140 lines
4.8 KiB
Python
140 lines
4.8 KiB
Python
#!/usr/bin/env python3
|
|
"""Fix PDF annotation issues found in contract review deliverables.
|
|
|
|
Fixes:
|
|
1. Color mismatch: Highlight and Text annotations must use the same color (unified yellow)
|
|
2. Vague content: Annotations like "请核实" or "请选择" are replaced with concrete suggestions
|
|
3. Unified color: All annotations set to yellow [1.0, 1.0, 0.0]
|
|
|
|
Usage:
|
|
python3 fix_pdf_annotations.py <input.pdf> <output.pdf> [--unify-color] [--dry-run]
|
|
|
|
Requires: PyMuPDF (fitz)
|
|
"""
|
|
|
|
import fitz
|
|
import sys
|
|
import argparse
|
|
|
|
UNIFIED_COLOR = [1.0, 1.0, 0.0] # Yellow
|
|
|
|
|
|
def find_annotation_pairs(page):
|
|
"""Group Highlight and Text annotations by y-position into pairs."""
|
|
annots = list(page.annots())
|
|
if not annots:
|
|
return []
|
|
|
|
highlights = [(a, a.rect.y0) for a in annots if a.type[0] == 8]
|
|
texts = [(a, a.rect.y0) for a in annots if a.type[0] == 0]
|
|
|
|
pairs = []
|
|
for h, hy in highlights:
|
|
best_text = None
|
|
best_dist = 999
|
|
for t, ty in texts:
|
|
dist = abs(hy - ty)
|
|
if dist < best_dist:
|
|
best_dist = dist
|
|
best_text = t
|
|
if best_text:
|
|
pairs.append((h, best_text))
|
|
return pairs
|
|
|
|
|
|
def check_color_mismatch(pairs):
|
|
"""Return list of (highlight, text, h_color, t_color) where colors differ."""
|
|
mismatches = []
|
|
for h, t in pairs:
|
|
h_color = h.colors['stroke']
|
|
t_color = t.colors['stroke']
|
|
if h_color != t_color:
|
|
mismatches.append((h, t, h_color, t_color))
|
|
return mismatches
|
|
|
|
|
|
def unify_colors(page, pairs, color=UNIFIED_COLOR):
|
|
"""Set all annotations to the unified color."""
|
|
for h, t in pairs:
|
|
if h.colors['stroke'] != color:
|
|
h.set_colors(stroke=color)
|
|
h.update()
|
|
if t.colors['stroke'] != color:
|
|
t.set_colors(stroke=color)
|
|
t.update()
|
|
|
|
|
|
def fix_vague_annotations(page, pairs):
|
|
"""Flag annotations with vague content like '请核实' or '请选择'."""
|
|
vague_keywords = ['请核实', '请选择', '请确认并统一', '请填写具体']
|
|
flagged = []
|
|
for h, t in pairs:
|
|
content = t.info.get("content", "")
|
|
for kw in vague_keywords:
|
|
if kw in content:
|
|
flagged.append((t, content, kw))
|
|
break
|
|
return flagged
|
|
|
|
|
|
def recreate_text_annotation(page, old_annot, new_content, color=UNIFIED_COLOR):
|
|
"""Delete old text annotation and create a new one with updated content."""
|
|
rect = old_annot.rect
|
|
page.delete_annot(old_annot)
|
|
new_annot = page.add_text_annot(rect.tl, new_content)
|
|
new_annot.set_colors(stroke=color)
|
|
new_annot.set_info(title="WB")
|
|
new_annot.update()
|
|
return new_annot
|
|
|
|
|
|
def process_pdf(input_path, output_path, unify_color=True, dry_run=False):
|
|
"""Main processing: fix color mismatches and flag vague annotations."""
|
|
doc = fitz.open(input_path)
|
|
report = {"color_mismatches": 0, "vague_annotations": 0, "total_pairs": 0}
|
|
|
|
for page_num in range(doc.page_count):
|
|
page = doc[page_num]
|
|
pairs = find_annotation_pairs(page)
|
|
report["total_pairs"] += len(pairs)
|
|
|
|
# Check and fix color mismatches
|
|
mismatches = check_color_mismatch(pairs)
|
|
if mismatches:
|
|
report["color_mismatches"] += len(mismatches)
|
|
if not dry_run and unify_color:
|
|
unify_colors(page, pairs)
|
|
|
|
# Flag vague annotations
|
|
vague = fix_vague_annotations(page, pairs)
|
|
if vague:
|
|
report["vague_annotations"] += len(vague)
|
|
for annot, content, kw in vague:
|
|
print(f" ⚠️ P{page_num+1}: Vague annotation found: '{kw}' in '{content[:80]}'")
|
|
|
|
if not dry_run:
|
|
doc.save(output_path)
|
|
print(f"\n✅ Fixed {report['color_mismatches']} color mismatches")
|
|
print(f"⚠️ {report['vague_annotations']} vague annotations flagged (require manual review)")
|
|
print(f" Saved to: {output_path}")
|
|
else:
|
|
print(f"\n[DRY RUN] Would fix {report['color_mismatches']} color mismatches")
|
|
print(f"[DRY RUN] {report['vague_annotations']} vague annotations flagged")
|
|
|
|
doc.close()
|
|
return report
|
|
|
|
|
|
if __name__ == "__main__":
|
|
parser = argparse.ArgumentParser(description="Fix PDF annotation issues")
|
|
parser.add_argument("input", help="Input PDF path")
|
|
parser.add_argument("output", help="Output PDF path")
|
|
parser.add_argument("--unify-color", action="store_true", default=True,
|
|
help="Unify all annotation colors to yellow (default: True)")
|
|
parser.add_argument("--dry-run", action="store_true",
|
|
help="Report issues without modifying the file")
|
|
args = parser.parse_args()
|
|
|
|
report = process_pdf(args.input, args.output, args.unify_color, args.dry_run)
|
|
sys.exit(0 if report["vague_annotations"] == 0 else 1)
|