doc-extract

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

doc-extract

文档文本提取

Shared document-to-text extraction. One script, no state: reads an input file, prints JSON to stdout, writes nothing to disk (PHI-safe — no caches, no temp files; callers own any caching).

共享式文档转文本提取工具。单脚本、无状态：读取输入文件，将JSON输出到标准输出，不向磁盘写入任何内容（符合PHI安全要求——无缓存、无临时文件；缓存由调用方自行管理）。

Setup (once)

初始化（仅需一次）

bash

cd <this skill dir> && bun install

This pulls liteparse (the

lit

bin, used for PDF/DOCX/XLSX/PPTX, OCR included) and rtf-to-text (RTF). Without it, PDFs still work via a

pdftotext -layout

fallback if poppler is installed; other binary formats require liteparse.

bash

cd <this skill dir> && bun install

该命令会拉取liteparse（

lit

可执行文件，支持PDF/DOCX/XLSX/PPTX格式，内置OCR功能）和rtf-to-text（用于RTF格式）。如果未安装这些依赖，若系统已安装poppler，PDF仍可通过

pdftotext -layout

作为备选方案处理；其他二进制格式则必须依赖liteparse。

Use

使用方法

bash

bun <this skill dir>/scripts/extract.ts <input-file> [--content-type <mime>]

Output on stdout:

json

{ "text": "...", "method": "liteparse | pdftotext | rtf-to-text | passthrough", "pages": 12 }

```
text
```
is page-anchored for paged formats:
```
=== [page N] ===
```
markers between pages.
```
pages
```
is present when page markers exist.
```
method
```
is the extractor that actually produced the text.
Format is taken from the file extension; pass
```
--content-type
```
(e.g.
```
application/pdf
```
) when the file has no useful extension, as with downloaded EHR attachments. Note liteparse refuses extension-less files, so those PDFs go through the
```
pdftotext
```
fallback.
Errors print
```
{"error": "..."}
```
to stderr and exit 1.

bash

bun <this skill dir>/scripts/extract.ts <input-file> [--content-type <mime>]

标准输出内容：

json

{ "text": "...", "method": "liteparse | pdftotext | rtf-to-text | passthrough", "pages": 12 }

对于分页格式，
```
text
```
包含页面锚点：页面之间会有
```
=== [page N] ===
```
标记。
当存在页面标记时，会输出
```
pages
```
字段。
```
method
```
字段表示实际生成文本所使用的提取器。
格式由文件扩展名识别；当文件无有效扩展名时（如下载的EHR附件），需传入
```
--content-type
```
参数（例如
```
application/pdf
```
）。注意liteparse不处理无扩展名的文件，此类PDF会通过
```
pdftotext
```
备选方案处理。
错误信息会以
```
{"error": "..."}
```
格式输出到标准错误流，并以状态码1退出。

Table caveat

表格提取注意事项

Tables with multiple value columns (option A vs option B, in-tier vs out-of-tier) can interleave columns line-by-line in the extracted text: fragments of adjacent cells alternate, and a cell's text can even land mid-sentence inside a neighboring column. Values usually survive, but which column a value belongs to can become ambiguous. When an answer comes from one column of a multi-column table and the document has no redundant restatement of the value elsewhere, verify it by reading the original page directly before treating it as ground truth. The extracted text's

=== [page N] ===

anchor tells you which page: pass it to the Read tool's

pages

parameter (e.g. pages: "37") to render just that page to vision instead of the whole document.

包含多值列的表格（如选项A vs 选项B、计划内 vs 计划外）在提取后的文本中可能会逐行交错显示列内容：相邻单元格的文本片段交替出现，甚至某个单元格的文本可能会插入到相邻列的句子中间。数值通常会保留，但数值所属的列可能会变得模糊。当答案来自多列表格中的某一列，且文档其他位置未重复该数值时，在将其视为真实数据前，请直接查看原始页面进行验证。提取文本中的

=== [page N] ===

锚点会告知对应的页码：可将该页码传入Read工具的

pages

参数（例如pages: "37"），仅渲染该页面而非整个文档。

For other skills

供其他技能调用

Import the functions instead of shelling out when you're already in bun TS:

import { extract, resolveLit } from "../doc-extract/scripts/extract";
const lit = resolveLit([myRoot]); // also checks myRoot/node_modules/.bin/lit
const text = extract(lit, "/path/to/file.pdf"); // string | null

The contracts skill consumes it this way (its ingest caching stays on the contracts side).

当您已在bun TS环境中时，可直接导入函数而非通过shell调用：

import { extract, resolveLit } from "../doc-extract/scripts/extract";
const lit = resolveLit([myRoot]); // also checks myRoot/node_modules/.bin/lit
const text = extract(lit, "/path/to/file.pdf"); // string | null

contracts技能就是通过这种方式调用的（其导入缓存由contracts侧自行管理）。