Function process_file

scripts/postprocess_fulltext.py:701–736

(path: Path, dry_run: bool)

# ── main processing ───────────────────────────────────────────────────────────

def process_file(path: Path, dry_run: bool) -> dict:
    original = path.read_text(encoding='utf-8')
    content = original

    content, toc_removed = remove_toc_tables(content)
    content, fn_count = reformat_footnotes(content)
    content, frag_merged = merge_fragmented_tables(content)
    content, dup_cols_fixed = deduplicate_table_columns(content)
    content = deduplicate_headers(content)
    
    content, hdr_removed = cleanup_page_headers_and_cover_images(content)
    content, lay_fixed = fix_complex_layouts(content)
    content, specific_fixed = apply_specific_fixes(content)
    content, bullets_fixed = fix_messy_bullets(content)
    content, paras_joined = join_split_paragraphs(content)
    
    content = collapse_blank_lines(content)

    changed = content != original
    stats = {
        'toc_removed': toc_removed,
        'footnotes': fn_count,
        'frag_merged': frag_merged,
        'dup_cols_fixed': dup_cols_fixed,
        'hdr_removed': hdr_removed,
        'lay_fixed': lay_fixed,
        'specific_fixed': specific_fixed,
        'bullets_fixed': bullets_fixed,
        'paras_joined': paras_joined,
        'changed': changed,
    }

    if changed and not dry_run:
        path.write_text(content, encoding='utf-8')

    return stats


def main():