Janady13/missionmatch / cmd/tag-missions/main.go

Source identity

Field	Recorded value
Source type	GitHub source
Original record	Janady13/missionmatch / cmd/tag-missions/main.go
Date	Not specified in source
Repository	Janady13/missionmatch
Revision	720d47fae476b0d38477ae4401697d372b1ea2bc
Branch	feature/opportunity-to-outcome
Path or archive member	cmd/tag-missions/main.go
Original SHA-256	1d3edeef1d0ed7fecf42ec3bd2eb9e52f9835fb02e88996ca2d6d8cce4a4602e
Git blob SHA	4868f432aa3e9ff47f30d2a3e7df946c79faa4b5
Reading-copy SHA-256	1d3edeef1d0ed7fecf42ec3bd2eb9e52f9835fb02e88996ca2d6d8cce4a4602e
Original source ↗
Recovered source transcript

Download complete source text ↓162 lines · 6.9 KB
// Command tag-missions runs the WS7 Phase 5a tenant mission tagging backfill. For each organization it
// classifies that tenant's candidates (headline plus skill names) and requirements (title plus labor
// category plus required skill names) with the pure deterministic missionclass classifier, and upserts
// structured mission tags into candidate_mission_tags and requirement_mission_tags. Those tags are the
// structured signal the Phase 5 retrieval upgrade and the Phase 6 Command Center read, replacing the WS6
// text only mission filter. It is a one shot job and a cron candidate like staffing-proactive-sync, and it
// is idempotent and safe to rerun: every upsert refreshes in place rather than duplicating.
//
// Mission tags are reference data, NEVER a scoring input, and the pass never reaches the eeo schema. The
// classifier is the SAME one the global federalsignal pass uses, so a tenant's SIGINT analyst gets a SIGINT
// tag from its headline and a COMSEC tag from a COMSEC skill; the curated WS2 alias layer adds synonyms
// beyond a literal keyword.
//
// It lists organization ids and reads the global alias pairs through the migrator role (the alias tables
// are global reference data). Every TAG WRITE, and the candidate and requirement reads it classifies, run
// through the NOBYPASSRLS application role inside each org's RunInTenantTx, so Postgres Row Level Security
// is exercised on every row and one tenant's tags can never read or write another's. Pass -org to limit the
// run to a single organization id.
//
// Env (server side only): MM_DATABASE_URL (the app DSN, NOBYPASSRLS, required), MM_MIGRATOR_DATABASE_URL
// (the migrator DSN, BYPASSRLS, used ONLY to enumerate org ids and read the global alias pairs; falls back
// to MM_DATABASE_URL).
//
// Build and run FIPS validated: CGO_ENABLED=0 GOFIPS140=v1.0.0 go build, run with GODEBUG=fips140=on.
package main

import (
	"context"
	"crypto/fips140"
	"crypto/pbkdf2"
	"crypto/sha256"
	"flag"
	"log"
	"os"
	"strings"
	"time"

	"github.com/google/uuid"
	"github.com/jackc/pgx/v5/pgxpool"

	"github.com/missionworkforce/missionmatch/internal/modules/federalsignal"
	"github.com/missionworkforce/missionmatch/internal/platform/database"
	"github.com/missionworkforce/missionmatch/internal/platform/database/dbgen"
	"github.com/missionworkforce/missionmatch/internal/platform/security"
	"github.com/missionworkforce/missionmatch/internal/platform/tenant"
)

func main() {
	orgFlag := flag.String("org", "", "limit the backfill to a single organization id (default: all orgs)")
	flag.Parse()

	// FIPS self test: derive one PBKDF2 key at the production cost under the active crypto module and fail
	// closed if it errors, so a build that dropped the validated module is caught at boot, not mid run. This
	// is the same in build FIPS gate the api and missionclass-sync binaries use.
	if err := fipsSelfTest(); err != nil {
		log.Fatalf("tag-missions: FIPS self test (PBKDF2 HMAC SHA 256) failed: %v", err)
	}
	log.Printf("tag-missions: fips_enabled=%v module=%s", fips140.Enabled(), fips140.Version())

	appDSN := strings.TrimSpace(os.Getenv("MM_DATABASE_URL"))
	if appDSN == "" {
		log.Fatal("tag-missions: MM_DATABASE_URL (the app DSN) is required")
	}
	migDSN := firstNonEmpty(os.Getenv("MM_MIGRATOR_DATABASE_URL"), appDSN)

