Обзор
LangExtract turns text into entities and attributes while preserving where extracted spans came from. It fits after OCR or parsing and before business validation. Your team still defines schemas, examples, entity merging, and acceptable error rates.
Основные функции
- Few-shot extraction
- Source alignment
- Long-document processing
- JSONL and visualization
- Provider flexibility
Требования, установка и быстрый старт
Использование
How it works
Instructions and examples guide extraction. Outputs can be saved as JSONL and visualized against source spans. An unaligned extraction may have no char_interval and should be reviewed or filtered.
Audience and requirements
Document and data engineering teams. Python, text inputs, labeled examples, and a supported model.
Practical use cases
Report structuring; entity extraction; annotation assistance.
Limitations and selection
Alignment does not guarantee semantic correctness or completeness. OCR remains a separate step. Chunking and provider limits affect quality and cost.
Related projects and selection
JaidedAI/EasyOCR：Complement: OCR supplies text; recognition errors propagate downstream.
explosion/spaCy：Comparison: spaCy serves fixed-label NLP pipelines; LangExtract uses task-specific LLM instructions.
Source review
Editorial analysis of upstream sources, without runtime or benchmark testing. Proposed workflows are editorial suggestions.
Совместимость моделей и варианты использования
README documents Gemini, other providers, and local Ollama; schema capabilities depend on the provider.
Лицензия и примечания о рисках
The repository page identifies Apache-2.0. Read LICENSE; model weights and datasets may have separate terms.
Editorial source review 2026-09-09T05:00:00.950Z. README and live repository page verified; current stars/forks from GitHub HTML. Last-push metadata retained from 2026-09-05 discovery snapshot. No runtime benchmark. Integration proposals are editorial analysis.
Релиз и сопровождение
Reviewed 2026-09-09. Counters come from repository pages; features are based on upstream documentation. See Releases in the source links. Editorial analysis of upstream sources, without runtime or benchmark testing. Proposed workflows are editorial suggestions.