Repository navigation
feat(WO19): exhaustive per-row tests for alalc-hye/zgh/chr (conformance depth) - #281
Merged
Merged
Conversation
Transcribe the full romanization tables of the Armenian (2022), Moroccan Tamazight (2012) and Cherokee (2012) ALA-LC sources into row-source provenance artifacts and generate exhaustive per-row embedded tests with scripts/gen-row-tests.mjs, mirroring the alalc-vai demonstration. - alalc-hye-Armn-Latn-2022: 128 rows (letter table pp.1-2, Note 3 soft-sign combinations p.2, letter numerals p.7) - alalc-zgh-Tfng-Latn-2012: 43 rows (single-page table, incl. the labialized digraphs and the tamatart) - alalc-chr-Cher-Latn-2012: 85 rows (complete syllabary grid) No map rules changed: every generated row agrees with the existing rules (0 mismatches). Un-expressible source rows are recorded as skipped in the artifact notes, not faked. Gates: rs test-all 11,671 passed / 0 failed (hye 140, zgh 57, chr 103); ts parse of the three maps clean; rs golden recapture byte-identical on all 12,337 stored rows, +502 rows (these 256 plus pre-existing uncaptured vai 240 and the six 3.2.2 escape rows).
ronaldtse
added a commit
that referenced
this pull request
Nov 14, 2026
256 exhaustive per-row tests transcribed from the ALA-LC Armenian, Tamazight, and Cherokee source tables (maps #281): hye 12->140, zgh 14->57, chr 18->103 embedded tests. Zero findings - all rows agree with existing map rules; transliteration output unchanged. Provenance JSONs ship in scripts/row-sources/.
This was referenced Nov 16, 2026
Merged
Merged
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
WO19-depth: the last executable item in the program (WO17 ledger line "hye 134/12, zgh 43/14, chr 85/18 as future work"). The full romanization tables of the three thin ALA-LC maps are transcribed from the locally-held source PDFs into row-source provenance artifacts, and exhaustive per-row embedded tests are generated with
scripts/gen-row-tests.mjs, mirroring the alalc-vai demonstration. No map rules changed — the tests pin the maps' tables against the sources; every generated row agrees with the existing rules (0 mismatches, 0 failures).Per-map counts (transcribed = generated; every generated test passes)
․․․placeholders — not romanization rowsArmenian exclusions recorded in the artifact note: the punctuation section (pp. 5-6) is instruction prose, not a table — «» and ։ are encoded map rules not row-transcribed here, ՞/՜ require sentence-end repositioning (un-encodable), ՛ has empty output (generator contract); the diacritics page (p. 4) repeats letter-table rows; the Ev row's printed lower-case vernacular cell is և (եվ is not a printed row; the map encodes it additionally); the printed Զ/զ and Ս/ս rows carry a "(see Note 3)" reference marker, transcribed as plain values.
Source pages: armenian.pdf pp. 1-2 (letter table), p. 2 (Note 3 combinations), p. 7 (numerals); tamazight.pdf p. 1 (single-page table — the text layer drops the U+0331 macron-below marks and the ⵕ ⵟ ⴺ value cells; those values are transcribed from the printed page image); cherokee.pdf p. 1 (syllabary grid; codepoints verified against the text layer).
Gates
iscrs test-all --maps <branch>/maps: 11,671 passed, 0 failed (hye 140, zgh 57, chr 103)iscrs capture --root <branch>/maps): byte-identical on all 12,337 stored rows; +502 rows = these 256 plus pre-existing uncaptured rows (vai 240 from the WO17 demo, the six 3.2.2 escape rows). The golden jsonl lives in interscript-rs, so it is not touched here (same state as after the vai and G6-item-4 landings; a corpus-wide refresh is a release-round activity, cf. rs PR Update README to include directions on how to develop maps #11).Test plan