addrkit: deterministic address resolution service (Go)

- domain: address normalization/parsing, matching with explicit statuses
- adapters: SQLite registry store, ArcGIS outages client, optional LLM rewriter, HTTP API
- seed pipeline with duplicate handling and unparsed reporting
- CLI (seed/verify/resolve), systemd unit, OpenAPI contract, tests
This commit is contained in:
addrkit
2026-10-09 20:05:32 +07:00
commit 3be331bb65
34 changed files with 3446 additions and 0 deletions
+146
View File
@@ -0,0 +1,146 @@
// Command addrkit is the operator CLI: it seeds the canonical registry from a
// registry dump, verifies a built database and resolves a single address for
// quick checks.
package main
import (
"context"
"encoding/json"
"flag"
"fmt"
"os"
"git.totmin.ru/en2zmax/addrkit/internal/adapters/registrystore"
"git.totmin.ru/en2zmax/addrkit/internal/domain/match"
"git.totmin.ru/en2zmax/addrkit/internal/seed"
)
func main() {
if len(os.Args) < 2 {
usage()
os.Exit(2)
}
switch os.Args[1] {
case "seed":
os.Exit(runSeed(os.Args[2:]))
case "verify":
os.Exit(runVerify(os.Args[2:]))
case "resolve":
os.Exit(runResolve(os.Args[2:]))
case "help", "-h", "--help":
usage()
os.Exit(0)
default:
fmt.Fprintf(os.Stderr, "unknown command %q\n", os.Args[1])
usage()
os.Exit(2)
}
}
func usage() {
fmt.Fprint(os.Stderr, `addrkit — address registry toolkit
Usage:
addrkit seed -in registry.ndjson -out registry.db [-version 2026-10-09]
addrkit verify -db registry.db
addrkit resolve -db registry.db "2-я Станционная улица 1/1 стр."
`)
}
func runSeed(args []string) int {
fs := flag.NewFlagSet("seed", flag.ExitOnError)
in := fs.String("in", "", "input registry dump (Bulk NDJSON or JSONL)")
out := fs.String("out", "", "output SQLite database")
version := fs.String("version", "", "data version (defaults to input file mtime)")
_ = fs.Parse(args)
if *in == "" || *out == "" {
fmt.Fprintln(os.Stderr, "seed: -in and -out are required")
return 2
}
if *version == "" {
if fi, err := os.Stat(*in); err == nil {
*version = fi.ModTime().UTC().Format("20060102T150405Z")
}
}
f, err := os.Open(*in)
if err != nil {
fmt.Fprintf(os.Stderr, "seed: open input: %v\n", err)
return 1
}
defer f.Close()
sources, err := seed.ReadBulkNDJSON(f)
if err != nil {
fmt.Fprintf(os.Stderr, "seed: read: %v\n", err)
return 1
}
entries, stats := seed.Build(sources)
if stats.Parsed == 0 {
fmt.Fprintln(os.Stderr, "seed: no parseable entries; refusing to write empty index")
return 1
}
if err := registrystore.Save(*out, *version, entries); err != nil {
fmt.Fprintf(os.Stderr, "seed: save: %v\n", err)
return 1
}
enc := json.NewEncoder(os.Stdout)
enc.SetIndent("", " ")
_ = enc.Encode(map[string]any{
"output": *out,
"version": *version,
"entries": len(entries),
"stats": stats,
})
return 0
}
func runVerify(args []string) int {
fs := flag.NewFlagSet("verify", flag.ExitOnError)
db := fs.String("db", "", "SQLite database")
_ = fs.Parse(args)
if *db == "" {
fmt.Fprintln(os.Stderr, "verify: -db is required")
return 2
}
version, entries, err := registrystore.Load(*db)
if err != nil {
fmt.Fprintf(os.Stderr, "verify: %v\n", err)
return 1
}
streets := make(map[string]struct{})
for _, e := range entries {
streets[e.StreetKey] = struct{}{}
}
enc := json.NewEncoder(os.Stdout)
enc.SetIndent("", " ")
_ = enc.Encode(map[string]any{
"version": version,
"entries": len(entries),
"distinct_streets": len(streets),
})
return 0
}
func runResolve(args []string) int {
fs := flag.NewFlagSet("resolve", flag.ExitOnError)
db := fs.String("db", "", "SQLite database")
_ = fs.Parse(args)
rest := fs.Args()
if *db == "" || len(rest) == 0 {
fmt.Fprintln(os.Stderr, `resolve: -db and an address are required`)
return 2
}
version, entries, err := registrystore.Load(*db)
if err != nil {
fmt.Fprintf(os.Stderr, "resolve: %v\n", err)
return 1
}
resolver := match.NewResolver(match.NewIndex(version, entries, match.DefaultOptions()), nil)
res := resolver.Resolve(context.Background(), rest[0])
enc := json.NewEncoder(os.Stdout)
enc.SetIndent("", " ")
_ = enc.Encode(res)
return 0
}
+120
View File
@@ -0,0 +1,120 @@
// Command addrkitd runs the address resolution service.
