edgeparse 0.3.0__tar.gz → 0.3.2__tar.gz

{edgeparse-0.3.0 → edgeparse-0.3.2}/Cargo.lock

@@ -184,7 +184,7 @@ dependencies = [
  "num-traits",
  "pastey",
  "rayon",
- "thiserror",
+ "thiserror 2.0.21",
  "v_frame",
  "y4m",
 ]
@@ -582,7 +582,7 @@ version = "1.0.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "10d60334b3b2e7c9d91ef8150abfb6fa4c1c39ebbcf4a81c2e346aad939fee3e"
 dependencies = [
- "thiserror",
+ "thiserror 2.0.21",
 ]
 
 [[package]]
@@ -655,7 +655,7 @@ dependencies = [
 
 [[package]]
 name = "edgeparse-cli"
-version = "0.3.0"
+version = "0.3.2"
 dependencies = [
  "anyhow",
  "clap",
@@ -669,13 +669,15 @@ dependencies = [
 
 [[package]]
 name = "edgeparse-core"
-version = "0.3.0"
+version = "0.3.2"
 dependencies = [
  "anyhow",
  "base64 0.22.1",
  "euclid",
+ "hayro-jbig2",
  "image",
  "indexmap",
+ "jpeg2k",
  "js-sys",
  "log",
  "ocrs",
@@ -688,7 +690,7 @@ dependencies = [
  "rten-imageproc",
  "serde",
  "serde_json",
- "thiserror",
+ "thiserror 2.0.21",
  "tokio",
  "ttf-parser",
  "unicode-normalization",
@@ -698,7 +700,7 @@ dependencies = [
 
 [[package]]
 name = "edgeparse-node"
-version = "0.3.0"
+version = "0.3.2"
 dependencies = [
  "edgeparse-core",
  "napi",
@@ -710,7 +712,7 @@ dependencies = [
 
 [[package]]
 name = "edgeparse-python"
-version = "0.3.0"
+version = "0.3.2"
 dependencies = [
  "edgeparse-core",
  "pyo3",
@@ -719,7 +721,7 @@ dependencies = [
 
 [[package]]
 name = "edgeparse-wasm"
-version = "0.3.0"
+version = "0.3.2"
 dependencies = [
  "console_error_panic_hook",
  "console_log",
@@ -1030,6 +1032,21 @@ version = "0.17.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "ed5909b6e89a2db4456e54cd5f673791d7eca6732202bbf2a9cc504fe2f9b84a"
 
+[[package]]
+name = "hayro-ccitt"
+version = "0.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9f4d0e94ddd48749f06bbe4e5389fb9799a0c45bcaf00495042076ef05e3241a"
+
+[[package]]
+name = "hayro-jbig2"
+version = "0.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "69374b3668dd45aeb3d3145cda68f2c7b4f223aaa2511e67d076f1c7d741388d"
+dependencies = [
+ "hayro-ccitt",
+]
+
 [[package]]
 name = "heck"
 version = "0.5.0"
@@ -1442,6 +1459,20 @@ dependencies = [
  "libc",
 ]
 
+[[package]]
+name = "jpeg2k"
+version = "0.9.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f5b1c0ab2091b7044e98ec1910d63f68cc51358ec0dfde83f0b1d7aab654adb9"
+dependencies = [
+ "anyhow",
+ "image",
+ "log",
+ "openjp2",
+ "openjpeg-sys",
+ "thiserror 1.0.69",
+]
+
 [[package]]
 name = "js-sys"
 version = "0.3.106"
@@ -1845,7 +1876,7 @@ dependencies = [
  "rten",
  "rten-imageproc",
  "rten-tensor",
- "thiserror",
+ "thiserror 2.0.21",
  "wasm-bindgen",
 ]
 
@@ -1867,6 +1898,29 @@ version = "11.1.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "d6790f58c7ff633d8771f42965289203411a5e5c68388703c06e14f24770b41e"
 
