Elamur Groupantsannotation-store-adapters
The groupant beside models/annotation-store-adapters.groupant 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 | forward |
| ast | not materialized | |
| composition | partial | composition |
| cost | symbolic source derived | cost |
| descriptant | present | descriptant |
| invariant | partial | invariant |
| tests | see source local evidence | checks |
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 |
representation-transport | representation_transport | partial source derived model | representation-transport |
identity-coordinate-join | identity_coordinate_join | partial source derived model | identity-coordinate-join |
effect-frontier | effect_frontier | partial source derived model | effect-frontier |
tagged-independent-paths | tagged_independent_paths | partial source derived model | tagged-independent-paths |
fixture-observation-limit | fixture_observation_limit | partial source derived model | fixture-observation-limit |
cost-composition | cost_composition | partial source derived model | cost-composition |
Model text
descriptant
annotation-store-adapters: wave31
Real joins: main->load/export/prag/qa/tales; load/export/prag/qa->Duck+Stage+Tsv; load/export/tales->en::conllu or tokenizer/annotator; util supplies SHA1/path/time. TSV write includes header while read expects COPY output without header: direct write/read is not identity unless header separated. Prag eight/nine-column input and QA eight-column input are distinct tagged schemas. Integration test fixtures exercise real CoNLL representation but SQL branches remain unexecuted. SQL include_str files are external dependencies, not selected Rust sources.
This source-bound partial group organizes existing Mova code. Native evidence and excluded effects are recorded in source sidecars and the root wave31 report. Registration is not a universal proof or whole-system runtime claim.
model
annotation-store-adapters
forward
CLI chooses one tagged command. load: CoNLL Doc -> identity match against prior state -> six TSV tables -> Duck exec -> post-count reconciliation. export: SQL TSV materialization -> raw Doc -> text. prag/qa: schema-specific TSV parser -> wanted identities -> state validation -> staged records -> exec -> postcounts. tales text segmentation is a separately tagged preprocessing path; production train/load path is not executed. Tests join supplied fixtures with actual Doc parsing, not a fake DuckDB interpreter.
invariant
TSV NULL/empty encoding remains distinct within admitted cells; row/table identity references agree only under state snapshot and SQL/schema assumptions. Loader sentence identity is id AND text, QA identity is doc/para/k plus anchor doc/para. Exact export additionally needs file-only filter, one file row, preserved ordering/comments/blanks/newline. A post-commit error must not be interpreted as rollback.
cost
Rust helper work is mostly linear in materialized bytes/rows with hash-map expected lookups, BTreeSet O(u log u) wanted identities, duplicate key suffix probing and ordered heading scans. Materialized Docs/TSV results occupy proportional memory. CLI startup, SQL execution, filesystem effects and annotation training/inference are separate costs; no throughput measurement or speedup claim.
composition
Real joins: main->load/export/prag/qa/tales; load/export/prag/qa->Duck+Stage+Tsv; load/export/tales->en::conllu or tokenizer/annotator; util supplies SHA1/path/time. TSV write includes header while read expects COPY output without header: direct write/read is not identity unless header separated. Prag eight/nine-column input and QA eight-column input are distinct tagged schemas. Integration test fixtures exercise real CoNLL representation but SQL branches remain unexecuted. SQL include_str files are external dependencies, not selected Rust sources.
domain
Finite CPU inputs; valid UTF-8 and representable signed IDs/counters; temp filesystem fixtures under isolated run TMPDIR; no production DB/corpus/jobs or annotator training. Database relations are conditional on real CLI/schema/SQL ordering and writer isolation, not established by this helper evidence.
limitations
DuckDB executable absent from PATH and bounded known bin paths; no download/install, no mock subprocess SQL evidence. Source-derived transaction statements are not observed database guarantees. Tests inspected before selection. All production prefixes preserved byte-for-byte; additions are cfg(test) checks only. Partial syntax excludes macro expansion/name/type resolution.
checks
- 16 exact target/name cases twice debug and twice release; sorted actual outcomes identical per profile
- 6 compiled production-behavior mutants assertion-red, exact restoration, positive green
- 12 source projections generated twice identically; 9 narrow fragments share parent projections
- Added blocks formatted separately; original production prefixes checked against base
catalog
behavior-complement
In the admitted helper trace universe, invalid TSV width/sentinel, invalid parser headers/k/anchors and invalid pre-dispatch CLI flags exclude successful adapter results. Database post-check failure does not exclude committed effects; helper exclusions cannot establish SQL rollback.
reverse-reconstruction
Output text/table rows induce compatible-input/state relations across each join, preserving unknown IDs/provenance/unused records. NULL and empty are distinguishable in staged TSV, but bool spellings, stripped anchor whitespace, paragraph whitespace and overwritten CLI duplicates form many-to-one fibers. Exact database recovery requires ordered raw rows/comments/blank flags and external SQL assumptions.
representation-transport
Tsv::row encodes Null as backslash-N, rejects sentinel string, preserves empty string and writes header at creation. COPY-to read has no header skip. JSON serializes anchor arrays and comments before TSV so control characters travel as JSON escapes.
identity-coordinate-join
load matches (sent_id,text) against pre-state while qa maps doc_key then insists each anchor has unique sid with same doc and Some(para). Equal annotation bytes alone cannot identify a sentence; matching sent_id alone can admit an incorrect text.
effect-frontier
Parse/duplicate/state checks and TSV validation precede write exec. Post TSV parsing/counts follow exec, and batch file loops preserve earlier effects after later failure. Partial TSV row buffers may contain prefixes even when count remains zero.
tagged-independent-paths
Prag label schema, QA question schema, CoNLL load/export and tales segmentation are separate dispatched paths sharing transport utilities. One schema's parser success does not imply another's admission or DB table constraints.
fixture-observation-limit
Existing DB integration tests are modeled but not run; new safe tests in those files exercise actual en::conllu Doc parsing and fixture layout only. They do not evaluate query strings, views, migrations or transactional behavior.
cost-composition
Sum actual invoked helper costs plus process/SQL costs on the selected tagged path; do not add all CLI branches as if executed together. Full input/result materialization and sequential per-file DB calls remain, with no measured optimization claim.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/models/annotation-store-adapters.groupant/