| |
| |
| """Relocate only plain above-chart captions with geometric/preservation guards.""" |
| import re |
| from bs4 import BeautifulSoup |
|
|
| BLOCK = re.compile( |
| r"(<x_([\d.]+)><y_([\d.]+)>)(.*?)(<x_([\d.]+)><y_([\d.]+)><class_([^>]+)>)", re.S |
| ) |
|
|
|
|
| def plain(text): |
| text = text.replace("<tbc>", "").strip() |
| if any(marker in text for marker in ["**", "__", "~~", "`"]): |
| return None |
| if re.search(r"(?<!\w)_[^_\n]+_(?!\w)|<[/a-zA-Z!]|\\[A-Za-z&%_]|\]\(", text): |
| return None |
| return text if text and "\n" not in text else None |
|
|
|
|
| def relocate_above_captions(raw, markdown, render_base): |
| matches = list(BLOCK.finditer(raw)) |
| eligible = {} |
| targets = {} |
| moved = set() |
| allowed = {"Chart"} |
| boxes = [tuple(map(float, [m[2], m[3], m[6], m[7]])) for m in matches] |
| for i, m in enumerate(matches): |
| if m[8] != "Caption" or plain(m[4]) is None: |
| continue |
| box = boxes[i] |
| candidates = [] |
| for j in [i - 1, i + 1]: |
| if not 0 <= j < len(matches) or matches[j][8] not in allowed: |
| continue |
| other = boxes[j] |
| if box[3] > other[1]: |
| continue |
| width = min(box[2] - box[0], other[2] - other[0]) |
| overlap = max(0, min(box[2], other[2]) - max(box[0], other[0])) |
| gap = max(other[1] - box[3], box[1] - other[3]) |
| if width > 0 and overlap / width >= 0.8 and 0 <= gap <= 0.03: |
| candidates.append((gap, j)) |
| candidates.sort() |
| if not candidates or ( |
| len(candidates) > 1 and abs(candidates[0][0] - candidates[1][0]) < 1e-9 |
| ): |
| continue |
| gap, j = candidates[0] |
| eligible.setdefault(j, []).append((gap, i)) |
| for j, caps in eligible.items(): |
| caps.sort() |
| if len(caps) > 1 and abs(caps[0][0] - caps[1][0]) < 1e-9: |
| continue |
| caption_index = caps[0][1] |
| label = plain(matches[caption_index][4]) |
| m = matches[j] |
| body = m[4] |
| if len(re.findall(r"<chart\b", body)) != 1: |
| continue |
| title = re.search(r"(?m)^title:[ \t]*([^\n]*)", body) |
| if title: |
| if label not in title[1]: |
| body = ( |
| body[: title.start()] |
| + "title: " |
| + title[1].strip() |
| + " — " |
| + label |
| + body[title.end() :] |
| ) |
| else: |
| body = re.sub( |
| r"(<chart\b[^>]*>)", lambda q: q[0] + "\ntitle: " + label, body, count=1 |
| ) |
| derived = m[1] + body + m[5] |
| rendered = render_base(derived) |
| |
| captions = BeautifulSoup(rendered, "html.parser").find_all("caption") |
| if not any(label in cap.get_text(" ", strip=True) for cap in captions): |
| continue |
| targets[m.start()] = derived |
| moved.add(matches[caption_index].start()) |
| if not targets: |
| return markdown |
| derived = BLOCK.sub( |
| lambda m: "" if m.start() in moved else targets.get(m.start(), m[0]), raw |
| ) |
| return render_base(derived) |
|
|
