{
  "generated": "2026-10-17T06:30:24.880Z",
  "source": {
    "title": "Moby-Dick; or, The Whale",
    "author": "Herman Melville",
    "year": 1851,
    "status": "public domain",
    "origin": "Project Gutenberg eBook #2701",
    "url": "https://www.gutenberg.org/cache/epub/2701/pg2701.txt",
    "rawSha256": "b5e88b71a1cda068e9ab1d94344c9d932338ba6da1a66def97a1aecd9cfcc87d",
    "sha256": "b8d4a465cfb1eb0c7c8a3bd597f0af7666d00caa5d3a40aa52f7af9819e31d7b",
    "characters": 1218974,
    "chapters": 135,
    "epilogueAt": 1217466
  },
  "poster": {
    "widthMm": 2000,
    "heightMm": 3000,
    "mmPerPx": 0.5,
    "margin": 0.025,
    "options": {
      "font": "roboto",
      "legibility": "medium",
      "color": "photo",
      "paper": "light",
      "unit": "char",
      "contrast": 1.9,
      "edges": 0.8,
      "lineSpacing": 1,
      "width": 3810,
      "aspect": 1.5249343832020998
    },
    "characters": 1214967,
    "lines": 886,
    "fontSizeMm": 3.2578125,
    "toneFidelity": 0.8964970737996707,
    "sections": [
      {
        "title": "Part I – Ashore and out to sea",
        "firstLine": 0,
        "lineCount": 313,
        "characters": 409272,
        "y": 0,
        "height": 2039.390625,
        "motif": "harbour",
        "firstChapter": "Etymology",
        "heightMm": 1019.6953125
      },
      {
        "title": "Part II – The whaling voyage",
        "firstLine": 313,
        "lineCount": 277,
        "characters": 400736,
        "y": 2039.390625,
        "height": 1804.828125,
        "motif": "hunt",
        "firstChapter": "Chapter 42",
        "heightMm": 902.4140625
      },
      {
        "title": "Part III – The white whale",
        "firstLine": 590,
        "lineCount": 296,
        "characters": 404957,
        "y": 3844.21875,
        "height": 1928.625,
        "motif": "whiteWhale",
        "firstChapter": "Chapter 86",
        "heightMm": 964.3125
      }
    ]
  },
  "tests": [
    {
      "id": "T1",
      "name": "Full text from the SVG equals the source (SHA-256)",
      "pass": true,
      "detail": {
        "source": "744a89261a75f10e4f5da3f37fd59eb8618c859147b902584f08dadf3ab69028",
        "extracted": "744a89261a75f10e4f5da3f37fd59eb8618c859147b902584f08dadf3ab69028",
        "characters": 1214967
      }
    },
    {
      "id": "T2",
      "name": "Every visible character of the print PDF equals the source (MuPDF)",
      "pass": true,
      "detail": {
        "extractor": "mutool draw -F txt",
        "visibleCharacters": 1002170,
        "exact": false,
        "bidiMarksInSource": 2,
        "characters": 1214964
      }
    },
    {
      "id": "T3",
      "name": "100 random 160-character passages are found at their original position",
      "pass": true,
      "detail": {
        "found": 100,
        "of": 100
      }
    },
    {
      "id": "T4",
      "name": "Chapter structure intact (table of contents and 135 chapters, in order)",
      "pass": true,
      "detail": {
        "headings": 271
      }
    },
    {
      "id": "T5",
      "name": "Opening and closing of the book intact",
      "pass": true,
      "detail": {}
    },
    {
      "id": "T6",
      "name": "Word frequencies (top 25) identical",
      "pass": true,
      "detail": {
        "top": [
          "the 14535",
          "of 6624",
          "and 6447",
          "a 4747",
          "to 4627",
          "in 4184",
          "that 3085",
          "his 2532"
        ]
      }
    },
    {
      "id": "T7",
      "name": "Question set: every answer has its evidence in the extracted text",
      "pass": true,
      "detail": {
        "grounded": 8,
        "of": 8
      }
    }
  ],
  "ocr": {
    "lines": 16,
    "segments": 128,
    "words": 3943,
    "recognised": 2895,
    "accuracy": 0.7342125285315749,
    "pxPerMm": 2
  },
  "poppler": {
    "exact": false,
    "characters": 1009382,
    "wordsInSource": 0.5445620141357259
  },
  "vision": {
    "model": "Claude (Anthropic, Opus 5.5), reading the segment images directly",
    "procedure": "The three segments with the lowest Tesseract scores were stacked into one image without looking at their ground truth; the transcription was written down before comparing. Cut-off fragments of neighbouring words at the left edge were left out; a cut-off letter at the right edge was transcribed as seen.",
    "segments": 3,
    "words": 146,
    "accuracy": 1,
    "tesseractOnSameSegments": 0.04794520547945205
  }
}