andreferraro/TextNLPClassifierApp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
main
TextNLPClassifierApp/graphify-out/cache/ast/v0.9.47-s2/aa17b59997ab6bb549a918fb1026ddf5a5c5143c75c234895dc32447e2c5b85a.json
T
andreferraro 6a45368cb0 feat(extractor): implement multi-engine article content extractor 
- Added scripts/extract_article_contents.py for batch scraping with stealth Foxcape and triple extraction (Trafilatura, Newspaper4k, Readability)
- Created unit, integration, and E2E test suite in tests/test_extract_article_contents.py (90/90 passing)
- Updated specs/003-article-content-extractor and README.md with usage documentation and CLI contracts
- Passed ruff linting/formatting and mypy type checking cleanly
2026-08-20 19:22:20 -03:00

1 line
45 KiB
JSON
Raw Permalink Blame History

{"nodes": [{"id": "$graphify-root$_scripts_extract_google_news_py", "label": "extract_google_news.py", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L1"}, {"id": "$graphify-root$_scripts_extract_google_news_searchquery", "label": "SearchQuery", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L32", "_callable": true, "_callable_class": true}, {"id": "$graphify-root$_scripts_extract_google_news_searchquery_post_init", "label": ".__post_init__()", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L40", "_callable": true}, {"id": "$graphify-root$_scripts_extract_google_news_searchquery_clean_keyword", "label": ".clean_keyword()", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L51", "_callable": true}, {"id": "$graphify-root$_scripts_extract_google_news_searchquery_clean_language", "label": ".clean_language()", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L55", "_callable": true}, {"id": "$graphify-root$_scripts_extract_google_news_searchquery_clean_locale", "label": ".clean_locale()", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L59", "_callable": true}, {"id": "$graphify-root$_scripts_extract_google_news_newsarticle", "label": "NewsArticle", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L64", "_callable": true, "_callable_class": true}, {"id": "$graphify-root$_scripts_extract_google_news_newsarticle_to_dict", "label": ".to_dict()", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L73", "_callable": true}, {"id": "any", "label": "Any", "file_type": "code", "source_file": "", "source_location": "", "origin_file": "$graphify-root$/scripts/extract_google_news.py"}, {"id": "$graphify-root$_scripts_extract_google_news_extractionresult", "label": "ExtractionResult", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L84", "_callable": true, "_callable_class": true}, {"id": "$graphify-root$_scripts_extract_google_news_extractionresult_to_dict", "label": ".to_dict()", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L95", "_callable": true}, {"id": "$graphify-root$_scripts_extract_google_news_get_hl_gl_ceid", "label": "get_hl_gl_ceid()", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L107", "_callable": true}, {"id": "$graphify-root$_scripts_extract_google_news_normalize_text_for_comparison", "label": "_normalize_text_for_comparison()", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L151", "_callable": true}, {"id": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "label": "parse_google_news_rss()", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L156", "_callable": true}, {"id": "$graphify-root$_scripts_extract_google_news_resolve_article_url", "label": "resolve_article_url()", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L207", "_callable": true}, {"id": "$graphify-root$_scripts_extract_google_news_resolve_articles_urls", "label": "resolve_articles_urls()", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L220", "_callable": true}, {"id": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "label": "_fetch_rss_content()", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L253", "_callable": true}, {"id": "$graphify-root$_scripts_extract_google_news_extract_google_news", "label": "extract_google_news()", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L297", "_callable": true}, {"id": "$graphify-root$_scripts_extract_google_news_build_parser", "label": "build_parser()", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L336", "_callable": true}, {"id": "argumentparser", "label": "ArgumentParser", "file_type": "code", "source_file": "", "source_location": "", "origin_file": "$graphify-root$/scripts/extract_google_news.py"}, {"id": "$graphify-root$_scripts_extract_google_news_main", "label": "main()", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L403", "_callable": true}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_33", "label": "Value Object com par\u00e2metros de busca validados.", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L33"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_65", "label": "Entidade que representa uma not\u00edcia extra\u00edda.", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L65"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_85", "label": "Resultado consolidado da extra\u00e7\u00e3o.", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L85"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_108", "label": "Mapeia idioma e locale para os par\u00e2metros hl, gl e ceid do Google News.", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L108"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_152", "label": "Remove pontua\u00e7\u00e3o e espa\u00e7os extras para compara\u00e7\u00e3o de redund\u00e2ncia.", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L152"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_157", "label": "Parseia o XML do RSS do Google News e extrai os itens estruturados.", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L157"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_208", "label": "Resolve a URL intermedi\u00e1ria do Google News para a URL real do ve\u00edculo.", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L208"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_223", "label": "Resolve em paralelo as URLs intermedi\u00e1rias do Google News para os links finais\u2026", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L223"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_254", "label": "Realiza a requisi\u00e7\u00e3o ao feed RSS utilizando a biblioteca Foxcape como motor\u2026", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L254"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_300", "label": "Orquestra a consulta ao Google News e retorna o resultado consolidado com URLs\u2026", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L300"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_337", "label": "Cria e configura o parser de argumentos CLI.", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L337"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_404", "label": "Ponto de entrada do script CLI.", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L404"}], "edges": [{"source": "$graphify-root$_scripts_extract_google_news_py", "target": "argparse", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L13", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "concurrent_futures", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L14", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "json", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L15", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "re", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L16", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "sys", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L17", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "urllib_error", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L18", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "urllib_parse", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L19", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "urllib_request", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L20", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "dataclasses", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L21", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "datetime", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L22", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "pathlib", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L23", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "typing", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L24", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "bs4", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L26", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "foxcape", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L27", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "googlenewsdecoder", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L28", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "$graphify-root$_scripts_extract_google_news_searchquery", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L32", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_searchquery", "target": "$graphify-root$_scripts_extract_google_news_searchquery_post_init", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L40", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_searchquery", "target": "$graphify-root$_scripts_extract_google_news_searchquery_clean_keyword", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L51", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_searchquery", "target": "$graphify-root$_scripts_extract_google_news_searchquery_clean_language", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L55", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_searchquery", "target": "$graphify-root$_scripts_extract_google_news_searchquery_clean_locale", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L59", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "$graphify-root$_scripts_extract_google_news_newsarticle", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L64", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_newsarticle", "target": "$graphify-root$_scripts_extract_google_news_newsarticle_to_dict", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L73", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_newsarticle_to_dict", "target": "any", "relation": "references", "context": "generic_arg", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L73", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "$graphify-root$_scripts_extract_google_news_extractionresult", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L84", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_extractionresult", "target": "$graphify-root$_scripts_extract_google_news_extractionresult_to_dict", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L95", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_extractionresult_to_dict", "target": "any", "relation": "references", "context": "generic_arg", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L95", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "$graphify-root$_scripts_extract_google_news_get_hl_gl_ceid", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L107", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "$graphify-root$_scripts_extract_google_news_normalize_text_for_comparison", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L151", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L156", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "target": "$graphify-root$_scripts_extract_google_news_newsarticle", "relation": "references", "context": "generic_arg", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L156", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "$graphify-root$_scripts_extract_google_news_resolve_article_url", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L207", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "$graphify-root$_scripts_extract_google_news_resolve_articles_urls", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L220", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_resolve_articles_urls", "target": "$graphify-root$_scripts_extract_google_news_newsarticle", "relation": "references", "context": "generic_arg", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L220", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_resolve_articles_urls", "target": "$graphify-root$_scripts_extract_google_news_newsarticle", "relation": "references", "context": "generic_arg", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L220", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L253", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "$graphify-root$_scripts_extract_google_news_extract_google_news", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L297", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_extract_google_news", "target": "$graphify-root$_scripts_extract_google_news_searchquery", "relation": "references", "context": "parameter_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L297", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_extract_google_news", "target": "$graphify-root$_scripts_extract_google_news_extractionresult", "relation": "references", "context": "return_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L297", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "$graphify-root$_scripts_extract_google_news_build_parser", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L336", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_build_parser", "target": "argumentparser", "relation": "references", "context": "return_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L336", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_py", "target": "$graphify-root$_scripts_extract_google_news_main", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L403", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "target": "$graphify-root$_scripts_extract_google_news_normalize_text_for_comparison", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L188", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "target": "$graphify-root$_scripts_extract_google_news_newsarticle", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L195", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_extract_google_news", "target": "$graphify-root$_scripts_extract_google_news_get_hl_gl_ceid", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L305", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_extract_google_news", "target": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L314", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_extract_google_news", "target": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L318", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_extract_google_news", "target": "$graphify-root$_scripts_extract_google_news_resolve_articles_urls", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L323", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_extract_google_news", "target": "$graphify-root$_scripts_extract_google_news_extractionresult", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L325", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_main", "target": "$graphify-root$_scripts_extract_google_news_build_parser", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L416", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_main", "target": "$graphify-root$_scripts_extract_google_news_searchquery", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L420", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_main", "target": "$graphify-root$_scripts_extract_google_news_extract_google_news", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L435", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_33", "target": "$graphify-root$_scripts_extract_google_news_searchquery", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L33", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_65", "target": "$graphify-root$_scripts_extract_google_news_newsarticle", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L65", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_85", "target": "$graphify-root$_scripts_extract_google_news_extractionresult", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L85", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_108", "target": "$graphify-root$_scripts_extract_google_news_get_hl_gl_ceid", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L108", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_152", "target": "$graphify-root$_scripts_extract_google_news_normalize_text_for_comparison", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L152", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_157", "target": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L157", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_208", "target": "$graphify-root$_scripts_extract_google_news_resolve_article_url", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L208", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_223", "target": "$graphify-root$_scripts_extract_google_news_resolve_articles_urls", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L223", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_254", "target": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L254", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_300", "target": "$graphify-root$_scripts_extract_google_news_extract_google_news", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L300", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_337", "target": "$graphify-root$_scripts_extract_google_news_build_parser", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L337", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_404", "target": "$graphify-root$_scripts_extract_google_news_main", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L404", "weight": 