lpgcast 0.1.1__tar.gz

lpgcast-0.1.1/.github/workflows/build-docs.yml

@@ -0,0 +1,65 @@
+name: Build GitHub Pages
+on:
+    push:
+        branches:
+        -   main
+        paths:
+        -   'docs/**'
+        -   'properdocs.yml'
+        -   'pyproject.toml'
+        -   'uv.lock'
+        -   'lpgcast/**'
+    # A pull request builds the site strictly but does not deploy it.
+    pull_request:
+        paths:
+        -   'docs/**'
+        -   'properdocs.yml'
+        -   'pyproject.toml'
+        -   'uv.lock'
+        -   'lpgcast/**'
+    workflow_dispatch:
+
+permissions:
+    contents: read
+
+# One deployment at a time; a newer push queues instead of cancelling one in flight.
+# Each pull request builds in its own group.
+concurrency:
+    group: pages-${{ github.event_name == 'pull_request' && github.ref || 'deploy' }}
+    cancel-in-progress: false
+
+jobs:
+    build:
+        runs-on: ubuntu-latest
+        steps:
+        -   name: Checkout code
+            uses: actions/checkout@v7
+        -   name: Install uv
+            uses: astral-sh/setup-uv@v10.2.0
+            with:
+                version: "0.12.21"
+                enable-cache: true
+        -   name: Install dependencies
+            run: uv sync --no-sources --group docs
+        -   name: Build site
+            run: uv run --no-sources properdocs build --strict --site-dir _site
+        -   name: Upload Pages artifact
+            if: github.event_name != 'pull_request'
+            uses: actions/upload-pages-artifact@v5
+            with:
+                path: _site
+
+    deploy:
+        needs: build
+        if: github.event_name != 'pull_request'
+        permissions:
+            pages: write
+            id-token: write
+        environment:
+            name: github-pages
+            url: ${{ steps.deployment.outputs.page_url }}
+        runs-on: ubuntu-latest
+        steps:
+        -   name: Deploy to GitHub Pages
+            id: deployment
+            uses: actions/deploy-pages@v5

lpgcast-0.1.1/.github/workflows/pre-commit.yml

@@ -0,0 +1,39 @@
+name: pre-commit
+
+on:
+    pull_request:
+    push:
+        branches: [main]
+
+concurrency:
+    group: ${{ github.workflow }}-${{ github.ref }}
+    cancel-in-progress: true
+
+permissions:
+    contents: read
+
+jobs:
+    pre-commit:
+        runs-on: ubuntu-latest
+        steps:
+        -   uses: actions/checkout@v7
+        -   name: Install uv
+            uses: astral-sh/setup-uv@v10.2.0
+            with:
+                version: "0.12.21"
+                enable-cache: true
+        # Hook environments live in ~/.cache/pre-commit, outside uv's cache;
+        # keyed on the config so a pin change rebuilds and nothing else does.
+        -   name: Cache hook environments
+            uses: actions/cache@v6
+            with:
+                path: ~/.cache/pre-commit
+                key: pre-commit-${{ runner.os }}-${{ hashFiles('.pre-commit-config.yaml') }}
+        # Lint and formatting only; type checking runs in tests.yml, where the
+        # whole program is installed.
+        -   name: Install dependencies
+            run: uv sync --no-sources --group dev --no-install-project
+        -   name: Run pre-commit
+            run: uv run --no-sources pre-commit run --all-files
+            env:
+                SKIP: ty

lpgcast-0.1.1/.github/workflows/pypi-publish.yml

@@ -0,0 +1,45 @@
+name: Upload Python Package to PyPI
+
+on:
+    release:
+        types: [published]
+
+permissions:
+    contents: read
+
+jobs:
+    pypi-publish:
+        # Off until the `lpgcast` name is reserved on PyPI: set the repository
+        # variable PYPI_PUBLISH to `true` to turn publishing on.
+        if: vars.PYPI_PUBLISH == 'true'
+        runs-on: ubuntu-latest
+        environment:
+            name: pypi
+            url: https://pypi.org/p/lpgcast
+        permissions:
+            id-token: write
+
+        steps:
+        -   uses: actions/checkout@v7
+
+        -   name: Install uv
+            uses: astral-sh/setup-uv@v10.2.0
+            with:
+                version: "0.12.21"
+        # A release whose tag differs from the project version fails instead
+        # of publishing a version nobody tagged.
+        -   name: Verify tag matches pyproject version
+            run: |
+                project_version="$(uv version --short)"
+                tag_version="${GITHUB_REF_NAME#v}"
+                echo "pyproject: ${project_version} / tag: ${tag_version}"
+                if [ "${project_version}" != "${tag_version}" ]; then
+                  echo "::error::Release tag ${GITHUB_REF_NAME} does not match pyproject version ${project_version}"
+                  exit 1
+                fi
+        -   name: Build and publish
+            run: |
+                uv build --no-sources
+                uv publish
+            env:
+                UV_PUBLISH_TOKEN: ${{ secrets.PYPI_API_TOKEN }}

