Elamur Groupantslangseg-corpus
The groupant beside tools/langseg/src/corpus.rs in the Mova repository: what the code computes, what it never does, how to go back from its outputs, its invariants, cost and checks.
Facts
Implementation material. A Mova code model describes this code at one source revision: partial models, finite checks and their stated scope. It is not a reviewed general exposition, and its labels are shown as written, never raised.
Components
| component | label | where |
|---|---|---|
| approximant | partial | approximant |
| ast | partial | syntax.json |
| composition | partial | composition |
| cost | symbolic source derived | invariant |
| descriptant | present | descriptant |
| invariant | partial | invariant |
| tests | executed bounded checks | tests |
Obernants
The typed relations around the forward model: the behaviour complement (what the code never does), the reverse reconstruction (every input compatible with an output) and specialised views.
| id | type | label | where |
|---|---|---|---|
behavior_complement | behavior_complement | partial source derived model | behavior_complement |
reverse_reconstruction | reverse_reconstruction | partial source derived model | reverse_reconstruction |
bracket_horizon | bracket_horizon | partial source derived model | bracket_horizon |
licence_routing | licence_routing | partial source derived model | licence_routing |
document_partition | document_partition | partial source derived model | document_partition |
partial_ingestion | partial_ingestion | partial source derived model | partial_ingestion |
Parts
Model text
descriptant
Descriptant
Prospective epoch
2026-10-31 source review predicted: Builder::build ingests MACULA, licensed registry routes, then UD; retains first occurrence by (Lang,text). push chunks, assigning ids only to surviving chunks. clean strips tags and references with60-character closer lookahead. split_of uses fnv(doc)%10 except explicit UD split. Talmud state expires after ten subsequent segments. Resource model: Cleaning O(B) with bounded60 lookahead; maps/sets O(E log E) string comparisons/storage. English quota rescans accumulated output per row, potentially O(rows*E). Whole XML file and output retained; allocation and disk errors outside pure domain. External prospective.md predates test additions and preserves APIs, effect order, hazards and interaction hypotheses.
Observed epoch
Bounded real helper checks in tests.md constrain these models; syntax is a partial lexical projection with no macro expansion, name resolution or proof. Production expressions unchanged; five inline test modules added. Compiled mutants are temporary and exactly restored. Existing malformed-input and nonfinite domains remain explicit, not silently repaired. Parent composes this scope with siblings without adding overlapping costs twice.
Builder, pool I/O and XML accumulation source-modeled only; no corpus read.
approximant
Forward relation
Builder::build ingests MACULA, licensed registry routes, then UD; retains first occurrence by (Lang,text). push chunks, assigning ids only to surviving chunks. clean strips tags and references with60-character closer lookahead. split_of uses fnv(doc)%10 except explicit UD split. Talmud state expires after ten subsequent segments.
Domain: Chunks require>=3 alphabetic words; max=0 drops every single-word flush. Dedup retains first licence/id, not unique ids. Missing sources often yield empty; XML error yields partial prefix. Registry filtering does not govern earlier MACULA route.
invariant
Invariant and cost
Invariant
Chunks require>=3 alphabetic words; max=0 drops every single-word flush. Dedup retains first licence/id, not unique ids. Missing sources often yield empty; XML error yields partial prefix. Registry filtering does not govern earlier MACULA route.
Cost
Cleaning O(B) with bounded60 lookahead; maps/sets O(E log E) string comparisons/storage. English quota rescans accumulated output per row, potentially O(rows*E). Whole XML file and output retained; allocation and disk errors outside pure domain.
Limits
Builder, pool I/O and XML accumulation source-modeled only; no corpus read. No percentage surrogate is implemented: categorical parsing/effect observations lack an admitted numerical error metric. Any future profile must define its metric/domain and use only1/2/5/10 percent caps.
composition
Local composition
Builder::build ingests MACULA, licensed registry routes, then UD; retains first occurrence by (Lang,text). push chunks, assigning ids only to surviving chunks. clean strips tags and references with60-character closer lookahead. split_of uses fnv(doc)%10 except explicit UD split. Talmud state expires after ten subsequent segments.
Children select narrow source functions with their exact byte intervals. Remaining branches stay in the file model; fragments are not exhaustive partitions. Calls and state joins follow the forward relation, not folder order. Shared scans/helpers are counted once in whole-file costs. Complement of composition is not union of complements; reverse composition retains compatible intermediates and caller premises.
models
Typed source-bound relations
Subject: exact source SHA-256 in manifest. Domain and resource premises are in invariant.md. Each model is partial, source-derived, over a single helper invocation or the described finite trace. Complements are proposed inner exclusions under those premises; reverse relations retain all compatible antecedents, with no unique inverse claim. The oracle is the named checks in tests.md; unexecuted cases remain hypotheses. No numerical approximation certificate.
behavior_complement
Within bounded helper domain, no emitted chunk has<3 alphabetic tokens; exact ud:dev cannot become hash train. Registry scope!=open, NC, unknown and non-JSON rows excluded; separate MACULA path is an explicit exception.
reverse_reconstruction
Cleaned output admits arbitrary removed tags/references and dropped short chunks. Retained (language,text) admits arbitrary later duplicates. Split admits all corresponding document hash residues, not a unique original.
bracket_horizon
Matching closer must occur among60 characters from opener; nesting is not balanced. Unmatched opener survives. Boundary helper test checks unmatched removal, not exhaustive horizon cases.
licence_routing
Admission conjunction precedes route match; MACULA ingestion independently precedes registry. Changing scope removes only that route's additions.
document_partition
Exact ud:train/dev/test bypass hash. Same document shares split; no statistical independence claim for FNV residues.
partial_ingestion
read_sefaria read/JSON failure returns empty, MACULA XML failure breaks then consumes collected verses. write_pool truncates before rows and sanitizes text tabs only; no atomicity.
tests
Checks and evidence
Named source/helper checks: corpus::wave31::cleaning_chunk_and_split_boundaries.
Wave31 on2026-10-31:30 distinct exact tests passed twice debug and twice release, with equal sorted named outcomes. This node claims only its listed subset; empty subset means source review only. Six actual source-behavior mutants compiled, failed assertions, restored exact bytes/hash, and passed focused green reruns. No assertion-only mutants; existing swapped-input controls are separate. Receipt bundle: $MOVA_DATA/runs/octopus-wave31-corpus-2026-10-31 (native-commands.json, source-test-map.json, mutation-results.json, .status/.log). No live corpus, teacher, frozen assessment, download or production data used.
Builder, pool I/O and XML accumulation source-modeled only; no corpus read.
This page in Markdown ā https://md.groupants.elamur.ai/free/en/mova/tools/langseg/src/corpus.rs.groupant/