCalculation(newFrameRate);
+ Paragraph previous = null;
+ var previousOriginalEndMs = 0d;
+ foreach (var p in Paragraphs)
+ {
+ var originalStartMs = p.StartTime.TotalMilliseconds;
+ var originalEndMs = p.EndTime.TotalMilliseconds;
+
+ var newStartMs = Math.Round(originalStartMs * factor, MidpointRounding.AwayFromZero);
+ var newDurationMs = Math.Round((originalEndMs - originalStartMs) * factor, MidpointRounding.AwayFromZero);
+ p.StartTime.TotalMilliseconds = newStartMs;
+ p.EndTime.TotalMilliseconds = newStartMs + newDurationMs;
+
+ // The two roundings can push the previous end one millisecond past this start,
+ // turning a clean join into an overlap the source never had. Overlaps that were
+ // already in the source are left as they were.
+ if (previous != null &&
+ previousOriginalEndMs <= originalStartMs &&
+ previous.EndTime.TotalMilliseconds > p.StartTime.TotalMilliseconds)
+ {
+ previous.EndTime.TotalMilliseconds = p.StartTime.TotalMilliseconds;
+ }
+
+ previous = p;
+ previousOriginalEndMs = originalEndMs;
+ }
+ }
+
public void AdjustDisplayTimeUsingPercent(double percent, List selectedIndexes, List shotChanges = null, bool enforceDurationLimits = true)
{
// List.Contains per paragraph made this O(paragraphs * selection) - quadratic with
@@ -412,8 +468,10 @@ public void AdjustDisplayTimeUsingPercent(double percent, List selectedInde
}
}
- // handle overlap with next
- if (newEndMilliseconds > nextStartMilliseconds)
+ // handle overlap with next - measured against the minimum gap, the way
+ // AdjustDisplayTimeUsingMilliseconds does. Testing only for a hard overlap
+ // let a new end land inside the gap (or exactly on the next start).
+ if (newEndMilliseconds > nextStartMilliseconds - Configuration.Settings.General.MinimumMillisecondsBetweenLines)
{
newEndMilliseconds = nextStartMilliseconds - Configuration.Settings.General.MinimumMillisecondsBetweenLines;
}
@@ -554,7 +612,16 @@ public void RecalculateDisplayTime(double maxCharactersPerSecond, int index, dou
if (p.GetCharactersPerSecond() > maxCharactersPerSecond)
{
var numberOfCharacters = (double)p.Text.CountCharacters(true);
- var maxDurationMilliseconds = (numberOfCharacters / maxCharactersPerSecond) * 1000.0;
+ // Whole milliseconds, rounded up: the fractional value truncates on save to one ms
+ // short, which is just over the maximum this branch exists to enforce (#14418).
+ // Verified with the same division GetCharactersPerSecond performs rather than a
+ // plain Math.Ceiling, so binary-fraction noise does not add a needless millisecond.
+ var maxDurationMilliseconds = Math.Floor(numberOfCharacters / maxCharactersPerSecond * 1000.0);
+ if (numberOfCharacters / (maxDurationMilliseconds / 1000.0) > maxCharactersPerSecond)
+ {
+ maxDurationMilliseconds += 1;
+ }
+
p.EndTime.TotalMilliseconds = p.StartTime.TotalMilliseconds + maxDurationMilliseconds;
}
diff --git a/src/libse/Common/SubtitlePositionToAssa.cs b/src/libse/Common/SubtitlePositionToAssa.cs
index efded0c47ba..f7460e1ff02 100644
--- a/src/libse/Common/SubtitlePositionToAssa.cs
+++ b/src/libse/Common/SubtitlePositionToAssa.cs
@@ -1,4 +1,4 @@
-using Nikse.SubtitleEdit.Core.SubtitleFormats;
+using Nikse.SubtitleEdit.Core.SubtitleFormats;
using System;
using System.Collections.Generic;
using System.Globalization;
@@ -69,6 +69,11 @@ public static bool ApplyPositions(Subtitle subtitle, string sourceHeader, int pl
return ApplyEbuTeletextRows(subtitle, sourceHeader, playResY, usePositions);
}
+ if (DvbTeletext.IsDvbTeletextHeader(sourceHeader))
+ {
+ return ApplyDvbTeletextRows(subtitle, playResY, usePositions);
+ }
+
var regions = GetTtmlRegions(sourceHeader);
if (regions != null)
{
@@ -121,28 +126,59 @@ private static bool IsEbuStlHeader(string header)
///
private static bool ApplyEbuTeletextRows(Subtitle subtitle, string header, int playResY, bool usePositions)
{
- var rows = GetEbuRowCount(header);
- var newLineRows = Math.Max(1, Configuration.Settings.SubtitleSettings.EbuStlNewLineRows);
+ return ApplyTeletextRows(
+ subtitle,
+ GetEbuRowCount(header),
+ Math.Max(1, Configuration.Settings.SubtitleSettings.EbuStlNewLineRows),
+ Configuration.Settings.SubtitleSettings.EbuStlMarginBottom,
+ GetEbuJustificationOffset(),
+ playResY,
+ usePositions);
+ }
+
+ ///
+ /// DVB teletext places subtitles the same way, but with the writer's fixed geometry
+ /// instead of the EBU save options: 23 rows, the default bottom row 22 (a double height
+ /// row covers 22 and 23), two rows per extra line, and the alignment coming off the
+ /// {\an} tags alone - see ManzanitaTeletextWriter.GetRowNumbers.
+ ///
+ private static bool ApplyDvbTeletextRows(Subtitle subtitle, int playResY, bool usePositions)
+ {
+ return ApplyTeletextRows(subtitle, rows: 23, newLineRows: 2, marginBottom: 1,
+ justificationOffset: 0, playResY, usePositions);
+ }
+
+ private static bool ApplyTeletextRows(Subtitle subtitle, int rows, int newLineRows, int marginBottom, int justificationOffset, int playResY, bool usePositions)
+ {
var applied = false;
foreach (var p in subtitle.Paragraphs)
{
- if (string.IsNullOrEmpty(p.MarginV))
+ var hasRow = int.TryParse(p.MarginV, NumberStyles.Integer, CultureInfo.InvariantCulture, out var row) &&
+ row >= 1 &&
+ row <= rows;
+
+ // A row number is not a pixel margin - leaving one behind would nudge the line by
+ // a near random amount.
+ p.MarginV = null;
+
+ if (!usePositions)
{
continue;
}
- if (!usePositions ||
- !int.TryParse(p.MarginV, NumberStyles.Integer, CultureInfo.InvariantCulture, out var row) ||
- row < 1 || row > rows)
+ var lineCount = Math.Max(1, Utilities.GetNumberOfLines(p.Text));
+ if (!hasRow)
{
- // A row number is not a pixel margin - leaving it would nudge every line
- // by a near random amount.
- p.MarginV = null;
- continue;
+ // Only a subtitle that was read from an STL file carries teletext rows. Anything
+ // typed in or converted from another format has none, and used to be skipped
+ // outright - so for the far more common case the justification and the vertical
+ // margins of the EBU options dialog changed nothing on screen at all. Put the
+ // line where Ebu.Save would put it: counted up from the bottom margin.
+ row = Math.Max(1, rows - marginBottom - (lineCount - 1) * newLineRows);
}
- var lastRow = row + (Math.Max(1, Utilities.GetNumberOfLines(p.Text)) - 1) * newLineRows;
+ var lastRow = row + (lineCount - 1) * newLineRows;
var alignment = GetLeadingAlignment(p.Text);
if (alignment == 0)
{
@@ -159,6 +195,11 @@ private static bool ApplyEbuTeletextRows(Subtitle subtitle, string header, int p
alignment = 5;
}
+ // The row says how far down the line sits, the justification says which way it
+ // is aligned - without this every line stays centered no matter what the EBU
+ // options dialog is set to.
+ alignment += justificationOffset;
+
p.Text = "{\\an" + alignment.ToString(CultureInfo.InvariantCulture) + "}" + p.Text;
}
@@ -181,6 +222,28 @@ private static bool ApplyEbuTeletextRows(Subtitle subtitle, string header, int p
return applied;
}
+ ///
+ /// Turns the EBU justification code into a shift of the ASSA alignment column: the codes
+ /// are 0 = unchanged, 1 = left, 2 = centered and 3 = right, and each centered alignment
+ /// (2, 5 and 8) has its left neighbour one below it and its right one above.
+ ///
+ ///
+ /// Read from the EBU STL settings, next to the margins and the teletext flags this method's
+ /// caller uses. It used to come off
+ private static int GetEbuJustificationOffset()
+ {
+ switch (Configuration.Settings.SubtitleSettings.EbuStlJustificationCode)
+ {
+ case 1: return -1;
+ case 3: return 1;
+ default: return 0;
+ }
+ }
+
private static int GetEbuRowCount(string header)
{
try
diff --git a/src/libse/Common/TextEffect/KaraokeWordTransform.cs b/src/libse/Common/TextEffect/KaraokeWordTransform.cs
index 08281744b65..6e2d2de8b65 100644
--- a/src/libse/Common/TextEffect/KaraokeWordTransform.cs
+++ b/src/libse/Common/TextEffect/KaraokeWordTransform.cs
@@ -26,8 +26,9 @@ public override string[] Transform(string text)
var karaoke = text.Substring(0, i) + fontClose + text.Substring(i);
lastFontCloseIdx = i;
result.Add(karaoke);
- // skip only whitespace
- while (i < len && char.IsWhiteSpace(text[i]))
+ // skip only whitespace - land on the last whitespace char so the loop's own
+ // i++ moves onto the next word (or tag) instead of past it
+ while (i + 1 < len && char.IsWhiteSpace(text[i + 1]))
{
i++;
}
diff --git a/src/libse/Common/TextSplit.cs b/src/libse/Common/TextSplit.cs
index 085cd038a21..8fe8bdd733c 100644
--- a/src/libse/Common/TextSplit.cs
+++ b/src/libse/Common/TextSplit.cs
@@ -195,7 +195,7 @@ public static List SplitText(string text, int numberOfParts)
for (var i = 0; i < numberOfParts - 1; i++)
{
- var endIndex = FindNearestSpaceIndex(text, startIndex + partLength);
+ var endIndex = FindNearestSpaceIndex(text, startIndex + partLength, startIndex);
if (endIndex == -1)
{
@@ -227,10 +227,62 @@ public static List SplitText(string text, int numberOfParts)
return parts;
}
- static int FindNearestSpaceIndex(string text, int startIndex)
+ ///
+ /// A space to break this chunk on: the last one at or before
+ static int FindNearestSpaceIndex(string text, int targetIndex, int minIndex)
{
- var spaceIndex = text.LastIndexOf(' ', Math.Min(startIndex, text.Length - 1));
- return spaceIndex;
+ var clamped = Math.Min(targetIndex, text.Length - 1);
+ var spaceIndex = text.LastIndexOf(' ', clamped);
+ if (spaceIndex > minIndex)
+ {
+ return spaceIndex;
+ }
+
+ // No space inside this chunk - take the next one after it rather than break early.
