{
  "schema": "pdfmd.public-paddle-component-comparison/v1",
  "unit": 36,
  "status": "blocked-resource-limit-accuracy-unmeasured",
  "recognition_deployed": false,
  "product_adapter_created": false,
  "retry_authorized": false,
  "next_trial_authorized": false,
  "decision": "The bounded PP-StructureV3 component trial stopped at its exact 14 GiB memory limit during the first page call. Zero pages completed and no page-level provider output exists, so Paddle accuracy and product compatibility remain unmeasured.",
  "readiness_audit": {
    "unit": 35,
    "status": "complete-comparative-benefit-unverified",
    "recognition_calls": 0,
    "finding": "PaddleOCR 3.7.0 / PP-StructureV3 on ctb1 was the only checked executable local comparator. Runner, configuration, no-MKLDNN amendment, and relevant model hashes matched retained bindings; historical package-metadata byte identity is unknown."
  },
  "candidate_evidence": [
    {
      "candidate": "PaddleOCR / PP-StructureV3",
      "actual_prior_testing": "Eight consumed synthetic pages completed after disabling MKLDNN; independent review passed 1/8 whole fixtures in raw/all-label output and 0/8 in default-ignore Markdown; 18:20.52 wall time and 12465468 KiB peak RSS.",
      "current_boundary": "Historical runtime evidence only; no prior frozen-source measurement and no product adapter.",
      "official_source": "https://github.com/PaddlePaddle/PaddleOCR/blob/main/docs/version3.x/pipeline_usage/PP-StructureV3.en.md"
    },
    {
      "candidate": "EasyOCR",
      "actual_prior_testing": "No configured runtime or frozen-source invocation surfaced in the scoped audit.",
      "current_boundary": "Published interface evidence only; no corpus result or product compatibility result.",
      "official_sources": [
        "https://github.com/JaidedAI/EasyOCR/blob/master/easyocr/easyocr.py",
        "https://github.com/JaidedAI/EasyOCR/blob/master/easyocr/utils.py"
      ]
    },
    {
      "candidate": "docTR",
      "actual_prior_testing": "No configured runtime or frozen-source invocation surfaced in the scoped audit.",
      "current_boundary": "Published interface evidence only; no corpus result or product compatibility result.",
      "official_sources": [
        "https://mindee.github.io/doctr/using_doctr/using_models.html",
        "https://mindee.github.io/doctr/latest/modules/io.html"
      ]
    },
    {
      "candidate": "Marker / Surya",
      "actual_prior_testing": "Six-case Marker 2.0.0 / Surya 0.22.1 native-text pilot with OCR disabled; layout/order assets remain present and the OCR GGUF is absent.",
      "current_boundary": "Native parsing and layout/order evidence only; not Marker OCR or standalone Surya OCR.",
      "official_sources": [
        "https://github.com/datalab-to/marker/blob/master/README.md",
        "https://github.com/datalab-to/surya/blob/master/surya/recognition/__init__.py"
      ]
    },
    {
      "candidate": "Mistral OCR",
      "actual_prior_testing": "No recorded call or frozen-source result surfaced in the scoped audit.",
      "current_boundary": "Official product documentation is research evidence only.",
      "official_source": "https://docs.mistral.ai/studio/document-processing/basic_ocr"
    },
    {
      "candidate": "LlamaParse",
      "actual_prior_testing": "No recorded call or frozen-source result surfaced in the scoped audit.",
      "current_boundary": "Official product documentation is research evidence only.",
      "official_source": "https://developers.llamaindex.ai/llamaparse/parse/examples/parse_granular_bboxes/"
    }
  ],
  "trial": {
    "planned_page_calls": 3,
    "started_page_calls": 1,
    "completed_page_calls": 0,
    "started_cases": [
      "vnat-page-08"
    ],
    "unrun_cases": [
      "internal-digit-positive",
      "digits-and-blank-confounder"
    ],
    "retries": 0,
    "initialization_seconds": 27.306094,
    "first_page_started_utc": "2026-09-11T02:24:46.314883Z",
    "oom_kill_utc": "2026-09-11T02:27:57.001644Z",
    "resource": {
      "systemd_result": "oom-kill",
      "memory_max_bytes": 15032385536,
      "memory_peak_bytes": 15032385536,
      "memory_peak_equal_to_limit": true,
      "memory_swap_peak_bytes": 0,
      "substage_attribution": "unresolved"
    },
    "runtime_interpretation": "The full PP-Structure wrapper initialized layout, text-line orientation, region detection, table, detection, and recognition stages. Time and memory cannot be attributed to raw OCR alone or compared directly with Tesseract timing."
  },
  "stage_boundary": [
    {
      "stage": "source",
      "result": "Three consumed image-only pages and independently authored exact expectations were frozen before output."
    },
    {
      "stage": "raw_paddle_ocr",
      "result": "No page-level raw OCR, recognized text, detector polygons, or recognized-text polygons returned; accuracy, order, geometry, and blank restraint are unmeasured."
    },
    {
      "stage": "ppstructure_parser",
      "result": "No page-level parser blocks returned; parser order and paragraph grouping are unmeasured."
    },
    {
      "stage": "product",
      "result": "No adapter, provider word identities, PDFMD, canonical output, or human Markdown was created."
    }
  ],
  "accuracy": {
    "status": "unmeasured",
    "paddle_accuracy_failure_finding": "none; no OCR result exists",
    "comparative_benefit": "unmeasured",
    "source_fidelity": "unmeasured",
    "blank_geometry": "unmeasured; cross-provider geometric equivalence would remain unresolved even with zero Paddle intersections",
    "english_text_conservation": "failed on earlier product evidence; Unit 36 produced no result that changes it",
    "acceptance": false
  },
  "distinct_provider_note": "RapidOCR results are not PaddleOCR / PP-StructureV3 results and are not counted as Paddle evidence.",
  "retained_bindings": [
    {
      "path": "artifacts/prd-02-semantic-document-storage/plan-02-meaning-first/foundational-gates-20260910/ocr-alternatives-readiness/audit.md",
      "sha256": "8b63bdf9fe1ed06d9435de96217b224c3ed0d6e7af5740cd61b89d84a34847c8"
    },
    {
      "path": "artifacts/prd-02-semantic-document-storage/plan-02-meaning-first/foundational-gates-20260910/ocr-alternatives-readiness/root-review.json",
      "sha256": "b327b5346737c923b70460cef856ff4cc61cffa811b8ce4d930f578228456003"
    },
    {
      "path": "artifacts/prd-02-semantic-document-storage/plan-02-meaning-first/foundational-gates-20260910/ocr-alternatives-readiness/result-disposition.json",
      "sha256": "adae9144c9b4f2ffce18ea3f87e35d33863b2c0a39a476298fdda7c57dc6bceb"
    },
    {
      "path": "artifacts/prd-02-semantic-document-storage/plan-02-meaning-first/foundational-gates-20260910/paddle-component-comparison/call-authority.json",
      "sha256": "f0668b5db62c83b4ca33bce2c4a3e8f755877e75dfda3c2b3a6cf0ec6e96e393"
    },
    {
      "path": "artifacts/prd-02-semantic-document-storage/plan-02-meaning-first/foundational-gates-20260910/paddle-component-comparison/independent-review/review.json",
      "sha256": "0e1270440b95caea7de54350d45ae2db36873584326e212cc307e277db75f3ca"
    },
    {
      "path": "artifacts/prd-02-semantic-document-storage/plan-02-meaning-first/foundational-gates-20260910/paddle-component-comparison/root-result-review.json",
      "sha256": "13fa5d6befd48c915200c5274c85abd01c4f74d0323991756373f410f1be3621"
    },
    {
      "path": "artifacts/prd-02-semantic-document-storage/plan-02-meaning-first/foundational-gates-20260910/paddle-component-comparison/result-advice.md",
      "sha256": "a85e4faed241cf1f3e6bb84ff1f80b6eb546601fac33c9150dbc4dadb77eb920"
    },
    {
      "path": "artifacts/prd-02-semantic-document-storage/plan-02-meaning-first/foundational-gates-20260910/paddle-component-comparison/result-disposition.json",
      "sha256": "16ca00e5eaa2499b9901910a47dbf3d3692e07ff0be4a6aba191938db7a71d23"
    },
    {
      "path": "artifacts/prd-02-semantic-document-storage/plan-02-meaning-first/foundational-gates-20260910/paddle-component-comparison/resource/systemd-final.txt",
      "sha256": "ac9d36c2279862a30c78f012610b3ad75ceeb811284c9a55f1928aceec2297b7"
    },
    {
      "path": "artifacts/prd-02-semantic-document-storage/plan-02-meaning-first/foundational-gates-20260910/paddle-component-comparison/termination-verification.txt",
      "sha256": "ce02056252958e871729f4d765fc7844d5c0ec92cb91bf58bd1cece880c09ae6"
    },
    {
      "path": "artifacts/prd-02-semantic-document-storage/plan-02-meaning-first/foundational-gates-20260910/paddle-component-comparison/execution/effective-config.json",
      "sha256": "e09bb98880a0aec7fd4385ceaa55b5e6c4d1505da2c9521972a3f1e6b95b3e49"
    }
  ],
  "limits": [
    "No page output exists, so the frozen evaluator did not run.",
    "The retained trace does not identify which model or PP-Structure substage caused the memory peak.",
    "The later inactive/dead MainPID=0 snapshot is the durable survivor evidence; no separately labeled process-scan transcript was retained.",
    "English recognition remains failed, production recognition is unchanged, and no product, canonical, human-output, generalization, merge, or deployment acceptance follows.",
    "A direct PP-OCR setup is only a prospective recommendation. Selection and reservation are deferred until verified publication; no next trial is authorized."
  ]
}