Merge pull request #313 from pymupdf/v0.1.8

Version 0.1.8

Author: JorjMcKie

pdf4llm/setup.py

@@ -6,7 +6,7 @@
 with open(os.path.join(setup_py_cwd, "README.md"), encoding="utf-8") as f:
     readme = f.read()
 
-version = "0.1.7"
+version = "0.1.8"
 
 classifiers = [
     "Development Status :: 5 - Production/Stable",

pymupdf4llm/pymupdf4llm/__init__.py

@@ -14,7 +14,7 @@
 version = VERSION
 version_tuple = tuple(map(int, version.split(".")))
 
-if not callable(pymupdf._get_layout):
+if pymupdf._get_layout is None:
     from .helpers.pymupdf_rag import IdentifyHeaders, TocHeaders, to_markdown
 
     pymupdf._warn_layout_once()  # recommend pymupdf_layout

pymupdf4llm/pymupdf4llm/helpers/document_layout.py

@@ -764,23 +764,31 @@ def parse_document(
                         if pymupdf.table._iou(tab.bbox, clip) > 0.6
                     ][0]
                     cells = [[c for c in row.cells] for row in table.rows]
-
+                    row_count = table.row_count
                     if table.header.external:  # if the header ioutside table
                         cells.insert(0, table.header.cells)  # insert a row
-                        table.row_count += 1  # increase row count
+                        row_count += 1  # increase row count
 
                     layoutbox.table = {
                         "bbox": list(table.bbox),
-                        "row_count": table.row_count,
+                        "row_count": row_count,
                         "col_count": table.col_count,
                         "cells": cells,
-                        "extract": table.extract(),
                     }
+
+                    layoutbox.table["extract"] = utils.table_extract(
+                        textpage,
+                        layoutbox,
+                    )
+
                     layoutbox.table["markdown"] = utils.table_to_markdown(
-                        textpage, layoutbox, markdown=True
+                        textpage,
+                        layoutbox,
+                        markdown=True,
                     )
+
                 except Exception as e:
-                    print(f"table detection error '{e}'")
+                    print(f"table detection error '{e}' on page {page.number+1}")
                     # table structure not detected: treat like an image
                     pix = page.get_pixmap(clip=clip, dpi=document.image_dpi)
                     layoutbox.image = pix.tobytes(document.image_format)

pymupdf4llm/pymupdf4llm/helpers/utils.py

@@ -514,25 +514,34 @@ def extract_cells(textpage, cell, markdown=False):
     Returns:
         A string with the text extracted from the cell.
     """
+
+    def outside_cell(bbox, cell):
+        return (
+            0
+            or bbox[0] >= cell[2]
+            or bbox[2] <= cell[0]
+            or bbox[1] >= cell[3]
+            or bbox[3] <= cell[1]
+        )
+
     text = ""
     for block in textpage.extractRAWDICT()["blocks"]:
         if block["type"] != 0:
             continue
+        if outside_cell(block["bbox"], cell):
+            continue
         for line in block["lines"]:
-            new_line = True
+            if outside_cell(line["bbox"], cell):
+                continue
             if text:  # must be a new line in the cell
-                if text.endswith("$"):
-                    text += " "
-                elif text.endswith("$ "):
-                    pass
-                else:
-                    text += "<br>" if markdown else "\n"
+                text += "<br>" if markdown else "\n"
 
             # strikeout detection only works with horizontal text
             horizontal = line["dir"] == (0, 1) or line["dir"] == (1, 0)
 
             for span in line["spans"]:
-                sbbox = span["bbox"]
+                if outside_cell(span["bbox"], cell):
+                    continue
                 # only include chars with more than 50% bbox overlap
                 span_text = ""
                 for char in span["chars"]:
@@ -576,7 +585,7 @@ def extract_cells(textpage, cell, markdown=False):
                         text += " "
                     else:
                         text += prefix + span_text + suffix
-
+    text = text.replace("$<br>", "$ ").replace(" $ <br>", "$ ")
     return text.strip()
 
 
@@ -635,3 +644,19 @@ def table_to_markdown(textpage, table_item, markdown=True):
         line += "\n"
         output += line
     return output + "\n"
+
+
+def table_extract(textpage, table_item):
+    table = table_item.table
+    row_count = table["row_count"]
+    col_count = table["col_count"]
+    cell_boxes = table["cells"]
+    # make empty cell text list
+    cells = [[None for i in range(col_count)] for j in range(row_count)]
+
+    for i, row in enumerate(cell_boxes):
+        for j, cell in enumerate(row):
+            if cell is not None:
+                cells[i][j] = extract_cells(textpage, cell_boxes[i][j], markdown=False)
+
+    return cells

pymupdf4llm/pymupdf4llm/versions_file.py

@@ -1,3 +1,3 @@
 # Generated file - do not edit.
 MINIMUM_PYMUPDF_VERSION = (1, 26, 6)
-VERSION = '0.1.7'
+VERSION = '0.1.8'

pymupdf4llm/setup.py

@@ -14,7 +14,7 @@
     "Topic :: Utilities",
 ]
 
-version = "0.1.7"
+version = "0.1.8"
 requires = ["pymupdf>=1.26.6", "tabulate"]
 
 text = requires[0].split("=")[1]