+ return text.IndexOf(' ', clamped);
+ }
+
+ ///
+ /// A period between two digits ("04.00 uur", "1.500 euro") is a decimal/time separator, not a
+ /// sentence ending - treating it as one broke a subtitle mid-number (issue #14230).
+ ///
+ private static bool IsSentenceEndingAt(string text, int index)
+ {
+ return text[index] != '.' ||
+ index <= 0 ||
+ index >= text.Length - 1 ||
+ !char.IsDigit(text[index - 1]) ||
+ !char.IsDigit(text[index + 1]);
+ }
+
+ private static int LastSentenceEndingIndex(string text)
+ {
+ for (var i = text.Length - 1; i >= 0; i--)
+ {
+ if (Array.IndexOf(PeriodQuestionExclamation, text[i]) >= 0 && IsSentenceEndingAt(text, i))
+ {
+ return i;
+ }
+ }
+
+ return -1;
+ }
+
+ private static int FirstSentenceEndingIndex(string text)
+ {
+ for (var i = 0; i < text.Length; i++)
+ {
+ if (Array.IndexOf(PeriodQuestionExclamation, text[i]) >= 0 && IsSentenceEndingAt(text, i))
+ {
+ return i;
+ }
+ }
+
+ return -1;
}
public static Subtitle TryForWholeSentences(Subtitle inputSubtitle, string language, int lineMaxLength)
@@ -246,7 +298,10 @@ public static Subtitle TryForWholeSentences(Subtitle inputSubtitle, string langu
if (p == null ||
next == null ||
- p.EndTime.TotalMilliseconds - next.StartTime.TotalMilliseconds > 100 ||
+ // The gap is next.Start - p.End; written the other way round it is positive
+ // only when the two overlap, so the "don't move a word across a long pause"
+ // guard never fired and a word could jump a minute of silence.
+ next.StartTime.TotalMilliseconds - p.EndTime.TotalMilliseconds > 100 ||
p.Text.Contains('<') ||
next.Text.Contains('<') ||
!(p.Text.Contains('.') || p.Text.Contains('?') || p.Text.Contains('!') || next.Text.Contains('.') || next.Text.Contains('?') || next.Text.Contains('!')) ||
@@ -263,7 +318,7 @@ public static Subtitle TryForWholeSentences(Subtitle inputSubtitle, string langu
}
// check for period in last part of current
- var lastPeriodIdx = p.Text.LastIndexOfAny(PeriodQuestionExclamation);
+ var lastPeriodIdx = LastSentenceEndingIndex(p.Text);
if (lastPeriodIdx > 3 && lastPeriodIdx > p.Text.Length - maxMoveChunkSize)
{
var newCurrentText = p.Text.Substring(0, lastPeriodIdx + 1).Trim();
@@ -304,7 +359,7 @@ public static Subtitle TryForWholeSentences(Subtitle inputSubtitle, string langu
}
// check for period in beginning of next
- var firstPeriodIdx = next.Text.IndexOfAny(PeriodQuestionExclamation);
+ var firstPeriodIdx = FirstSentenceEndingIndex(next.Text);
if (firstPeriodIdx >= 3 && firstPeriodIdx < maxMoveChunkSize)
{
var newCurrentText = next.Text.Substring(0, firstPeriodIdx + 1).Trim();
diff --git a/src/libse/Common/TextSplitResult.cs b/src/libse/Common/TextSplitResult.cs
index f54839ea100..af6aae899a1 100644
--- a/src/libse/Common/TextSplitResult.cs
+++ b/src/libse/Common/TextSplitResult.cs
@@ -184,6 +184,14 @@ public double DiffFromAverage()
public double DiffFromAveragePixel()
{
var pixels = LengthPixels;
+ // With "auto break by pixel width" off no pixel lengths are measured, so this used
+ // to score every candidate 0 and the "best" selectors just kept the first one.
+ // Fall back to the character-based diff so the most balanced split still wins.
+ if (pixels.Count == 0)
+ {
+ return DiffFromAverage();
+ }
+
var avg = (double)(SumPixels(pixels) - SpaceLengthPixels) / Lines.Count;
double diff = 0;
for (var i = 0; i < pixels.Count; i++)
diff --git a/src/libse/Common/TimeCodesBeautifier.cs b/src/libse/Common/TimeCodesBeautifier.cs
deleted file mode 100644
index 932e237730a..00000000000
--- a/src/libse/Common/TimeCodesBeautifier.cs
+++ /dev/null
@@ -1,138 +0,0 @@
-using System;
-using System.Collections.Generic;
-using System.Linq;
-
-namespace Nikse.SubtitleEdit.Core.Common
-{
- public class TimeCodesBeautifier
- {
- private readonly List _paragraphs;
- private readonly double _frameRate;
- private readonly List _shotChanges;
- private readonly BeautifySettings _settings;
-
- public TimeCodesBeautifier(List paragraphs, double frameRate, List shotChanges, BeautifySettings settings)
- {
- _paragraphs = paragraphs;
- _frameRate = frameRate;
- _shotChanges = shotChanges ?? new List();
- _settings = settings;
- }
-
- public List Beautify()
- {
- var result = new List();
-
- for (int i = 0; i < _paragraphs.Count; i++)
- {
- var p = new Paragraph(_paragraphs[i]);
-
- // Frame alignment
- if (_settings.SnapToFrames)
- {
- p.StartTime = AlignToFrame(p.StartTime, _frameRate);
- p.EndTime = AlignToFrame(p.EndTime, _frameRate);
- }
-
- // Shot change snapping
- if (_shotChanges.Count > 0)
- {
- p.StartTime = SnapStartToShotChange(p.StartTime, _settings.ShotChangeThresholdMs, _settings.ShotChangeOffsetFrames, _frameRate);
- p.EndTime = SnapEndToShotChange(p.EndTime, _settings.ShotChangeThresholdMs, _settings.ShotChangeOffsetFrames, _frameRate);
- }
-
- // Gap management with previous paragraph
- if (i > 0)
- {
- var previousEnd = result[i - 1].EndTime.TotalMilliseconds;
- var frameGapMs = FramesToMilliseconds(_settings.FrameGap, _frameRate);
- var minStart = previousEnd + frameGapMs;
-
- if (p.StartTime.TotalMilliseconds < minStart)
- {
- p.StartTime = new TimeCode(minStart);
- }
- }
-
- // Validation: Ensure End > Start
- if (p.EndTime.TotalMilliseconds <= p.StartTime.TotalMilliseconds)
- {
- p.EndTime = new TimeCode(p.StartTime.TotalMilliseconds + _settings.MinDurationMs);
- }
-
- // Validation: Ensure minimum duration
- var duration = p.EndTime.TotalMilliseconds - p.StartTime.TotalMilliseconds;
- if (duration < _settings.MinDurationMs)
- {
- p.EndTime = new TimeCode(p.StartTime.TotalMilliseconds + _settings.MinDurationMs);
- }
-
- result.Add(p);
- }
-
- return result;
- }
-
- private TimeCode AlignToFrame(TimeCode timeCode, double fps)
- {
- var totalMilliseconds = timeCode.TotalMilliseconds;
- var frameNumber = Math.Round(totalMilliseconds * fps / 1000.0);
- var alignedMilliseconds = frameNumber * 1000.0 / fps;
- return new TimeCode(alignedMilliseconds);
- }
-
- private TimeCode SnapStartToShotChange(TimeCode timeCode, int thresholdMs, int offsetFrames, double fps)
- {
- var timeSeconds = timeCode.TotalMilliseconds / 1000.0;
- var thresholdSeconds = thresholdMs / 1000.0;
-
- foreach (var shotChange in _shotChanges)
- {
- var diff = Math.Abs(timeSeconds - shotChange);
- if (diff < thresholdSeconds)
- {
- // Snap to offsetFrames AFTER the shot change
- var offsetSeconds = offsetFrames / fps;
- var snappedSeconds = shotChange + offsetSeconds;
- return new TimeCode(snappedSeconds * 1000.0);
- }
- }
-
- return timeCode;
- }
-
- private TimeCode SnapEndToShotChange(TimeCode timeCode, int thresholdMs, int offsetFrames, double fps)
- {
- var timeSeconds = timeCode.TotalMilliseconds / 1000.0;
- var thresholdSeconds = thresholdMs / 1000.0;
-
- foreach (var shotChange in _shotChanges)
- {
- var diff = Math.Abs(timeSeconds - shotChange);
- if (diff < thresholdSeconds)
- {
- // Snap to offsetFrames BEFORE the shot change
- var offsetSeconds = offsetFrames / fps;
- var snappedSeconds = shotChange - offsetSeconds;
- return new TimeCode(Math.Max(0, snappedSeconds * 1000.0));
- }
- }
-
- return timeCode;
- }
-
- private double FramesToMilliseconds(int frames, double fps)
- {
- return frames * 1000.0 / fps;
- }
- }
-
- public class BeautifySettings
- {
- public bool SnapToFrames { get; set; } = true;
- public int FrameGap { get; set; } = 2;
- public int ShotChangeThresholdMs { get; set; } = 250;
- public int ShotChangeOffsetFrames { get; set; } = 2;
- public double MinDurationMs { get; set; } = 800;
- }
-}
diff --git a/src/libse/Common/UUEncoding.cs b/src/libse/Common/UUEncoding.cs
index cf7b3071e01..f8a79eb796c 100644
--- a/src/libse/Common/UUEncoding.cs
+++ b/src/libse/Common/UUEncoding.cs
@@ -28,8 +28,12 @@ public static string UUEncode(byte[] bytes)
{
if (i + 3 > length)
{
+ // Copy the remaining 1 or 2 bytes and zero-pad the rest. The count used to be
+ // "i + 2 - length", which is the remainder only by accident when one byte is
+ // left and zero when two are - silently dropping the last two bytes of every
+ // attachment whose length % 3 == 2.