lpgcast-0.1.1/.github/workflows/tests.yml

@@ -0,0 +1,76 @@
+name: tests
+
+on:
+    pull_request:
+    push:
+        branches: [main]
+
+concurrency:
+    group: ${{ github.workflow }}-${{ github.ref }}
+    cancel-in-progress: true
+
+permissions:
+    contents: read
+
+# Development uses graflo from a sibling checkout (`tool.uv.sources`); CI has
+# none, so every sync passes `--no-sources` and resolves graflo from PyPI.
+
+jobs:
+    # Guards the packaged artifact: a plain install must import and stay light.
+    import-smoke:
+        runs-on: ubuntu-latest
+        steps:
+        -   uses: actions/checkout@v7
+        -   name: Install uv
+            uses: astral-sh/setup-uv@v10.2.0
+            with:
+                version: "0.12.21"
+        -   name: Build wheel
+            run: uv build --wheel --no-sources --out-dir dist
+        -   name: Install the wheel
+            run: |
+                uv venv --python 3.12 /tmp/base-install
+                VIRTUAL_ENV=/tmp/base-install uv pip install dist/*.whl
+        -   name: Import the documented entry points, loading nothing heavy
+            run: |
+                /tmp/base-install/bin/python -c "
+                import sys
+                from lpgcast import LPGCaster, load_schema, read_episodes, write_cast
+                forbidden = ['langchain', 'langgraph', 'torch', 'numpy', 'pandas', 'rdflib']
+                leaked = [name for name in forbidden if name in sys.modules]
+                assert not leaked, f'imported on a plain install: {leaked}'
+                print('import OK')
+                "
+        -   name: Console scripts resolve
+            run: |
+                /tmp/base-install/bin/lpgcast --help
+                /tmp/base-install/bin/lpgcast-bench --help
+                /tmp/base-install/bin/lpgcast sheet --schema examples/tickets/schema.yaml > /dev/null
+
+    unit:
+        runs-on: ubuntu-latest
+        strategy:
+            fail-fast: false
+            matrix:
+                # PRs and non-main pushes: 3.12 only. Pushes to main also run 3.11 and 3.13.
+                python-version: ${{ github.event_name == 'push' && github.ref == 'refs/heads/main' && fromJSON('["3.11", "3.12", "3.13"]') || fromJSON('["3.12"]') }}
+        env:
+            UV_PYTHON: ${{ matrix.python-version }}
+        steps:
+        -   uses: actions/checkout@v7
+        -   name: Install uv
+            uses: astral-sh/setup-uv@v10.2.0
+            with:
+                version: "0.12.21"
+                enable-cache: true
+        -   name: Install dependencies
+            run: uv sync --no-sources --group dev
+        -   name: Type check
+            if: matrix.python-version == '3.12'
+            run: uv run --no-sources ty check lpgcast test
+        # No model is called: tests use a scripted endpoint, and the `llm` and
+        # `neo4j` markers are skipped without --run-llm / --run-neo4j.
+        -   name: Run tests
+            run: uv run --no-sources pytest
+            env:
+                PYTHONUNBUFFERED: '1'

