release: promote pre/beta (2.2.0-beta.6) to main - #1128
Merged
Conversation
Add SmartScraperMultiBatchGraph that uses the OpenAI Batch API for LLM
calls, providing ~50% cost savings when real-time results aren't needed.
Key features:
- SmartScraperMultiBatchGraph: 3-phase pipeline (fetch/parse → batch
submit → merge) that separates HTML fetching from LLM generation
- BatchGenerateAnswerNode: collects prompts from all URLs and submits
them as a single OpenAI Batch API request
- utils/batch_api.py: helpers for creating, polling, and retrieving
batch results with doc_id → URL mapping
- Per-document error handling: partial failures don't break the batch
- Configurable polling interval and max wait time
- OpenAI-only validation (rejects non-OpenAI providers gracefully)
- Results sorted by custom_id for consistent ordering
- 18 unit tests with 100% pass rate
Usage:
graph = SmartScraperMultiBatchGraph(
prompt='Extract key points',
source=['https://url1.com', 'https://url2.com'],
config={'llm': {'model': 'openai/gpt-4o-mini'}}
)
result = graph.run()
Closes #1036
## [1.60.0-beta.2](v1.60.0-beta.1...v1.60.0-beta.2) (2026-02-24) ### Features * add OpenAI Batch API support for SmartScraperMultiGraph ([#1036](#1036)) ([9d4eba1](9d4eba1))
- Replace removed ScrapeGraph with SmartScraperGraph in scrape_graph_test.py - Replace renamed convert_to_csv/convert_to_json with export_to_csv/export_to_json in xml_scraper_openai_test.py
## [1.60.0-beta.3](v1.60.0-beta.2...v1.60.0-beta.3) (2026-03-15) ### Bug Fixes * update broken test imports to match current API ([536e5ad](536e5ad))
Resolved conflicts in CHANGELOG.md, pyproject.toml, xml_scraper_openai_test.py, and accepted deletion of scrape_graph_test.py from main. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## [1.76.0-beta.1](v1.75.1...v1.76.0-beta.1) (2026-04-07) ### Features * add OpenAI Batch API support for SmartScraperMultiGraph ([#1036](#1036)) ([9d4eba1](9d4eba1)) ### Bug Fixes * update broken test imports to match current API ([536e5ad](536e5ad)) ### CI * **release:** 1.60.0-beta.2 [skip ci] ([54d1473](54d1473)), closes [#1036](#1036) * **release:** 1.60.0-beta.3 [skip ci] ([637c696](637c696)) * reduce GitHub Actions costs by ~85% on PRs ([403080a](403080a))
Resolve conflicts in CHANGELOG.md and pyproject.toml by keeping main's version. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
## [2.1.0-beta.1](v2.0.0...v2.1.0-beta.1) (2026-04-19) ### Features * add OpenAI Batch API support for SmartScraperMultiGraph ([#1036](#1036)) ([9d4eba1](9d4eba1)) ### Bug Fixes * update broken test imports to match current API ([536e5ad](536e5ad)) ### CI * **release:** 1.60.0-beta.2 [skip ci] ([54d1473](54d1473)), closes [#1036](#1036) * **release:** 1.60.0-beta.3 [skip ci] ([637c696](637c696)) * **release:** 1.76.0-beta.1 [skip ci] ([35ec272](35ec272)), closes [#1036](#1036) [#1036](#1036)
langchain 1.x removed langchain.prompts; import from langchain_core to fix ModuleNotFoundError causing all test collection to fail.
Resolves conflicts in CHANGELOG.md (combined entries) and pyproject.toml (keep 2.1.1 from main; next release will bump).