src = new byte[3];
- Array.Copy(bytes, i, src, 0, i + 2 - length);
+ Array.Copy(bytes, i, src, 0, length - i);
}
else
{
@@ -46,10 +50,20 @@ public static string UUEncode(byte[] bytes)
dest[2] += 33;
dest[3] += 33;
- sb.Append(Encoding.ASCII.GetString(dest));
+ // The final group emits only as many characters as it carries data for -
+ // min(remaining + 1, 4), per the ASS/Aegisub convention that UUDecode below
+ // already implements. Always writing 4 made every embedded font/attachment
+ // 1-2 bytes longer than the original, decoding back with padding zeros.
+ var charsToWrite = 4;
+ if (i + 3 > length)
+ {
+ charsToWrite = Math.Min(length - i + 1, 4);
+ }
+
+ sb.Append(Encoding.ASCII.GetString(dest, 0, charsToWrite));
- lineElements += 4;
- if (lineElements == 80)
+ lineElements += charsToWrite;
+ if (lineElements >= 80)
{
sb.AppendLine();
lineElements = 0;
@@ -78,7 +92,11 @@ public static byte[] UUDecode(string text)
for (var i = 0; i < 4 && pos < len; ++pos)
{
var c = text[pos];
- if (c != '\n' && c != '\r')
+ // Encoded data is always ASCII (a 6-bit value plus 33, so 33..96). Anything
+ // above U+00FF cannot be part of it, and Convert.ToByte threw OverflowException
+ // on it - a single stray non-ASCII character in an [Fonts]/[Graphics] block
+ // took down the attachments dialog. Skip it like a line break.
+ if (c != '\n' && c != '\r' && c <= byte.MaxValue)
{
src[i++] = (byte)(Convert.ToByte(c) - 33);
bytes++;
diff --git a/src/libse/Common/UnknownFormatImporter.cs b/src/libse/Common/UnknownFormatImporter.cs
index a3e560bd11c..fe52a6e21d5 100644
--- a/src/libse/Common/UnknownFormatImporter.cs
+++ b/src/libse/Common/UnknownFormatImporter.cs
@@ -55,6 +55,16 @@ public Subtitle AutoGuessImport(List lines, string fileName)
}
}
+ var firstNonEmptyLine = lines.FirstOrDefault(l => !string.IsNullOrWhiteSpace(l));
+ if (firstNonEmptyLine != null && firstNonEmptyLine.TrimStart('', ' ', '\t').StartsWith('<'))
+ {
+ var xmlSubtitle = new UnknownFormatImporterXml().AutoGuessImport(lines);
+ if (xmlSubtitle.Paragraphs.Count >= 2)
+ {
+ return xmlSubtitle;
+ }
+ }
+
var subtitle = ImportTimeCodesOnSameSeparateLine(lines);
if (subtitle.Paragraphs.Count < 2)
{
diff --git a/src/libse/Common/UnknownFormatImporterCsv.cs b/src/libse/Common/UnknownFormatImporterCsv.cs
index e386577e5b5..19bd41ab4ce 100644
--- a/src/libse/Common/UnknownFormatImporterCsv.cs
+++ b/src/libse/Common/UnknownFormatImporterCsv.cs
@@ -31,6 +31,10 @@ public Subtitle AutoGuessImport(List lines)
if (string.IsNullOrWhiteSpace(lines[0]) || lines[0].Trim() == "Dialogue List,,")
{
+ // On a copy: the caller (UnknownFormatImporter) keeps using the list it passed in
+ // for the other parsers when this one finds nothing, and dropping a line from
+ // under it lost a subtitle.
+ lines = new List(lines);
lines.RemoveAt(0);
}
@@ -41,7 +45,13 @@ public Subtitle AutoGuessImport(List lines)
var separator = DetectSeparator(lines);
- var headers = lines[0].RemoveChar('"').ToLowerInvariant().Split(separator).ToList();
+ // Split the header exactly like the data rows (quote-aware) and trim each cell:
+ // a naive Split broke on a quoted separator inside a header name - skewing every
+ // column index - and an unpadded "Start, End, Text" failed to match at all.
+ var headers = CsvUtil.CsvSplitLines(new List { lines[0] }, separator)
+ .FirstOrDefault()?
+ .Select(p => p.Trim().Trim('"').ToLowerInvariant())
+ .ToList() ?? new List();
if (!HasValidHeader(headers))
{
return new Subtitle();
@@ -127,6 +137,14 @@ private Subtitle MakeSubtitle(List csvLines)
private void FixStartTimeWithoutMs(List csvLines, bool isStartTimeFrames)
{
+ if (csvLines.Count == 0)
+ {
+ // No data rows at all - e.g. an unterminated quote in the header line makes the
+ // header swallow the rest of the file. The averages below divide by this count,
+ // and the file-open path has no catch for the exception (DivideByZeroException).
+ return;
+ }
+
var isEndTimeNull = csvLines.All(p => string.IsNullOrEmpty(p.End));
if (isStartTimeFrames || !isEndTimeNull)
{
@@ -147,7 +165,10 @@ private void FixStartTimeWithoutMs(List csvLines, bool isStartTimeFrame
var startGaps = GetGaps(csvLines.Select(p => p.Start).ToArray());
var avarageStartGap = startGaps.Sum() / startGaps.Length;
- var avarageTextLength = csvLines.Select(p => p.Text.Length).Sum() / csvLines.Count;
+ // Text stays null when no text column was recognised (HasValidHeader only needs two
+ // hits across all five name lists, so a text column is not required) - dereferencing
+ // it threw straight out of the file-open path, which has no catch.
+ var avarageTextLength = csvLines.Select(p => p.Text?.Length ?? 0).Sum() / csvLines.Count;
if (avarageStartGap >= 500 || avarageTextLength <= 90)
{
return;
@@ -179,13 +200,19 @@ private double[] GetGaps(string[] strings)
private bool DetectIsFrames(List toList)
{
+ // Blank values carry no information either way - a single empty cell, or the blank row
+ // a trailing newline produces, used to veto the whole column and silently reinterpret
+ // HH:MM:SS:FF as HH:MM:SS.ms (12 frames became 12 ms). Decide on the values that are
+ // actually present. Same rule as the import window's own detection.
