langgraph-stream-parser 0.1.7__tar.gz → 0.2.0__tar.gz

{langgraph_stream_parser-0.1.7 → langgraph_stream_parser-0.2.0}/.gitignore

@@ -1,208 +1,208 @@
-# 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
-.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
-
-# Sphinx documentation
-docs/_build/
-
-# 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
-
-# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
-__pypackages__/
-
-# Celery stuff
-celerybeat-schedule
-celerybeat.pid
-
-# 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 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/
-
-# Ruff stuff:
-.ruff_cache/
-
-# PyPI configuration file
-.pypirc
-
-# Cursor
-#  Cursor is an AI-powered code editor. `.cursorignore` specifies files/directories to
-#  exclude from AI features like autocomplete and code analysis. Recommended for sensitive data
-#  refer to https://docs.cursor.com/context/ignore-files
-.cursorignore
-.cursorindexingignore
-
-# Marimo
-marimo/_static/
-marimo/_lsp/
-__marimo__/
-.jupyter_ystore.db
+# 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
+.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
+
+# Sphinx documentation
+docs/_build/
+
+# 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
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# 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 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/
+
+# Ruff stuff:
+.ruff_cache/
+
+# PyPI configuration file
+.pypirc
+
+# Cursor
+#  Cursor is an AI-powered code editor. `.cursorignore` specifies files/directories to
+#  exclude from AI features like autocomplete and code analysis. Recommended for sensitive data
+#  refer to https://docs.cursor.com/context/ignore-files
+.cursorignore
+.cursorindexingignore
+
+# Marimo
+marimo/_static/
+marimo/_lsp/
+__marimo__/
+.jupyter_ystore.db

