andreferraro
6a45368cb0
- 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
1 line
62 KiB
JSON
1 line
62 KiB
JSON
{"nodes": [{"id": "$graphify-root$_scripts_extract_article_contents_py", "label": "extract_article_contents.py", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L1"}, {"id": "$graphify-root$_scripts_extract_article_contents_inputarticle", "label": "InputArticle", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L37", "_callable": true, "_callable_class": true}, {"id": "$graphify-root$_scripts_extract_article_contents_inputarticle_to_dict", "label": ".to_dict()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L46", "_callable": true}, {"id": "any", "label": "Any", "file_type": "code", "source_file": "", "source_location": "", "origin_file": "$graphify-root$/scripts/extract_article_contents.py"}, {"id": "$graphify-root$_scripts_extract_article_contents_trafilaturadata", "label": "TrafilaturaData", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L57", "_callable": true, "_callable_class": true}, {"id": "$graphify-root$_scripts_extract_article_contents_trafilaturadata_to_dict", "label": ".to_dict()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L71", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_newspaperdata", "label": "NewspaperData", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L87", "_callable": true, "_callable_class": true}, {"id": "$graphify-root$_scripts_extract_article_contents_newspaperdata_to_dict", "label": ".to_dict()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L101", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_readabilitydata", "label": "ReadabilityData", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L117", "_callable": true, "_callable_class": true}, {"id": "$graphify-root$_scripts_extract_article_contents_readabilitydata_to_dict", "label": ".to_dict()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L126", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_extractedarticle", "label": "ExtractedArticle", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L137", "_callable": true, "_callable_class": true}, {"id": "$graphify-root$_scripts_extract_article_contents_extractedarticle_to_dict", "label": ".to_dict()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L150", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_extractionbatchreport", "label": "ExtractionBatchReport", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L165", "_callable": true, "_callable_class": true}, {"id": "$graphify-root$_scripts_extract_article_contents_extractionbatchreport_to_dict", "label": ".to_dict()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L175", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_trafilaturaextractor", "label": "TrafilaturaExtractor", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L191", "_callable": true, "_callable_class": true}, {"id": "$graphify-root$_scripts_extract_article_contents_trafilaturaextractor_extract", "label": ".extract()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L195", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_newspaperextractor", "label": "NewspaperExtractor", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L260", "_callable": true, "_callable_class": true}, {"id": "$graphify-root$_scripts_extract_article_contents_newspaperextractor_extract", "label": ".extract()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L264", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_readabilityextractor", "label": "ReadabilityExtractor", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L315", "_callable": true, "_callable_class": true}, {"id": "$graphify-root$_scripts_extract_article_contents_readabilityextractor_extract", "label": ".extract()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L319", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_extract_all_engines", "label": "extract_all_engines()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L347", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_articlecrawler", "label": "ArticleCrawler", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L402", "_callable": true, "_callable_class": true}, {"id": "$graphify-root$_scripts_extract_article_contents_articlecrawler_init", "label": ".__init__()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L405", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_articlecrawler_start", "label": ".start()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L410", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_articlecrawler_close", "label": ".close()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L416", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_articlecrawler_enter", "label": ".__enter__()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L424", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_articlecrawler_exit", "label": ".__exit__()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L428", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_articlecrawler_crawl", "label": ".crawl()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L431", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_log_info", "label": "log_info()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L453", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_load_search_json", "label": "load_search_json()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L460", "_callable": true}, {"id": "path", "label": "Path", "file_type": "code", "source_file": "", "source_location": "", "origin_file": "$graphify-root$/scripts/extract_article_contents.py"}, {"id": "$graphify-root$_scripts_extract_article_contents_save_extracted_json", "label": "save_extracted_json()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L488", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_process_batch", "label": "process_batch()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L495", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "label": "parse_arguments()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L608", "_callable": true}, {"id": "namespace", "label": "Namespace", "file_type": "code", "source_file": "", "source_location": "", "origin_file": "$graphify-root$/scripts/extract_article_contents.py"}, {"id": "$graphify-root$_scripts_extract_article_contents_main", "label": "main()", "file_type": "code", "source_file": "scripts/extract_article_contents.py", "source_location": "L662", "_callable": true}, {"id": "$graphify-root$_scripts_extract_article_contents_rationale_38", "label": "Metadados originais da not\u00edcia contida no JSON de entrada.", "file_type": "rationale", "source_file": "scripts/extract_article_contents.py", "source_location": "L38"}, {"id": "$graphify-root$_scripts_extract_article_contents_rationale_58", "label": "Dados extra\u00eddos pelo motor Trafilatura.", "file_type": "rationale", "source_file": "scripts/extract_article_contents.py", "source_location": "L58"}, {"id": "$graphify-root$_scripts_extract_article_contents_rationale_88", "label": "Dados extra\u00eddos e enriquecidos com NLP pelo motor Newspaper4k.", "file_type": "rationale", "source_file": "scripts/extract_article_contents.py", "source_location": "L88"}, {"id": "$graphify-root$_scripts_extract_article_contents_rationale_118", "label": "Dados higienizados pelo algoritmo Readability.", "file_type": "rationale", "source_file": "scripts/extract_article_contents.py", "source_location": "L118"}, {"id": "$graphify-root$_scripts_extract_article_contents_rationale_138", "label": "Resultado