+[[package]]
+name = "openjp2"
+version = "0.5.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4c582a03232f5f2f2f0107a7327753df3c32f5505a195cd79cad6aa97d6bddf1"
+dependencies = [
+ "bitflags 1.3.2",
+ "libc",
+ "log",
+ "smallvec",
+ "sprintf",
+]
+
+[[package]]
+name = "openjpeg-sys"
+version = "1.0.12"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "92682cb92e7b01e2c020cf8ec12f3374558924bb24621df1112d8a7c93d419ef"
+dependencies = [
+ "cc",
+ "libc",
+]
+
 [[package]]
 name = "option-ext"
 version = "0.2.0"
@@ -1950,7 +2004,7 @@ dependencies = [
  "shellexpand",
  "stringprep",
  "tempfile",
- "thiserror",
+ "thiserror 2.0.21",
  "time",
  "tokio",
  "ttf-parser",
@@ -2180,7 +2234,7 @@ dependencies = [
  "rustc-hash",
  "rustls",
  "socket2",
- "thiserror",
+ "thiserror 2.0.21",
  "tokio",
  "tracing",
  "web-time",
@@ -2202,7 +2256,7 @@ dependencies = [
  "rustls",
  "rustls-pki-types",
  "slab",
- "thiserror",
+ "thiserror 2.0.21",
  "tinyvec",
  "tracing",
  "web-time",
@@ -2362,7 +2416,7 @@ dependencies = [
  "rand 0.9.5",
  "rand_chacha",
  "simd_helpers",
- "thiserror",
+ "thiserror 2.0.21",
  "v_frame",
  "wasm-bindgen",
 ]
@@ -2433,7 +2487,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "60dc65c0ff1a7ae1294b0c67b9f14baf70b644404010370171787bfac1038fc0"
 dependencies = [
  "libredox",
- "thiserror",
+ "thiserror 2.0.21",
 ]
 
 [[package]]
@@ -2891,6 +2945,12 @@ dependencies = [
  "windows-sys 0.61.2",
 ]
 
+[[package]]
+name = "sprintf"
+version = "0.1.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6c0cdea5a20a06e7c57f627094e7b1618e5665592cd88f2d45fa4014e348db58"
+
 [[package]]
 name = "stable_deref_trait"
 version = "1.2.1"
@@ -2981,13 +3041,33 @@ dependencies = [
  "windows-sys 0.61.2",
 ]
 
+[[package]]
+name = "thiserror"
+version = "1.0.69"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b6aaf5339b578ea85b50e080feb250a3e8ae8cfcdff9a461c9ec2904bc923f52"
+dependencies = [
+ "thiserror-impl 1.0.69",
+]
+
 [[package]]
 name = "thiserror"
 version = "2.0.21"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "09e52cb86a36cede5cb101bf8908837b3e4c6e5e59fe7fd85c23fb56200d189e"
 dependencies = [
- "thiserror-impl",
+ "thiserror-impl 2.0.21",
+]
+
+[[package]]
+name = "thiserror-impl"
+version = "1.0.69"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4fee6c4efc90059e10f81e6d42c60a18f76588c3d74cb83a0b242a2b6c7504c1"
+dependencies = [
+ "proc-macro2 1.0.107",
+ "quote 1.0.47",
+ "syn 2.0.119",
 ]
 
 [[package]]
@@ -3771,7 +3851,7 @@ dependencies = [
  "flate2",
  "indexmap",
  "memchr",
- "thiserror",
+ "thiserror 2.0.21",
  "zopfli",
 ]
 

{edgeparse-0.3.0 → edgeparse-0.3.2}/Cargo.toml

@@ -3,7 +3,7 @@ resolver = "2"
 members = ["crates/edgeparse-core", "crates/pdf-cos", "crates/edgeparse-python"]
 
 [workspace.package]
-version = "0.3.0"
+version = "0.3.2"
 edition = "2021"
 rust-version = "1.85"
 license = "Apache-2.0"

{edgeparse-0.3.0 → edgeparse-0.3.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: edgeparse
-Version: 0.3.0
+Version: 0.3.2
 Classifier: Development Status :: 4 - Beta
 Classifier: Intended Audience :: Developers
 Classifier: License :: OSI Approved :: Apache Software License

{edgeparse-0.3.0 → edgeparse-0.3.2}/crates/edgeparse-core/Cargo.toml

@@ -40,6 +40,8 @@ rten = { version = "0.26", optional = true, default-features = false, features =
 rten-imageproc = { version = "0.26", optional = true }
 js-sys = { version = "0.3", optional = true }
 wasm-bindgen = { version = "0.2", optional = true }
+hayro-jbig2 = { version = "0.3", optional = true, default-features = false, features = ["std"] }
+jpeg2k = { version = "0.9", optional = true }
 
