andreferraro
6e3d57619b
- Add standalone CLI script scripts/extract_google_news.py for Google News RSS scraping - Integrate foxcape in headless mode as primary stealth anti-bot engine - Implement parallel article URL resolution using googlenewsdecoder and ThreadPoolExecutor - Support language and regional locale mapping (-l, --lang, --locale) - Implement real-time progress logging in stderr and --silent flag - Add unit, integration, and live E2E tests in tests/test_extract_google_news.py - Add full SpecKit documentation (specs/002-google-news-extractor/) - Create comprehensive README.md covering both NLP Classifier and Google News Extractor
1 line
36 KiB
JSON
1 line
36 KiB
JSON
{"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": "L29", "_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": "L37", "_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": "L50", "_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": "L54", "_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": "L58", "_callable": true}, {"id": "$graphify-root$_scripts_extract_google_news_newsarticle", "label": "NewsArticle", "file_type": "code", "source_file": "scripts/extract_google_news.py", "source_location": "L63", "_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": "L72", "_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": "L83", "_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": "L94", "_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": "L106", "_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": "L150", "_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": "L155", "_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": "L206", "_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": "L240", "_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": "L262", "_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": "L314", "_callable": true}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_30", "label": "Value Object com par\u00e2metros de busca validados.", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L30"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_64", "label": "Entidade que representa uma not\u00edcia extra\u00edda.", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L64"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_84", "label": "Resultado consolidado da extra\u00e7\u00e3o.", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L84"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_107", "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": "L107"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_151", "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": "L151"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_156", "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": "L156"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_207", "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": "L207"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_241", "label": "Orquestra a consulta ao Google News e retorna o resultado consolidado.", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L241"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_263", "label": "Cria e configura o parser de argumentos CLI.", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L263"}, {"id": "$graphify-root$_scripts_extract_google_news_rationale_315", "label": "Ponto de entrada do script CLI.", "file_type": "rationale", "source_file": "scripts/extract_google_news.py", "source_location": "L315"}], "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": "L12", "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": "L13", "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": "L14", "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": "L15", "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": "L16", "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": "L17", "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": "L18", "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": "L19", "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": "L20", "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": "L21", "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": "L22", "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": "L24", "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": "L25", "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": "L29", "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": "L37", "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": "L50", "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": "L54", "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": "L58", "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": "L63", "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": "L72", "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": "L72", "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": "L83", "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": "L94", "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": "L94", "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": "L106", "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": "L150", "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": "L155", "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": "L155", "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": "L206", "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": "L240", "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": "L240", "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": "L240", "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": "L262", "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": "L262", "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": "L314", "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": "L187", "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": "L194", "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": "L245", "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": "L248", "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": "L249", "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": "L251", "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": "L327", "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": "L331", "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": "L344", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_30", "target": "$graphify-root$_scripts_extract_google_news_searchquery", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L30", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_64", "target": "$graphify-root$_scripts_extract_google_news_newsarticle", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L64", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_84", "target": "$graphify-root$_scripts_extract_google_news_extractionresult", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L84", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_107", "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": "L107", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_151", "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": "L151", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_156", "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": "L156", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_207", "target": "$graphify-root$_scripts_extract_google_news_fetch_rss_content", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L207", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_241", "target": "$graphify-root$_scripts_extract_google_news_extract_google_news", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L241", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_263", "target": "$graphify-root$_scripts_extract_google_news_build_parser", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L263", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_google_news_rationale_315", "target": "$graphify-root$_scripts_extract_google_news_main", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_google_news.py", "source_location": "L315", "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": "L38", "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": "L39", "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": "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": "ValueError", "is_member_call": false, "source_file": "scripts/extract_google_news.py", "source_location": "L47", "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": "L51", "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": "L55", "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": "L55", "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": "L59", "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": "L59", "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": "L102", "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": "L110", "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": "L110", "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": "L133", "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": "L135", "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": "L135", "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": "L137", "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": "L140", "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": "L140", "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": "L152", "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": "L152", "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": "L152", "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": "L160", "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": "L162", "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": "L164", "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": "L171", "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": "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": "get_text", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L176", "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": "L177", "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": "L178", "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": "L179", "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": "L184", "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": "L185", "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": "L193", "receiver": "articles"}, {"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": "L212", "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": "L219", "receiver": null}, {"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": "L232", "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": "L233", "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": "L233", "receiver": "response"}, {"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": "L235", "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": "L244", "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": "L257", "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": "L257", "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": "L264", "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": "L268", "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": "L276", "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": "L284", "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": "L291", "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": "L295"}, {"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": "L299", "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": "L306", "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": "L318", "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": "L323", "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": "L330", "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": "L337"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "write", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L338", "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": "L341"}, {"caller_nid": "$graphify-root$_scripts_extract_google_news_main", "callee": "dumps", "is_member_call": true, "source_file": "scripts/extract_google_news.py", "source_location": "L345", "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": "L346", "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": "L352", "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": "L353", "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": "L354", "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": "L356", "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": "L360", "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": "L363", "receiver": null}]} |