consolidado da extra\u00e7\u00e3o de um artigo.", "file_type": "rationale", "source_file": "scripts/extract_article_contents.py", "source_location": "L138"}, {"id": "$graphify-root$_scripts_extract_article_contents_rationale_166", "label": "Relat\u00f3rio consolidado de sa\u00edda do processamento de um lote.", "file_type": "rationale", "source_file": "scripts/extract_article_contents.py", "source_location": "L166"}, {"id": "$graphify-root$_scripts_extract_article_contents_rationale_192", "label": "Motor de extra\u00e7\u00e3o baseado na biblioteca Trafilatura.", "file_type": "rationale", "source_file": "scripts/extract_article_contents.py", "source_location": "L192"}, {"id": "$graphify-root$_scripts_extract_article_contents_rationale_261", "label": "Motor de extra\u00e7\u00e3o baseado no Newspaper4k com NLP.", "file_type": "rationale", "source_file": "scripts/extract_article_contents.py", "source_location": "L261"}, {"id": "$graphify-root$_scripts_extract_article_contents_rationale_316", "label": "Motor de extra\u00e7\u00e3o baseado no algoritmo Readability (readability-lxml).", "file_type": "rationale", "source_file": "scripts/extract_article_contents.py", "source_location": "L316"}, {"id": "$graphify-root$_scripts_extract_article_contents_rationale_350", "label": "Executa a extra\u00e7\u00e3o simult\u00e2nea pelos tr\u00eas motores de conte\u00fado com isolamento\u2026", "file_type": "rationale", "source_file": "scripts/extract_article_contents.py", "source_location": "L350"}, {"id": "$graphify-root$_scripts_extract_article_contents_rationale_403", "label": "Gerenciador de ciclo de vida e requisi\u00e7\u00f5es via Foxcape Headless.", "file_type": "rationale", "source_file": "scripts/extract_article_contents.py", "source_location": "L403"}, {"id": "$graphify-root$_scripts_extract_article_contents_rationale_432", "label": "Navega at\u00e9 a URL, aguarda o carregamento do DOM e retorna (html, page_title,\u2026", "file_type": "rationale", "source_file": "scripts/extract_article_contents.py", "source_location": "L432"}, {"id": "$graphify-root$_scripts_extract_article_contents_rationale_454", "label": "Escreve mensagem informativa no stderr.", "file_type": "rationale", "source_file": "scripts/extract_article_contents.py", "source_location": "L454"}, {"id": "$graphify-root$_scripts_extract_article_contents_rationale_461", "label": "Carrega o arquivo JSON gerado pelo extrator de not\u00edcias.", "file_type": "rationale", "source_file": "scripts/extract_article_contents.py", "source_location": "L461"}, {"id": "$graphify-root$_scripts_extract_article_contents_rationale_489", "label": "Salva o relat\u00f3rio consolidado em formato JSON com UTF-8.", "file_type": "rationale", "source_file": "scripts/extract_article_contents.py", "source_location": "L489"}, {"id": "$graphify-root$_scripts_extract_article_contents_rationale_503", "label": "Executa o pipeline completo de extra\u00e7\u00e3o em lote para o arquivo de entrada.", "file_type": "rationale", "source_file": "scripts/extract_article_contents.py", "source_location": "L503"}], "edges": [{"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "argparse", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L13", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "json", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L14", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "logging", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L15", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "os", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L16", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "sys", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L17", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "time", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L18", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "dataclasses", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L19", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "datetime", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L20", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "pathlib", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L21", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "typing", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L22", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "bs4", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L24", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "foxcape", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L25", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "newspaper", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L26", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "readability", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L27", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "trafilatura", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L28", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_inputarticle", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L37", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_inputarticle", "target": "$graphify-root$_scripts_extract_article_contents_inputarticle_to_dict", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L46", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_inputarticle_to_dict", "target": "any", "relation": "references", "context": "generic_arg", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L46", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_trafilaturadata", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L57", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_trafilaturadata", "target": "$graphify-root$_scripts_extract_article_contents_trafilaturadata_to_dict", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L71", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_trafilaturadata_to_dict", "target": "any", "relation": "references", "context": "generic_arg", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L71", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_newspaperdata", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L87", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_newspaperdata", "target": "$graphify-root$_scripts_extract_article_contents_newspaperdata_to_dict", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L101", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_newspaperdata_to_dict", "target": "any", "relation": "references", "context": "generic_arg", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L101", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_readabilitydata", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L117", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_readabilitydata", "target": "$graphify-root$_scripts_extract_article_contents_readabilitydata_to_dict", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L126", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_readabilitydata_to_dict", "target": "any", "relation": "references", "context": "generic_arg", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L126", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_extractedarticle", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L137", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_extractedarticle", "target": "$graphify-root$_scripts_extract_article_contents_extractedarticle_to_dict", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L150", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_extractedarticle_to_dict", "target": "any", "relation": "references", "context": "generic_arg", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L150", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_extractionbatchreport", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L165", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_extractionbatchreport", "target": "$graphify-root$_scripts_extract_article_contents_extractionbatchreport_to_dict", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L175", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_extractionbatchreport_to_dict", "target": "any", "relation": "references", "context": "generic_arg", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L175", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_trafilaturaextractor", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L191", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_trafilaturaextractor", "target": "$graphify-root$_scripts_extract_article_contents_trafilaturaextractor_extract", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L195", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_trafilaturaextractor_extract", "target": "$graphify-root$_scripts_extract_article_contents_trafilaturadata", "relation": "references", "context": "return_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L195", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_newspaperextractor", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L260", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_newspaperextractor", "target": "$graphify-root$_scripts_extract_article_contents_newspaperextractor_extract", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L264", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_newspaperextractor_extract", "target": "$graphify-root$_scripts_extract_article_contents_newspaperdata", "relation": "references", "context": "return_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L264", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_readabilityextractor", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L315", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_readabilityextractor", "target": "$graphify-root$_scripts_extract_article_contents_readabilityextractor_extract", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L319", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_readabilityextractor_extract", "target": "$graphify-root$_scripts_extract_article_contents_readabilitydata", "relation": "references", "context": "return_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L319", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_extract_all_engines", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L347", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_extract_all_engines", "target": "$graphify-root$_scripts_extract_article_contents_trafilaturadata", "relation": "references", "context": "generic_arg", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L347", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_extract_all_engines", "target": "$graphify-root$_scripts_extract_article_contents_newspaperdata", "relation": "references", "context": "generic_arg", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L347", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_extract_all_engines", "target": "$graphify-root$_scripts_extract_article_contents_readabilitydata", "relation": "references", "context": "generic_arg", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L347", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_articlecrawler", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L402", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_articlecrawler", "target": "$graphify-root$_scripts_extract_article_contents_articlecrawler_init", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L405", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_articlecrawler", "target": "$graphify-root$_scripts_extract_article_contents_articlecrawler_start", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L410", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_articlecrawler", "target": "$graphify-root$_scripts_extract_article_contents_articlecrawler_close", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L416", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_articlecrawler", "target": "$graphify-root$_scripts_extract_article_contents_articlecrawler_enter", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L424", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_articlecrawler_enter", "target": "$graphify-root$_scripts_extract_article_contents_articlecrawler", "relation": "references", "context": "return_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L424", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_articlecrawler", "target": "$graphify-root$_scripts_extract_article_contents_articlecrawler_exit", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L428", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_articlecrawler_exit", "target": "any", "relation": "references", "context": "parameter_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L428", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_articlecrawler_exit", "target": "any", "relation": "references", "context": "parameter_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L428", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_articlecrawler_exit", "target": "any", "relation": "references", "context": "parameter_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L428", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_articlecrawler", "target": "$graphify-root$_scripts_extract_article_contents_articlecrawler_crawl", "relation": "method", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L431", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_log_info", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L453", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_load_search_json", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L460", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_load_search_json", "target": "path", "relation": "references", "context": "parameter_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L460", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_load_search_json", "target": "$graphify-root$_scripts_extract_article_contents_inputarticle", "relation": "references", "context": "generic_arg", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L460", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_save_extracted_json", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L488", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_save_extracted_json", "target": "$graphify-root$_scripts_extract_article_contents_extractionbatchreport", "relation": "references", "context": "parameter_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L488", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_save_extracted_json", "target": "path", "relation": "references", "context": "parameter_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L488", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_process_batch", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L495", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_process_batch", "target": "path", "relation": "references", "