lpgcast-0.1.1/.gitignore

@@ -0,0 +1,225 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[codz]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#   Usually these files are written by a python script from a template
+#   before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py.cover
+*.lcov
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+# Pipfile.lock
+
+# uv
+#   Similar to Pipfile.lock, it is generally recommended to include uv.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+# uv.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+# poetry.lock
+# poetry.toml
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#   pdm recommends including project-wide configuration in pdm.toml, but excluding .pdm-python.
+#   https://pdm-project.org/en/latest/usage/project/#working-with-version-control
+# pdm.lock
+# pdm.toml
+.pdm-python
+.pdm-build/
+
+# pixi
+#   Similar to Pipfile.lock, it is generally recommended to include pixi.lock in version control.
+# pixi.lock
+#   Pixi creates a virtual environment in the .pixi directory, just like venv module creates one
+#   in the .venv directory. It is recommended not to include this directory in version control.
+.pixi/*
+!.pixi/config.toml
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule*
+celerybeat.pid
+
+# Redis
+*.rdb
+*.aof
+*.pid
+
+# RabbitMQ
+mnesia/
+rabbitmq/
+rabbitmq-data/
+
+# ActiveMQ
+activemq-data/
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.envrc
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs/Zensical documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#   JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#   be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#   and can be added to the global gitignore or merged into this file.  For a more nuclear
+#   option (not recommended) you can uncomment the following to ignore the entire idea folder.
+# .idea/
+
+# Abstra
+#   Abstra is an AI-powered process automation framework.
+#   Ignore directories containing user credentials, local state, and settings.
+#   Learn more at https://abstra.io/docs
+.abstra/
+
+# Visual Studio Code
+#   Visual Studio Code specific template is maintained in a separate VisualStudioCode.gitignore that
+#   can be found at https://github.com/github/gitignore/blob/main/Global/VisualStudioCode.gitignore
+#   and can be added to the global gitignore or merged into this file. However, if you prefer, you
+#   could uncomment the following to ignore the entire vscode folder
+# .vscode/
+# Temporary file for partial code execution
+tempCodeRunnerFile.py
+
+# Ruff stuff:
+.ruff_cache/
+
+# PyPI configuration file
+.pypirc
+
+# Marimo
+marimo/_static/
+marimo/_lsp/
+__marimo__/
+
+# Streamlit
+.streamlit/secrets.toml
+
+# Response cache written by tests run without LPGCAST_CACHE_DIR
+.test_cache/
+
+# Benchmark reports; results are published elsewhere, never committed
+bench-results/
+
+.idea/

lpgcast-0.1.1/.pre-commit-config.yaml

@@ -0,0 +1,58 @@
+repos:
+  # Repository hygiene and JSON formatting. Env files, fixture data and vendored
+  # font files are excluded: their bytes are kept as committed.
+-   repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v6.0.0
+    hooks:
+    -   id: check-merge-conflict
+    -   id: check-added-large-files
+        args: [--maxkb=2000]
+    -   id: end-of-file-fixer
+        exclude: (^|/)(\.env|data/)|^lpgcast/plot/fonts/|\.svg$
+    -   id: trailing-whitespace
+        args: [--markdown-linebreak-ext=md]
+        exclude: (^|/)(\.env|data/)|^lpgcast/plot/fonts/|\.svg$
+    -   id: pretty-format-json
+        args: [--autofix, --indent, '4', --no-sort-keys]
+
+  # Python linting and formatting
+-   repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.16.10
+    hooks:
+    -   id: ruff-check
+        args: [--fix]
+    -   id: ruff-format
+
+-   repo: local
+    hooks:
+    # Type checking
+    -   id: ty
+        name: ty check
+        entry: uv run --frozen ty check lpgcast test
+        language: system
+        types: [python]
+        pass_filenames: false
+
+    # Package docs must not point outside this repository
+    -   id: no-external-doc-paths
+        name: no external doc paths
+        language: pygrep
+        entry: '(?i)docs/(design|reports|ISSUES|ROADMAP|PLANNING)|benchmarking/|papers/|platform-demos/|\bworkspace\b'
+        files: \.md$
+
+  # YAML formatting
+-   repo: https://github.com/macisamuele/language-formatters-pre-commit-hooks
+    rev: v2.16.0
+    hooks:
+    -   id: pretty-format-yaml
+        args: [--autofix, --indent, '4', --preserve-quotes]
+
+  # TOML formatting and sorting. One tool only: pretty-format-toml is built on the
+  # toml-sort library, and its hook environment pins whatever toml-sort was current
+  # when it was installed. Once that differs from the rev below, the two rewrite
+  # inline tables back and forth ({a = 1} vs { a = 1 }) on every run.
+-   repo: https://github.com/pappasam/toml-sort
+    rev: v0.25.0
+    hooks:
+    -   id: toml-sort
+        args: [-ia]