langgraph_stream_parser-0.2.0/CHANGELOG.md

@@ -0,0 +1,142 @@
+# Changelog
+
+## [0.2.0] - 2026-06-02
+
+Repositions langgraph-stream-parser as the **shared runtime substrate** for the deep-agent host family (`cowork-dash`, `deepagent-lab`, `deepagent-code`, `deepagent-vscode`).
+
+### Added
+- **`host/` submodule** — shared host conventions:
+  - `load_agent_spec("path.py:var" | "module:var")` — strict agent-spec loader (the `:object` suffix is required).
+  - `HostConfig` — layered config resolver: `defaults < deepagents.toml < DEEPAGENT_* env < overrides`, with per-field source tracking. Subclass and extend the `_ENV` / `_TOML` maps (merged across the MRO) to add host-specific keys. `DEEPAGENT_AGENT_SPEC` is the canonical agent-spec env var.
+  - `load_toml_config()` — loads + deep-merges global `~/.deepagents/config.toml` (override dir via `DEEPAGENTS_CONFIG_HOME`) and the nearest project `deepagents.toml`.
+  - `Workspace` — workspace-root wrapper with traversal-safe `subpath()`.
+  - `python -m langgraph_stream_parser.host` (and `HostConfig.describe()`) — prints each resolved value, its source, and the env var / TOML key that sets it.
+- **`adapters.SessionAdapter`** — session-scoped streaming for web hosts: per-session event queue, cancellation, side-channel `push_event()`, and persistent SSE that survives client reconnects.
+- **`demo.create_default_agent()`** — shared filesystem-backed default agent factory (behind the `[demo]` extra; lazy-imports `deepagents`).
+- **Four built-in extractors** for the agentskills.io / Hermes pattern, wired into the default set so every host gets them through `compat`: `SkillManageExtractor` (`skill_manage`), `SkillViewExtractor` (`skill_view`), `CompressionExtractor` (`__compression__`), `MemoryExtractor` (`memory`).
+- `event_to_dict(event, *, max_result_len=...)` — lets hosts drop bespoke serializer shims.
+
+### Fixed
+- `skip_tools` previously suppressed a tool's **extractor** as well as its lifecycle events, silently dropping `todo_list` / `reflection`. Extractors now run for skipped tools; only the lifecycle (start/end) events are suppressed.
+
+## [0.1.9] - 2026-05-19
+
+Compatibility refresh for **langgraph 1.2**, **langchain-core 1.4**, and **deepagents 0.6**.
+
+### Added
+- `UsageEvent.cache_read_tokens` and `UsageEvent.cache_creation_tokens` — populated from `usage_metadata.input_token_details` (`cache_read`, `cache_creation`). Default 0; omitted from `to_dict()` when zero.
+- `ContentEvent.is_subagent` and `ReasoningEvent.is_subagent` — set to `True` when stream metadata carries `ls_agent_type == "subagent"` (deepagents >= 0.6). Lets consumers distinguish subagent output even when `lc_agent_name` is absent.
+- `"respond"` decision support in `InterruptEvent.build_decisions()` / `create_resume()`: pass `response="..."` to send a text reply in place of the tool call. Matches the deepagents 0.6 decision verb set.
+- `InterruptEvent.build_decisions(..., use_edited_action=False)` escape hatch for runtimes that still expect the legacy `{"type": "edit", "args": ...}` resume shape.
+
+### Changed
+- `extract_message_content()` now skips the full set of non-text content blocks defined by langchain-core 1.4 standard content blocks: `tool_call`, `tool_use`, `tool_call_chunk`, `server_tool_call`, `server_tool_call_chunk`, `server_tool_result`, `invalid_tool_call`, `image`, `audio`, `video`, `file` (plus the existing `reasoning` / `thinking`). Previously these could leak as stringified dicts into `ContentEvent.content`. Tool lifecycle and reasoning events are unchanged.
+- `InterruptEvent.build_decisions("edit", ...)` now emits the modern `{"type": "edit", "edited_action": {"name", "args"}}` shape by default, matching LangGraph 1.1+ / deepagents 0.5+. Set `use_edited_action=False` for the legacy shape.
+- `InterruptEvent.allowed_decisions` defaults to `{"approve", "reject", "edit", "respond"}` when no review configs are present (was `{"approve", "reject"}`).
+- Dev dependency bumped: `langgraph>=1.1.0`, `langchain-core>=1.4.0`.
+
+### Notes
+- **No breaking change for default tool extractors**: deepagents 0.6 ships new built-in tools (`glob_search`, `grep_search`, `execute`, `start_async_task` / `check_async_task` / `update_async_task` / `cancel_async_task` / `list_async_tasks`, plus QuickJS `CodeInterpreterMiddleware`). These flow through the regular `ToolCallStartEvent` / `ToolCallEndEvent` lifecycle — no parser change required.
+- **v3 `stream_events` typed projections** (LangGraph 1.2 beta) are not yet supported. v2 `StreamPart` parsing remains the recommended path for `stream()` / `astream()`.
+
+## [0.1.8] - 2026-04-18
+
+### Added
+- `ReasoningEvent` dataclass for reasoning / thinking content; emitted from langchain-core `reasoning` and `thinking` content blocks, and from `think_tool` reflections. Carries a `source` field (`"content_block"` or `"think_tool"`) so UIs can distinguish provenance.
+- `DisplayEvent` dataclass for rich inline content (dataframes, images, plotly, html, json) from `display_inline`-style tools. Carries `display_type`, `data`, `title`, `status`, `error`, `tool_name`, `tool_call_id`, `node`, `namespace`.
+- `extract_reasoning_content()` helper in `extractors.messages` for parsing reasoning blocks from `AIMessageChunk.content`.