context": "parameter_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L495", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_process_batch", "target": "path", "relation": "references", "context": "parameter_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L495", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_process_batch", "target": "$graphify-root$_scripts_extract_article_contents_extractionbatchreport", "relation": "references", "context": "return_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L495", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L608", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "target": "namespace", "relation": "references", "context": "return_type", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L608", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_py", "target": "$graphify-root$_scripts_extract_article_contents_main", "relation": "contains", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L662", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_trafilaturaextractor_extract", "target": "$graphify-root$_scripts_extract_article_contents_trafilaturadata", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L233", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_newspaperextractor_extract", "target": "$graphify-root$_scripts_extract_article_contents_newspaperdata", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L288", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_readabilityextractor_extract", "target": "$graphify-root$_scripts_extract_article_contents_readabilitydata", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L330", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_extract_all_engines", "target": "$graphify-root$_scripts_extract_article_contents_trafilaturadata", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L354", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_extract_all_engines", "target": "$graphify-root$_scripts_extract_article_contents_newspaperdata", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L370", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_extract_all_engines", "target": "$graphify-root$_scripts_extract_article_contents_readabilitydata", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L386", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_articlecrawler_enter", "target": "$graphify-root$_scripts_extract_article_contents_articlecrawler_start", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L425", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_articlecrawler_exit", "target": "$graphify-root$_scripts_extract_article_contents_articlecrawler_close", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L429", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_articlecrawler_crawl", "target": "$graphify-root$_scripts_extract_article_contents_articlecrawler_start", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L436", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_load_search_json", "target": "$graphify-root$_scripts_extract_article_contents_inputarticle", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L476", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_process_batch", "target": "$graphify-root$_scripts_extract_article_contents_load_search_json", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L506", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_process_batch", "target": "$graphify-root$_scripts_extract_article_contents_log_info", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L513", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_process_batch", "target": "$graphify-root$_scripts_extract_article_contents_articlecrawler", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L527", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_process_batch", "target": "$graphify-root$_scripts_extract_article_contents_extract_all_engines", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L539", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_process_batch", "target": "$graphify-root$_scripts_extract_article_contents_extractedarticle", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L543", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_process_batch", "target": "$graphify-root$_scripts_extract_article_contents_extractionbatchreport", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L584", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_process_batch", "target": "$graphify-root$_scripts_extract_article_contents_save_extracted_json", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L593", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_main", "target": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L664", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_main", "target": "path", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L665", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_main", "target": "$graphify-root$_scripts_extract_article_contents_process_batch", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L672", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_rationale_38", "target": "$graphify-root$_scripts_extract_article_contents_inputarticle", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L38", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_rationale_58", "target": "$graphify-root$_scripts_extract_article_contents_trafilaturadata", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L58", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_rationale_88", "target": "$graphify-root$_scripts_extract_article_contents_newspaperdata", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L88", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_rationale_118", "target": "$graphify-root$_scripts_extract_article_contents_readabilitydata", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L118", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_rationale_138", "target": "$graphify-root$_scripts_extract_article_contents_extractedarticle", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L138", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_rationale_166", "target": "$graphify-root$_scripts_extract_article_contents_extractionbatchreport", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L166", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_rationale_192", "target": "$graphify-root$_scripts_extract_article_contents_trafilaturaextractor", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L192", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_rationale_261", "target": "$graphify-root$_scripts_extract_article_contents_newspaperextractor", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L261", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_rationale_316", "target": "$graphify-root$_scripts_extract_article_contents_readabilityextractor", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L316", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_rationale_350", "target": "$graphify-root$_scripts_extract_article_contents_extract_all_engines", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L350", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_rationale_403", "target": "$graphify-root$_scripts_extract_article_contents_articlecrawler", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L403", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_rationale_432", "target": "$graphify-root$_scripts_extract_article_contents_articlecrawler_crawl", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L432", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_rationale_454", "target": "$graphify-root$_scripts_extract_article_contents_log_info", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L454", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_rationale_461", "target": "$graphify-root$_scripts_extract_article_contents_load_search_json", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L461", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_rationale_489", "target": "$graphify-root$_scripts_extract_article_contents_save_extracted_json", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L489", "weight": 