 [features]
 default = ["native"]
@@ -49,5 +51,9 @@ native = ["dep:rayon", "image", "dep:zip", "lopdf/rayon"]
 wasm = ["lopdf/wasm_js", "image", "dep:js-sys", "dep:wasm-bindgen"]
 hybrid = ["dep:reqwest", "dep:tokio"]
 ocr-ocrs = ["dep:ocrs", "dep:rten", "dep:rten-imageproc", "image"]
+# Optional PDF image codecs for the OCR path (scanned / hybrid pages).
+codecs-jbig2 = ["dep:hayro-jbig2", "image"]
+codecs-jpx = ["dep:jpeg2k", "image"]
+codecs = ["codecs-jbig2", "codecs-jpx"]
 
 [dev-dependencies]

{edgeparse-0.3.0 → edgeparse-0.3.2}/crates/edgeparse-core/src/api/config.rs

@@ -116,6 +116,9 @@ pub struct ProcessingConfig {
     pub image_dir: Option<String>,
     /// Enable raster table OCR recovery on image-based tables
     pub raster_table_ocr: bool,
+    /// Soft wall-clock budget for raster OCR recovery (milliseconds).
+    /// `None` = unlimited. When exhausted, remaining images are skipped.
+    pub ocr_budget_ms: Option<u64>,
     /// Pages to extract (e.g., "1,3,5-7")
     pub pages: Option<String>,
     /// Include headers/footers in output
@@ -153,6 +156,7 @@ impl Default for ProcessingConfig {
             image_format: ImageFormat::Png,
             image_dir: None,
             raster_table_ocr: true,
+            ocr_budget_ms: None,
             pages: None,
             include_header_footer: false,
             hybrid: HybridBackend::Off,
@@ -178,6 +182,11 @@ impl ProcessingConfig {
     pub fn raster_table_ocr_enabled(&self) -> bool {
         self.raster_table_ocr
     }
+
+    /// OCR wall-clock budget for this document.
+    pub fn ocr_budget(&self) -> crate::pdf::image_region::OcrBudget {
+        crate::pdf::image_region::OcrBudget::from_millis(self.ocr_budget_ms)
+    }
 }
 
 #[cfg(test)]

{edgeparse-0.3.0 → edgeparse-0.3.2}/crates/edgeparse-core/src/hybrid.rs

@@ -381,11 +381,10 @@ fn typographic_heading_deficit(doc: &PdfDocument) -> bool {
                     body_weights.push(w);
                 }
             }
-            ContentElement::Heading(h) => {
-                if h.base.base.semantic_type != SemanticType::Caption {
+            ContentElement::Heading(h)
+                if h.base.base.semantic_type != SemanticType::Caption => {
                     recovered += 1;
                 }
-            }
             ContentElement::NumberHeading(_) => {
                 recovered += 1;
             }

{edgeparse-0.3.0 → edgeparse-0.3.2}/crates/edgeparse-core/src/lib.rs

@@ -31,9 +31,50 @@ use crate::pdf::raster_table_ocr::{
 };
 use crate::pipeline::orchestrator::{run_pipeline, PipelineState};
 use crate::pipeline::stages::heading_detector::refine_heading_hierarchy;
-use crate::tagged::struct_tree::build_mcid_map;
+use crate::tagged::struct_tree::{build_mcid_map, should_use_struct_tree};
 use std::time::Instant;
 
