120 lines
4.3 KiB
Python
120 lines
4.3 KiB
Python
#!/usr/bin/env python3
|
|
# -*- coding: utf-8 -*-
|
|
|
|
import re
|
|
from pathlib import Path
|
|
|
|
TARGET = Path("manga-translator.py")
|
|
|
|
def cut_after_first_entrypoint(text: str) -> str:
|
|
"""
|
|
Keep only first full __main__ block and remove duplicated tail if present.
|
|
"""
|
|
m = re.search(r'(?m)^if __name__ == "__main__":\s*$', text)
|
|
if not m:
|
|
return text
|
|
|
|
start = m.start()
|
|
# Keep entrypoint block plus indented lines after it
|
|
lines = text[start:].splitlines(True)
|
|
keep = []
|
|
keep.append(lines[0]) # if __name__...
|
|
i = 1
|
|
while i < len(lines):
|
|
ln = lines[i]
|
|
if ln.strip() == "":
|
|
keep.append(ln)
|
|
i += 1
|
|
continue
|
|
# if dedented back to col 0 => end of block
|
|
if not ln.startswith((" ", "\t")):
|
|
break
|
|
keep.append(ln)
|
|
i += 1
|
|
|
|
cleaned = text[:start] + "".join(keep)
|
|
return cleaned
|
|
|
|
def replace_bad_vars(text: str) -> str:
|
|
text = text.replace(
|
|
"merge_micro_boxes_relaxed(bubbles, bubble_boxes, bubble_quads, bubble_indices, ocr, image_bgr)",
|
|
"merge_micro_boxes_relaxed(bubbles, bubble_boxes, bubble_quads, bubble_indices, filtered, image)"
|
|
)
|
|
text = text.replace(
|
|
"reattach_orphan_short_tokens(bubbles, bubble_boxes, bubble_quads, bubble_indices, ocr)",
|
|
"reattach_orphan_short_tokens(bubbles, bubble_boxes, bubble_quads, bubble_indices, filtered)"
|
|
)
|
|
return text
|
|
|
|
def ensure_autofix_chain(text: str) -> str:
|
|
old = (
|
|
" # ── Auto-fix (split + merge) ──────────────────────────────────────────\n"
|
|
" if auto_fix_bubbles:\n"
|
|
" bubbles, bubble_boxes, bubble_quads, bubble_indices = merge_micro_boxes_relaxed(bubbles, bubble_boxes, bubble_quads, bubble_indices, filtered, image)\n"
|
|
)
|
|
new = (
|
|
" # ── Auto-fix (split + merge) ──────────────────────────────────────────\n"
|
|
" if auto_fix_bubbles:\n"
|
|
" bubbles, bubble_boxes, bubble_quads, bubble_indices = auto_fix_bubble_detection(\n"
|
|
" bubble_boxes, bubble_indices, bubble_quads, bubbles, filtered, image)\n"
|
|
" bubbles, bubble_boxes, bubble_quads, bubble_indices = merge_micro_boxes_relaxed(\n"
|
|
" bubbles, bubble_boxes, bubble_quads, bubble_indices, filtered, image)\n"
|
|
)
|
|
return text.replace(old, new)
|
|
|
|
def ensure_split_commit(text: str) -> str:
|
|
marker = " # ── Remove nested / duplicate boxes ──────────────────────────────────\n"
|
|
if marker not in text:
|
|
return text
|
|
|
|
if "bubbles = new_bubbles" in text:
|
|
return text
|
|
|
|
inject = (
|
|
" bubbles = new_bubbles\n"
|
|
" bubble_boxes = new_bubble_boxes\n"
|
|
" bubble_quads = new_bubble_quads\n"
|
|
" bubble_indices = new_bubble_indices\n\n"
|
|
)
|
|
return text.replace(marker, inject + marker)
|
|
|
|
def ensure_rescue_pipeline(text: str) -> str:
|
|
anchor = ' print(f"Kept: {len(filtered)} | Skipped: {skipped}")\n'
|
|
if anchor not in text:
|
|
return text
|
|
|
|
if "rescue_name_and_short_tokens(raw" in text:
|
|
return text
|
|
|
|
block = (
|
|
' print(f"Kept: {len(filtered)} | Skipped: {skipped}")\n'
|
|
' # Protect short dialogue tokens confidence\n'
|
|
' tmp = []\n'
|
|
' for bbox, t, conf in filtered:\n'
|
|
' tmp.append((bbox, t, maybe_conf_floor_for_protected(t, conf, floor=0.40)))\n'
|
|
' filtered = tmp\n'
|
|
' # Rescue names/short tokens dropped by strict filters\n'
|
|
' rescued = rescue_name_and_short_tokens(raw, min_conf=0.20)\n'
|
|
' filtered = merge_rescued_items(filtered, rescued, iou_threshold=0.55)\n'
|
|
)
|
|
return text.replace(anchor, block)
|
|
|
|
def main():
|
|
if not TARGET.exists():
|
|
raise FileNotFoundError(f"Not found: {TARGET}")
|
|
|
|
src = TARGET.read_text(encoding="utf-8")
|
|
out = src
|
|
|
|
out = cut_after_first_entrypoint(out)
|
|
out = replace_bad_vars(out)
|
|
out = ensure_autofix_chain(out)
|
|
out = ensure_split_commit(out)
|
|
out = ensure_rescue_pipeline(out)
|
|
|
|
TARGET.write_text(out, encoding="utf-8")
|
|
print("✅ Patched manga-translator.py")
|
|
|
|
if __name__ == "__main__":
|
|
main()
|