Restore tracker mapping linking, drop only create

The tracker-mapping worker had been reduced to catalog resolution only,
which removed not just the auto-creation of an org ThirdParty but also
the auto-linking of an existing one. Only the creation needed to go: it
raced the load-then-create check and produced duplicate vendors.

Restore the full org ThirdParty resolution (exact common-id link,
sibling direct-link, high-confidence heuristic, and the disambiguation
agent) and remove only the CreateFromCommon branch and its
categorisation gate. When nothing matches, the worker now leaves
third_party_id unset rather than creating a vendor; creation happens
exclusively through the explicit ImportFromCommon action. Drop the
now-dead CreateFromCommon helper and rename match.go to common_match.go.

Fix a latent test bug surfaced by actually running the DB-backed suite
(skipped in CI without Postgres): the heuristic-match candidate lacked
Level 1, so the level-filtered candidate loader excluded it and the old
fallback create masked the miss.

Signed-off-by: Émile Ré <emile@probo.com>
This commit is contained in:
Émile Ré
2026-06-11 10:55:29 +02:00
parent 9a7bc26d49
commit 145aeaf402
17 changed files with 1323 additions and 238 deletions

View File

@@ -213,6 +213,17 @@ func (b *Builder) Build() (*probodconfig.FullConfig, error) {
Temperature: b.getEnvFloatPtr("AGENT_THIRD_PARTY_VETTER_TEMPERATURE"),
MaxTokens: b.getEnvIntPtr("AGENT_THIRD_PARTY_VETTER_MAX_TOKENS"),
},
ThirdPartyDisambiguation: probodconfig.LLMAgentConfig{
Provider: b.getEnvOrDefault("AGENT_THIRD_PARTY_DISAMBIGUATION_PROVIDER", ""),
ModelName: b.getEnvOrDefault("AGENT_THIRD_PARTY_DISAMBIGUATION_MODEL_NAME", ""),
// The disambiguation agent emits a single id plus a
// short rationale, but the budget must leave headroom
// for reasoning models whose reasoning tokens count
// against max_tokens; too small a budget truncates the
// JSON.
Temperature: b.getEnvFloatPtr("AGENT_THIRD_PARTY_DISAMBIGUATION_TEMPERATURE"),
MaxTokens: new(b.getEnvIntOrDefault("AGENT_THIRD_PARTY_DISAMBIGUATION_MAX_TOKENS", 4096)),
},
TrackerMapping: probodconfig.LLMAgentConfig{
Provider: b.getEnvOrDefault("AGENT_TRACKER_MAPPING_PROVIDER", ""),
ModelName: b.getEnvOrDefault("AGENT_TRACKER_MAPPING_MODEL_NAME", ""),
@@ -267,11 +278,12 @@ func (b *Builder) Build() (*probodconfig.FullConfig, error) {
MaxConcurrency: b.getEnvIntOrDefault("THIRD_PARTY_VETTING_MAX_CONCURRENCY", 1),
},
TrackerMappingWorker: probodconfig.TrackerMappingWorkerConfig{
Interval: b.getEnvIntOrDefault("TRACKER_MAPPING_INTERVAL", 10),
MaxConcurrency: b.getEnvIntOrDefault("TRACKER_MAPPING_MAX_CONCURRENCY", 3),
StaleAfter: b.getEnvIntOrDefault("TRACKER_MAPPING_STALE_AFTER", 600),
AgentTimeout: b.getEnvIntOrDefault("TRACKER_MAPPING_AGENT_TIMEOUT", 45),
AgentMaxTurns: b.getEnvIntOrDefault("TRACKER_MAPPING_AGENT_MAX_TURNS", 10),
Interval: b.getEnvIntOrDefault("TRACKER_MAPPING_INTERVAL", 10),
MaxConcurrency: b.getEnvIntOrDefault("TRACKER_MAPPING_MAX_CONCURRENCY", 3),
StaleAfter: b.getEnvIntOrDefault("TRACKER_MAPPING_STALE_AFTER", 600),
AgentTimeout: b.getEnvIntOrDefault("TRACKER_MAPPING_AGENT_TIMEOUT", 45),
AgentMaxTurns: b.getEnvIntOrDefault("TRACKER_MAPPING_AGENT_MAX_TURNS", 10),
DisambiguationAgentTimeout: b.getEnvIntOrDefault("TRACKER_MAPPING_DISAMBIGUATION_AGENT_TIMEOUT", 45),
},
CommonPatternEnrichmentWorker: probodconfig.CommonPatternEnrichmentWorkerConfig{
Interval: b.getEnvIntOrDefault("COMMON_PATTERN_ENRICHMENT_INTERVAL", 10),

