deproc-core

deproc-core provides the language-agnostic contracts and runtime used by the deproc source-analysis plugins.

It supplies the shared pipeline for discovering source files, registering entities, linking package hierarchies, and resolving symbols. Language-specific parsing and semantics live in separate plugin packages such as deproc-python and deproc-java.

Installation

python -m pip install deproc-core

deproc-core supports Python 3.12 and newer.

Core concepts

• Context is the central registry for language plugins, parsers, linkers, resolvers, symbol caches, and entities.
• AnalysisScope describes the permitted analysis universe, including project, source, generated, declaration, and dependency roots, selected languages and extensions, exclusions, and root provenance.
• EntityRegistry stores entities by ID and maintains a fully qualified name-to-ID index for entities with FQNs.
• SourceParser, Linker, Resolver, and SymbolCache are the protocols implemented by language plugins.

Example

from deproc.core.context import Context
from deproc.core.discovery import find_source_files
from deproc.core.scope import AnalysisScope

scope = AnalysisScope(project_roots=["/path/to/project"])
context = Context(scope=scope)
context.set_language("python", [".py", ".pyi"], aliases=["py"])

files = find_source_files(context)

Plugins register their parser, linker, and resolver implementations with the context, and may optionally register a symbol cache.

Related packages

• deproc-python — Python parsing, linking, and semantic resolution.
• deproc-java — Java parsing, linking, and semantic resolution.
• deproc-utils-tree-sitter — shared tree-sitter utilities.
• deproc-utils-python-env — Python environment discovery utilities.

Development

Source code and tests are maintained in the deproc repository.

uv sync
uv run pytest deproc-core/tests

deproc-core is released under the Apache-2.0 license.