updated the visualize script to use the pdf-parser

Signed-off-by: Peter Staar <[email protected]>

docling_parse/document.py

@@ -142,9 +142,10 @@ def to_top_left_origin(self, page_height: float) -> "BoundingRectangle":
 class PdfBaseElement(BaseModel):
     ordering: int
 
+
 class PdfColoredElement(PdfBaseElement):
-    rgba: ColorRGBA = ColorRGBA(r=0, g=0, b=0, a=255)    
-    
+    rgba: ColorRGBA = ColorRGBA(r=0, g=0, b=0, a=255)
+
 
 class PdfCell(PdfColoredElement):
 
@@ -166,6 +167,7 @@ def to_bottom_left_origin(self, page_height: float):
     def to_top_left_origin(self, page_height: float):
         self.rect = self.rect.to_top_left_origin(page_height=page_height)
 
+
 class PdfBitmapResource(PdfBaseElement):
 
     rect: BoundingRectangle
@@ -176,13 +178,14 @@ def to_bottom_left_origin(self, page_height: float):
 
     def to_top_left_origin(self, page_height: float):
         self.rect = self.rect.to_top_left_origin(page_height=page_height)
-
+
+
 class PdfLine(PdfColoredElement):
 
     parent_id: int
     points: List[Tuple[float, float]]
     width: float = 1.0
-    
+
     coord_origin: CoordOrigin = CoordOrigin.BOTTOMLEFT
 
     def __len__(self) -> int:
@@ -316,6 +319,13 @@ def crop_text(self, bbox: BoundingBox, eps: float = 1.0):
                     text += " "
                     text += cell.text
 
+    def export_to_textlines(self, add_location: bool = True) -> List[str]:
+        lines: List[str] = []
+        for cell in self.cells:
+            lines.append(f"{cell.text}")
+
+        return lines
+
     def render(
         self,
         boundary_type: PageBoundaryType = PageBoundaryType.CROP_BOX,  # media_box
@@ -513,17 +523,6 @@ def _draw_text_in_bounding_bbox(
         return result
 
 
-# class ParsedPageLabel(str, Enum):
-#     """ParsedPageLabel."""
-#
-#     ORIGINAL = "orginal"
-#     SANITIZED = "sanitized"
-#
-#     def __str__(self):
-#         """Get string value."""
-#         return str(self.value)
-
-
 class ParsedPage(BaseModel):
 
     original: SegmentedPage

docling_parse/pdf_parser.py

@@ -71,10 +71,11 @@ def get_page(self, page_no: int) -> ParsedPage:
                     self._pages[page_no] = self._to_parsed_page(page)  # put on cache
                     return self._pages[page_no]
 
-            else:
-                raise ValueError(
-                    f"incorrect page_no: {page_no} for key={self._key} (min:1, max:{self.number_of_pages()})"
-                )
+        raise ValueError(
+            f"incorrect page_no: {page_no} for key={self._key} (min:1, max:{self.number_of_pages()})"
+        )
+
+        return ParsedPage()
 
     def load_all_pages(self):
         doc_dict = self._parser.parse_pdf_from_key(
@@ -238,7 +239,7 @@ def _to_lines(self, data: dict) -> List[PdfLine]:
                 line = PdfLine(
                     ordering=ind,
                     parent_id=l,
-                    points=points,                    
+                    points=points,
                 )
                 result.append(line)
 

docling_parse/visualize.py

@@ -4,15 +4,14 @@
 import os
 from typing import Dict, Optional
 
+from docling_parse.document import ParsedPage
+from docling_parse.pdf_parser import DoclingPdfParser, PdfDocument
 from docling_parse.pdf_parsers import (  # type: ignore[import]
     pdf_parser_v1,
     pdf_parser_v2,
 )
 from docling_parse.utils import create_pil_image_of_page_v1, create_pil_image_of_page_v2
 
-from docling_parse.document import PageBoundaryType, ParsedPage
-from docling_parse.pdf_parser import DoclingPdfParser, PdfDocument
-
 # Configure logging
 logging.basicConfig(
     level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s"
@@ -277,15 +276,26 @@ def visualise_py(
 
     pdf_page: ParsedPage = pdf_doc.get_page(page_no=page_num)
 
-    if category=="both":
-        pdf_page.original.render(draw_cells_bbox=(not display_text), draw_cells_text=display_text).show()
-        pdf_page.sanitized.render(draw_cells_bbox=(not display_text), draw_cells_text=display_text).show()
-    elif category=="sanitized":
-        pdf_page.sanitized.render(draw_cells_bbox=(not display_text), draw_cells_text=display_text).show()
-    elif category=="original":
-        pdf_page.original.render(draw_cells_bbox=(not display_text), draw_cells_text=display_text).show()
-
-
+    if category == "both":
+        pdf_page.original.render(
+            draw_cells_bbox=(not display_text), draw_cells_text=display_text
+        ).show()
+        pdf_page.sanitized.render(
+            draw_cells_bbox=(not display_text), draw_cells_text=display_text
+        ).show()
+    elif category == "sanitized":
+        pdf_page.sanitized.render(
+            draw_cells_bbox=(not display_text), draw_cells_text=display_text
+        ).show()
+    elif category == "original":
+        pdf_page.original.render(
+            draw_cells_bbox=(not display_text), draw_cells_text=display_text
+        ).show()
+
+    lines = pdf_page.sanitized.export_to_textlines()
+    print("\n".join(lines))
+
+
 def main():
 
     (
@@ -335,7 +345,6 @@ def main():
         )
     else:
         return -1
-
 
 
 if __name__ == "__main__":

src/v2/pdf_states/grph.h

@@ -211,23 +211,35 @@ namespace pdflib
     QPDFObjectHandle arr = instructions[0].obj;
 
     //assert(arr.isArray());
-    if(not arr.isArray()) { LOG_S(ERROR) << "instructions[0].obj is not an array"; return; }
-
-    for(int l=0; l<arr.getArrayNItems(); l++)
+    //if(not arr.isArray()) { LOG_S(ERROR) << "instructions[0].obj is not an array"; return; }
+    if(arr.isArray())
       {
-	QPDFObjectHandle item = arr.getArrayItem(l);
-
-	//assert(item.isNumber());
-	if(item.isNumber())
+	for(int l=0; l<arr.getArrayNItems(); l++)
 	  {
-	    double val = item.getNumericValue();
-	    dash_array.push_back(val);
-	  }
-	else
-	  {
-	    LOG_S(WARNING) << "skipping items for dash_array ...";
+	    QPDFObjectHandle item = arr.getArrayItem(l);
+
+	    //assert(item.isNumber());
+	    if(item.isNumber())
+	      {
+		double val = item.getNumericValue();
+		dash_array.push_back(val);
+	      }
+	    else
+	      {
+		LOG_S(WARNING) << "skipping items for dash_array ...";
+	      }
 	  }
       }
+    else if(arr.isNull())
+      {
+	LOG_S(WARNING) << "instructions[0].obj is null, re-interpreting it as an empty array";
+	dash_array = {};
+      }
+    else
+      {
+	LOG_S(ERROR) << "instructions[0].obj is not an array nor null, defualting to an empty array";
+	dash_array = {};
+      }
 
     if(instructions[1].is_integer())
       {