+/// Run dominant-image text + raster-table recovery for one page (native only).
+///
+/// Single gate for OCR enablement + wall-clock budget so callers stay DRY.
+#[cfg(not(target_arch = "wasm32"))]
+fn recover_page_image_content(
+    input_path: &std::path::Path,
+    config: &ProcessingConfig,
+    page_info_by_number: &[Option<&page_info::PageInfo>],
+    page_num: u32,
+    page_chunks: &crate::pdf::chunk_parser::PageChunks,
+    ocr_budget: &mut crate::pdf::image_region::OcrBudget,
+) -> (
+    Vec<crate::models::chunks::TextChunk>,
+    Vec<crate::models::table::TableBorder>,
+) {
+    if !config.raster_table_ocr_enabled() || ocr_budget.exhausted() {
+        return (Vec::new(), Vec::new());
+    }
+    let Some(Some(page_info)) = page_info_by_number.get(page_num as usize) else {
+        return (Vec::new(), Vec::new());
+    };
+    let text = recover_dominant_image_text_chunks(
+        input_path,
+        &page_info.crop_box,
+        page_num,
+        &page_chunks.text_chunks,
+        &page_chunks.image_chunks,
+    );
+    if ocr_budget.exhausted() {
+        return (text, Vec::new());
+    }
+    let tables = recover_raster_table_borders(
+        input_path,
+        &page_info.crop_box,
+        page_num,
+        &page_chunks.text_chunks,
+        &page_chunks.image_chunks,
+    );
+    (text, tables)
+}
+
 /// Main entry point: convert a PDF file to structured data.
 ///
 /// # Arguments
@@ -81,28 +122,17 @@ pub fn convert(
     let mut page_contents = Vec::with_capacity(pages_map.len());
 
     let phase_start = Instant::now();
+    let mut ocr_budget = config.ocr_budget();
     for (&page_num, &page_id) in &pages_map {
         let page_chunks = extract_page_chunks(&raw_doc.document, page_num, page_id)?;
-        let mut recovered_text_chunks = Vec::new();
-        let mut recovered_tables = Vec::new();
-        if config.raster_table_ocr_enabled() {
-            if let Some(Some(page_info)) = page_info_by_number.get(page_num as usize) {
-                recovered_text_chunks = recover_dominant_image_text_chunks(
-                    input_path,
-                    &page_info.crop_box,
-                    page_num,
-                    &page_chunks.text_chunks,
-                    &page_chunks.image_chunks,
-                );
-                recovered_tables = recover_raster_table_borders(
-                    input_path,
-                    &page_info.crop_box,
-                    page_num,
-                    &page_chunks.text_chunks,
-                    &page_chunks.image_chunks,
-                );
-            }
-        }
+        let (recovered_text_chunks, recovered_tables) = recover_page_image_content(
+            input_path,
+            config,
+            &page_info_by_number,
+            page_num,
+            &page_chunks,
+            &mut ocr_budget,
+        );
         let mut elements: Vec<ContentElement> = page_chunks
             .text_chunks
             .into_iter()
@@ -142,9 +172,12 @@ pub fn convert(
     }
     log_phase_duration(timing_enabled, "extract_page_chunks", phase_start);
 
-    // Run the processing pipeline
     let phase_start = Instant::now();
-    let mcid_map = build_mcid_map(&raw_doc.document);
+    let mcid_map = if should_use_struct_tree(&raw_doc.document, config.use_struct_tree) {
+        build_mcid_map(&raw_doc.document)
+    } else {
+        Default::default()
+    };
     let mut pipeline_state = PipelineState::with_mcid_map(page_contents, config.clone(), mcid_map)
         .with_page_info(page_info_list);
     run_pipeline(&mut pipeline_state)?;
@@ -248,6 +281,10 @@ pub struct ExtractSession {
     pub creation_date: Option<String>,
     /// Modification date metadata.
     pub modification_date: Option<String>,
+    /// Producer metadata.
+    pub producer: Option<String>,
+    /// Creator metadata.
+    pub creator: Option<String>,
     /// Page count.
     pub number_of_pages: u32,
     /// Per-page content elements (tables not yet recovered from OCR candidates).
@@ -302,11 +339,18 @@ pub fn extract_session(
 
         #[cfg(feature = "image")]
         if config.raster_table_ocr_enabled() {
+            let (page_w, page_h) = page_info_list
+                .iter()
+                .find(|p| p.page_number == page_num)
+                .map(|p| (p.width, p.height))
+                .unwrap_or((612.0, 792.0));
             let page_cands = pdf::inmem_raster::collect_raster_candidates(
                 &raw_doc.document,
                 page_id,
                 &page_chunks.image_chunks,
                 &page_chunks.text_chunks,
+                page_w,
+                page_h,
                 &mut next_id,
             );
             candidates.extend(page_cands);
@@ -344,7 +388,11 @@ pub fn extract_session(
         }
     }
 
