docling-parse 4.0.2__tar.gz → 4.0.4__tar.gz

{docling_parse-4.0.2 → docling_parse-4.0.4}/CHANGELOG.md

@@ -1,3 +1,15 @@
+## [v4.0.4](https://github.com/docling-project/docling-parse/releases/tag/v4.0.4) - 2025-06-10
+
+### Fix
+
+* Fix cropbox if it is larger than mediabox ([#126](https://github.com/docling-project/docling-parse/issues/126)) ([`a157d5a`](https://github.com/docling-project/docling-parse/commit/a157d5a6bbfd338c74022f0bfe842ff67feb82f5))
+
+## [v4.0.3](https://github.com/docling-project/docling-parse/releases/tag/v4.0.3) - 2025-06-05
+
+### Fix
+
+* Filenames with unicode chars on Windows ([#124](https://github.com/docling-project/docling-parse/issues/124)) ([`ec6556b`](https://github.com/docling-project/docling-parse/commit/ec6556b988d696e09227a9f8c5bf7a65bc843cae))
+
 ## [v4.0.2](https://github.com/docling-project/docling-parse/releases/tag/v4.0.2) - 2025-06-04
 
 ### Fix

{docling_parse-4.0.2 → docling_parse-4.0.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: docling-parse
-Version: 4.0.2
+Version: 4.0.4
 Summary: Simple package to extract text with coordinates from programmatic PDFs
 License: MIT
 Keywords: docling,pdf,parser

{docling_parse-4.0.2 → docling_parse-4.0.4}/docling_parse/pdf_parser.py

@@ -486,7 +486,7 @@ class DoclingPdfParser:
         Returns:
             bool: True if the document was successfully loaded, False otherwise.)")
         """
-        return self.parser.load_document(key=key, filename=filename)
+        return self.parser.load_document(key=key, filename=filename.encode("utf8"))
 
     def _load_document_from_bytesio(self, key: str, data: BytesIO) -> bool:
         """Load a document by key from a BytesIO-like object.

{docling_parse-4.0.2 → docling_parse-4.0.4}/docling_parse/visualize.py

@@ -144,6 +144,11 @@ def visualise_py(
 
         pdf_page: SegmentedPdfPage = pdf_doc.get_page(page_no=page_no)
 
+        if os.path.exists(str(output_dir)):
+            pdf_page.save_as_json(
+                Path(f"{output_dir}/{os.path.basename(pdf_path)}.page_{page_no}.json")
+            )
+
         if category in ["all", "char"]:
 
             img = pdf_page.render_as_image(

{docling_parse-4.0.2 → docling_parse-4.0.4}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "docling-parse"
-version = "4.0.2"
+version = "4.0.4"
 description = "Simple package to extract text with coordinates from programmatic PDFs"
 authors = ["Peter Staar <taa@zurich.ibm.com>"]
 maintainers = [

{docling_parse-4.0.2 → docling_parse-4.0.4}/src/pybind/docling_parser_v2.h

@@ -3,6 +3,11 @@
 #ifndef PYBIND_PDF_PARSER_V2_H
 #define PYBIND_PDF_PARSER_V2_H
 
+#ifdef _WIN32
+#include <locale>
+#include <codecvt>
+#endif
+
 #include <pybind/docling_resources.h>
 
 #include <v2.h>
@@ -178,7 +183,16 @@ namespace docling
 
   bool docling_parser_v2::load_document(std::string key, std::string filename)
   {
-    if (std::filesystem::exists(filename))
+#ifdef _WIN32
+    // Convert UTF-8 string to UTF-16 wstring
+    std::wstring_convert<std::codecvt_utf8_utf16<wchar_t>> converter;
+    std::wstring wide_filename = converter.from_bytes(filename);
+    std::filesystem::path path_filename(wide_filename);
+#else
+    std::filesystem::path path_filename(filename);
+#endif
+
+    if (std::filesystem::exists(path_filename))
       {
         //key2doc[key] = std::filesystem::path(filename);
 	key2doc[key] = std::make_shared<decoder_type>();

{docling_parse-4.0.2 → docling_parse-4.0.4}/src/v2/pdf_resources/page_dimension.h

@@ -246,6 +246,29 @@ namespace pdflib
         crop_bbox = media_bbox;
       }
 
+    if(crop_bbox[0]<media_bbox[0] or
+       crop_bbox[2]>media_bbox[2] or
+       crop_bbox[1]<media_bbox[1] or
+       crop_bbox[3]>media_bbox[3])
+      {
+        LOG_S(ERROR) << "The crop-box is larger than the media-box, \n"
+		     << "crop-box: {"
+		     << crop_bbox[0] << ", "
+		     << crop_bbox[1] << ", "
+		     << crop_bbox[2] << ", "
+		     << crop_bbox[3] << "}\n"
+		     << "media-box: {"
+		     << media_bbox[0] << ", "
+		     << media_bbox[1] << ", "
+		     << media_bbox[2] << ", "
+		     << media_bbox[3] << "}\n";
+
+	crop_bbox[0] = std::max(crop_bbox[0], media_bbox[0]);
+	crop_bbox[1] = std::max(crop_bbox[1], media_bbox[1]);
+	crop_bbox[2] = std::min(crop_bbox[2], media_bbox[2]);
+	crop_bbox[3] = std::min(crop_bbox[3], media_bbox[3]);	
+      }
+    
     if(json_resources.count("/BleedBox"))
       {        
         for(int d=0; d<4; d++)