{
  "id": "doc-qa-with-citations",
  "title": "Document QA With Citations and LangChain",
  "level": 2,
  "levelName": "Builder",
  "tagline": "A citation receipt that separates exact source support from relevance to the question.",
  "summary": "A citation receipt that separates exact source support from relevance to the question. Import query(directory, question, model). A model callback returns JSON {source:chunk_id,quote:exact_text}. --response accepts a recorded response and applies the same gate.",
  "hours": 8,
  "languages": ["Python"],
  "status": "ready",
  "source": "core",
  "author": {
    "name": "AI Engineering from Scratch",
    "github": "rohitg00"
  },
  "youWillBuild": "A citation receipt that separates exact source support from relevance to the question.",
  "usefulFor": [
    "Import query(directory, question, model). A model callback returns JSON {source:chunk_id,quote:exact_text}. --response accepts a recorded response and applies the same gate."
  ],
  "skills": [
    "Load local documents with stable provenance",
    "Rank chunks with an inspectable keyword score",
    "Accept only answers grounded in retrieved spans",
    "Use a framework splitter without losing offsets"
  ],
  "prerequisites": [
    "Python 3.10+",
    "pathlib, strings, offsets, dictionaries and callback functions. Optional LangChain dependencies are pinned separately."
  ],
  "figures": true,
  "overviewFigure": "pj-doc-qa-with-citations-1",
  "languageWhy": {
    "Python": "Stdlib data structures, explicit algorithms and deterministic local tests."
  },
  "demo": {
    "cwd": "solution",
    "command": [
      "python3",
      "cli.py",
      "samples/docs",
      "When does cache expire?",
      "--output",
      "answer.json",
      "--html",
      "answer.html"
    ]
  },
  "stages": [
    {
      "id": "01-documents",
      "title": "Load local documents with stable provenance",
      "summary": "The second overlapping chunk of abcdef starts at offset 3 when size=4 and overlap=1.",
      "language": "python",
      "languageWhy": "Python standard-library collections, bytes and file operations expose this stage directly.",
      "hours": 2,
      "difficulty": "starter",
      "figure": "pj-doc-qa-with-citations-1",
      "concepts": ["load", "local", "documents", "with"]
    },
    {
      "id": "02-retrieval",
      "title": "Rank chunks with an inspectable keyword score",
      "summary": "A socket query selects the socket chunk, while an unknown term abstains.",
      "language": "python",
      "languageWhy": "Python standard-library collections, bytes and file operations expose this stage directly.",
      "hours": 2,
      "difficulty": "core",
      "figure": "pj-doc-qa-with-citations-2",
      "concepts": ["rank", "chunks", "with", "an"]
    },
    {
      "id": "03-answer",
      "title": "Accept only answers grounded in retrieved spans",
      "summary": "A model response containing words absent from its cited chunk is rejected.",
      "language": "python",
      "languageWhy": "Python standard-library collections, bytes and file operations expose this stage directly.",
      "hours": 2,
      "difficulty": "core",
      "figure": "pj-doc-qa-with-citations-3",
      "concepts": ["accept", "only", "answers", "grounded"]
    },
    {
      "id": "04-adapter",
      "title": "Use a framework splitter without losing offsets",
      "summary": "Framework chunks are converted to the same source-span contract used by the baseline.",
      "language": "python",
      "languageWhy": "Python standard-library collections, bytes and file operations expose this stage directly.",
      "hours": 2,
      "difficulty": "stretch",
      "figure": "pj-doc-qa-with-citations-4",
      "concepts": ["use", "a", "framework", "splitter"],
      "runners": [
        {
          "language": "python"
        },
        {
          "language": "python",
          "optional": true,
          "requires": ["python:langchain_text_splitters"],
          "argv": [
            "{python}",
            "-m",
            "unittest",
            "discover",
            "-s",
            "{stage}/tests-framework",
            "-p",
            "test_framework.py"
          ]
        }
      ]
    }
  ],
  "demos": [
    {
      "gif": "media/run.gif",
      "poster": "media/run.png",
      "title": "Initialize and grade",
      "caption": "Recorded starter failure followed by all reference-solution stages passing. Reference passes do not earn learner certificates."
    },
    {
      "gif": "media/output.gif",
      "poster": "media/output.png",
      "title": "Inspect the output",
      "caption": "Actual output from the local deterministic demo command. See the README for fixture and live-adapter limits."
    },
    {
      "gif": "media/framework.gif",
      "poster": "media/framework.png",
      "title": "Verify the real SDK",
      "caption": "Actual optional SDK tests with installed dependencies and deterministic local models. No live cloud or model provider calls."
    }
  ]
}