//
// It loads the canonical registry into memory, wires the optional adapters
// (ArcGIS portal, LLM rewriter) and serves the versioned JSON API. The process
// fails closed: if the registry cannot be loaded it exits rather than serving
// an empty index.
package main
import (
"context"
"errors"
"flag"
"log/slog"
"net/http"
"os"
"os/signal"
"syscall"
"time"
"git.totmin.ru/en2zmax/addrkit/internal/adapters/arcgis"
"git.totmin.ru/en2zmax/addrkit/internal/adapters/httpapi"
"git.totmin.ru/en2zmax/addrkit/internal/adapters/llm"
"git.totmin.ru/en2zmax/addrkit/internal/adapters/registrystore"
"git.totmin.ru/en2zmax/addrkit/internal/config"
"git.totmin.ru/en2zmax/addrkit/internal/domain/match"
)
func main() {
cfgPath := flag.String("config", "config.json", "path to the JSON config file")
flag.Parse()
cfg, err := config.Load(*cfgPath)
if err != nil {
slog.Error("config", "error", err)
os.Exit(1)
}
logger := newLogger(cfg.LogLevel)
version, entries, err := registrystore.Load(cfg.DataPath)
if err != nil {
// Fail closed: no registry, no service.
logger.Error("load registry", "error", err, "path", cfg.DataPath)
os.Exit(1)
}
logger.Info("registry loaded", "version", version, "entries", len(entries))
index := match.NewIndex(version, entries, match.Options{
FuzzyThreshold: cfg.Resolve.FuzzyThreshold,
AmbiguityMargin: cfg.Resolve.AmbiguityMargin,
MaxCandidates: cfg.Resolve.MaxCandidates,
})
var rewriter match.Rewriter
if cfg.LLM.Enabled {
rewriter = llm.New(cfg.LLM)
logger.Info("llm rewriter enabled", "url", cfg.LLM.URL, "model", cfg.LLM.Model)
}
resolver := match.NewResolver(index, rewriter)
portal := arcgis.New(
cfg.Portal.URL,
time.Duration(cfg.Portal.TimeoutSeconds)*time.Second,
cfg.Portal.Retries,
)
loc := operationalLocation()
srv := httpapi.New(resolver, portal, logger, loc)
httpServer := &http.Server{
Addr: cfg.Addr,
Handler: srv.Handler(),
ReadHeaderTimeout: 5 * time.Second,
}
ctx, stop := signal.NotifyContext(context.Background(), syscall.SIGINT, syscall.SIGTERM)
defer stop()
go func() {
logger.Info("listening", "addr", cfg.Addr, "tz", loc.String())
if err := httpServer.ListenAndServe(); err != nil && !errors.Is(err, http.ErrServerClosed) {
logger.Error("serve", "error", err)
stop()
}
}()
<-ctx.Done()
logger.Info("shutting down")
shutdownCtx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
defer cancel()
if err := httpServer.Shutdown(shutdownCtx); err != nil {
logger.Error("shutdown", "error", err)
}
}
func newLogger(level string) *slog.Logger {
var lvl slog.Level
switch level {
case "debug":
lvl = slog.LevelDebug
case "warn":
lvl = slog.LevelWarn
case "error":
lvl = slog.LevelError
default:
lvl = slog.LevelInfo
}
return slog.New(slog.NewJSONHandler(os.Stdout, &slog.HandlerOptions{Level: lvl}))
}
// operationalLocation returns Asia/Novosibirsk when the tz database is
// available and falls back to a fixed UTC+7 offset otherwise (containers
// without tzdata).
func operationalLocation() *time.Location {
if loc, err := time.LoadLocation("Asia/Novosibirsk"); err == nil {
return loc
}
return time.FixedZone("Asia/Novosibirsk", 7*3600)
}