Agent skill
pdf-text-extractor
Download PDFs (when available) and extract plain text to support full-text evidence, writing `papers/fulltext_index.jsonl` and `papers/fulltext/*.txt`. **Trigger**: PDF download, fulltext, extract text, papers/pdfs, 全文抽取, 下载PDF. **Use when**: `queries.md` 设置 `evidence_mode: fulltext`(或你明确需要全文证据)并希望为 paper notes/claims 提供更强 evidence。 **Skip if**: `evidence_mode: abstract`(默认);或你不希望进行下载/抽取(成本/权限/时间)。 **Network**: fulltext 下载通常需要网络(除非你手工提供 PDF 缓存在 `papers/pdfs/`)。 **Guardrail**: 缓存下载到 `papers/pdfs/`;默认不覆盖已有抽取文本(除非显式要求重抽)。
Install this agent skill to your Project
npx add-skill https://github.com/WILLOSCAR/research-units-pipeline-skills/tree/main/.codex/skills/pdf-text-extractor
SKILL.md
PDF Text Extractor
Optionally collect full-text snippets to deepen evidence beyond abstracts.
This skill is intentionally conservative: in many survey runs, abstract/snippet mode is enough and avoids heavy downloads.
Inputs
papers/core_set.csv(expectspaper_id,title, and ideallypdf_url/arxiv_id/url)- Optional:
outline/mapping.tsv(to prioritize mapped papers)
Outputs
papers/fulltext_index.jsonl(one record per attempted paper)- Side artifacts:
papers/pdfs/<paper_id>.pdf(cached downloads)papers/fulltext/<paper_id>.txt(extracted text)
Decision: evidence mode
queries.mdcan setevidence_mode: "abstract" | "fulltext".abstract(default template): do not download; write an index that clearly records skipping.fulltext: download PDFs (when possible) and extract text topapers/fulltext/.
Local PDFs Mode
When you cannot/should not download PDFs (restricted network, rate limits, no permission), provide PDFs manually and run in “local PDFs only” mode.
- PDF naming convention:
papers/pdfs/<paper_id>.pdfwhere<paper_id>matchespapers/core_set.csv. - Set
- evidence_mode: "fulltext"inqueries.md. - Run:
python .codex/skills/pdf-text-extractor/scripts/run.py --workspace <ws> --local-pdfs-only
If PDFs are missing, the script writes a to-do list:
output/MISSING_PDFS.md(human-readable summary)papers/missing_pdfs.csv(machine-readable list)
Workflow (heuristic)
- Read
papers/core_set.csv. - If
outline/mapping.tsvexists, prioritize mapped papers first. - For each selected paper (fulltext mode):
- resolve
pdf_url(usepdf_url, else derive fromarxiv_id/urlwhen possible) - download to
papers/pdfs/<paper_id>.pdfif missing - extract a reasonable prefix of text to
papers/fulltext/<paper_id>.txt - append/update a JSONL record in
papers/fulltext_index.jsonlwith status + stats
- resolve
- Never overwrite existing extracted text unless explicitly requested (delete the
.txtto re-extract).
Quality checklist
-
papers/fulltext_index.jsonlexists and is non-empty. - If
evidence_mode: "fulltext": at least a small but non-trivial subset has extracted text (strict mode blocks if extraction coverage is near-zero). - If
evidence_mode: "abstract": the index records clearly reflect skip status (no downloads attempted).
Script
Quick Start
python .codex/skills/pdf-text-extractor/scripts/run.py --helppython .codex/skills/pdf-text-extractor/scripts/run.py --workspace <workspace_dir>
All Options
--max-papers <n>: cap number of papers processed (can be overridden byqueries.md)--max-pages <n>: extract at most N pages per PDF--min-chars <n>: minimum extracted chars to count as OK--sleep <sec>: delay between downloads--local-pdfs-only: do not download; only usepapers/pdfs/<paper_id>.pdfif presentqueries.mdsupports:evidence_mode,fulltext_max_papers,fulltext_max_pages,fulltext_min_chars
Examples
- Abstract mode (no downloads):
- Set
- evidence_mode: "abstract"inqueries.md, then run the script (it will emitpapers/fulltext_index.jsonlwith skip statuses)
- Set
- Fulltext mode with local PDFs only:
- Set
- evidence_mode: "fulltext"inqueries.md, put PDFs underpapers/pdfs/, then run:python .codex/skills/pdf-text-extractor/scripts/run.py --workspace <ws> --local-pdfs-only
- Set
- Fulltext mode with smaller budget:
python .codex/skills/pdf-text-extractor/scripts/run.py --workspace <ws> --max-papers 20 --max-pages 4 --min-chars 1200
Notes
- Downloads are cached under
papers/pdfs/; extracted text is cached underpapers/fulltext/. - The script does not overwrite existing extracted text unless you delete the
.txtfile.
Troubleshooting
Issue: no PDFs are available to download
Fix:
- Use
evidence_mode: abstract(default) or provide local PDFs underpapers/pdfs/and rerun with--local-pdfs-only.
Issue: extracted text is empty/garbled
Fix:
- Try a different extraction backend if supported; otherwise mark the paper as
abstractevidence level and avoid strong fulltext claims.
Didn't find tool you were looking for?