feat(pdf): return positioned image files with Markdown

Cargo.toml

@@ -28,7 +28,7 @@ csv = "1.4.0"
 flate2 = "1"
 encoding_rs = "0.8.35"
 log = "0.4"
-pdf-inspector = "0.1.8"
+pdf-inspector = "1.14.1"
 quick-xml = "0.41.0"
 zip = { version = "8.6.0", default-features = false, features = ["deflate"] }
 

src/formats/pdf.rs

@@ -5,14 +5,18 @@
 //! image-only PDFs need OCR, which is out of scope here; they error as
 //! unsupported. Pages flagged for OCR in an otherwise text-based document
 //! degrade with a log, consistent with the crate-wide recovery policy.
+//! Embedded images are retained as positioned Markdown placeholders.
 //!
 //! [pdf-inspector]: https://github.com/firecrawl/pdf-inspector
 
 use crate::error::ConvertError;
-use pdf_inspector::PdfError;
+use pdf_inspector::{MarkdownOptions, PdfError, PdfOptions};
 
 pub fn to_markdown(bytes: &[u8]) -> Result<String, ConvertError> {
-    let result = pdf_inspector::process_pdf_mem(bytes).map_err(map_error)?;
+    let markdown = MarkdownOptions { include_images: true, ..MarkdownOptions::default() };
+    let result =
+        pdf_inspector::process_pdf_mem_with_options(bytes, PdfOptions::new().markdown(markdown))
+            .map_err(map_error)?;
     if !result.pages_needing_ocr.is_empty() {
         log::warn!(
             "{} of {} pages need OCR and were not extracted",

tests/snapshots/snapshots__pdf__text.pdf.snap

@@ -30,6 +30,8 @@ External link to example. Relative link to <u>a sibling file</u>. This plain par
 
 ## Objects
 
+![Image: Im107](image)
+
 Inline image: done.
 
 ### Inside the text box.