1.0}], "raw_calls": [{"caller_nid": "$graphify-root$_scripts_extract_google_news_searchquery_post_init", "callee": "strip", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L41", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_searchquery_post_init", "callee": "ValueError", "is_member_call": false, "source_file": "scripts/extract_google_news.py", "source_location": "L42", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_searchquery_post_init", "callee": "strip", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L44", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_searchquery_post_init", "callee": "ValueError", "is_member_call": false, "source_file": "scripts/extract_google_news.py", "source_location": "L45", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_searchquery_post_init", "callee": "ValueError", "is_member_call": false, "source_file": "scripts/extract_google_news.py", "source_location": "L48", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_searchquery_clean_keyword", "callee": "strip", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L52", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_searchquery_clean_language", "callee": "lower", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L56", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_searchquery_clean_language", "callee": "strip", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L56", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_searchquery_clean_locale", "callee": "upper", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L60", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_searchquery_clean_locale", "callee": "strip", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L60", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_extractionresult_to_dict", "callee": "to_dict", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L103", "receiver": "item"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_get_hl_gl_ceid", "callee": "replace", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L111", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_get_hl_gl_ceid", "callee": "lower", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L111", "receiver": "lang_raw"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_get_hl_gl_ceid", "callee": "split", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L134", "receiver": "lang_clean"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_get_hl_gl_ceid", "callee": "upper", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L136", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_get_hl_gl_ceid", "callee": "upper", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L136", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_get_hl_gl_ceid", "callee": "upper", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L138", "receiver": "lang_clean"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_get_hl_gl_ceid", "callee": "upper", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L141", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_get_hl_gl_ceid", "callee": "strip", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L141", "receiver": "locale"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_normalize_text_for_comparison", "callee": "strip", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L153", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_normalize_text_for_comparison", "callee": "lower", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L153", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_normalize_text_for_comparison", "callee": "sub", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L153", "receiver": "re"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "callee": "BeautifulSoup", "is_member_call": false, "source_file": "scripts/extract_google_news.py", "source_location": "L161", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "callee": "BeautifulSoup", "is_member_call": false, "source_file": "scripts/extract_google_news.py", "source_location": "L163", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "callee": "find_all", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L165", "receiver": "soup"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "callee": "find", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L172", "receiver": "item"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "callee": "find", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L173", "receiver": "item"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "callee": "find", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L174", "receiver": "item"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "callee": "find", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L175", "receiver": "item"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "callee": "get_text", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L177", "receiver": "title_tag"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "callee": "get_text", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L178", "receiver": "link_tag"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "callee": "get_text", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L179", "receiver": "pubdate_tag"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "callee": "get_text", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L180", "receiver": "desc_tag"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "callee": "BeautifulSoup", "is_member_call": false, "source_file": "scripts/extract_google_news.py", "source_location": "L185", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "callee": "get_text", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L186", "receiver": "desc_soup"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_parse_google_news_rss", "callee": "append", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L194", "receiver": "articles"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_resolve_article_url", "callee": "gnewsdecoder", "is_member_call": false, "source_file": "scripts/extract_google_news.py", "source_location": "L212", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_resolve_article_url", "callee": "dict", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_google_news.py", "source_location": "L213"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_resolve_article_url", "callee": "get", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L213", "receiver": "res"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_resolve_article_url", "callee": "get", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L213", "receiver": "res"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_resolve_article_url", "callee": "ValueError", "is_member_call": false, "indirect": true, "context": "collection", "source_file": "scripts/extract_google_news.py", "source_location": "L215"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_resolve_article_url", "callee": "OSError", "is_member_call": false, "indirect": true, "context": "collection", "source_file": "scripts/extract_google_news.py", "source_location": "L215"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_resolve_article_url", "callee": "RuntimeError", "is_member_call": false, "indirect": true, "context": "collection", "source_file": "scripts/extract_google_news.py", "source_location": "L215"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_resolve_article_url", "callee": "TypeError", "is_member_call": false, "indirect": true, "context": "collection", "source_file": "scripts/extract_google_news.py", "source_location": "L215"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_resolve_articles_urls", "callee": "write", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L228", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_resolve_articles_urls", "callee": "ThreadPoolExecutor", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L241", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_resolve_articles_urls", "callee": "_resolve_single", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_google_news.py", "source_location": "L242"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_resolve_articles_urls", "callee": "write", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L246", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "callee": "FoxcapeConfig", "is_member_call": false, "source_file": "scripts/extract_google_news.py", "source_location": "L258", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "callee": "fetch", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L263", "receiver": "Foxcape"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "callee": "RuntimeError", "is_member_call": false, "source_file": "scripts/extract_google_news.py", "source_location": "L271", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "callee": "RuntimeError", "is_member_call": false, "indirect": true, "context": "collection", "source_file": "scripts/extract_google_news.py", "source_location": "L272"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "callee": "OSError", "is_member_call": false, "indirect": true, "context": "collection", "source_file": "scripts/extract_google_news.py", "source_location": "L272"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "callee": "TimeoutError", "is_member_call": false, "indirect": true, "context": "collection", "source_file": "scripts/extract_google_news.py", "source_location": "L272"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "callee": "ValueError", "is_member_call": false, "indirect": true, "context": "collection", "source_file": "scripts/extract_google_news.py", "source_location": "L272"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "callee": "urlopen", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L284", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "callee": "decode", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L285", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "callee": "read", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L285", "receiver": "response"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "callee": "OSError", "is_member_call": false, "indirect": true, "context": "collection", "source_file": "scripts/extract_google_news.py", "source_location": "L289"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "callee": "TimeoutError", "is_member_call": false, "indirect": true, "context": "collection", "source_file": "scripts/extract_google_news.py", "source_location": "L290"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "callee": "RuntimeError", "is_member_call": false, "source_file": "scripts/extract_google_news.py", "source_location": "L292", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_extract_google_news", "callee": "quote_plus", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L304", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_extract_google_news", "callee": "write", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L309", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_extract_google_news", "callee": "write", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L316", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_extract_google_news", "callee": "write", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L320", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_extract_google_news", "callee": "isoformat", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L331", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_extract_google_news", "callee": "now", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L331", "receiver": "datetime"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_build_parser", "callee": "ArgumentParser", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L338", "receiver": "argparse"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_build_parser", "callee": "add_argument", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L342", "receiver": "parser"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_build_parser", "callee": "add_argument", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L350", "receiver": "parser"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_build_parser", "callee": "add_argument", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L358", "receiver": "parser"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_build_parser", "callee": "add_argument", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L365", "receiver": "parser"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_build_parser", "callee": "int", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_google_news.py", "source_location": "L369"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_build_parser", "callee": "add_argument", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L373", "receiver": "parser"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_build_parser", "callee": "add_argument", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L380", "receiver": "parser"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_build_parser", "callee": "add_argument", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L385", "receiver": "parser"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_build_parser", "callee": "add_argument", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L392", "receiver": "parser"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "reconfigure", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L407", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "reconfigure", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L412", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "parse_args", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L419", "receiver": "parser"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "ValueError", "is_member_call": false, "indirect": true, "context": "collection", "source_file": "scripts/extract_google_news.py", "source_location": "L426"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "write", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L427", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "int", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_google_news.py", "source_location": "L430"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "silent", "is_member_call": false, "indirect": true, "context": "getattr", "source_file": "scripts/extract_google_news.py", "source_location": "L432"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "dumps", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L436", "receiver": "json"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "to_dict", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L437", "receiver": "result"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "Path", "is_member_call": false, "source_file": "scripts/extract_google_news.py", "source_location": "L443", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "mkdir", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L444", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "write_text", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L445", "receiver": "out_path"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "write", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L447", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "write", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L451", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "write", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L455", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "write", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L458", "receiver": null}]}
Reference in New Issue View Git Blame Copy Permalink