+- `UpdatesHandler._event_from_extraction()` routes extractor output to typed events; unknown `extracted_type` values still flow through `ToolExtractedEvent` for custom extractors.
+- README sections: "Reasoning & Thinking" and "Rich Inline Display" with typed matching examples.
+
+### Changed
+- `think_tool` output is now a `ReasoningEvent(source="think_tool")` instead of `ToolExtractedEvent(extracted_type="reflection")`. Legacy dict API (`stream_graph_updates`) still produces `{"chunk": text}` for backward compatibility.
+- `display_inline` tool output is now a `DisplayEvent` instead of `ToolExtractedEvent(extracted_type="display_inline")`.
+- `extract_message_content()` now skips reasoning blocks so they can be surfaced as `ReasoningEvent` separately.
+
+### Fixed
+- Removed dead `has_messages` variable in `_parse_v2`.
+
+## [0.1.7] - 2026-04-18
+
+### Added
+- `FastAPIAdapter` for streaming LangGraph events over WebSocket and Server-Sent Events; stateless by design — conversation state is keyed by `session_id` used as LangGraph `thread_id`
+- Per-session asyncio lock with refcounted cleanup to serialize concurrent turns on the same thread
+- `BaseAdapter._text_prompt_interrupt()` helper, shared by `PrintAdapter` and `JupyterDisplay`
+- `BaseAdapter._truncate()` helper for preview-length capping
+- `fastapi` optional dependency group (`pip install langgraph-stream-parser[fastapi]`)
+
+### Changed
+- Hoisted `_last_rendered_count` incremental-render cursor from Print/CLI into `BaseAdapter`
+- Slimmed `examples/fastapi_websocket.py` from ~455 to ~234 lines by using the new adapter
+
+### Fixed
+- `UsageEvent` now has an explicit case in `BaseAdapter._process_event` instead of silently falling through
+
+## [0.1.6] - 2026-03-28
+
+### Added
+- v2 StreamPart parsing (`stream_mode="v2"`) with auto-detection of `{"type", "ns", "data"}` dict format
+- `ValuesEvent` for full state snapshots from `stream_mode="values"` (v2)
+- `DebugEvent` for debug, checkpoint, and task trace data from v2 streaming
+- Routing for v2 stream types: updates, messages, custom, values, debug, checkpoints, tasks
+
+## [0.1.5] - 2026-02-06
+
+### Added
+- Subgraph namespace preservation on events (`namespace` field on `ContentEvent`, `ToolCallStartEvent`, `ToolCallEndEvent`, `ToolExtractedEvent`, `InterruptEvent`, `StateUpdateEvent`, `UsageEvent`)
+- `agent_name` field on `ContentEvent`, extracted from `lc_agent_name` metadata in messages mode (for deep agent subagents)
+- `CustomEvent` for data emitted via `get_stream_writer()` (`stream_mode="custom"`)
+- `stream_mode="custom"` support in single and multi-mode parsing
+
+## [0.1.4] - 2026-02-06
+
+### Added
+- `context_parts` parameter on `prepare_agent_input()` for prepending context lines (e.g., timestamp, working directory) to user messages
+
+## [0.1.3] - 2026-02-09
+
+### Fixed
+- Handle multi-element interrupt tuples from LangGraph subgraphs
+- Aggregate `action_requests` and `review_configs` across all Interrupt objects in a tuple
+
+## [0.1.2] - 2026-02-08
+
+### Added
+- Subgraph namespace stripping for `subgraphs=True` streams
+- Automatic handling of single-mode `(namespace, data)` and multi-mode `(namespace, mode, data)` chunk formats
+- All parent and subgraph chunks processed uniformly with namespace stripped
+
+## [0.1.1] - 2026-02-07
+
+### Added
+- Dual stream mode support (`stream_mode=["updates", "messages"]`) with automatic deduplication
+- Auto-detection mode (`stream_mode="auto"`) that inspects the first chunk
+- `MessagesHandler` for token-level content streaming from `stream_mode="messages"`
+- `UsageEvent` for token usage metadata from AIMessage `usage_metadata`
+- `DisplayInlineExtractor` for extracting inline display artifacts
+- Event serialization helpers (`InterruptEvent.build_decisions()`, `InterruptEvent.create_resume()`)
+
+### Changed
+- `stream_mode` is now a constructor parameter on `StreamParser` (moved from `parse()`/`aparse()`)
+- `UpdatesHandler` accepts `suppress_content` flag for dual-mode deduplication
+
+## [0.1.0] - 2026-02-01
+
+### Initial Release
+
+- Add `StreamParser` for parsing LangGraph stream outputs into typed events
+- Add typed event classes: `ContentEvent`, `ToolCallStartEvent`, `ToolCallEndEvent`, `ToolExtractedEvent`, `InterruptEvent`, `StateUpdateEvent`, `CompleteEvent`, `ErrorEvent`
+- Add tool lifecycle tracking (start → end)
+- Add extensible extractor system with built-in `ThinkToolExtractor` and `TodoExtractor`
+- Add interrupt handling with `create_resume_input()` and `prepare_agent_input()`
+- Add async support via `aparse()`
+- Add legacy dict-based API for backward compatibility (`stream_graph_updates`, `resume_graph_from_interrupt`)
+
+### Display Adapters
+
+- Add `BaseAdapter` abstract class for building custom display adapters
+- Add `PrintAdapter` for plain text output in any Python environment
+- Add `CLIAdapter` for styled terminal output with ANSI colors and spinner animation
+- Add `JupyterDisplay` for rich notebook display with live updates
+- Add configurable `reflection_types` and `todo_types` for custom tool rendering
+- Add `**stream_kwargs` pass-through to `graph.stream()`

{langgraph_stream_parser-0.1.7 → langgraph_stream_parser-0.2.0}/LICENSE

@@ -1,21 +1,21 @@
-MIT License
-
-Copyright (c) 2026 Kedar Dabhadkar
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.
+MIT License
+
+Copyright (c) 2026 Kedar Dabhadkar
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.