word-analysis

Word Analysis — .docx / .doc

Environment

from docx import Document
import os

# python-docx is available; for .doc (old format) convert via libreoffice first
def load_doc(path):
    """Load .docx directly; convert .doc to .docx first if needed."""
    if path.lower().endswith('.doc'):
        import subprocess
        out_dir = os.path.dirname(path)
        subprocess.run(
            ['libreoffice', '--headless', '--convert-to', 'docx', '--outdir', out_dir, path],
            check=True, capture_output=True
        )
        path = path.rsplit('.', 1)[0] + '.docx'
    return Document(path)

Core Method 1: Full Text Extraction

def extract_full_text(doc_path):
    """Extract all text: paragraphs + table cells, in document order."""
    doc = load_doc(doc_path)
    lines = []

    # Iterate paragraphs and tables in body order
    from docx.oxml.ns import qn
    for block in doc.element.body:
        tag = block.tag.split('}')[-1]
        if tag == 'p':
            # Paragraph
            from docx.text.paragraph import Paragraph
            para = Paragraph(block, doc)
            text = para.text.strip()
            if text:
                lines.append(text)
        elif tag == 'tbl':
            # Table
            from docx.table import Table
            tbl = Table(block, doc)
            for row in tbl.rows:
                row_text = '\t'.join(cell.text.strip() for cell in row.cells)
                if row_text.strip():
                    lines.append(row_text)

    return '\n'.join(lines)

# Usage
text = extract_full_text("/mnt/data/doc.docx")
print(text[:2000])  # preview first 2000 chars

Core Method 2: Table Extraction (Structured)

import pandas as pd

def extract_all_tables(doc_path):
    """Extract all tables from a Word document as list of DataFrames."""
    doc = load_doc(doc_path)
    tables = []

    for i, tbl in enumerate(doc.tables):
        rows = []
        for row in tbl.rows:
            rows.append([cell.text.strip() for cell in row.cells])
        if not rows:
            continue
        # Use first row as header if it looks like a header
        df = pd.DataFrame(rows[1:], columns=rows[0]) if rows else pd.DataFrame()
        tables.append((i, df))
        print(f"Table {i}: {df.shape[0]} rows × {df.shape[1]} cols")
        print(df.head(3))

    return tables

# Usage
tables = extract_all_tables("/mnt/data/doc.docx")