	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Minute)
	defer cancel()

	// App pool: the tenant tag reads and writes run here, under RLS, never the migrator bypass.
	appPool, err := database.NewPool(ctx, appDSN)
	if err != nil {
		log.Fatalf("tag-missions: app pool: %v", err)
	}
	defer appPool.Close()
	// Fail closed unless this is the NOBYPASSRLS app role, so the backfill can never bypass tenant RLS.
	if err := database.AssertRole(ctx, appPool, database.AppRoleName()); err != nil {
		log.Fatalf("tag-missions: app pool role check: %v", err)
	}
	appDB := tenant.New(appPool)

	// Migrator pool: enumerate org ids and read the global alias pairs ONLY. No tenant rows are written here.
	migPool, err := pgxpool.New(ctx, migDSN)
	if err != nil {
		log.Fatalf("tag-missions: migrator pool: %v", err)
	}
	defer migPool.Close()
	if err := database.AssertMigratorRole(ctx, migPool); err != nil {
		log.Fatalf("tag-missions: %v", err)
	}

	// Build the AliasIndex ONCE from the global alias pairs (read via the migrator) and share it across every
	// org, so the alias compile happens one time, not per org. The reader is the same MissionAliasPairs query
	// the global pass uses, so both passes classify against an identical curated alias layer.
	idx, err := federalsignal.BuildAliasIndex(ctx, federalsignal.NewRepo(migPool))
	if err != nil {
		log.Fatalf("tag-missions: build alias index: %v", err)
	}

	orgIDs, err := listOrgs(ctx, migPool, *orgFlag)
	if err != nil {
		log.Fatalf("tag-missions: list orgs: %v", err)
	}
	if len(orgIDs) == 0 {
		log.Print("tag-missions: no organizations to process")
		return
	}

	svc := federalsignal.NewTenantTagService(federalsignal.NewTenantTagRepo(appDB), idx)

	var totalCandTags, totalReqTags int
	for _, orgID := range orgIDs {
		orgCtx := tenant.WithOrg(ctx, orgID.String())

		cRes, cerr := svc.TagCandidatesForOrg(orgCtx)
		if cerr != nil {
			log.Fatalf("tag-missions: org %s tag candidates: %v", orgID, cerr)
		}
		rRes, rerr := svc.TagRequirementsForOrg(orgCtx)
		if rerr != nil {
			log.Fatalf("tag-missions: org %s tag requirements: %v", orgID, rerr)
		}

		log.Printf("tag-missions: org %s scanned %d candidates wrote %d candidate tags, scanned %d requirements wrote %d requirement tags",
			orgID, cRes.CandidatesScanned, cRes.CandidateTags, rRes.RequirementsScanned, rRes.RequirementTags)
		totalCandTags += cRes.CandidateTags
		totalReqTags += rRes.RequirementTags
	}
	log.Printf("tag-missions: DONE, %d org(s), wrote %d candidate mission tags and %d requirement mission tags",
		len(orgIDs), totalCandTags, totalReqTags)
}

// listOrgs returns the organization ids to process. It reads them through the migrator role (which sees
// every org) so one run covers the cluster; the per org tagging then runs under the app role and RLS. A
// -org value restricts the run to a single org for a targeted backfill.
func listOrgs(ctx context.Context, migPool *pgxpool.Pool, only string) ([]uuid.UUID, error) {
	if only != "" {
		id, err := uuid.Parse(strings.TrimSpace(only))
		if err != nil {
			return nil, err
		}
		return []uuid.UUID{id}, nil
	}
	return dbgen.New(migPool).ListAllOrganizationIDs(ctx)
}

// fipsSelfTest derives one PBKDF2 key at the production iteration count under the active crypto module. It
// is the in build FIPS gate, identical to the api and missionclass-sync binaries, so this backfill binary
// is held to the same validated crypto posture.
func fipsSelfTest() error {
	_, err := pbkdf2.Key(sha256.New, "missionmatch fips self test", []byte("self test salt"), security.PBKDF2Iterations, 32)
	return err
}

func firstNonEmpty(vals ...string) string {
	for _, v := range vals {
		if s := strings.TrimSpace(v); s != "" {
			return s
		}
	}
	return ""
}