1.0}, {"source": "$graphify-root$_scripts_extract_article_contents_rationale_503", "target": "$graphify-root$_scripts_extract_article_contents_process_batch", "relation": "rationale_for", "confidence": "EXTRACTED", "source_file": "scripts/extract_article_contents.py", "source_location": "L503", "weight": 1.0}], "raw_calls": [{"caller_nid": "$graphify-root$_scripts_extract_article_contents_extractedarticle_to_dict", "callee": "to_dict", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L152", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_extractedarticle_to_dict", "callee": "to_dict", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L158", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_extractedarticle_to_dict", "callee": "to_dict", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L159", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_extractedarticle_to_dict", "callee": "to_dict", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L160", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_extractionbatchreport_to_dict", "callee": "to_dict", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L182", "receiver": "a"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_trafilaturaextractor_extract", "callee": "extract_metadata", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L198", "receiver": "trafilatura"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_trafilaturaextractor_extract", "callee": "extract", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L208", "receiver": "trafilatura"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_trafilaturaextractor_extract", "callee": "extract", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L218", "receiver": "trafilatura"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_trafilaturaextractor_extract", "callee": "loads", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L227", "receiver": "json"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_trafilaturaextractor_extract", "callee": "strip", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L241", "receiver": "raw_text"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_trafilaturaextractor_extract", "callee": "e", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_article_contents.py", "source_location": "L256"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_newspaperextractor_extract", "callee": "lower", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L267", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_newspaperextractor_extract", "callee": "split", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L267", "receiver": "language"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_newspaperextractor_extract", "callee": "Article", "is_member_call": false, "source_file": "scripts/extract_article_contents.py", "source_location": "L269", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_newspaperextractor_extract", "callee": "download", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L270", "receiver": "article"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_newspaperextractor_extract", "callee": "parse", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L271", "receiver": "article"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_newspaperextractor_extract", "callee": "nlp", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L275", "receiver": "article"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_newspaperextractor_extract", "callee": "isoformat", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L283", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_newspaperextractor_extract", "callee": "e", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_article_contents.py", "source_location": "L311"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_readabilityextractor_extract", "callee": "Document", "is_member_call": false, "source_file": "scripts/extract_article_contents.py", "source_location": "L321", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_readabilityextractor_extract", "callee": "title", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L322", "receiver": "doc"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_readabilityextractor_extract", "callee": "short_title", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L323", "receiver": "doc"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_readabilityextractor_extract", "callee": "summary", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L324", "receiver": "doc"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_readabilityextractor_extract", "callee": "BeautifulSoup", "is_member_call": false, "source_file": "scripts/extract_article_contents.py", "source_location": "L327", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_readabilityextractor_extract", "callee": "get_text", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L328", "receiver": "soup"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_readabilityextractor_extract", "callee": "e", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_article_contents.py", "source_location": "L343"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_extract_all_engines", "callee": "extract", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L352", "receiver": "TrafilaturaExtractor"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_extract_all_engines", "callee": "exc", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_article_contents.py", "source_location": "L364"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_extract_all_engines", "callee": "extract", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L368", "receiver": "NewspaperExtractor"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_extract_all_engines", "callee": "exc", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_article_contents.py", "source_location": "L380"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_extract_all_engines", "callee": "extract", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L384", "receiver": "ReadabilityExtractor"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_extract_all_engines", "callee": "exc", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_article_contents.py", "source_location": "L391"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_articlecrawler_start", "callee": "FoxcapeConfig", "is_member_call": false, "source_file": "scripts/extract_article_contents.py", "source_location": "L412", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_articlecrawler_start", "callee": "Foxcape", "is_member_call": false, "source_file": "scripts/extract_article_contents.py", "source_location": "L413", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_articlecrawler_start", "callee": "start", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L414", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_articlecrawler_close", "callee": "close", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L419", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_articlecrawler_crawl", "callee": "get", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L439", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_log_info", "callee": "write", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L456", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_log_info", "callee": "flush", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L457", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_load_search_json", "callee": "exists", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L462", "receiver": "file_path"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_load_search_json", "callee": "FileNotFoundError", "is_member_call": false, "source_file": "scripts/extract_article_contents.py", "source_location": "L463", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_load_search_json", "callee": "load", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L466", "receiver": "json"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_load_search_json", "callee": "f", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_article_contents.py", "source_location": "L466"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_load_search_json", "callee": "get", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L468", "receiver": "data"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_load_search_json", "callee": "get", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L469", "receiver": "data"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_load_search_json", "callee": "get", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L470", "receiver": "data"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_load_search_json", "callee": "dict", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_article_contents.py", "source_location": "L474"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_load_search_json", "callee": "append", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L475", "receiver": "articles"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_load_search_json", "callee": "get", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L479", "receiver": "item"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_load_search_json", "callee": "get", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L480", "receiver": "item"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_load_search_json", "callee": "get", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L481", "receiver": "item"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_save_extracted_json", "callee": "mkdir", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L490", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_save_extracted_json", "callee": "dump", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L492", "receiver": "json"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_save_extracted_json", "callee": "f", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_article_contents.py", "source_location": "L492"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_save_extracted_json", "callee": "to_dict", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L492", "receiver": "report"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_process_batch", "callee": "time", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L525", "receiver": "time"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_process_batch", "callee": "crawl", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L533", "receiver": "crawler"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_process_batch", "callee": "exc", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_article_contents.py", "source_location": "L562"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_process_batch", "callee": "append", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L579", "receiver": "extracted_list"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_process_batch", "callee": "time", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L581", "receiver": "time"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_process_batch", "callee": "isoformat", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L582", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_process_batch", "callee": "now", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L582", "receiver": "datetime"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "callee": "ArgumentParser", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L609", "receiver": "argparse"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "callee": "add_argument", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L613", "receiver": "parser"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "callee": "str", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_article_contents.py", "source_location": "L617"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "callee": "add_argument", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L620", "receiver": "parser"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "callee": "str", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_article_contents.py", "source_location": "L624"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "callee": "add_argument", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L628", "receiver": "parser"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "callee": "int", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_article_contents.py", "source_location": "L632"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "callee": "add_argument", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L636", "receiver": "parser"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "callee": "str", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_article_contents.py", "source_location": "L641"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "callee": "add_argument", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L645", "receiver": "parser"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "callee": "int", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "scripts/extract_article_contents.py", "source_location": "L649"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "callee": "add_argument", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L653", "receiver": "parser"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_parse_arguments", "callee": "parse_args", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L659", "receiver": "parser"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_main", "callee": "exists", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L668", "receiver": "input_path"}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_main", "callee": "write", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L669", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_main", "callee": "write", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L682", "receiver": null}, {"caller_nid": "$graphify-root$_scripts_extract_article_contents_main", "callee": "write", "is_member_call": true, "source_file": "scripts/extract_article_contents.py", "source_location": "L685", "receiver": null}]} |