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
11 KiB
JSON
1 line
11 KiB
JSON
{"nodes": [{"id": "$graphify-root$_classify_py", "label": "classify.py", "file_type": "code", "source_file": "classify.py", "source_location": "L1"}, {"id": "$graphify-root$_classify_parse_args", "label": "parse_args()", "file_type": "code", "source_file": "classify.py", "source_location": "L27", "_callable": true}, {"id": "namespace", "label": "Namespace", "file_type": "code", "source_file": "", "source_location": "", "origin_file": "$graphify-root$/classify.py"}, {"id": "$graphify-root$_classify_emit_error", "label": "emit_error()", "file_type": "code", "source_file": "classify.py", "source_location": "L72", "_callable": true}, {"id": "errorcode", "label": "ErrorCode", "file_type": "code", "source_file": "", "source_location": "", "origin_file": "$graphify-root$/classify.py"}, {"id": "$graphify-root$_classify_main", "label": "main()", "file_type": "code", "source_file": "classify.py", "source_location": "L97", "_callable": true}], "edges": [{"source": "$graphify-root$_classify_py", "target": "argparse", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "classify.py", "source_location": "L6", "weight": 1.0}, {"source": "$graphify-root$_classify_py", "target": "sys", "relation": "imports", "context": "import", "confidence": "EXTRACTED", "source_file": "classify.py", "source_location": "L7", "weight": 1.0}, {"source": "$graphify-root$_classify_py", "target": "pathlib", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "classify.py", "source_location": "L8", "weight": 1.0}, {"source": "$graphify-root$_classify_py", "target": "src", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "classify.py", "source_location": "L10", "weight": 1.0}, {"source": "$graphify-root$_classify_py", "target": "src_classifier", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "classify.py", "source_location": "L11", "weight": 1.0}, {"source": "$graphify-root$_classify_py", "target": "src_models", "relation": "imports_from", "context": "import", "confidence": "EXTRACTED", "source_file": "classify.py", "source_location": "L12", "weight": 1.0}, {"source": "$graphify-root$_classify_py", "target": "$graphify-root$_classify_parse_args", "relation": "contains", "confidence": "EXTRACTED", "source_file": "classify.py", "source_location": "L27", "weight": 1.0}, {"source": "$graphify-root$_classify_parse_args", "target": "namespace", "relation": "references", "context": "return_type", "confidence": "EXTRACTED", "source_file": "classify.py", "source_location": "L27", "weight": 1.0}, {"source": "$graphify-root$_classify_py", "target": "$graphify-root$_classify_emit_error", "relation": "contains", "confidence": "EXTRACTED", "source_file": "classify.py", "source_location": "L72", "weight": 1.0}, {"source": "$graphify-root$_classify_emit_error", "target": "errorcode", "relation": "references", "context": "parameter_type", "confidence": "EXTRACTED", "source_file": "classify.py", "source_location": "L72", "weight": 1.0}, {"source": "$graphify-root$_classify_py", "target": "$graphify-root$_classify_main", "relation": "contains", "confidence": "EXTRACTED", "source_file": "classify.py", "source_location": "L97", "weight": 1.0}, {"source": "$graphify-root$_classify_main", "target": "$graphify-root$_classify_parse_args", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "classify.py", "source_location": "L99", "weight": 1.0}, {"source": "$graphify-root$_classify_main", "target": "$graphify-root$_classify_emit_error", "relation": "calls", "context": "call", "confidence": "EXTRACTED", "source_file": "classify.py", "source_location": "L108", "weight": 1.0}], "raw_calls": [{"caller_nid": "$graphify-root$_classify_parse_args", "callee": "ArgumentParser", "is_member_call": true, "source_file": "classify.py", "source_location": "L28", "receiver": "argparse"}, {"caller_nid": "$graphify-root$_classify_parse_args", "callee": "add_argument", "is_member_call": true, "source_file": "classify.py", "source_location": "L32", "receiver": "parser"}, {"caller_nid": "$graphify-root$_classify_parse_args", "callee": "str", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "classify.py", "source_location": "L34"}, {"caller_nid": "$graphify-root$_classify_parse_args", "callee": "add_argument", "is_member_call": true, "source_file": "classify.py", "source_location": "L38", "receiver": "parser"}, {"caller_nid": "$graphify-root$_classify_parse_args", "callee": "str", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "classify.py", "source_location": "L40"}, {"caller_nid": "$graphify-root$_classify_parse_args", "callee": "add_argument", "is_member_call": true, "source_file": "classify.py", "source_location": "L44", "receiver": "parser"}, {"caller_nid": "$graphify-root$_classify_parse_args", "callee": "str", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "classify.py", "source_location": "L47"}, {"caller_nid": "$graphify-root$_classify_parse_args", "callee": "add_argument", "is_member_call": true, "source_file": "classify.py", "source_location": "L51", "receiver": "parser"}, {"caller_nid": "$graphify-root$_classify_parse_args", "callee": "add_argument", "is_member_call": true, "source_file": "classify.py", "source_location": "L57", "receiver": "parser"}, {"caller_nid": "$graphify-root$_classify_parse_args", "callee": "add_argument", "is_member_call": true, "source_file": "classify.py", "source_location": "L63", "receiver": "parser"}, {"caller_nid": "$graphify-root$_classify_parse_args", "callee": "parse_args", "is_member_call": true, "source_file": "classify.py", "source_location": "L69", "receiver": "parser"}, {"caller_nid": "$graphify-root$_classify_emit_error", "callee": "ClassificationError", "is_member_call": false, "source_file": "classify.py", "source_location": "L78", "receiver": null}, {"caller_nid": "$graphify-root$_classify_emit_error", "callee": "to_json_str", "is_member_call": true, "source_file": "classify.py", "source_location": "L83", "receiver": "err"}, {"caller_nid": "$graphify-root$_classify_emit_error", "callee": "Path", "is_member_call": false, "source_file": "classify.py", "source_location": "L87", "receiver": null}, {"caller_nid": "$graphify-root$_classify_emit_error", "callee": "mkdir", "is_member_call": true, "source_file": "classify.py", "source_location": "L88", "receiver": null}, {"caller_nid": "$graphify-root$_classify_emit_error", "callee": "write_text", "is_member_call": true, "source_file": "classify.py", "source_location": "L89", "receiver": "out_file"}, {"caller_nid": "$graphify-root$_classify_emit_error", "callee": "write", "is_member_call": true, "source_file": "classify.py", "source_location": "L93", "receiver": null}, {"caller_nid": "$graphify-root$_classify_main", "callee": "Path", "is_member_call": false, "source_file": "classify.py", "source_location": "L103", "receiver": null}, {"caller_nid": "$graphify-root$_classify_main", "callee": "Path", "is_member_call": false, "source_file": "classify.py", "source_location": "L104", "receiver": null}, {"caller_nid": "$graphify-root$_classify_main", "callee": "is_file", "is_member_call": true, "source_file": "classify.py", "source_location": "L107", "receiver": "ecp_path"}, {"caller_nid": "$graphify-root$_classify_main", "callee": "read_text", "is_member_call": true, "source_file": "classify.py", "source_location": "L116", "receiver": "ecp_path"}, {"caller_nid": "$graphify-root$_classify_main", "callee": "e", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "classify.py", "source_location": "L121"}, {"caller_nid": "$graphify-root$_classify_main", "callee": "from_json_str", "is_member_call": true, "source_file": "classify.py", "source_location": "L126", "receiver": "ECPSnapshot"}, {"caller_nid": "$graphify-root$_classify_main", "callee": "e", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "classify.py", "source_location": "L128"}, {"caller_nid": "$graphify-root$_classify_main", "callee": "is_file", "is_member_call": true, "source_file": "classify.py", "source_location": "L144", "receiver": "content_path"}, {"caller_nid": "$graphify-root$_classify_main", "callee": "read_text", "is_member_call": true, "source_file": "classify.py", "source_location": "L153", "receiver": "content_path"}, {"caller_nid": "$graphify-root$_classify_main", "callee": "e", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "classify.py", "source_location": "L158"}, {"caller_nid": "$graphify-root$_classify_main", "callee": "strip", "is_member_call": true, "source_file": "classify.py", "source_location": "L162", "receiver": "content_raw"}, {"caller_nid": "$graphify-root$_classify_main", "callee": "strip", "is_member_call": true, "source_file": "classify.py", "source_location": "L166", "receiver": "content_raw"}, {"caller_nid": "$graphify-root$_classify_main", "callee": "InherenceClassifier", "is_member_call": false, "source_file": "classify.py", "source_location": "L171", "receiver": null}, {"caller_nid": "$graphify-root$_classify_main", "callee": "classify", "is_member_call": true, "source_file": "classify.py", "source_location": "L177", "receiver": "classifier"}, {"caller_nid": "$graphify-root$_classify_main", "callee": "e", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "classify.py", "source_location": "L181"}, {"caller_nid": "$graphify-root$_classify_main", "callee": "e", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "classify.py", "source_location": "L189"}, {"caller_nid": "$graphify-root$_classify_main", "callee": "to_json_str", "is_member_call": true, "source_file": "classify.py", "source_location": "L193", "receiver": "result"}, {"caller_nid": "$graphify-root$_classify_main", "callee": "Path", "is_member_call": false, "source_file": "classify.py", "source_location": "L198", "receiver": null}, {"caller_nid": "$graphify-root$_classify_main", "callee": "mkdir", "is_member_call": true, "source_file": "classify.py", "source_location": "L199", "receiver": null}, {"caller_nid": "$graphify-root$_classify_main", "callee": "write_text", "is_member_call": true, "source_file": "classify.py", "source_location": "L200", "receiver": "out_file"}, {"caller_nid": "$graphify-root$_classify_main", "callee": "e", "is_member_call": false, "indirect": true, "context": "argument", "source_file": "classify.py", "source_location": "L205"}, {"caller_nid": "$graphify-root$_classify_main", "callee": "write", "is_member_call": true, "source_file": "classify.py", "source_location": "L208", "receiver": null}]} |