+ var considered = 0;
foreach (var s in toList)
{
- if (s == null)
+ if (string.IsNullOrWhiteSpace(s))
{
- return false;
+ continue;
}
+ considered++;
var parts = s.Split(TimeCode.TimeSplitChars, StringSplitOptions.RemoveEmptyEntries);
if (parts.Length != 4 || parts[3].Trim().Length != 2)
{
@@ -193,7 +220,7 @@ private bool DetectIsFrames(List toList)
}
}
- return true;
+ return considered > 0;
}
private List ReadCsvLines(List headers, List lines, char separator)
diff --git a/src/libse/Common/UnknownFormatImporterJson.cs b/src/libse/Common/UnknownFormatImporterJson.cs
index ebe2e9fd267..cbe319f0dd5 100644
--- a/src/libse/Common/UnknownFormatImporterJson.cs
+++ b/src/libse/Common/UnknownFormatImporterJson.cs
@@ -1,4 +1,4 @@
-using Nikse.SubtitleEdit.Core.SubtitleFormats;
+using Nikse.SubtitleEdit.Core.SubtitleFormats;
using System;
using System.Collections.Generic;
using System.Globalization;
@@ -9,6 +9,56 @@ namespace Nikse.SubtitleEdit.Core.Common
{
public class UnknownFormatImporterJson
{
+ private static readonly string[] StartTags =
+ {
+ "start", "in", "begin",
+ "startTime", "start_time", "starttime",
+ "startMillis", "start_Millis", "startmillis",
+ "startMs", "start_ms", "startms",
+ "startMilliseconds", "start_millisesonds", "startmilliseconds",
+ "from", "fromTime", "from_ms","fromms", "fromMilliseconds", "from_milliseconds", "show",
+ "tStartMs", "displayTimeOffset", "milliseconds", "timestamp", "timestamp_begin", "time",
+ "t1", "st", "s"
+ };
+
+ private static readonly string[] EndTags =
+ {
+ "end", "out", "stop",
+ "endTime", "end_time", "endtime",
+ "endMillis", "end_Millis", "endmillis",
+ "endMs", "end_ms", "endms",
+ "endMilliseconds", "end_millisesonds", "endmilliseconds",
+ "to", "toTime", "to_ms", "toms", "toMilliseconds", "to_milliseconds", "hide",
+ "tEndMs", "timestamp_end",
+ "t2", "et", "e"
+ };
+
+ private static readonly string[] DurationTags =
+ {
+ "duration",
+ "durationMs",
+ "dDurationMs",
+ "dur",
+ "d",
+ };
+
+ private static readonly string[] TextTags =
+ {
+ "text", "content", "value", "caption", "sentence", "dialog", "dialogue",
+ "line", "utf8", "tt", "t", "n"
+ };
+
+ private class Fields
+ {
+ public string Start { get; set; }
+ public string End { get; set; }
+ public string Duration { get; set; }
+ public string Text { get; set; }
+ public string Source { get; set; }
+
+ public bool HasTimes => Start != null && (End != null || Duration != null);
+ }
+
public Subtitle AutoGuessImport(List lines)
{
var sb = new StringBuilder();
@@ -23,15 +73,18 @@ public Subtitle AutoGuessImport(List lines)
return new Subtitle();
}
- var subtitle1 = new Subtitle();
+ var subtitle1 = ImportFromSegments(allText.Split('{', '}', '[', ']'));
+ var subtitle2 = ImportFromSegments(allText.Split('{', '}'));
+
+ var subtitle3 = new Subtitle();
try
{
var count = 0;
- foreach (var line in allText.Split('{', '}', '[', ']'))
+ foreach (var line in Json.ReadObjectArray(allText))
{
count++;
- ReadParagraph(line, subtitle1);
- if (count > 20 && subtitle1.Paragraphs.Count == 0)
+ ReadParagraph(line, subtitle3);
+ if (count > 20 && subtitle3.Paragraphs.Count == 0)
{
break;
}
@@ -42,34 +95,72 @@ public Subtitle AutoGuessImport(List lines)
// ignored
}
- var subtitle2 = new Subtitle();
- try
+ if (subtitle1.Paragraphs.Count >= subtitle2.Paragraphs.Count && subtitle1.Paragraphs.Count >= subtitle3.Paragraphs.Count)
{
- var count = 0;
- foreach (var line in allText.Split('{', '}'))
- {
- count++;
- ReadParagraph(line, subtitle2);
- if (count > 20 && subtitle2.Paragraphs.Count == 0)
- {
- break;
- }
- }
+ subtitle1.Renumber();
+ return FixTimeCodeMsOrSeconds(FixMissingEndTimes(subtitle1));
}
- catch
+
+ if (subtitle2.Paragraphs.Count >= subtitle1.Paragraphs.Count && subtitle2.Paragraphs.Count >= subtitle3.Paragraphs.Count)
{
- // ignored
+ subtitle2.Renumber();
+ return FixTimeCodeMsOrSeconds(FixMissingEndTimes(subtitle2));
}
- var subtitle3 = new Subtitle();
+ subtitle3.Renumber();
+ return FixTimeCodeMsOrSeconds(FixMissingEndTimes(subtitle3));
+ }
+
+ private static Subtitle ImportFromSegments(string[] segments)
+ {
+ var subtitle = new Subtitle();
try
{
var count = 0;
- foreach (var line in Json.ReadObjectArray(allText))
+ Fields pendingTimes = null;
+ var pendingAge = 0;
+ foreach (var segment in segments)
{
count++;
- ReadParagraph(line, subtitle3);
- if (count > 20 && subtitle3.Paragraphs.Count == 0)
+ var fields = ReadFields(segment);
+ if (fields == null)
+ {
+ if (count > 20 && subtitle.Paragraphs.Count == 0)
+ {
+ break;
+ }
+
+ continue;
+ }
+
+ if (fields.Start != null && fields.Text != null)
+ {
+ AddParagraph(subtitle, fields, fields.Text);
+ pendingTimes = null;
+ }
+ else if (fields.HasTimes && fields.Text == null)
+ {
+ // nested object layout: the time codes and the text live in different
+ // segments after splitting (e.g. {"start":..,"end":..,"metadata":{"text":..}})
+ pendingTimes = fields;
+ pendingAge = 0;
+ }
+ else if (fields.Text != null && fields.Start == null && pendingTimes != null)
+ {
+ AddParagraph(subtitle, pendingTimes, fields.Text);
+ pendingTimes = null;
+ }
+
+ if (pendingTimes != null)
+ {
+ pendingAge++;
+ if (pendingAge > 3)
+ {
+ pendingTimes = null;
+ }
+ }
+
+ if (count > 20 && subtitle.Paragraphs.Count == 0)
{
break;
}
@@ -80,54 +171,62 @@ public Subtitle AutoGuessImport(List lines)
// ignored
}
- if (subtitle1.Paragraphs.Count >= subtitle2.Paragraphs.Count && subtitle1.Paragraphs.Count >= subtitle3.Paragraphs.Count)
+ return subtitle;
+ }
+
+ private static Subtitle FixMissingEndTimes(Subtitle subtitle)
+ {
+ if (subtitle.Paragraphs.Count < 2 ||
+ !subtitle.Paragraphs.All(p => Math.Abs(p.EndTime.TotalMilliseconds) < 0.001))
{
- subtitle1.Renumber();
- return FixTimeCodeMsOrSeconds(subtitle1);
+ return subtitle;
}
- if (subtitle2.Paragraphs.Count >= subtitle1.Paragraphs.Count && subtitle2.Paragraphs.Count >= subtitle3.Paragraphs.Count)
+ // start-time-only format (e.g. {"milliseconds":1500,"line":"..."})
+ for (var index = 0; index < subtitle.Paragraphs.Count; index++)
{
- subtitle2.Renumber();
- return FixTimeCodeMsOrSeconds(subtitle2);
+ var paragraph = subtitle.Paragraphs[index];
+ var next = subtitle.GetParagraphOrDefault(index + 1);
+ if (next != null)
+ {
+ paragraph.EndTime.TotalMilliseconds = next.StartTime.TotalMilliseconds - Configuration.Settings.General.MinimumMillisecondsBetweenLines;
+ }
+ else
+ {
+ paragraph.EndTime.TotalMilliseconds = paragraph.StartTime.TotalMilliseconds + Utilities.GetOptimalDisplayMilliseconds(paragraph.Text);
+ }
}
- subtitle3.Renumber();
- return FixTimeCodeMsOrSeconds(subtitle3);
+ return subtitle;
}
private static Subtitle FixTimeCodeMsOrSeconds(Subtitle subtitle)
{
- if (subtitle == null || subtitle.Paragraphs.Count < 5)
+ if (subtitle == null || subtitle.Paragraphs.Count < 2)
{
return subtitle;
}
+ var msKeys = new[]
+ {
+ "start", "startMs", "start_ms", "startMillis", "start_millis",
+ "startMilliseconds", "start_millisecondsMs", "tStartMs", "milliseconds",
+ "fromMs", "from_ms", "fromms", "fromMillis", "fromMilliseconds", "from_milliseconds", "show"
+ };
+
double totalDuration = 0;
var msFound = 0;
foreach (var p in subtitle.Paragraphs)
{
totalDuration += p.DurationTotalMilliseconds;
- if (p.Style.Contains("\"start\"") ||
- p.Style.Contains("\"startMs\"") ||
- p.Style.Contains("\"start_ms\"") ||
- p.Style.Contains("\"startMillis\"") ||
- p.Style.Contains("\"start_millis\"") ||
- p.Style.Contains("\"startMilliseconds\"") ||
- p.Style.Contains("\"start_millisecondsMs\"") ||
- p.Style.Contains("\"fromMs\"") ||
- p.Style.Contains("\"from_ms\"") ||
- p.Style.Contains("\"fromms\"") ||
- p.Style.Contains("\"fromMillis\"") ||
- p.Style.Contains("\"fromMilliseconds\"") ||
- p.Style.Contains("\"from_milliseconds\"") ||
- p.Style.Contains("\"show\""))
+ if (msKeys.Any(key => ReadKeyValue(p.Style, key, out _) != null))
{
msFound++;
}
}
- if (totalDuration / subtitle.Paragraphs.Count > 1000000 || msFound == subtitle.Paragraphs.Count)
+ var averageDuration = totalDuration / subtitle.Paragraphs.Count;
+ if (averageDuration > 1000000 || msFound == subtitle.Paragraphs.Count && averageDuration > 100000)
{
// Time codes were read as seconds, but they are actually milliseconds,
// so all time codes are divided by 1000.
@@ -143,63 +242,95 @@ private static Subtitle FixTimeCodeMsOrSeconds(Subtitle subtitle)
private static void ReadParagraph(string s, Subtitle subtitle)
{
- s = s.Trim();
- if (s.Length < 7)
+ var fields = ReadFields(s);
+ if (fields?.Start == null || fields.Text == null)
{
return;
}
+ AddParagraph(subtitle, fields, fields.Text);
+ }
+
+ private static Fields ReadFields(string s)
+ {
+ s = s.Trim();
+ if (s.Length < 5)
+ {
+ return null;
+ }
+
if (!s.EndsWith('}'))
{
s += '}';
}
- var start = ReadStartTag(s);
- var end = ReadEndTag(s);
- var duration = ReadDurationTag(s);
+ var start = NormalizeTime(ReadFirstTimeTag(s, StartTags));
+ var end = NormalizeTime(ReadFirstTimeTag(s, EndTags));
+ var duration = ReadFirstTimeTag(s, DurationTags);
var text = ReadTextTag(s);
- var originalStart = start;
+ if (start == null && end == null && duration == null && text == null)
+ {
+ return null;
+ }
+
+ return new Fields { Start = start, End = end, Duration = duration, Text = text, Source = s };
+ }
- if (start != null && start.Contains(":") && start.Length >= 11 && start.Length <= 12 && start.Split(new[] { ':', ',', '.' }, StringSplitOptions.RemoveEmptyEntries).Length == 4)
+ private static void AddParagraph(Subtitle subtitle, Fields times, string text)
+ {
+ var start = times.Start?.TrimEnd('s');
+ var end = times.End?.TrimEnd('s');
+ var duration = times.Duration?.TrimEnd('s');
+
+ if (start == null)
{
- start = DecodeFormatToSeconds(start);
+ return;
}
- else if (start != null && start.Contains(":") && start.Length == 8 && start.Split(new[] { ':', ',', '.' }, StringSplitOptions.RemoveEmptyEntries).Length == 3)
+
+ if (!double.TryParse(start, NumberStyles.AllowDecimalPoint, CultureInfo.InvariantCulture, out var startSeconds))
{
- start = DecodeFormatHourMinuteSecondsToSeconds(start);
+ return;
}
- if (end != null && end.Contains(":") && end.Length >= 11 && end.Length <= 12 && end.Split(new[] { ':', ',', '.' }, StringSplitOptions.RemoveEmptyEntries).Length == 4)
+ if (end != null &&
+ double.TryParse(end, NumberStyles.AllowDecimalPoint, CultureInfo.InvariantCulture, out var endSeconds))
{
- end = DecodeFormatToSeconds(end);
+ subtitle.Paragraphs.Add(new Paragraph(Json.DecodeJsonText(text), startSeconds * TimeCode.BaseUnit, endSeconds * TimeCode.BaseUnit) { Extra = times.Start, Style = times.Source });
+ return;
}
- else if (end != null && end.Contains(":") && end.Length == 8 && end.Split(new[] { ':', ',', '.' }, StringSplitOptions.RemoveEmptyEntries).Length == 3)
+
+ if (duration != null &&
+ double.TryParse(duration, NumberStyles.AllowDecimalPoint, CultureInfo.InvariantCulture, out var durationSeconds))
{
- end = DecodeFormatHourMinuteSecondsToSeconds(end);
+ subtitle.Paragraphs.Add(new Paragraph(Json.DecodeJsonText(text), startSeconds * TimeCode.BaseUnit, (startSeconds + durationSeconds) * TimeCode.BaseUnit) { Extra = times.Start, Style = times.Source });
+ return;
}
- if (start != null && end != null && text != null)
+ if (end == null && duration == null)
{
- start = start.TrimEnd('s');
- end = end.TrimEnd('s');
- if (double.TryParse(start, NumberStyles.AllowDecimalPoint, CultureInfo.InvariantCulture, out var startSeconds) &&
- double.TryParse(end, NumberStyles.AllowDecimalPoint, CultureInfo.InvariantCulture, out var endSeconds))
- {
- var p = new Paragraph(Json.DecodeJsonText(text), startSeconds * TimeCode.BaseUnit, endSeconds * TimeCode.BaseUnit) { Extra = originalStart, Style = s };
- subtitle.Paragraphs.Add(p);
- }
+ // start-time-only - end times are computed at the end of the import
+ subtitle.Paragraphs.Add(new Paragraph(Json.DecodeJsonText(text), startSeconds * TimeCode.BaseUnit, 0) { Extra = times.Start, Style = times.Source });
}
- else if (start != null && duration != null && text != null)
+ }
+
+ private static string NormalizeTime(string value)
+ {
+ if (value == null)
{
- start = start.TrimEnd('s');
- duration = duration.TrimEnd('s');
- if (double.TryParse(start, NumberStyles.AllowDecimalPoint, CultureInfo.InvariantCulture, out double startSeconds) &&
- double.TryParse(duration, NumberStyles.AllowDecimalPoint, CultureInfo.InvariantCulture, out double durationSeconds))
- {
- var p = new Paragraph(Json.DecodeJsonText(text), startSeconds * TimeCode.BaseUnit, (startSeconds + durationSeconds) * TimeCode.BaseUnit) { Extra = originalStart, Style = s };
- subtitle.Paragraphs.Add(p);
- }
+ return null;
}
+
+ if (value.Contains(":") && value.Length >= 11 && value.Length <= 12 && value.Split(new[] { ':', ',', '.' }, StringSplitOptions.RemoveEmptyEntries).Length == 4)
+ {
+ return DecodeFormatToSeconds(value);
+ }
+
+ if (value.Contains(":") && value.Length == 8 && value.Split(new[] { ':', ',', '.' }, StringSplitOptions.RemoveEmptyEntries).Length == 3)
+ {
+ return DecodeFormatHourMinuteSecondsToSeconds(value);
+ }
+
+ return value;
}
private static string DecodeFormatToSeconds(string s)
@@ -213,140 +344,245 @@ private static string DecodeFormatHourMinuteSecondsToSeconds(string s)
return DecodeFormatToSeconds(s + ":00");
}
- private static string ReadStartTag(string s)
+ private static string ReadFirstTimeTag(string s, string[] tags)
{
- return ReadFirstMultiTag(s, new[]
- {
- "start", "in", "begin",
- "startTime", "start_time", "starttime",
- "startMillis", "start_Millis", "startmillis",
- "startMs", "start_ms", "startms",
- "startMilliseconds", "start_Millisesonds", "startmilliseconds",
- "from", "fromTime", "from_ms","fromms", "fromMilliseconds", "from_milliseconds", "show"
- });
- }
-
- private static string ReadEndTag(string s)
- {
- return ReadFirstMultiTag(s, new[]
- {
- "end", "out", "stop",
- "endTime", "end_time", "endtime",
- "endMillis", "end_Millis", "endmillis",
- "endMs", "end_ms", "endms",
- "endMilliseconds", "end_Millisesonds", "endmilliseconds",
- "to", "toTime", "to_ms", "toms", "toMilliseconds", "to_milliseconds", "hide"
- });
- }
-
- private static string ReadDurationTag(string s)
- {
- return ReadFirstMultiTag(s, new[]
+ foreach (var tag in tags)
{
- "duration",
- "durationMs",
- "dur",
- });
+ var res = ReadKeyValue(s, tag, out var isStringValue);
+ if (!string.IsNullOrEmpty(res) && IsTimeLikeValue(res))
+ {
+ return res;
+ }
+ }
+
+ return null;
}
- private static string ReadTextTag(string s)
+ private static bool IsTimeLikeValue(string value)
{
- var idx = s.IndexOf("\"text", StringComparison.OrdinalIgnoreCase);
- if (idx < 0)
+ var v = value.Trim().TrimEnd('s');
+ if (v.Length == 0 || v.Length > 20)
{
- idx = s.IndexOf("\"content", StringComparison.OrdinalIgnoreCase);
+ return false;
}
- if (idx < 0)
- {
- idx = s.IndexOf("\"value", StringComparison.OrdinalIgnoreCase);
- }
- if (idx < 0)
- {
- idx = s.IndexOf("\"caption", StringComparison.OrdinalIgnoreCase);
- }
- if (idx < 0)
+
+ var hasDigit = false;
+ foreach (var ch in v)
{
- idx = s.IndexOf("\"sentence", StringComparison.OrdinalIgnoreCase);
+ if (char.IsDigit(ch))
+ {
+ hasDigit = true;
+ }
+ else if (ch != ':' && ch != '.' && ch != ',' && ch != ';' && ch != '-')
+ {
+ return false;
+ }
}
- if (idx < 0)
+ return hasDigit;
+ }
+
+ ///
+ /// Reads a key whose value is an array of strings; null when the key is missing,
+ /// the value is not an array, or the array holds anything but strings.
+ ///
+ private static List ReadStringArrayValue(string s, string tag)
+ {
+ var from = 0;
+ while (from < s.Length)
{
+ var idx = s.IndexOf("\"" + tag + "\"", from, StringComparison.OrdinalIgnoreCase);
+ if (idx < 0)
+ {
+ return null;
+ }
+
+ var i = idx + tag.Length + 2;
+ while (i < s.Length && char.IsWhiteSpace(s[i]))
+ {
+ i++;
+ }
+
+ if (i >= s.Length || s[i] != ':')
+ {
+ from = idx + 1;
+ continue;
+ }
+
+ i++;
+ while (i < s.Length && char.IsWhiteSpace(s[i]))
+ {
+ i++;
+ }
+
+ if (i >= s.Length || s[i] != '[')
+ {
+ return null;
+ }
+
+ i++;
+ var list = new List();
+ while (i < s.Length)
+ {
+ while (i < s.Length && (char.IsWhiteSpace(s[i]) || s[i] == ','))
+ {
+ i++;
+ }
+
+ if (i >= s.Length)
+ {
+ return null; // unterminated
+ }
+
+ if (s[i] == ']')
+ {
+ return list;
+ }
+
+ if (s[i] != '"')
+ {
+ return null; // not an array of strings
+ }
+
+ i++;
+ var sb = new StringBuilder();
+ while (i < s.Length && s[i] != '"')
+ {
+ if (s[i] == '\\' && i + 1 < s.Length)
+ {
+ sb.Append(s[i]);
+ sb.Append(s[i + 1]);
+ i += 2;
+ continue;
+ }
+
+ sb.Append(s[i]);
+ i++;
+ }
+
+ if (i >= s.Length)
+ {
+ return null;
+ }
+
+ i++; // closing quote
+ list.Add(sb.ToString());
+ }
+
return null;
}
- s = s.Substring(idx);
- idx = s.IndexOf(']');
- if (idx > 0)
- {
- s = s.Substring(0, idx + 1);
- }
+ return null;
+ }
- var text = Json.ReadTag(s, "text");
- if (text == null)
- {
- text = Json.ReadTag(s, "content");
- }
- if (text == null)
- {
- text = Json.ReadTag(s, "value");
- }
- if (text == null)
- {
- text = Json.ReadTag(s, "caption");
- }
- if (text == null)
- {
- text = Json.ReadTag(s, "sentence");
- }
- if (text == null)
+ ///
+ /// Reads the value of a json key. Unlike a plain IndexOf, the tag must actually be in
+ /// key position (followed by a colon) - otherwise string VALUES that happen to match a
+ /// tag name hijack the lookup (e.g. "lineAlign":"start" matching the tag "start").
