AI Skill Hub 推荐使用:数字藏品管道 是一款优质的AI工具。AI 综合评分 7.5 分,在同类工具中表现稳健。如果你正在寻找可靠的AI工具解决方案,这是一个值得深入了解的选择。
数字藏品管道 是一款基于 Python 开发的开源工具,专注于 文化遗产、数字化、目录 等核心功能。作为 GitHub 开源项目,它拥有活跃的社区支持和持续的版本迭代,代码完全透明可审计,支持本地部署以保护数据隐私。无论是个人使用还是集成到企业工作流,都能提供稳定可靠的解决方案。
数字藏品管道 是一款基于 Python 开发的开源工具,专注于 文化遗产、数字化、目录 等核心功能。作为 GitHub 开源项目,它拥有活跃的社区支持和持续的版本迭代,代码完全透明可审计,支持本地部署以保护数据隐私。无论是个人使用还是集成到企业工作流,都能提供稳定可靠的解决方案。
# 方式一:pip 安装(推荐)
pip install directory-pipeline
# 方式二:虚拟环境安装(推荐生产环境)
python -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\activate
pip install directory-pipeline
# 方式三:从源码安装(获取最新功能)
git clone https://github.com/hadro/directory-pipeline
cd directory-pipeline
pip install -e .
# 验证安装
python -c "import directory_pipeline; print('安装成功')"
# 命令行使用
directory-pipeline --help
# 基本用法
directory-pipeline input_file -o output_file
# Python 代码中调用
import directory_pipeline
# 示例
result = directory_pipeline.process("input")
print(result)
# directory-pipeline 配置文件示例(config.yml) app: name: "directory-pipeline" debug: false log_level: "INFO" # 运行时指定配置文件 directory-pipeline --config config.yml # 或通过环境变量配置 export DIRECTORY_PIPELINE_API_KEY="your-key" export DIRECTORY_PIPELINE_OUTPUT_DIR="./output"
Requires Python 3.11+ and uv.
uv sync # core: Gemini OCR + entry extraction
uv sync --extra gpu # add Surya OCR (GPU or Apple Silicon recommended)
uv sync --extra geo # add geocoding + map generation
uv sync --all-extras # everything
This installs the pipeline command (run pipeline --help for all subcommands):
pipeline run <URL> # automated: download → OCR → extract → explore
pipeline guided <URL> # human-in-loop: page selection + alignment review
pipeline ingest <URL> # download only
pipeline calibrate <URL|DIR> # select sample pages + generate prompts (once per collection type)
pipeline ocr <DIR> # Surya OCR + Gemini OCR + align bboxes
pipeline review <DIR> # interactive alignment review (browser UI)
pipeline extract <DIR> # NER extraction + explorer
pipeline geo <DIR> # geocode entries + build map (needs address fields)
pipeline postprocess <DIR> # fix + combine volumes + rebuild explorer
Each subcommand wraps the underlying python main.py <URL> [flags] stage interface — see docs/pipeline-stages.md for the flag-level reference and every artifact each stage produces.
Set your API keys (or copy .env.template to .env):
export GEMINI_API_KEY=your_key_here
export GOOGLE_MAPS_API_KEY=your_key_here # optional; enables address-level geocoding
---
Requires Python 3.11+ and uv.
```bash uv sync # installs dependencies and the pipeline command
Two published collections built with this pipeline:
<p> <a href="https://hadro.github.io/tulsa-city-directories/1921#about"><img src="docs/screenshots/tulsa-city-directory-entry.png" width="49%" alt="Tulsa 1921 city directory explorer: entry detail with a crop of the source scan highlighting the matched line"></a> <a href="https://hadro.github.io/green-books/explorer#about"><img src="docs/screenshots/green-books-entry.png" width="49%" alt="Green Book explorer: faceted entry list with a detail panel showing cross-year listings and a location map"></a> </p>
Entry detail views from the published explorers. Left: each Tulsa entry renders the exact line from the source scan via its canvas_fragment URI. Right: a Green Book establishment with its listing history across volumes.
Both explorers received additional front-end design work beyond what the pipeline generates. The pipeline produces the entry CSVs, IIIF manifests, and a baseline HTML explorer; these published sites build on that output.
---
Turn a public digital archive URL into a structured, browsable CSV — no manual transcription, no custom code per collection type.
Give it a URL from the Library of Congress, Internet Archive, or any institution that publishes a public IIIF manifest (CONTENTdm repositories included). It downloads the scans, OCRs them, and extracts entries into a structured CSV. With the enrichment steps, every row links back to the exact location in the original scan.
Built for digitized historical directories — city directories, gazetteers, trade directories — but works on just about any historical document with regular entry-like structure.

The auto-generated data explorer. Categorical facets generated from the data, full-text search, IIIF page thumbnails, and a "View in source document" deep link for every row.
---
一个有用的数字藏品结构化数据转化工具
AI Skill Hub 为第三方内容聚合平台,本页面信息基于公开数据整理,不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后,再部署至生产环境,并做好必要的安全评估。
✅ MIT 协议 — 最宽松的开源协议之一,可自由商用、修改、分发,仅需保留版权声明。
总体来看,数字藏品管道 是一款质量良好的AI工具,在同类工具中具备一定竞争力。AI Skill Hub 将持续追踪其更新动态,建议收藏备用,结合自身场景选择合适时机引入使用。
| 原始名称 | directory-pipeline |
| 原始描述 | 开源AI工具:A pipeline for turning digital collections into structured data -- an LLM assist。⭐10 · Python |
| Topics | 文化遗产数字化目录 |
| GitHub | https://github.com/hadro/directory-pipeline |
| License | MIT |
| 语言 | Python |
收录时间:2026-06-11 · 更新时间:2026-06-11 · License:MIT · AI Skill Hub 不对第三方内容的准确性作法律背书。