## [2.2.0-beta.1](v2.1.1...v2.2.0-beta.1) (2026-05-16) ### Features * add OpenAI Batch API support for SmartScraperMultiGraph ([#1036](#1036)) ([9d4eba1](9d4eba1)) ### Bug Fixes * update broken test imports to match current API ([536e5ad](536e5ad)) * **batch:** use langchain_core.prompts for PromptTemplate import ([24127da](24127da)) ### CI * **release:** 1.60.0-beta.2 [skip ci] ([54d1473](54d1473)), closes [#1036](#1036) * **release:** 1.60.0-beta.3 [skip ci] ([637c696](637c696)) * **release:** 1.76.0-beta.1 [skip ci] ([35ec272](35ec272)), closes [#1036](#1036) [#1036](#1036) * **release:** 2.1.0-beta.1 [skip ci] ([a2ea9eb](a2ea9eb)), closes [#1036](#1036) [#1036](#1036) [#1036](#1036) [#1036](#1036)
- Add MiniMax-M3 to the model selection list (524288 context window) - Set MiniMax-M3 as the new default model (first in the dict) - Retain MiniMax-M2.7 and MiniMax-M2.7-highspeed as legacy options - Remove deprecated older versions (M2.5 / M2.5-highspeed / M2 / M1 / M1-40k) - Update unit tests to reflect the new default and removed models
## [2.2.0-beta.2](v2.2.0-beta.1...v2.2.0-beta.2) (2026-06-01) ### Features * upgrade MiniMax default model to M3 ([#1080](#1080)) ([1b16c26](1b16c26))
Resolve conflicts: - pyproject.toml: keep beta version 2.2.0b2 - CHANGELOG.md: keep both 2.2.0-beta.2 and 2.1.2 release entries - README.md: adopt main's standardized relative doc links (#1074) - docs/italian.md: adopt main's standardized version for consistency Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
## [2.2.0-beta.3](v2.2.0-beta.2...v2.2.0-beta.3) (2026-06-01) ### Bug Fixes * **nodes:** update outdated ChatOllama import path to langchain_ollama ([#1076](#1076)) ([e6054cb](e6054cb)) ### Docs * 📚 Standardize and fix links across translated READMEs ([#1074](#1074)) ([458d36a](458d36a)) ### CI * **release:** 2.1.2 [skip ci] ([210c992](210c992)), closes [#1076](#1076) [#1074](#1074)
…eek (#1085) The default GenerateAnswerNode format_instructions show the expected shape as {{"content": ...}} (LangChain's escaped braces). Strongly instruction-following models emit single braces, but some models (notably DeepSeek) copy the doubled braces verbatim, yielding {{"content": ...}} which JsonOutputParser rejects with 'Invalid json output'. Add TolerantJsonOutputParser, a JsonOutputParser subclass that retries once with a single layer of wrapping braces removed, only on the parse-failure path. Behaviour is unchanged for any model already returning valid JSON. Use it in the schema-less branch of GenerateAnswerNode. Adds unit tests.
## [2.2.0-beta.4](v2.2.0-beta.3...v2.2.0-beta.4) (2026-06-11) ### Bug Fixes * **nodes:** tolerate doubled-brace JSON output from models like DeepSeek ([#1085](#1085)) ([aaa5d2c](aaa5d2c))
…1089) Remove BaseLoader inheritance from ChromiumLoader and use lazy imports for PyPDFLoader and AsyncChromiumLoader to avoid triggering the sentence_transformers -> torchcodec -> FFmpeg native DLL loading chain at import time, which crashes on systems where FFmpeg DLLs are not available. Also add torchcodec mock to conftest.py for the test suite. Changes: - chromium.py: remove BaseLoader import/inheritance, add load()/aload() - docloaders/__init__.py: lazy __getattr__ for ChromiumLoader, PlasmateLoader - fetch_node.py: lazy PyPDFLoader import - robots_node.py: lazy AsyncChromiumLoader import - conftest.py: torchcodec module mock
## [2.2.0-beta.5](v2.2.0-beta.4...v2.2.0-beta.5) (2026-06-23) ### Bug Fixes * lazy imports to prevent torchcodec FFmpeg DLL crash on Windows ([#1089](#1089)) ([2f55377](2f55377))
Document the workflow AI coding agents must follow in this repository: branch from and open every PR against pre/beta (never main), uv-based setup, the lint/type-check/test commands, Conventional Commits for semantic-release, files that must not be hand-edited (CHANGELOG.md, version in pyproject.toml, build artifacts) and the repository layout. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
docs: add AGENTS.md with contribution rules for AI agents
dosubot
Bot
added
size:L