+ ///
+ private static string ReadKeyValue(string s, string tag, out bool isStringValue)
+ {
+ isStringValue = false;
+ var from = 0;
+ while (from < s.Length)
{
- text = Json.ReadTag(s, "dialog");
- }
+ var idx = s.IndexOf("\"" + tag + "\"", from, StringComparison.OrdinalIgnoreCase);
+ if (idx < 0)
+ {
+ return null;
+ }
- if (text != null)
- {
- text = Json.DecodeJsonText(text);
- }
+ var i = idx + tag.Length + 2;
+ while (i < s.Length && char.IsWhiteSpace(s[i]))
+ {
+ i++;
+ }
- var textLines = Json.ReadArray(s, "text");
- if (textLines == null || textLines.Count == 0)
- {
- textLines = Json.ReadArray(s, "content");
- }
+ if (i >= s.Length || s[i] != ':')
+ {
+ from = idx + 1;
+ continue;
+ }
- var isArray = s.Contains("[");
- if (isArray && textLines.Any(p => p == "end_time" || p == "endTime" || p == "end" || p == "endMs" || p == "endMilliseconds" || p == "end_ms" || p == "endms" || p == "to" || p == "to_ms" || p == "toms" || p == "from" || p == "from_ms" | p == "hide"))
- {
- isArray = false;
- }
+ i++;
+ while (i < s.Length && char.IsWhiteSpace(s[i]))
+ {
+ i++;
+ }
- if (!isArray && !string.IsNullOrEmpty(text))
- {
- return text
- .Replace("'", "'")
- .Replace("
", Environment.NewLine)
- .Replace("
", Environment.NewLine)
- ;
- }
+ if (i >= s.Length)
+ {
+ return null;
+ }
- if (textLines != null && textLines.Count > 0)
- {
- return string.Join(Environment.NewLine, textLines);
+ if (s[i] == '"')
+ {
+ isStringValue = true;
+ var sb = new StringBuilder();
+ i++;
+ while (i < s.Length)
+ {
+ var ch = s[i];
+ if (ch == '\\' && i + 1 < s.Length)
+ {
+ sb.Append(ch);
+ sb.Append(s[i + 1]);
+ i += 2;
+ continue;
+ }
+
+ if (ch == '"')
+ {
+ break;
+ }
+
+ sb.Append(ch);
+ i++;
+ }
+
+ return sb.ToString();
+ }
+
+ var end = i;
+ while (end < s.Length && s[end] != ',' && s[end] != '}' && s[end] != ']')
+ {
+ end++;
+ }
+
+ return s.Substring(i, end - i).Trim();
}
- return ReadFirstMultiTag(s, new[] { "text", "content" });
+ return null;
}
- private static string ReadFirstMultiTag(string s, string[] tags)
+ private static string ReadTextTag(string s)
{
- foreach (var tag in tags)
+ // text as an array of lines, e.g. "text": ["line 1", "line 2"]
+ foreach (var arrayTag in new[] { "text", "content", "lines" })
{
- var res = Json.ReadTag(s, tag);
- if (!string.IsNullOrEmpty(res))
+ var textLines = ReadStringArrayValue(s, arrayTag);
+ if (textLines != null && textLines.Count > 0)
{
- return res;
+ var joined = string.Join(Environment.NewLine, textLines.Where(p => !string.IsNullOrEmpty(p)));
+ if (!string.IsNullOrWhiteSpace(joined))
+ {
+ return joined;
+ }
}
}
+
+ foreach (var tag in TextTags)
+ {
+ var res = ReadKeyValue(s, tag, out var isStringValue);
+ if (isStringValue && !string.IsNullOrEmpty(res))
+ {
+ return res
+ .Replace("'", "'")
+ .Replace("
", Environment.NewLine)
+ .Replace("
", Environment.NewLine);
+ }
+ }
+
return null;
}
}
diff --git a/src/libse/Common/UnknownFormatImporterOds.cs b/src/libse/Common/UnknownFormatImporterOds.cs
index eb1b78ee899..ef1836e0568 100644
--- a/src/libse/Common/UnknownFormatImporterOds.cs
+++ b/src/libse/Common/UnknownFormatImporterOds.cs
@@ -70,22 +70,24 @@ private static List ReadTable(XElement table)
foreach (var row in table.Elements(TableNs + "table-row"))
{
var cells = new List();
- foreach (var cell in row.Elements(TableNs + "table-cell"))
+
+ // covered-table-cell too, not just table-cell: a merged cell is written as one
+ // table-cell plus one covered-table-cell per column it swallows, and skipping the
+ // placeholders pulled every later column one to the left - so the text column of
+ // a row with a merge was read as the end time.
+ foreach (var cell in row.Elements().Where(e => e.Name == TableNs + "table-cell" || e.Name == TableNs + "covered-table-cell"))
{
- var value = GetCellValue(cell);
+ var value = cell.Name == TableNs + "covered-table-cell" ? string.Empty : GetCellValue(cell);
var repeat = ParseRepeat(cell.Attribute(TableNs + "number-columns-repeated")?.Value);
- if (string.IsNullOrEmpty(value))
- {
- // Trailing empties with huge repeat counts are common — only emit a single empty cell
- cells.Add(string.Empty);
- }
- else
+ // Honour the repeat count for empty cells too: LibreOffice writes an
+ // *interior* run of blanks that way, and collapsing it to one cell shifted
+ // every following column left, so a row with a gap lost its text column.
+ // Trailing blank runs are still handled by the trim below, and ParseRepeat
+ // already caps the count, so this cannot blow up on a "blank to column 1024" row.
+ for (var i = 0; i < repeat; i++)
{
- for (var i = 0; i < repeat; i++)
- {
- cells.Add(value);
- }
+ cells.Add(value ?? string.Empty);
}
}
diff --git a/src/libse/Common/UnknownFormatImporterXlsx.cs b/src/libse/Common/UnknownFormatImporterXlsx.cs
index 2df7242209d..ce4e373862a 100644
--- a/src/libse/Common/UnknownFormatImporterXlsx.cs
+++ b/src/libse/Common/UnknownFormatImporterXlsx.cs
@@ -11,6 +11,8 @@ namespace Nikse.SubtitleEdit.Core.Common
public class UnknownFormatImporterXlsx
{
private static readonly XNamespace Ns = "http://schemas.openxmlformats.org/spreadsheetml/2006/main";
+ private static readonly XNamespace RelationshipsNs = "http://schemas.openxmlformats.org/officeDocument/2006/relationships";
+ private static readonly XNamespace PackageRelationshipsNs = "http://schemas.openxmlformats.org/package/2006/relationships";
public Subtitle AutoGuessImport(string fileName)
{
@@ -36,11 +38,7 @@ public List ReadLinesFromFile(string fileName)
using (var archive = new ZipArchive(stream, ZipArchiveMode.Read))
{
var sharedStrings = ReadSharedStrings(archive);
- var sheetEntry = archive.Entries
- .Where(e => e.FullName.StartsWith("xl/worksheets/sheet", StringComparison.OrdinalIgnoreCase) &&
- e.FullName.EndsWith(".xml", StringComparison.OrdinalIgnoreCase))
- .OrderBy(e => e.FullName, StringComparer.OrdinalIgnoreCase)
- .FirstOrDefault();
+ var sheetEntry = FindFirstSheet(archive);
if (sheetEntry == null)
{
return null;
@@ -55,6 +53,70 @@ public List ReadLinesFromFile(string fileName)
}
}
+ ///
+ /// The worksheet the user sees first: the first <sheet> in xl/workbook.xml, resolved
+ /// through the workbook relationships. That is not necessarily sheet1.xml - moving a sheet
+ /// in Excel reorders the workbook entries but never renames the parts - so picking the
+ /// first worksheet file imported the wrong sheet. Falls back to the first worksheet part
+ /// when the workbook or its relationships cannot be read.
+ ///
+ private static ZipArchiveEntry FindFirstSheet(ZipArchive archive)
+ {
+ var worksheets = archive.Entries
+ .Where(e => e.FullName.StartsWith("xl/worksheets/sheet", StringComparison.OrdinalIgnoreCase) &&
+ e.FullName.EndsWith(".xml", StringComparison.OrdinalIgnoreCase))
+ .OrderBy(e => e.FullName, StringComparer.OrdinalIgnoreCase)
+ .ToList();
+ if (worksheets.Count <= 1)
+ {
+ return worksheets.FirstOrDefault();
+ }
+
+ try
+ {
+ var workbookEntry = archive.GetEntry("xl/workbook.xml");
+ var relsEntry = archive.GetEntry("xl/_rels/workbook.xml.rels");
+ if (workbookEntry != null && relsEntry != null)
+ {
+ string relationshipId;
+ using (var stream = workbookEntry.Open())
+ {
+ relationshipId = (string)XDocument.Load(stream).Root?
+ .Element(Ns + "sheets")?
+ .Elements(Ns + "sheet")
+ .FirstOrDefault()?
+ .Attribute(RelationshipsNs + "id");
+ }
+
+ if (!string.IsNullOrEmpty(relationshipId))
+ {
+ using (var stream = relsEntry.Open())
+ {
+ var target = XDocument.Load(stream).Root?
+ .Elements(PackageRelationshipsNs + "Relationship")
+ .FirstOrDefault(r => (string)r.Attribute("Id") == relationshipId)?