View File

@@ -220,6 +220,10 @@ func TestBuilder_Build_Defaults(t *testing.T) {
assert.Empty(t, cfg.Probod.Agents.ThirdPartyVetter.ModelName)
assert.Nil(t, cfg.Probod.Agents.ThirdPartyVetter.Temperature)
assert.Nil(t, cfg.Probod.Agents.ThirdPartyVetter.MaxTokens)
assert.Empty(t, cfg.Probod.Agents.ThirdPartyDisambiguation.Provider)
assert.Empty(t, cfg.Probod.Agents.ThirdPartyDisambiguation.ModelName)
assert.Nil(t, cfg.Probod.Agents.ThirdPartyDisambiguation.Temperature)
assert.Equal(t, new(4096), cfg.Probod.Agents.ThirdPartyDisambiguation.MaxTokens)
assert.Empty(t, cfg.Probod.Agents.TrackerMapping.Provider)
assert.Empty(t, cfg.Probod.Agents.TrackerMapping.ModelName)
assert.Nil(t, cfg.Probod.Agents.TrackerMapping.Temperature)
@@ -235,6 +239,7 @@ func TestBuilder_Build_Defaults(t *testing.T) {
assert.Equal(t, 600, cfg.Probod.TrackerMappingWorker.StaleAfter)
assert.Equal(t, 45, cfg.Probod.TrackerMappingWorker.AgentTimeout)
assert.Equal(t, 10, cfg.Probod.TrackerMappingWorker.AgentMaxTurns)
assert.Equal(t, 45, cfg.Probod.TrackerMappingWorker.DisambiguationAgentTimeout)
assert.Equal(t, 10, cfg.Probod.CommonPatternEnrichmentWorker.Interval)
assert.Equal(t, 2, cfg.Probod.CommonPatternEnrichmentWorker.MaxConcurrency)
assert.Equal(t, 600, cfg.Probod.CommonPatternEnrichmentWorker.StaleAfter)
@@ -337,6 +342,11 @@ func TestBuilder_Build_CustomValues(t *testing.T) {
env["AGENT_THIRD_PARTY_VETTER_MODEL_NAME"] = "gpt-4o"
env["AGENT_THIRD_PARTY_VETTER_TEMPERATURE"] = "0.3"
env["AGENT_THIRD_PARTY_VETTER_MAX_TOKENS"] = "8192"
// Agents — third-party-disambiguation override
env["AGENT_THIRD_PARTY_DISAMBIGUATION_PROVIDER"] = "anthropic"
env["AGENT_THIRD_PARTY_DISAMBIGUATION_MODEL_NAME"] = "claude-sonnet-4-20250514"
env["AGENT_THIRD_PARTY_DISAMBIGUATION_TEMPERATURE"] = "0.4"
env["AGENT_THIRD_PARTY_DISAMBIGUATION_MAX_TOKENS"] = "2048"
// Agents — tracker-mapping override
env["AGENT_TRACKER_MAPPING_PROVIDER"] = "openai"
env["AGENT_TRACKER_MAPPING_MODEL_NAME"] = "gpt-4o-mini"
@@ -353,6 +363,7 @@ func TestBuilder_Build_CustomValues(t *testing.T) {
env["TRACKER_MAPPING_STALE_AFTER"] = "1200"
env["TRACKER_MAPPING_AGENT_TIMEOUT"] = "30"
env["TRACKER_MAPPING_AGENT_MAX_TURNS"] = "6"
env["TRACKER_MAPPING_DISAMBIGUATION_AGENT_TIMEOUT"] = "35"
env["COMMON_PATTERN_ENRICHMENT_INTERVAL"] = "15"
env["COMMON_PATTERN_ENRICHMENT_MAX_CONCURRENCY"] = "4"
env["COMMON_PATTERN_ENRICHMENT_STALE_AFTER"] = "900"
@@ -452,6 +463,11 @@ func TestBuilder_Build_CustomValues(t *testing.T) {
assert.Equal(t, "gpt-4o", cfg.Probod.Agents.ThirdPartyVetter.ModelName)
assert.Equal(t, new(0.3), cfg.Probod.Agents.ThirdPartyVetter.Temperature)
assert.Equal(t, new(8192), cfg.Probod.Agents.ThirdPartyVetter.MaxTokens)
// Agents — third-party-disambiguation overrides
assert.Equal(t, "anthropic", cfg.Probod.Agents.ThirdPartyDisambiguation.Provider)
assert.Equal(t, "claude-sonnet-4-20250514", cfg.Probod.Agents.ThirdPartyDisambiguation.ModelName)
assert.Equal(t, new(0.4), cfg.Probod.Agents.ThirdPartyDisambiguation.Temperature)
assert.Equal(t, new(2048), cfg.Probod.Agents.ThirdPartyDisambiguation.MaxTokens)
// Agents — tracker-mapping overrides
assert.Equal(t, "openai", cfg.Probod.Agents.TrackerMapping.Provider)
assert.Equal(t, "gpt-4o-mini", cfg.Probod.Agents.TrackerMapping.ModelName)
@@ -468,6 +484,7 @@ func TestBuilder_Build_CustomValues(t *testing.T) {
assert.Equal(t, 1200, cfg.Probod.TrackerMappingWorker.StaleAfter)
assert.Equal(t, 30, cfg.Probod.TrackerMappingWorker.AgentTimeout)
assert.Equal(t, 6, cfg.Probod.TrackerMappingWorker.AgentMaxTurns)
assert.Equal(t, 35, cfg.Probod.TrackerMappingWorker.DisambiguationAgentTimeout)
assert.Equal(t, 15, cfg.Probod.CommonPatternEnrichmentWorker.Interval)
assert.Equal(t, 4, cfg.Probod.CommonPatternEnrichmentWorker.MaxConcurrency)
assert.Equal(t, 900, cfg.Probod.CommonPatternEnrichmentWorker.StaleAfter)