eric/narratio
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 15 Wiki Activity

Updated transcript artifact names and canonical paths to use a consistent, role-based nomenclature

Browse Source
This commit is contained in:
Eric Rakestraw
2026-05-22 19:05:23 -05:00
parent e920f3a8d5
commit cee52aa092
56 changed files with 991 additions and 463 deletions
Download Patch File Download Diff File Expand all files Collapse all files

44
internal/config/defaults.go
View File

@@ -1,5 +1,7 @@
package config
import "gitea.maximumdirect.net/eric/narratio/internal/artifactmodel"
// Default filesystem locations for config lookup when config path flags are
// omitted. Order is highest to lowest precedence.
const (
@@ -40,32 +42,32 @@ const (
DefaultTrimBoundsTimeout = "10m"
DefaultTrimSeriatimReport = false
DefaultNormalizeOutputPath = "transcripts/normalized.json"
DefaultNormalizeOutputPath = artifactmodel.TranscriptPathFinal
DefaultNormalizeOutputSchema = "seriatim-intermediate"
DefaultNormalizeReport = true
DefaultArchiveEnabled = true
DefaultArchiveUploadRun = true
PathWorkDirSegment = "work"
PathInputsDirSegment = "inputs"
PathAudioDirSegment = "audio"
PathTranscriptsSegment = "transcripts"
PathTranscriptsRaw = "transcripts/raw"
PathTranscriptsTrimmed = "transcripts/trimmed"
PathArtifactsDirSegment = "artifacts"
PathReportsDirSegment = "reports"
PathConfigDirSegment = "config"
PathLogsDirSegment = "logs"
PathCurrentDirSegment = "current"
PathRunsDirSegment = "runs"
PathPreviousDirSegment = "previous"
PathManifestFile = "manifest.json"
PathLockFile = ".lock"
PathTranscriptMerged = "transcripts/merged.json"
PathTranscriptProcessed = "transcripts/processed.json"
PathTranscriptNormalized = "transcripts/normalized.json"
PathTranscriptTrimmed = "transcripts/trimmed.json"
PathWorkDirSegment = "work"
PathInputsDirSegment = "inputs"
PathAudioDirSegment = "audio"
PathTranscriptsSegment = "transcripts"
PathTranscriptsRaw = "transcripts/raw"
PathTranscriptsTrimmed = "transcripts/trimmed"
PathArtifactsDirSegment = "artifacts"
PathReportsDirSegment = "reports"
PathConfigDirSegment = "config"
PathLogsDirSegment = "logs"
PathCurrentDirSegment = "current"
PathRunsDirSegment = "runs"
PathPreviousDirSegment = "previous"
PathManifestFile = "manifest.json"
PathLockFile = ".lock"
PathTranscriptBase = artifactmodel.TranscriptPathBase
PathTranscriptPolished = artifactmodel.TranscriptPathPolished
PathTranscriptFinal = artifactmodel.TranscriptPathFinal
PathTranscriptFinalTrimmed = artifactmodel.TranscriptPathFinalTrimmed
S3CampaignsSegment = "campaigns"
S3SessionsSegment = "sessions"
@@ -78,7 +80,7 @@ const (
// DefaultArchivePromoteArtifacts defines the default archive promotion rules.
// Callers should copy this slice before mutating.
var DefaultArchivePromoteArtifacts = []ArchivePromotionRule{
{Source: "narratio.transcript.trimmed", Dest: PathTranscriptTrimmed},
{Source: artifactmodel.SourceTranscriptFinalTrimmed, Dest: PathTranscriptFinalTrimmed},
}
// DefaultPipelineConfigSearchPaths defines the default search order for

4
internal/config/load_validate_test.go
View File

@@ -853,8 +853,8 @@ inputs:
if cfg.Pipeline.Normalize == nil {
t.Fatal("normalize config should be present via defaults")
}
if cfg.Pipeline.Normalize.OutputPath != "transcripts/normalized.json" {
t.Fatalf("normalize.output_path = %q, want %q", cfg.Pipeline.Normalize.OutputPath, "transcripts/normalized.json")
if cfg.Pipeline.Normalize.OutputPath != "transcripts/final.json" {
t.Fatalf("normalize.output_path = %q, want %q", cfg.Pipeline.Normalize.OutputPath, "transcripts/final.json")
}
if cfg.Pipeline.Normalize.OutputSchema != "seriatim-intermediate" {
t.Fatalf("normalize.output_schema = %q, want %q", cfg.Pipeline.Normalize.OutputSchema, "seriatim-intermediate")