Resolve the four conflicts that blocked #1128 (pre/beta -> main): - scrapegraphai/helpers/models_tokens.py: keep main's MiniMax numbers (M3 1000000, M2.7/M2.7-highspeed 204800) from #1103, which corrects the values introduced by #1080 on pre/beta. - tests/test_minimax_models.py: keep main's file — it asserts the corrected limits and adds the cost/endpoint fixtures and tests. - pyproject.toml: keep the prerelease version 2.2.0b5; semantic-release bumps it on the next release from pre/beta. - CHANGELOG.md: keep both histories, main's stable entries (2.1.3 -> 2.1.7) above the beta entries, in date order. Also drop two orphan "=======" lines left over from older bad conflict resolutions. Verified: the deterministic unit suites run by .github/workflows/test-suite.yml pass (66 passed), plus tests/test_atlascloud_model.py and tests/utils/output_parser_test.py (24 passed). Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
tests/test_atlascloud_model.py (added on main by #1104) and tests/utils/output_parser_test.py (added on pre/beta by #1085) were both introduced without being wired into the Test Suite workflow, so neither branch actually ran them. Both are mock-based and network-free. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Dependency Review✅ No vulnerabilities or license issues or OpenSSF Scorecard issues found.Snapshot WarningsEnsure that dependencies are being submitted on PR branches and consider enabling retry-on-snapshot-warnings. See the documentation for more information and troubleshooting advice. Scanned FilesNone |
## [2.2.0-beta.6](v2.2.0-beta.5...v2.2.0-beta.6) (2026-08-19) ### Bug Fixes * **graph:** expose when the 8192 token fallback was used ([470da9d](470da9d)), closes [#1121](#1121) * lazy imports to prevent torchcodec FFmpeg DLL crash on Windows ([#1089](#1089)) ([#1092](#1092)) ([e5c2a42](e5c2a42)) * pop model_tokens so it is not forwarded to the model client ([#1100](#1100)) ([d2b970c](d2b970c)) * **search:** restore SearchGraph by migrating DuckDuckGo backend to ddgs ([#1083](#1083)) ([2139e37](2139e37)), closes [#1082](#1082) [#1082](#1082) * update MiniMax model metadata and endpoints ([#1103](#1103)) ([e5f8f2b](e5f8f2b)) ### Docs * add AGENTS.md with contribution rules for AI agents ([3b2f986](3b2f986)) * **readme:** add Open Source vs Managed API comparison ([#1091](#1091)) ([ef3523b](ef3523b)) * swap Integrations infographic for new API banner; fix CTA links ([71ab440](71ab440)) ### CI * **release:** 2.1.3 [skip ci] ([cfb815f](cfb815f)), closes [#1082](#1082) [#1082](#1082) * **release:** 2.1.4 [skip ci] ([7fc9c57](7fc9c57)), closes [#1089](#1089) [#1092](#1092) * **release:** 2.1.5 [skip ci] ([c9e0bd0](c9e0bd0)), closes [#1100](#1100) [#1091](#1091) [#1095](#1095) * **release:** 2.1.6 [skip ci] ([27d9d28](27d9d28)), closes [#1103](#1103) * **release:** 2.1.7 [skip ci] ([ca112db](ca112db)), closes [#1121](#1121) * run only deterministic unit suites in Test Suite workflow ([#1095](#1095)) ([037a42e](037a42e)) * run the two new deterministic unit suites ([23e3d06](23e3d06)), closes [#1104](#1104) [#1085](#1085)
|
🎉 This PR is included in version 2.2.0-beta.6 🎉 The release is available on:
Your semantic-release bot 📦🚀 |
|
🎉 This PR is included in version 2.2.0 🎉 The release is available on:
Your semantic-release bot 📦🚀 |
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
8 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.
Promotion of
pre/betaintomain. Conflicts withmainare resolved (seeeed377cb), the branch is now conflict-free and the Test Suite is green.What lands on main
SmartScraperMultiGraph(OpenAI Batch API support for SmartScraperMultiGraph #1036 / feat: add OpenAI Batch API support for SmartScraperMultiGraph #1039) —SmartScraperMultiBatchGraph,BatchGenerateAnswerNode,utils/batch_api.py.TolerantJsonOutputParser.PromptTemplatefromlangchain_core.prompts(langchain 1.x).AGENTS.md— contribution rules for AI agents (docs: add AGENTS.md with contribution rules for AI agents #1127).tests/test_atlascloud_model.pyandtests/utils/output_parser_test.py, which were added on the two branches but never wired into the Test Suite workflow.Conflict resolution
scrapegraphai/helpers/models_tokens.py1000000, M2.7 / M2.7-highspeed204800) from #1103, which supersede the values #1080 introduced onpre/beta.tests/test_minimax_models.pypyproject.toml2.2.0b5at merge time, now2.2.0b6); semantic-release owns the bump.CHANGELOG.md=======lines left over from older bad conflict resolutions.🤖 Generated with Claude Code