- process-level tests: seed a real EДДС Bulk NDJSON, start the daemon, run
the full-registry selfcheck, canonical HTTP sample, real address-log
queries, strict house identity, input limits, load and graceful shutdown
- production data stays outside the repo (ADDRKIT_E2E_REGISTRY/LOG); all
temporary files live in t.TempDir()
- add `make e2e` and `make check` (vet + race tests + dependency guard)
Another repository can import pkg/addrkit and use the parser, index and resolver without HTTP or SQLite. The package only re-exports domain types and constructors and performs no I/O.
- remove the events/outages module, ArcGIS adapter, portal config and event statuses
- add a configurable city dictionary/parser and make City part of StreetKey
- seed: read city/city_name, namespace IDs by city, keep structures distinct
- fix reliability: schema v2 with fail-closed load, deterministic ordering,
input limits, config validation, atomic DB publish, LLM response caps
- add read-only shadow mode comparing the old ES path with addrkit
- breaking: registry.db must be re-seeded; config has no portal section
Portal date fields arrive as numbers; fmt's %v rendered them in scientific
notation, so the "current outage" filter could not parse them and kept past
outages (a 'Прошлые' record leaked through). Format integer-valued attributes
explicitly.
likePattern padded every token containing a digit, including ordinals like
"3-я". When the ordinal starts the address, " 3-я " never matches, so
"3-я Механическая улица 24а" returned no outages. Pad only pure-numeric
house tokens; precision is enforced by the local house filter regardless.
The outages portal appends the district in parentheses, e.g.
"Медкадры улица, 1 (Заельцовский район)". That suffix hid the house number
from the parser, so every portal address failed house matching and /v1/outages
always returned empty.
- address: NormalizeText drops a trailing "(...)" group (root fix, benefits
every source)
- outages: introduce HouseUnit; the ArcGIS adapter parses the (cleaned) address
at the boundary and stores the structured unit; MatchingHouse matches on the
unit (with a parse fallback for sources that only provide text)
- config: portal.url is now optional (core runs without the events module)
- httpapi: always return "outages": [] (no more omitempty); report
"detail" when the events module is not configured
- tests: regression coverage for the portal district format
Exhaustive check of all 72,825 indexed addresses: 72,811 exact (99.98%),
14 partial (malformed house units, e.g. '27с1', '32 к 1/4-1/6'), 0 not_found,
0 ambiguous.
- match: require candidate ordinal to equal the query ordinal in fuzzy search
(deterministic typo recall 73% -> 80%, no wrong houses)
- resolver: compatibleVariant guard — LLM may only fix street spelling, keeping
the same words, ordinal and house unit; damerau distance for transpositions
- llm: prompt moved out of code (go:embed default_prompt.txt) with
llm.system_prompt / llm.system_prompt_file overrides; guided JSON parsing
tolerant of thinking blocks
- address: recognize ordinals appearing after the street name
- tests: first coverage for the LLM adapter and config prompt loading; README
Measured (sample 200): deterministic 94.7-95.8% exact / 0 wrong across 4 seeds;
LLM+guard typo 90.7-89.8% / 0 wrong. LLM fallback latency ~1.2-1.9s.
Maksim Totmin