-    let mcid_map = build_mcid_map(&raw_doc.document);
+    let mcid_map = if should_use_struct_tree(&raw_doc.document, config.use_struct_tree) {
+        build_mcid_map(&raw_doc.document)
+    } else {
+        Default::default()
+    };
     let bookmarks = extract_bookmarks(&raw_doc.document);
 
     Ok(ExtractSession {
@@ -354,6 +402,8 @@ pub fn extract_session(
         title: raw_doc.metadata.title,
         creation_date: raw_doc.metadata.creation_date,
         modification_date: raw_doc.metadata.modification_date,
+        producer: raw_doc.metadata.producer,
+        creator: raw_doc.metadata.creator,
         number_of_pages: pages_map.len() as u32,
         page_contents,
         page_info_list,
@@ -389,6 +439,27 @@ impl ExtractSession {
     }
 }
 
+/// Drop the source Image (by XObject index) and inject OCR words as text chunks.
+#[cfg(feature = "image")]
+fn inject_ocr_text_blocks(
+    session: &mut ExtractSession,
+    page_idx: usize,
+    cand: &pdf::inmem_raster::RasterCandidate,
+    words: &[pdf::ocr::OcrWord],
+) {
+    if let Some(page) = session.page_contents.get_mut(page_idx) {
+        if let Some(idx) = cand.image_index {
+            page.retain(|el| match el {
+                ContentElement::Image(img) => img.index != Some(idx),
+                _ => true,
+            });
+        }
+        for chunk in pdf::inmem_raster::ocr_words_to_text_chunks(cand, words) {
+            page.push(ContentElement::TextChunk(chunk));
+        }
+    }
+}
+
 /// Run the pipeline after injecting OCR-derived tables from session candidates.
 pub fn assemble(
     mut session: ExtractSession,
@@ -397,9 +468,11 @@ pub fn assemble(
     #[cfg(feature = "image")]
     if session.config.raster_table_ocr_enabled() {
         use image::GrayImage;
+        use pdf::inmem_raster::RasterCandidateKind;
         use pdf::ocr::{default_engine, OcrEngine};
 
-        for cand in &session.candidates {
+        let candidates = std::mem::take(&mut session.candidates);
+        for cand in &candidates {
             let words = match &mode {
                 OcrAssembleMode::SyncEngine => {
                     if let Some(gray) =
@@ -423,15 +496,27 @@ pub fn assemble(
                     }
                 }
             };
-            if let Some(table) = pdf::inmem_raster::build_table_from_candidate(cand, &words) {
-                if pdf::inmem_raster::bordered_table_is_plausible(&table) {
-                    let page_idx = cand.page.saturating_sub(1) as usize;
-                    if let Some(page) = session.page_contents.get_mut(page_idx) {
-                        page.push(ContentElement::TableBorder(table));
+            let page_idx = cand.page.saturating_sub(1) as usize;
+            match cand.kind {
+                RasterCandidateKind::Table => {
+                    let table = pdf::inmem_raster::build_table_from_candidate(cand, &words)
+                        .filter(pdf::inmem_raster::bordered_table_is_plausible);
+                    if let Some(table) = table {
+                        if let Some(page) = session.page_contents.get_mut(page_idx) {
+                            page.push(ContentElement::TableBorder(table));
+                        }
+                    } else if !words.is_empty() {
+                        // OCR found glyphs but no lattice — assemble as text,
+                        // not discard (right assembly from evidence).
+                        inject_ocr_text_blocks(&mut session, page_idx, cand, &words);
                     }
                 }
+                RasterCandidateKind::TextBlocks => {
+                    inject_ocr_text_blocks(&mut session, page_idx, cand, &words);
+                }
             }
         }
+        session.candidates = candidates;
     }
 