+ .Attribute("Target")?.Value;
+ if (!string.IsNullOrEmpty(target))
+ {
+ var fullName = "xl/" + target.TrimStart('/').Replace("../", string.Empty);
+ var match = worksheets.FirstOrDefault(e => e.FullName.Equals(fullName, StringComparison.OrdinalIgnoreCase));
+ if (match != null)
+ {
+ return match;
+ }
+ }
+ }
+ }
+ }
+ }
+ catch
+ {
+ // fall through to the first worksheet part
+ }
+
+ return worksheets[0];
+ }
+
private static List ReadSharedStrings(ZipArchive archive)
{
var result = new List();
diff --git a/src/libse/Common/UnknownFormatImporterXml.cs b/src/libse/Common/UnknownFormatImporterXml.cs
new file mode 100644
index 00000000000..3969e53cf6e
--- /dev/null
+++ b/src/libse/Common/UnknownFormatImporterXml.cs
@@ -0,0 +1,594 @@
+using Nikse.SubtitleEdit.Core.SubtitleFormats;
+using System;
+using System.Collections.Generic;
+using System.Globalization;
+using System.Linq;
+using System.Text;
+using System.Xml;
+
+namespace Nikse.SubtitleEdit.Core.Common
+{
+ ///
+ /// Generic importer for unknown XML subtitle dialects: finds the repeated element that
+ /// carries time codes (as attributes or child elements) plus text, without knowing the
+ /// schema. Handles TTML-like ("begin"/"end" attributes with inline text), wrapper
+ /// elements with a text child (D-Cinema like), child-element time codes (Matroska
+ /// chapters like), time codes on an empty child element, text in an attribute, and
+ /// start-only samples where an empty element closes the previous one (GPAC TTXT like).
+ ///
+ public class UnknownFormatImporterXml
+ {
+ private class Candidate
+ {
+ public double StartMs { get; set; }
+ public double? EndMs { get; set; }
+ public double? DurationMs { get; set; }
+ public string Text { get; set; }
+ }
+
+ private bool _use250TickTimeCodes;
+
+ public Subtitle AutoGuessImport(List lines)
+ {
+ var allText = string.Join(Environment.NewLine, lines).Trim().TrimStart('').TrimStart();
+ if (!allText.StartsWith('<') || !allText.EndsWith('>'))
+ {
+ return new Subtitle();
+ }
+
+ var doc = new XmlDocument { XmlResolver = null };
+ try
+ {
+ var settings = new XmlReaderSettings { DtdProcessing = DtdProcessing.Ignore, XmlResolver = null };
+ using (var stringReader = new System.IO.StringReader(allText))
+ using (var reader = XmlReader.Create(stringReader, settings))
+ {
+ doc.Load(reader);
+ }
+ }
+ catch
+ {
+ return new Subtitle();
+ }
+
+ if (doc.DocumentElement == null)
+ {
+ return new Subtitle();
+ }
+
+ // D-Cinema interop writes the last time code part in 250ths of a second ("editable units")
+ _use250TickTimeCodes = doc.DocumentElement.Name == "DCSubtitle";
+
+ // group candidates per element name - the subtitle element is the name that
+ // repeats with time codes, everything else (styles, regions, metadata) won't
+ var candidatesByName = new Dictionary>();
+ Walk(doc.DocumentElement, candidatesByName);
+
+ List best = null;
+ var bestScore = 0;
+ var bestTextCount = 0;
+ foreach (var kvp in candidatesByName)
+ {
+ var textCount = kvp.Value.Count(c => !string.IsNullOrWhiteSpace(c.Text));
+ // a group that also knows when subtitles end beats one with bare start times
+ // (e.g. a layout/config element with an index that parses as a time)
+ var score = textCount * 2 + kvp.Value.Count(c => c.EndMs.HasValue || c.DurationMs.HasValue);
+ if (score > bestScore)
+ {
+ bestScore = score;
+ bestTextCount = textCount;
+ best = kvp.Value;
+ }
+ }
+
+ if (best == null || bestTextCount < 2)
+ {
+ return new Subtitle();
+ }
+
+ return MakeSubtitle(best);
+ }
+
+ private static Subtitle MakeSubtitle(List candidates)
+ {
+ var subtitle = new Subtitle();
+ Paragraph last = null;
+ foreach (var c in candidates)
+ {
+ var end = c.EndMs ?? (c.DurationMs.HasValue ? c.StartMs + c.DurationMs.Value : 0);
+ if (string.IsNullOrWhiteSpace(c.Text))
+ {
+ // start-only sample with no text: it closes the previous paragraph
+ if (last != null && !c.EndMs.HasValue && !c.DurationMs.HasValue &&
+ Math.Abs(last.EndTime.TotalMilliseconds) < 0.01 && c.StartMs > last.StartTime.TotalMilliseconds)
+ {
+ last.EndTime.TotalMilliseconds = c.StartMs;
+ }
+
+ continue;
+ }
+
+ var p = new Paragraph(c.Text.Trim(), c.StartMs, end);
+ subtitle.Paragraphs.Add(p);
+ last = p;
+ }
+
+ // fill in missing end times
+ for (var i = 0; i < subtitle.Paragraphs.Count; i++)
+ {
+ var p = subtitle.Paragraphs[i];
+ if (Math.Abs(p.EndTime.TotalMilliseconds) > 0.01)
+ {
+ continue;
+ }
+
+ var optimalDurationMs = Utilities.GetOptimalDisplayMilliseconds(p.Text);
+ var next = subtitle.GetParagraphOrDefault(i + 1);
+ if (next != null && next.StartTime.TotalMilliseconds < p.StartTime.TotalMilliseconds + optimalDurationMs + 2000)
+ {
+ p.EndTime.TotalMilliseconds = next.StartTime.TotalMilliseconds - Configuration.Settings.General.MinimumMillisecondsBetweenLines;
+ }
+ else
+ {
+ p.EndTime.TotalMilliseconds = p.StartTime.TotalMilliseconds + optimalDurationMs;
+ }
+ }
+
+ subtitle.Renumber();
+ return subtitle;
+ }
+
+ private void Walk(XmlElement element, Dictionary> candidatesByName)
+ {
+ var candidate = TryGetCandidate(element);
+ if (candidate != null)
+ {
+ if (!candidatesByName.TryGetValue(element.LocalName, out var list))
+ {
+ list = new List();
+ candidatesByName.Add(element.LocalName, list);
+ }
+
+ list.Add(candidate);
+ }
+
+ foreach (XmlNode child in element.ChildNodes)
+ {
+ if (child is XmlElement childElement)
+ {
+ Walk(childElement, candidatesByName);
+ }
+ }
+ }
+
+ private Candidate TryGetCandidate(XmlElement element)
+ {
+ double? start = null;
+ double? end = null;
+ double? duration = null;
+ var timeChildNames = new List();
+
+ foreach (XmlAttribute attribute in element.Attributes)
+ {
+ ReadTimeField(attribute.LocalName, attribute.Value, ref start, ref end, ref duration);
+ }
+
+ foreach (XmlNode child in element.ChildNodes)
+ {
+ if (!(child is XmlElement el))
+ {
+ continue;
+ }
+
+ var before = (start, end, duration);
+
+ // simple child element whose value is a time code, e.g. 00:00:01.500
+ if (!el.HasChildNodes || el.ChildNodes.Count == 1 && el.FirstChild is XmlText)
+ {
+ ReadTimeField(el.LocalName, el.InnerText, ref start, ref end, ref duration);
+ }
+
+ // empty child element carrying the time codes as attributes, e.g.
+ if (el.Attributes.Count > 0 && !el.HasChildNodes)
+ {
+ foreach (XmlAttribute attribute in el.Attributes)
+ {
+ ReadTimeField(attribute.LocalName, attribute.Value, ref start, ref end, ref duration);
+ }
+ }
+
+ if (before != (start, end, duration))
+ {
+ timeChildNames.Add(el.LocalName);
+ }
+ }
+
+ if (!start.HasValue)
+ {
+ return null;
+ }
+
+ var text = GetText(element, timeChildNames);
+ if (string.IsNullOrWhiteSpace(text) && !end.HasValue && !duration.HasValue)
+ {
+ // keep as an end marker only when it's a completely empty element
+ // (GPAC TTXT like) - a start-time-only element with non-time children
+ // is a container we shouldn't consume
+ if (element.ChildNodes.Count > 0)
+ {
+ return null;
+ }
+ }
+
+ return new Candidate { StartMs = start.Value, EndMs = end, DurationMs = duration, Text = text };
+ }
+
+ private void ReadTimeField(string name, string value, ref double? start, ref double? end, ref double? duration)
+ {
+ var kind = ClassifyTimeName(name);
+ if (kind == 0)
+ {
+ return;
+ }
+
+ var ms = TryParseTime(value, _use250TickTimeCodes);
+ if (!ms.HasValue)
+ {
+ return;
+ }
+
+ if (kind == 1 && !start.HasValue)
+ {
+ start = ms;
+ }
+ else if (kind == 2 && !end.HasValue)
+ {
+ end = ms;
+ }
+ else if (kind == 3 && !duration.HasValue)
+ {
+ duration = ms;
+ }
+ }
+
+ /// 0 = not a time name, 1 = start, 2 = end, 3 = duration
+ private static int ClassifyTimeName(string rawName)
+ {
+ var name = Normalize(rawName);
+ if (name.Length == 0)
+ {
+ return 0;
+ }
+
+ if (name == "d" || name.Contains("dur", StringComparison.Ordinal))
+ {
+ return 3;
+ }
+
+ if (name == "out" ||
+ name == "to" ||
+ name == "hide" ||
+ name == "et" ||
+ name == "secout" ||
+ name == "clear" ||
+ name.Contains("end", StringComparison.Ordinal) ||
+ name.Contains("stop", StringComparison.Ordinal) ||
+ name.Contains("timeout", StringComparison.Ordinal) ||
+ name.Contains("tcout", StringComparison.Ordinal))
+ {
+ return 2;
+ }
+
+ if (name == "in" ||
+ name == "from" ||
+ name == "show" ||
+ name == "time" ||
+ name == "timestamp" ||
+ name == "offset" ||
+ name == "t" ||
+ name == "st" ||
+ name == "secin" ||
+ name == "display" ||
+ name == "position" ||
+ name == "value" ||
+ name.Contains("start", StringComparison.Ordinal) ||
+ name.Contains("begin", StringComparison.Ordinal) ||
+ name.Contains("timein", StringComparison.Ordinal) ||
+ name.Contains("tcin", StringComparison.Ordinal) ||
+ name.Contains("sampletime", StringComparison.Ordinal))
+ {
+ return 1;
+ }
+
+ return 0;
+ }
+
+ private static string Normalize(string name)
+ {
+ var sb = new StringBuilder(name.Length);
+ foreach (var ch in name)
+ {
+ if (char.IsLetterOrDigit(ch))
+ {
+ sb.Append(char.ToLowerInvariant(ch));
+ }
+ }
+
+ return sb.ToString();
+ }
+
+ internal static double? TryParseTime(string input, bool use250TickTimeCodes = false)
+ {
+ if (string.IsNullOrWhiteSpace(input))
+ {
+ return null;
+ }
+
+ var s = input.Trim();
+ if (s.Length > 30 || s.Contains('%'))
+ {
+ return null;
+ }
+
+ // rational seconds, e.g. "3003/24000s" (Final Cut like)
+ if (s.EndsWith('s') && s.Contains('/'))
+ {
+ var arr = s.TrimEnd('s').Split('/');
+ if (arr.Length == 2 &&
+ double.TryParse(arr[0], NumberStyles.AllowDecimalPoint, CultureInfo.InvariantCulture, out var numerator) &&
+ double.TryParse(arr[1], NumberStyles.AllowDecimalPoint, CultureInfo.InvariantCulture, out var denominator) &&
+ denominator > 0)
+ {
+ return numerator / denominator * TimeCode.BaseUnit;
+ }
+
+ return null;
+ }
+
+ // TTML ticks, e.g. "15000000t" (default tick rate is 10,000,000/s)
+ if (s.EndsWith('t') && long.TryParse(s.TrimEnd('t'), NumberStyles.None, CultureInfo.InvariantCulture, out var ticks))
+ {
+ return ticks / 10000.0;
+ }
+
+ // "150ms"
+ if (s.EndsWith("ms", StringComparison.Ordinal) &&
+ double.TryParse(s.Remove(s.Length - 2), NumberStyles.AllowDecimalPoint, CultureInfo.InvariantCulture, out var milliseconds))
+ {
+ return milliseconds;
+ }
+
+ // seconds with unit, e.g. "1.5s"
+ if (s.EndsWith('s') &&
+ double.TryParse(s.TrimEnd('s'), NumberStyles.AllowDecimalPoint, CultureInfo.InvariantCulture, out var seconds))
+ {
+ return seconds * TimeCode.BaseUnit;
+ }
+
+ if (s.Contains(':'))
+ {
+ var parts = s.Split(':', '.', ',', ';');
+ if (parts.Length < 2 || parts.Length > 4 || parts.Any(p => p.Length == 0 || p.Length > 9 || !p.All(char.IsDigit)))
+ {
+ return null;
+ }
+
+ if (parts.Length == 2)
+ {
+ // MM:SS
+ var m = int.Parse(parts[0]);
+ var sec = int.Parse(parts[1]);
+ if (sec >= 60)
+ {
+ return null;
+ }
+
+ return (m * 60 + sec) * TimeCode.BaseUnit;
+ }
+
+ int hours;
+ int minutes;
+ int secs;
+ string last;
+ if (parts.Length == 3)
+ {
+ if (s.Contains('.') || s.Contains(','))
+ {
+ // MM:SS.mmm
+ hours = 0;
+ minutes = int.Parse(parts[0]);
+ secs = int.Parse(parts[1]);
+ last = parts[2];
+ }
+ else
+ {
+ // HH:MM:SS
+ hours = int.Parse(parts[0]);
+ minutes = int.Parse(parts[1]);
+ secs = int.Parse(parts[2]);
+ if (minutes >= 60 || secs >= 60 || hours > 99)
+ {
+ return null;
+ }
+
+ return new TimeCode(hours, minutes, secs, 0).TotalMilliseconds;
+ }
+ }
+ else
+ {
+ hours = int.Parse(parts[0]);
+ minutes = int.Parse(parts[1]);
+ secs = int.Parse(parts[2]);
+ last = parts[3];
+ }
+
+ if (minutes >= 60 || secs >= 60 || hours > 99)
+ {
+ return null;
+ }
+
+ double lastMs;
+ if (last.Length <= 3 && use250TickTimeCodes)
+ {
+ // D-Cinema interop: last part is in 4 ms units
+ lastMs = int.Parse(last) * 4;
+ }
+ else if (last.Length == 2 && !s.Contains('.') && !s.Contains(','))
+ {
+ // HH:MM:SS:FF
+ lastMs = SubtitleFormat.FramesToMillisecondsMax999(int.Parse(last));
+ }
+ else
+ {
+ // fraction of a second - "5", "50", "500", "500000000" all mean the same lead digits
+ lastMs = double.Parse("0." + last, CultureInfo.InvariantCulture) * TimeCode.BaseUnit;
+ }
+
+ return new TimeCode(hours, minutes, secs, 0).TotalMilliseconds + lastMs;
+ }
+
+ // plain number: integers are milliseconds, decimals are seconds
+ if (double.TryParse(s, NumberStyles.AllowDecimalPoint, CultureInfo.InvariantCulture, out var number))
+ {
+ if (s.Contains('.'))
+ {
+ return number * TimeCode.BaseUnit;
+ }
+
+ return number;
+ }
+
+ return null;
+ }
+
+ private static readonly string[] TextElementNames = { "text", "string", "content", "caption", "dialog", "dialogue", "sentence", "line", "sub" };
+ private static readonly string[] WeakTextElementNames = { "comment", "name", "label" };
+
+ private static string GetText(XmlElement element, List timeChildNames)
+ {
+ // inline text (TTML like) - text nodes plus inline markup like
and
+ var sb = new StringBuilder();
+ AppendInlineText(element, sb);
+ var text = sb.ToString().Trim();
+ if (!string.IsNullOrWhiteSpace(text))
+ {
+ return text;
+ }
+
+ // named text descendants (D-Cinema "Text", Matroska "ChapterString", ...) -
+ // multi-line subtitles repeat the element, so join all of them
+ var textElements = new List();
+ FindTextElements(element, timeChildNames, TextElementNames, textElements);
+ if (textElements.Count == 0)
+ {
+ FindTextElements(element, timeChildNames, WeakTextElementNames, textElements);
+ }
+
+ if (textElements.Count > 0)
+ {
+ sb.Clear();
+ foreach (var textElement in textElements)
+ {
+ var lineSb = new StringBuilder();
+ AppendInlineText(textElement, lineSb);
+ var line = lineSb.ToString().Trim();
+ if (string.IsNullOrWhiteSpace(line))
+ {
+ // text wrapped in unknown markup (styling runs etc.)
+ line = textElement.InnerText.Trim();
+ }
+
+ if (line.Length > 0)
+ {
+ if (sb.Length > 0)
+ {
+ sb.AppendLine();
+ }
+
+ sb.Append(line);
+ }
+ }
+
+ text = sb.ToString().Trim();
+ if (!string.IsNullOrWhiteSpace(text))
+ {
+ return text;
+ }
+ }
+
+ // text in an attribute, e.g. timeChildNames, string[] textNames, List result)
+ {
+ foreach (XmlNode child in element.ChildNodes)
+ {
+ if (!(child is XmlElement childElement) || timeChildNames.Contains(childElement.LocalName))
+ {
+ continue;
+ }
+
+ var name = Normalize(childElement.LocalName);
+ var isTextElement = false;
+ foreach (var textName in textNames)
+ {
+ if (name.Contains(textName, StringComparison.Ordinal))
+ {
+ isTextElement = true;
+ break;
+ }
+ }
+
+ if (isTextElement)
+ {
+ result.Add(childElement);
+ }
+ else
+ {
+ FindTextElements(childElement, timeChildNames, textNames, result);
+ }
+ }
+ }
+ }
+}
diff --git a/src/libse/Common/Utilities.cs b/src/libse/Common/Utilities.cs
index 94090fa9cf4..f4730679725 100644
--- a/src/libse/Common/Utilities.cs
+++ b/src/libse/Common/Utilities.cs
@@ -176,6 +176,14 @@ public static SubtitleFormat GetSubtitleFormatByFriendlyName(string friendlyName
}
}
+ foreach (var format in SubtitleFormat.AllSubtitleFormats)
+ {
+ if (format.AlternateNames.Contains(friendlyName))
+ {
+ return format;
+ }
+ }
+
return null;
}
@@ -1567,6 +1575,9 @@ public static int CountTagInText(string text, string tag)
public static int CountTagInText(string text, char tag)
{
+#if NET8_0_OR_GREATER
+ return text.AsSpan().Count(tag);
+#else
int count = 0;
int index = text.IndexOf(tag);
while (index >= 0)
@@ -1580,6 +1591,7 @@ public static int CountTagInText(string text, char tag)
index = text.IndexOf(tag, index + 1);
}
return count;
+#endif
}
///
@@ -2244,9 +2256,12 @@ public static SKColor GetColorFromFontString(string text, SKColor defaultColor)
}
// f starts at 'start' within s, so offset f-relative index back into s
var colorStart = start + f.IndexOf(" color=", StringComparison.OrdinalIgnoreCase);
- if (s.IndexOf('"', colorStart + " color=".Length + 1) > 0)
+ var quoteIndex = s.IndexOf('"', colorStart + " color=".Length + 1);
+ if (quoteIndex > 0 && quoteIndex < end)
{
- end = s.IndexOf('"', colorStart + " color=".Length + 1);
+ // only a quote inside the font tag itself ends the value - a quotation mark
+ // in the dialogue text must not hijack an unquoted color attribute
+ end = quoteIndex;
}
s = s.Substring(colorStart, end - colorStart);
s = s.Replace(" color=", string.Empty);
@@ -3094,7 +3109,7 @@ public static string RemoveSpaceBeforeAfterTag(string input, string openTag)
// e.g.: \r\n
" + Encode(summary) + "
"); + sb.AppendLine(" "); + sb.AppendLine("