Modernize crawley to Python 3 (asyncio + httpx)#20
Merged
Conversation
Full port of the legacy Python 2 framework to a modern Python 3 (3.9+) crawling/scraping library, keeping the declarative crawler/scraper API. Core - Replace eventlet green pools with an asyncio-based concurrency model (AsyncPool bounded by a semaphore); crawlers are now coroutines with a synchronous `run()` convenience wrapper. - Rewrite the HTTP layer on top of httpx.AsyncClient: cookies, proxies, timeouts, retries and randomized request delays. - Extractors: XPath and PyQuery ported, new CSSExtractor (cssselect). - Logging instead of print; type-friendly, lazy public API in crawley/__init__. Persistence - Relational layer rewritten on SQLAlchemy 2.x (declarative Entity/UrlEntity, Field/Unicode shims, connectors for sqlite/postgres/mysql/oracle), replacing the dead elixir + SQLAlchemy 0.7.8 stack. - Document storages (JSON/CSV/XML) modernized; Mongo on current pymongo and CouchDB via a small httpx client (the couchdb lib is unmaintained on py3). Framework - CLI manager (startproject/run/syncdb/migratedb/shell/browser) and project types ported; DSL parser/compiler ported to py3 + SQLAlchemy. - GUI scraping browser ported from PyQt4/QtWebKit to PySide6/QtWebEngine (optional extra, lazily imported). - smtp sender rewritten with email.message.EmailMessage. Packaging & tooling - pyproject.toml (PEP 621) with optional extras [sql], [mongo], [gui], [shell], [http2], [dev]; console_scripts entry point. - Remove setup.py/setup.cfg/requirements.txt/.travis.yml and other legacy build scripts. - GitHub Actions CI (py 3.9-3.12, ruff + pytest) replacing Travis. Tests - New hermetic pytest suite (31 tests) running against a local HTTP server: extractors, matchers, url finder, http layer, async crawl flow, scrapers, persistence and CLI integration. No network access required.
Modern scraping API (crawley.scraping) - New high-level, ergonomic scraping interface (parsel / requests-html flavoured) on top of httpx + lxml: Document / Element with css(), css_first(), xpath(), links(), title and declarative extract(). - CSS selectors accept ::text and ::attr(name) suffixes; links are resolved to absolute urls. - Network helpers: fetch() / afetch() / afetch_all() (concurrent) and a one-call scrape(url, rules); plus parse() for raw html. - Response gains .doc plus .css / .css_first / .extract shortcuts so the same API is available inside a crawler's scrape(). - Exposed from the top-level crawley package. Tests (31 -> 133, ~92% core coverage) - Hermetic pytest suite against a local HTTP server, covering: the scraping API, extractors, url matchers/finder, async HTTP layer, cookies, the AsyncPool, crawler flow + events/login/error handling, Fast/OffLine/Smart crawlers and scrapers, relational + document + NoSQL persistence (with fake drivers), the DSL parser/compiler, the INI config parser, the CLI commands (startproject/syncdb/migratedb/run), utils and the SMTP sender. Tooling - coverage config (omitting the headless-only PySide6 GUI), pytest-cov in the dev extra, CI runs with coverage reporting. - README documents the modern scraping API; new examples/modern_scraping.py.
Politeness & robustness - RetryPolicy (crawley.http.retry): retries on network errors and retryable HTTP statuses (429/5xx) with exponential backoff + jitter, honouring the Retry-After header; configurable via crawler attrs max_retries / retry_backoff / retry_statuses. - HostRateLimiter (crawley.http.throttle): per-host minimum delay between requests and optional per-host concurrency cap (crawl_delay / max_concurrency_per_host). - RobotsPolicy (crawley.http.robots): fetches and caches robots.txt per host, skips disallowed urls (new on_robots_blocked event) and applies the Crawl-delay directive. Opt-in via respect_robots. - Wired into RequestManager (retry + throttle) and BaseCrawler; FastCrawler reuses the same policies. Documentation (MkDocs Material + mkdocstrings) - mkdocs.yml and docs/: home, installation, scraping API, crawlers & scrapers, politeness, persistence, CLI & framework, and an auto-generated API reference. Builds clean with `mkdocs build --strict`. - New docs extra; docs deploy workflow; README updated (politeness section, docs pointer); .gitignore refreshed (site/, caches, coverage). Tests (133 -> 154) - test_retry, test_throttle, test_robots covering the new policies and their integration; conftest server now serves robots.txt, a flaky endpoint and a 503-with-Retry-After endpoint. Core coverage ~92%.
