fixed parameter-operator confusion in the qpdf-stream, added regression test files

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

Author: PeterStaar-IBM

docling_parse/visualize.py

@@ -273,29 +273,31 @@ def visualise_py(
 
     pdf_doc: PdfDocument = parser.load(path_or_stream=pdf_path, lazy=True)
 
-    pdf_page: ParsedPdfPage = pdf_doc.get_page(page_no=page_num)
+    page_nos = [page_num]
+    if page_num==-1:
+        page_nos = [(page_ind+1) for page_ind in range(0, pdf_doc.number_of_pages())]
 
-    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.original.export_to_textlines(add_fontkey=True)
-    print("text-lines (original): \n", "\n".join(lines))
+    for page_no in page_nos:
+        print(f"parsing {pdf_path} on page: {page_no}")
+        
+        pdf_page: ParsedPdfPage = pdf_doc.get_page(page_no=page_no)
+
+        if category in ["sanitized", "both"]:
+            pdf_page.sanitized.render(
+                draw_cells_bbox=(not display_text), draw_cells_text=display_text
+            ).show()
+        elif category in ["original", "both"]:
+            pdf_page.original.render(
+                draw_cells_bbox=(not display_text), draw_cells_text=display_text
+            ).show()
+
+        lines = pdf_page.original.export_to_textlines(add_fontkey=True)
+        print(f"text-lines (original, page_no: {page_no}):")
+        print("\n".join(lines))
 
-    lines = pdf_page.sanitized.export_to_textlines(add_fontkey=True)
-    print("text-lines (sanitized): \n", "\n".join(lines))
+        lines = pdf_page.sanitized.export_to_textlines(add_fontkey=True)
+        print(f"text-lines (sanitized, page_no: {page_no}):")
+        print("\n".join(lines))
 
     """
     lines = pdf_page.original.export_to_textlines(add_fontkey=True)

src/v2/qpdf/stream_decoder.h

@@ -32,11 +32,15 @@ namespace pdflib
   private:
 
     std::vector<qpdf_instruction>& stream;
+
+    std::regex value_pattern_0;    
   };
 
   qpdf_stream_decoder::qpdf_stream_decoder(std::vector<qpdf_instruction>& stream_):
     QPDFObjectHandle::ParserCallbacks(),
-    stream(stream_)
+    stream(stream_),
+
+    value_pattern_0(R"(^(\d\.\d+)(\-\d+)$)")
   {}
 
   qpdf_stream_decoder::~qpdf_stream_decoder()
@@ -95,6 +99,7 @@ namespace pdflib
       row.key = obj.getTypeName();
       row.val = obj.unparse();
       row.obj = obj;
+
       //LOG_S(INFO) << std::setw(12) << row.key << " | " << row.val;
     }
 
@@ -105,6 +110,8 @@ namespace pdflib
       }
     */
 
+    std::smatch match;
+    
     // if the row is null, reinterprete it as an empty array. We encountered
     // this usecase for a parameter of the d operator (see Table 56) that is
     // null but in reality should be an empty array.
@@ -113,6 +120,19 @@ namespace pdflib
 	row.key = "parameter";
 	row.val = "[]";
       }
+    else if (std::regex_match(row.val, match, value_pattern_0))
+      {
+	LOG_S(WARNING) << std::setw(12) << row.key << " | " << row.val << " => new matched value: " << match[1];
+
+	double value = std::stod(match[1].str());
+	
+	// Creating a real (floating-point) QPDFObjectHandle
+	QPDFObjectHandle new_obj = QPDFObjectHandle::newReal(value);
+
+	row.key = new_obj.getTypeName();
+	row.val = new_obj.unparse();
+	row.obj = new_obj;
+      }
 
     stream.push_back(row);
   }