lpgcast-0.1.1/CHANGELOG.md

@@ -0,0 +1,67 @@
+# Changelog
+
+All notable changes to this project will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
+and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [0.1.1] - 2026-10-11
+
+### Added
+
+- **Package skeleton**: `lpgcast` on Python ≥ 3.11 with six runtime dependencies (`graflo`, `openai`,
+  `pydantic`, `pydantic-settings`, `click`, `pyyaml`).
+- **Settings** from `LPGCAST_LLM_*`, `LPGCAST_CACHE_*` and `LPGCAST_*` variables only; no `.env` file
+  is read. The default model is `gpt-5.6-luna`.
+- **`LLMClient.acomplete_json`**: one JSON object per request from any OpenAI-compatible endpoint;
+  strict `json_schema` output with a `json_object` fallback, tolerant parsing, re-ask on malformed
+  JSON, one re-issue on timeout, typed configuration errors.
+- **Response cache** keyed on the logical request, written atomically; read-only mode and
+  `LPGCAST_CACHE_REQUIRE_HIT` for replays that make no request. Without an API key only cached
+  responses are served.
+- **Usage accounting**: calls, failures and tokens, with billed tokens kept apart from tokens
+  replayed from the cache; `track_usage()` scopes it per task.
+- **CLI** `lpgcast` (alias `text2lpg`) with `sheet`: the type sheet, or with `--json-schema` the
+  reply schema, for a schema or manifest file.
+- **Episodes** (`Episode`, `read_episodes`): text, source id, reference time, metadata (never shown
+  to the model) and thread context, from JSON Lines.
+- **Schema selection** (`select`): exact-form, phrase and number-unit lanes fused by rank, a relative
+  relevance gate, and closure by graflo's `subschema()` under a budget; a schema under budget passes
+  whole, and a large schema nothing matches is a `SelectionError` with a reason.
+- **Extraction prompt** (`build_messages`): byte-deterministic, type sheet and rules first, the text
+  inside a tagged block it cannot close.
+- **`extract`**: one call per episode in the flat reply format, parsed into wire models; an episode
+  over `LPGCAST_MAX_EPISODE_TOKENS` is refused before any call.
+- **CLI** `lpgcast select`: the slice each text would be cast against, as JSON lines or type sheets.
+- **Validation** (`validate`): unknown types and relations, properties not declared on their type,
+  values that do not coerce to their field type, insufficient identities, dangling refs and quotes
+  not found in the text are findings; values are coerced (declared units stripped), reversed edges
+  flipped and quotes matched ignoring case, accents and whitespace, each recorded as a fix.
+- **Typed repair**: at most two follow-up requests listing the findings by kind; the attempt that
+  keeps the most items wins.
+- **Quarantine sink** (`JsonlQuarantineSink`): left-out items, failed episodes and `unmapped`
+  entries, with the raw reply.
+- **Identity resolution and assembly** (`resolve`): natural keys, digests for hash and funnel types,
+  UUIDs derived from schema, episode, type and ref for assigned and blank types; mentions are never
+  written as vertices, and an edge to a mention known only by a secondary identity waits for a lookup.
+- **Groundings**: one per vertex, edge and property, with character offsets found from the quote.
+- **`LPGCaster`**: the whole pipeline over many episodes, concurrently; a misconfigured model or a
+  cache miss under replay stops the run.
+- **`write_cast`**: defines the schema on the target (reusing an existing one), looks up
+  secondary-identity mentions and upserts through graflo's writer.
+- **CLI** `lpgcast cast`: graph, groundings, quarantine, episode results and a run manifest under
+  `--out`; `--db` writes to a graflo connection; `--replay` makes no request.
+- **`LPGCaster.cast_episode`** returns an `EpisodeOutcome` that also carries the slice, the
+  first reply as checked and the attempt kept.
+- **Benchmark** `lpgcast-bench`: a versioned corpus of seven families in four domains with a
+  dev/test split and a pinned digest; exact scoring with no model as judge; an oracle row; a
+  fresh-call re-cast; JSON and Markdown reports with token counts and optional prices.
+- **Top-level API**: `from lpgcast import LPGCaster, load_schema, read_episodes, write_cast, ...`;
+  names load on first use, so `import lpgcast` stays light.
+- **Documentation site**: installation, quick start, how a cast works, identities and mentions,
+  groundings and quarantine, guides to schemas, replay, databases and the benchmark, and a
+  reference whose settings, command-line and Python pages are generated from the code.
+- **Examples**: `examples/tickets/` (a support-desk schema and five texts) and
+  `examples/connections/` (file backend and Neo4j).
+- **CI**: tests (with a check that a plain install imports and stays light), pre-commit, docs to
+  GitHub Pages, and PyPI publishing on release once enabled.