2654400439/Paper2Seminar / skills/paper-ppt-orchestrator
paper-ppt-orchestrator
VerifiedBuild complete fixed-layout academic-paper presentations from a PDF with bundled PPTX and one-page HTML templates. Use for full paper reading, complexity-aware slide budgeting, subsection coverage mapping, figure extraction, TikZ or Matplotlib evidence visuals, deterministic PPTX assembly, and final QA. Trigger for requests to turn a paper PDF into a discussion-group deck, especially 12-16+ page systems, security, networking, or measurement papers.
scripts/crop_author_block.py
View on GitHub#!/usr/bin/env python3
"""Crop the author and affiliation band below an academic paper title."""
from __future__ import annotations
import argparse
import hashlib
import json
import re
import sys
from dataclasses import dataclass
from pathlib import Path
import fitz
BODY_ANCHORS = (
"abstract",
"摘要",
"index terms",
"keywords",
"key words",
"introduction",
)
class CropError(RuntimeError):
pass
@dataclass(frozen=True)
class TextLine:
text: str
bbox: fitz.Rect
max_size: float
direction: tuple[float, float]
def normalize(text: str) -> str:
return re.sub(r"[^0-9a-z\u2e80-\u9fff]+", "", text.casefold())
def horizontal_lines(page: fitz.Page) -> list[TextLine]:
lines: list[TextLine] = []
for block in page.get_text("dict").get("blocks", []):
if block.get("type") != 0:
continue
for line in block.get("lines", []):
spans = line.get("spans", [])
text = "".join(str(span.get("text", "")) for span in spans).strip()
direction = tuple(line.get("dir", (1.0, 0.0)))
if not text or len(direction) != 2 or abs(direction[0]) < 0.9 or abs(direction[1]) > 0.2:
continue
lines.append(
TextLine(
text=text,
bbox=fitz.Rect(line["bbox"]),
max_size=max(float(span.get("size", 0.0)) for span in spans),
direction=(float(direction[0]), float(direction[1])),
)
)
return sorted(lines, key=lambda item: (item.bbox.y0, item.bbox.x0))
def detect_title_lines(lines: list[TextLine], page: fitz.Page, expected_title: str) -> tuple[list[TextLine], bool]:
upper = [line for line in lines if page.rect.y0 + 25 <= line.bbox.y0 <= page.rect.y0 + page.rect.height * 0.32]
if not upper:
raise CropError("no horizontal text lines found in the upper page region")
expected = normalize(expected_title)
matched: list[TextLine] = []
if expected:
for line in upper:
candidate = normalize(line.text)
if len(candidate) >= 6 and candidate in expected:
matched.append(line)
if matched:
matched.sort(key=lambda item: item.bbox.y0)
return matched, True
largest = max(line.max_size for line in upper)
threshold = max(13.0, largest - 1.5)
candidates = [line for line in upper if line.max_size >= threshold]
if not candidates:
raise CropError("could not identify title lines by font size")
candidates.sort(key=lambda item: item.bbox.y0)
cluster = [candidates[0]]
for line in candidates[1:]:
gap = line.bbox.y0 - cluster[-1].bbox.y1
if gap <= max(14.0, largest * 0.9):
cluster.append(line)
elif len(cluster) >= 1:
break
return cluster, False
def detect_body_start(lines: list[TextLine], title_bottom: float, page: fitz.Page) -> tuple[float, str | None]:
candidates = [line for line in lines if line.bbox.y0 > title_bottom + 18 and line.bbox.y0 < page.rect.y0 + page.rect.height * 0.58]
for line in candidates:
normalized = re.sub(r"\s+", " ", line.text).strip().casefold()
if any(normalized.startswith(anchor) for anchor in BODY_ANCHORS):
return line.bbox.y0, line.text[:80]
raise CropError("could not find Abstract, keywords, or introduction below the title; use --bbox")
def clamp_rect(rect: fitz.Rect, page_rect: fitz.Rect) -> fitz.Rect:
clipped = rect & page_rect
if clipped.is_empty or clipped.width < 40 or clipped.height < 20:
raise CropError(f"detected crop is too small: {tuple(round(value, 2) for value in clipped)}")
return clipped