Examples - New examples/ scripts that double as documentation, each parametrized by a base_url (real site by default) so they run standalone and under tests: 01 scraping API, 02 crawler with pagination, 03 polite crawler (robots/rate-limit/retries), 04 persistence to JSON, 05 concurrent fetch. - examples/README.md indexes them; README and docs link to them. Tests (154 -> 166) - tests/test_examples.py loads every example by path and runs it against a new hermetic `quotes_server` fixture (a paginated quotes site), asserting the scraped results; plus an import smoke test for each. Housekeeping - Remove the legacy Sphinx `doc/` tree (superseded by the MkDocs `docs/`). - ruff: exclude the legacy sample projects under examples/. `ruff check .`, `pytest` and `mkdocs build --strict` all pass.
Crawler - Visited-url de-duplication (unique_urls, on by default): skips already-seen urls, avoiding redundant fetches and preventing crawl loops even with unbounded max_depth. The seen-set resets on each start(). Library hygiene - Attach a NullHandler to the "crawley" logger so importing the package never emits "No handlers could be found" warnings; applications stay in control of logging configuration. Docs & meta - Document unique_urls in the crawler reference. - Add CHANGELOG.md summarizing the 0.3.0 modernization. Tests (166 -> 169) - test_dedupe: de-dup on/off and a cycle (loop-a <-> loop-b) that only terminates thanks to de-duplication; new loop endpoints in the test server.
Typing - Type hints on the public-facing modules (scraping, extractors, http response/retry/throttle/robots, toolbox) using PEP 563 deferred annotations. - PEP 561 `py.typed` marker shipped in the wheel so downstream code gets the types. - mypy config ([tool.mypy], follow_imports=silent over the typed modules) and a `mypy` step in CI; passes clean. Project - Add the GPL-3.0 LICENSE text. - README points to the published docs site; CHANGELOG notes the typing/LICENSE additions.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Full port of the legacy Python 2 framework to a modern Python 3 (3.9+) crawling / scraping library, keeping the declarative crawler/scraper API.
Highlights
Core
eventletgreen pools to asyncio (AsyncPool); crawlers are coroutines with a synchronousrun()wrapper.AsyncClient): cookies, proxies, timeouts.Modern scraping API (
crawley.scraping)fetch/afetch/afetch_all/scrape/parseandDocument/Elementwith CSS selectors (::text,::attr(name)), XPath,links()and declarativeextract().css,css_first,extract,doc) available on the crawlerresponse.Politeness & robustness
robots.txt(respect_robots,on_robots_blocked,Crawl-delay).crawl_delay,max_concurrency_per_host).Retry-After(max_retries,retry_backoff,retry_statuses).unique_urls) preventing redundant fetches and crawl loops.Persistence
elixir+ SQLAlchemy 0.7.8 to SQLAlchemy 2.x; MongoDB on currentpymongo; CouchDB via a small httpx client; JSON / XML / CSV document exports.Framework
startproject/run/syncdb/migratedb/shell/browser), project types and the DSL parser/compiler ported to Python 3.email.message.EmailMessage.Packaging & tooling
pyproject.toml(PEP 621) with optional extras (sql,mongo,gui,shell,http2,dev,docs) and acrawleyconsole script.py.typed; mypy in CI.setup.py/setup.cfg/requirements.txt/.travis.yml, the deadeventlet/elixirdeps and the old Sphinxdoc/tree.LICENSEfile (GPL-3.0).Docs, examples & tests
docs/), published via GitHub Pages.examples/.pytestsuite (~170 tests, ~92% core coverage) running against a local server — no network access required.See
CHANGELOG.mdfor the full list.https://claude.ai/code/session_0159tWs4APGGyjhKYQGYiFWY
Generated by Claude Code