     #[cfg(not(feature = "image"))]
@@ -452,6 +537,8 @@ pub fn assemble(
     doc.title = session.title;
     doc.creation_date = session.creation_date;
     doc.modification_date = session.modification_date;
+    doc.producer = session.producer;
+    doc.creator = session.creator;
 
     for page in pipeline_state.pages {
         doc.kids.extend(page);
@@ -776,4 +863,105 @@ mod tests {
             let _ = count_table_borders(&doc);
         }
     }
+
+    /// Pattern-painted page raster (Skia/Penpot style) must emit an OCR
+    /// candidate, skip page-edge lines, and not render artifact metadata titles.
+    #[cfg(feature = "image")]
+    #[test]
+    fn pattern_form_fixture_emits_text_blocks_candidate() {
+        let pdf =
+            std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")).join("testdata/pattern_form.pdf");
+        assert!(pdf.exists(), "missing fixture {}", pdf.display());
+        let data = std::fs::read(&pdf).unwrap();
+        let config = ProcessingConfig::default();
+        let session = extract_session(&data, "pattern_form.pdf", &config, None).unwrap();
+        assert!(
+            !session.candidates.is_empty(),
+            "expected ≥1 OCR candidate from pattern-painted image"
+        );
+        assert!(
+            session
+                .candidates
+                .iter()
+                .any(|c| c.kind == pdf::inmem_raster::RasterCandidateKind::TextBlocks),
+            "expected TextBlocks candidate for page-sized raster"
+        );
+
+        // No OCR words → still assemble; markdown must skip "Penpot - Render".
+        let doc = assemble(session, OcrAssembleMode::Provided).unwrap();
+        assert_eq!(doc.producer.as_deref(), Some("Skia/PDF m151"));
+        let md = output::markdown::to_markdown(&doc).unwrap();
+        assert!(
+            !md.contains("# Penpot - Render"),
+            "artifact metadata title must not become H1, got:\n{md}"
+        );
+        // Native text from the fixture should still appear.
+        assert!(
+            md.contains("NAME VALUE") || md.contains("ACME"),
+            "expected native text in markdown, got:\n{md}"
+        );
+        // Reading order: title (higher on page) before value text.
+        if let (Some(a), Some(b)) = (md.find("ACME"), md.find("NAME VALUE")) {
+            assert!(a < b, "expected ACME before NAME VALUE in:\n{md}");
+        }
+
+        // Page-edge lines should not dominate kids.
+        let edge_lines = doc
+            .kids
+            .iter()
+            .filter(|e| {
+                matches!(
+                    e,
+                    models::content::ContentElement::Line(l)
+                        if l.bbox.width() < 2.0 || l.bbox.height() < 2.0
+                )
+            })
+            .count();
+        assert_eq!(edge_lines, 0, "pattern fill must not emit page-edge lines");
+        assert!(
+            !doc.kids.iter().any(|e| matches!(
+                e,
+                models::content::ContentElement::Image(_)
+                    | models::content::ContentElement::Figure(_)
+            )),
+            "page-sized TextBlocks source image should be dropped from layout"
+        );
+    }
+
+    #[test]
+    fn artifact_metadata_title_skipped_in_markdown() {
+        let mut doc = PdfDocument::new("x.pdf".into());
+        doc.title = Some("Penpot - Render".into());
+        doc.producer = Some("Skia/PDF m151".into());
+        doc.creator = Some("Chromium".into());
+        doc.number_of_pages = 1;
+        // Body text present but does not corroborate Info.Title → skip H1.
+        doc.kids.push(models::content::ContentElement::TextChunk(
+            models::chunks::TextChunk {
+                value: "MANSUY RAPHAEL".into(),
+                bbox: models::bbox::BoundingBox::new(Some(1), 10.0, 100.0, 200.0, 120.0),
+                font_name: "Helvetica".into(),
+                font_size: 12.0,
+                font_weight: 400.0,
+                italic_angle: 0.0,
+                font_color: "#000000".into(),
+                contrast_ratio: 21.0,
+                symbol_ends: vec![200.0],
+                text_format: models::enums::TextFormat::Normal,
+                text_type: models::enums::TextType::Regular,
+                pdf_layer: models::enums::PdfLayer::Main,
+                ocg_visible: true,
+                index: Some(0),
+                page_number: Some(1),
+                level: None,
+                mcid: None,
+            },
+        ));
+        let md = output::markdown::to_markdown(&doc).unwrap();
+        assert!(
+            !md.contains("# Penpot - Render"),
+            "uncorroborated metadata title must be skipped when body text exists, got:\n{md}"
+        );
+        assert!(md.contains("MANSUY RAPHAEL"), "got:\n{md}");
+    }
 }