def auto_bbox(page: fitz.Page, expected_title: str) -> tuple[fitz.Rect, dict[str, object]]:
lines = horizontal_lines(page)
title_lines, title_matched = detect_title_lines(lines, page, expected_title)
title_bottom = max(line.bbox.y1 for line in title_lines)
body_start, body_anchor = detect_body_start(lines, title_bottom, page)
y0 = title_bottom + 7
y1 = body_start - 8
band_lines = [line for line in lines if line.bbox.y0 >= y0 - 2 and line.bbox.y1 <= y1 + 2]
if not band_lines:
raise CropError("no author or affiliation text found between title and body")
x0 = max(page.rect.x0 + 24, min(line.bbox.x0 for line in band_lines) - 12)
x1 = min(page.rect.x1 - 24, max(line.bbox.x1 for line in band_lines) + 12)
rect = clamp_rect(fitz.Rect(x0, y0, x1, y1), page.rect)
return rect, {
"method": "title_to_body_anchor",
"title_matched": title_matched,
"title_lines": [line.text for line in title_lines],
"body_anchor": body_anchor,
"confidence": "high" if title_matched and body_anchor else "medium",
}
def crop_author_block(
pdf_path: Path,
output_path: Path,
*,
page_number: int = 1,
title: str = "",
dpi: int = 260,
bbox: tuple[float, float, float, float] | None = None,
) -> dict[str, object]:
if dpi < 120 or dpi > 400:
raise CropError("dpi must be between 120 and 400")
if not pdf_path.is_file():
raise CropError(f"PDF does not exist: {pdf_path}")
document = fitz.open(pdf_path)
try:
if page_number < 1 or page_number > document.page_count:
raise CropError(f"page must be between 1 and {document.page_count}")
page = document[page_number - 1]
if bbox is None:
rect, detection = auto_bbox(page, title)
else:
rect = clamp_rect(fitz.Rect(*bbox), page.rect)
detection = {"method": "manual_bbox", "confidence": "manual"}
output_path.parent.mkdir(parents=True, exist_ok=True)
pixmap = page.get_pixmap(matrix=fitz.Matrix(dpi / 72, dpi / 72), clip=rect, alpha=False)
pixmap.save(output_path)
finally:
document.close()
manifest = {
"tool": "PaperAuthorCrop/0.1",
"pdf": str(pdf_path.resolve()),
"pdf_sha256": hashlib.sha256(pdf_path.read_bytes()).hexdigest(),
"page": page_number,
"dpi": dpi,
"crop_bbox": [round(value, 2) for value in rect],
"output": str(output_path.resolve()),
"width_px": pixmap.width,
"height_px": pixmap.height,
"detection": detection,
}
output_path.with_suffix(".manifest.json").write_text(
json.dumps(manifest, ensure_ascii=False, indent=2), encoding="utf-8"
)
return manifest
def parse_bbox(value: str) -> tuple[float, float, float, float]:
try:
values = tuple(float(item.strip()) for item in value.split(","))
except ValueError as exc:
raise argparse.ArgumentTypeError("bbox must be x0,y0,x1,y1 in PDF points") from exc
if len(values) != 4:
raise argparse.ArgumentTypeError("bbox must contain exactly four comma-separated values")
return values # type: ignore[return-value]
def main() -> int:
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("pdf", type=Path)
parser.add_argument("-o", "--output", type=Path, required=True)
parser.add_argument("--page", type=int, default=1, help="1-based PDF page number")
parser.add_argument("--title", default="", help="expected paper title for stronger title matching")
parser.add_argument("--dpi", type=int, default=260)
parser.add_argument("--bbox", type=parse_bbox, help="manual x0,y0,x1,y1 override in PDF points")
args = parser.parse_args()
try:
manifest = crop_author_block(
args.pdf.resolve(),
args.output.resolve(),
page_number=args.page,
title=args.title,
dpi=args.dpi,
bbox=args.bbox,
)
except (CropError, OSError, RuntimeError) as exc:
print(f"ERROR: {exc}", file=sys.stderr)
return 1
print(f"Author block: {manifest['output']}")
print(f"Crop bbox: {manifest['crop_bbox']} | confidence: {manifest['detection']['confidence']}")
return 0
if __name__ == "__main__":
raise SystemExit(main())
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |