Added PDF table extraction feature with aligned Markdown tables and deduplication

Author: Ashok

packages/markitdown/src/markitdown/converters/_pdf_converter.py

@@ -1,22 +1,18 @@
 import sys
 import io
-
 from typing import BinaryIO, Any
 
-
 from .._base_converter import DocumentConverter, DocumentConverterResult
 from .._stream_info import StreamInfo
 from .._exceptions import MissingDependencyException, MISSING_DEPENDENCY_MESSAGE
 
-
-# Try loading optional (but in this case, required) dependencies
-# Save reporting of any exceptions for later
+# Load dependencies
 _dependency_exc_info = None
 try:
     import pdfminer
     import pdfminer.high_level
+    import pdfplumber
 except ImportError:
-    # Preserve the error and stack trace for later
     _dependency_exc_info = sys.exc_info()
 
 
@@ -28,16 +24,43 @@
 ACCEPTED_FILE_EXTENSIONS = [".pdf"]
 
 
+def _to_markdown_table(table: list[list[str]]) -> str:
+    """Convert a 2D list (rows/columns) into a nicely aligned Markdown table."""
+    if not table:
+        return ""
+
+    # Normalize None → ""
+    table = [[cell if cell is not None else "" for cell in row] for row in table]
+
+    # Column widths
+    col_widths = [max(len(str(cell)) for cell in col) for col in zip(*table)]
+
+    def fmt_row(row):
+        return "| " + " | ".join(
+            str(cell).ljust(width) for cell, width in zip(row, col_widths)
+        ) + " |"
+
+    header, *rows = table
+    md = [fmt_row(header)]
+    md.append("| " + " | ".join("-" * w for w in col_widths) + " |")
+    for row in rows:
+        md.append(fmt_row(row))
+
+    return "\n".join(md)
+
+
 class PdfConverter(DocumentConverter):
     """
-    Converts PDFs to Markdown. Most style information is ignored, so the results are essentially plain-text.
+    Converts PDFs to Markdown.
+    Supports extracting tables into aligned Markdown format (via pdfplumber).
+    Falls back to pdfminer if pdfplumber is missing or fails.
     """
 
     def accepts(
         self,
         file_stream: BinaryIO,
         stream_info: StreamInfo,
-        **kwargs: Any,  # Options to pass to the converter
+        **kwargs: Any,
     ) -> bool:
         mimetype = (stream_info.mimetype or "").lower()
         extension = (stream_info.extension or "").lower()
@@ -55,23 +78,59 @@ def convert(
         self,
         file_stream: BinaryIO,
         stream_info: StreamInfo,
-        **kwargs: Any,  # Options to pass to the converter
+        **kwargs: Any,
     ) -> DocumentConverterResult:
-        # Check the dependencies
         if _dependency_exc_info is not None:
             raise MissingDependencyException(
                 MISSING_DEPENDENCY_MESSAGE.format(
                     converter=type(self).__name__,
                     extension=".pdf",
                     feature="pdf",
                 )
-            ) from _dependency_exc_info[
-                1
-            ].with_traceback(  # type: ignore[union-attr]
-                _dependency_exc_info[2]
-            )
-
-        assert isinstance(file_stream, io.IOBase)  # for mypy
-        return DocumentConverterResult(
-            markdown=pdfminer.high_level.extract_text(file_stream),
-        )
+            ) from _dependency_exc_info[1].with_traceback(_dependency_exc_info[2])  # type: ignore[union-attr]
+
+        assert isinstance(file_stream, io.IOBase)
+
+        markdown_chunks: list[str] = []
+
+        try:
+            with pdfplumber.open(file_stream) as pdf:
+                for page in pdf.pages:
+                    text = page.extract_text() or ""
+                    page_tables = page.extract_tables()
+
+                    # Remove table rows from text to avoid duplication
+                    for table in page_tables:
+                        if not table:
+                            continue
+                        header_line = " ".join(table[0])
+                        if header_line in text:
+                            text = text.replace(header_line, "")
+                        for row in table[1:]:
+                            row_line = " ".join(row)
+                            if row_line in text:
+                                text = text.replace(row_line, "")
+
+                    # Normalize whitespace: collapse multiple blank lines
+                    lines = [line.strip() for line in text.splitlines() if line.strip()]
+                    clean_text = "\n".join(lines)
+                    if clean_text:
+                        markdown_chunks.append(clean_text)
+
+                    # Append tables as aligned Markdown
+                    for table in page_tables:
+                        md_table = _to_markdown_table(table)
+                        if md_table:
+                            markdown_chunks.append(md_table)
+
+            markdown = "\n\n".join(markdown_chunks).strip()
+
+        except Exception:
+            # Fallback if pdfplumber fails
+            markdown = pdfminer.high_level.extract_text(file_stream)
+
+        # Fallback if still empty
+        if not markdown:
+            markdown = pdfminer.high_level.extract_text(file_stream)
+
+        return DocumentConverterResult(markdown=markdown)