How explosion/spacy works
spaCy is an NLP library: users process text through a language-specific tokenizer and configurable pipeline, producing annotated Docs for downstream use. Its separate training workflow converts and initializes data, trains components, evaluates them, and packages deployable pipelines; project commands support end-to-end workflows. The graph preserves major documented NLP capabilities and the website’s documentation/universe surfaces. Edges are limited to README statements and relationships visible in the supplied excerpts or source index; unsampled implementation details are not inferred.
Main entry points: language.py Language Pipeline · train.py Training CLI · factories.py Component Factories · index.js Website Layout
Components 24
NLP Runtime
- Language Pipeline
spacy/language.py - Language Rules
spacy/lang - Doc Data
spacy/tokens/_serialize.py - Lookup Data
spacy/lookups.py - Component Factories
spacy/pipeline/factories.py - Registrations
spacy/registrations.py
Pipeline Models
- Tagging and Parsing
spacy/ml/models/parser.py - Named Entities
spacy/pipeline/factories.py - Text Classification
spacy/pipeline/textcat.py - Span Categorization
spacy/pipeline/spancat.py - Entity Linking
spacy/pipeline/entity_linker.py - Lemmatization
spacy/pipeline/edit_tree_lemmatizer.py - Scoring
spacy/scorer.py
Training and Projects
- Data Conversion
spacy/training/converters - Model Initialization
spacy/training/initialize.py - Training Loop
spacy/training/loop.py - Training CLI
spacy/cli/train.py - Project Workflows
spacy/cli/project - Model Packaging
spacy/cli/package.py
Documentation Site
- Website Layout
website/src/templates/index.js - Landing Page
website/pages/index.tsx - Universe Catalog
website/src/templates/universe.js - Site Links
website/src/components/link.js
Other components
- Application Developer
Connections 23
- Application Developer → Language Pipeline: processes text
- Language Pipeline → Language Rules: uses language rules
spacy/language.py - Language Pipeline → Component Factories: loads factories
spacy/language.py - Language Pipeline → Doc Data: serializes Docs
- Component Factories → Named Entities: registers entity ruler
spacy/pipeline/factories.py - Component Factories → Span Categorization: constructs categorizer
spacy/pipeline/factories.py - Component Factories → Entity Linking: constructs linker
spacy/pipeline/factories.py - Language Pipeline → Lookup Data: uses lookup data
spacy/language.py - Language Pipeline → Scoring: uses scorer
spacy/language.py - Language Pipeline → Registrations: uses registrations
spacy/__init__.py - Language Pipeline → Tagging and Parsing: runs model pipeline
- Language Pipeline → Text Classification: runs classifier
- Language Pipeline → Lemmatization: runs lemmatizer
- Training CLI → Data Conversion: converts training data
- Training CLI → Model Initialization: initializes pipeline
- Model Initialization → Language Pipeline: configures pipeline
spacy/training/initialize.py - Training CLI → Training Loop: starts training
- Training Loop → Language Pipeline: updates components
spacy/training/loop.py - Project Workflows → Training CLI: orchestrates workflows
- Training CLI → Model Packaging: packages trained model
- Landing Page → Website Layout: renders in layout
website/pages/index.tsx - Website Layout → Site Links: uses links
website/src/templates/index.js - Website Layout → Universe Catalog: renders universe pages
website/src/templates/index.js