8
internal/config/normalize_test.go
View File

@@ -21,8 +21,8 @@ func TestNormalizeLoadAndValidate(t *testing.T) {
if cfg.Pipeline.Normalize == nil {
t.Fatal("normalize config should be present via defaults")
}
if cfg.Pipeline.Normalize.OutputPath != "transcripts/normalized.json" {
t.Fatalf("normalize.output_path = %q, want %q", cfg.Pipeline.Normalize.OutputPath, "transcripts/normalized.json")
if cfg.Pipeline.Normalize.OutputPath != "transcripts/final.json" {
t.Fatalf("normalize.output_path = %q, want %q", cfg.Pipeline.Normalize.OutputPath, "transcripts/final.json")
}
if cfg.Pipeline.Normalize.OutputSchema != "seriatim-intermediate" {
t.Fatalf("normalize.output_schema = %q, want %q", cfg.Pipeline.Normalize.OutputSchema, "seriatim-intermediate")
@@ -58,7 +58,7 @@ func TestNormalizeLoadAndValidate(t *testing.T) {
{
name: "invalid normalize output schema fails",
normalizeYAML: `normalize:
output_path: transcripts/normalized.json
output_path: transcripts/final.json
output_schema: not-a-schema
report: true
`,
@@ -76,7 +76,7 @@ func TestNormalizeLoadAndValidate(t *testing.T) {
{
name: "unknown normalize field fails strict decoding",
normalizeYAML: `normalize:
output_path: transcripts/normalized.json
output_path: transcripts/final.json
output_schema: seriatim-intermediate
report: true
bogus: true

40
internal/config/scriptorium_test.go
View File

@@ -198,7 +198,7 @@ func TestScriptoriumLoadAndValidate(t *testing.T) {
output_path: artifacts/session_recap.md
inputs:
transcript:
source: narratio.transcript.trimmed
source: narratio.transcript.final_trimmed
required: true
`,
},
@@ -287,7 +287,7 @@ func TestScriptoriumLoadAndValidate(t *testing.T) {
output_path: artifacts/session_recap.md
inputs:
transcript:
source: narratio.transcript.trimmed
source: narratio.transcript.final_trimmed
required: true
player_handout:
enabled: true
@@ -507,6 +507,42 @@ func TestScriptoriumLoadAndValidate(t *testing.T) {
}
}
func TestScriptoriumLegacyTranscriptSourcesRejected(t *testing.T) {
legacyTranscriptSources := []string{
"narratio.transcript." + "merged",
"narratio.transcript." + "full",
"narratio.transcript." + "trimmed",
}
for _, source := range legacyTranscriptSources {
t.Run(source, func(t *testing.T) {
pipelineYAML := testPipelineBaseYAML + `
scriptorium:
binary: scriptorium
artifacts:
session_recap:
enabled: true
prompt_id: dnd.session_recap
output_path: artifacts/session_recap.md
inputs:
transcript:
source: ` + source + `
required: true
`
pipelinePath, sessionPath := writeConfigFiles(t, pipelineYAML, testSessionBaseYAML)
cfg, err := Load(pipelinePath, sessionPath)
if err != nil {
t.Fatalf("Load() error = %v", err)
}
err = Validate(cfg)
wantErr := `pipeline.scriptorium.artifacts.session_recap.inputs.transcript.source "` + source + `" is unsupported`
if err == nil || !strings.Contains(err.Error(), wantErr) {
t.Fatalf("Validate() error = %v, want to contain %q", err, wantErr)
}
})
}
}
const testPipelineBaseYAML = `workspace:
root: /tmp/narratio
whisperx:

117
internal/config/storage_archive_test.go
View File

@@ -186,11 +186,11 @@ func TestSpoolAndArchiveDefaults(t *testing.T) {
if item.Required == nil || !*item.Required {
t.Fatalf("archive.promote_artifacts[0].required = %#v, want true", item.Required)
}
if item.Source != "narratio.transcript.trimmed" {
t.Fatalf("archive.promote_artifacts[0].source = %q, want narratio.transcript.trimmed", item.Source)
if item.Source != "narratio.transcript.final_trimmed" {
t.Fatalf("archive.promote_artifacts[0].source = %q, want narratio.transcript.final_trimmed", item.Source)
}
if item.Dest != "transcripts/trimmed.json" {
t.Fatalf("archive.promote_artifacts[0].dest = %q, want transcripts/trimmed.json", item.Dest)
if item.Dest != "transcripts/final.trimmed.json" {
t.Fatalf("archive.promote_artifacts[0].dest = %q, want transcripts/final.trimmed.json", item.Dest)
}
}
@@ -204,8 +204,8 @@ func TestArchivePromotionValidation(t *testing.T) {
name: "absolute dest path rejected",
ruleYML: `archive:
promote_artifacts:
- source: "narratio.transcript.trimmed"
dest: "/transcripts/trimmed.json"
- source: "narratio.transcript.final_trimmed"
dest: "/transcripts/final.trimmed.json"
`,
wantErr: "must be a relative path",
},
@@ -213,7 +213,7 @@ func TestArchivePromotionValidation(t *testing.T) {
name: "traversal dest path rejected",
ruleYML: `archive:
promote_artifacts:
- source: "narratio.transcript.trimmed"
- source: "narratio.transcript.final_trimmed"
dest: "../trimmed.json"
`,
wantErr: "must not contain path traversal",
@@ -223,7 +223,7 @@ func TestArchivePromotionValidation(t *testing.T) {
ruleYML: `archive:
promote_artifacts:
- source: "narratio.unknown"
dest: "transcripts/trimmed.json"
dest: "transcripts/final.trimmed.json"
`,
wantErr: "source \"narratio.unknown\" is unsupported",
},
@@ -231,9 +231,9 @@ func TestArchivePromotionValidation(t *testing.T) {
name: "duplicate destination rejected",
ruleYML: `archive:
promote_artifacts:
- source: "narratio.transcript.trimmed"
- source: "narratio.transcript.final_trimmed"
dest: "artifacts/shared.md"
- source: "narratio.transcript.full"
- source: "narratio.transcript.final"
dest: "artifacts/shared.md"
`,
wantErr: "duplicates another archive promotion destination",
@@ -278,6 +278,35 @@ archive:
}
}
func TestArchivePromotionLegacyTranscriptSourcesRejected(t *testing.T) {
legacyTranscriptSources := []string{
"narratio.transcript." + "merged",
"narratio.transcript." + "full",
"narratio.transcript." + "trimmed",
}
for _, source := range legacyTranscriptSources {
t.Run(source, func(t *testing.T) {
pipelineYAML := testPipelineBaseYAML + `
archive:
promote_artifacts:
- source: ` + source + `
dest: transcripts/final.trimmed.json
`
pipelinePath, sessionPath := writeConfigFiles(t, pipelineYAML, testSessionBaseYAML)
cfg, err := Load(pipelinePath, sessionPath)
if err != nil {
t.Fatalf("Load() error = %v", err)
}
err = Validate(cfg)
wantErr := `source "` + source + `" is unsupported`
if err == nil || !strings.Contains(err.Error(), wantErr) {
t.Fatalf("Validate() error = %v, want to contain %q", err, wantErr)
}
})
}
}
func TestArchivePromotionDerivesDestinationWhenOmitted(t *testing.T) {
tests := []struct {
name string
@@ -289,9 +318,9 @@ func TestArchivePromotionDerivesDestinationWhenOmitted(t *testing.T) {
pipelineYML: testPipelineBaseYAML + `
archive:
promote_artifacts:
- source: narratio.transcript.full
- source: narratio.transcript.final
`,
wantDest: "transcripts/normalized.json",
wantDest: "transcripts/final.json",
},
{
name: "configured source derives configured output path",
@@ -341,7 +370,7 @@ func TestArchiveLockValidation(t *testing.T) {
pipelineYML: testPipelineBaseYAML + `
archive:
locks:
- source: narratio.transcript.trimmed
- source: narratio.transcript.final_trimmed
reason: reviewed transcript
`,
},
@@ -382,8 +411,8 @@ archive:
pipelineYML: testPipelineBaseYAML + `
archive:
locks:
- source: narratio.transcript.trimmed
- source: " narratio.transcript.trimmed "
- source: narratio.transcript.final_trimmed
- source: " narratio.transcript.final_trimmed "
`,
wantErr: "duplicates another archive lock source",
},
@@ -410,12 +439,40 @@ archive:
}
}
func TestArchiveLockLegacyTranscriptSourcesRejected(t *testing.T) {
legacyTranscriptSources := []string{
"narratio.transcript." + "merged",
"narratio.transcript." + "full",
"narratio.transcript." + "trimmed",
}
for _, source := range legacyTranscriptSources {
t.Run(source, func(t *testing.T) {
pipelineYAML := testPipelineBaseYAML + `
archive:
locks:
- source: ` + source + `
`
pipelinePath, sessionPath := writeConfigFiles(t, pipelineYAML, testSessionBaseYAML)
cfg, err := Load(pipelinePath, sessionPath)
if err != nil {
t.Fatalf("Load() error = %v", err)
}
err = Validate(cfg)
wantErr := `source "` + source + `" is unsupported`
if err == nil || !strings.Contains(err.Error(), wantErr) {
t.Fatalf("Validate() error = %v, want to contain %q", err, wantErr)
}
})
}
}
func TestArchiveLockUnknownFieldFailsStrictDecode(t *testing.T) {
pipelineYAML := testPipelineBaseYAML + `
archive:
locks:
- source: narratio.transcript.trimmed
dest: transcripts/trimmed.json
- source: narratio.transcript.final_trimmed
dest: transcripts/final.trimmed.json
`
pipelinePath, sessionPath := writeConfigFiles(t, pipelineYAML, testSessionBaseYAML)
_, err := Load(pipelinePath, sessionPath)
@@ -428,8 +485,8 @@ func TestArchiveLegacyFromToFailsStrictDecode(t *testing.T) {
pipelineYAML := testPipelineBaseYAML + `
archive:
promote_artifacts:
- from: transcripts/trimmed.json
to: transcripts/trimmed.json
- from: transcripts/final.trimmed.json
to: transcripts/final.trimmed.json
`
pipelinePath, sessionPath := writeConfigFiles(t, pipelineYAML, testSessionBaseYAML)
_, err := Load(pipelinePath, sessionPath)
@@ -440,27 +497,27 @@ archive:
func TestArchiveLockStoreBytesStrictDecodeAndValidation(t *testing.T) {
store, err := LoadArchiveLockStoreBytes("locks.yml", []byte(`locks:
- source: narratio.transcript.trimmed
- source: narratio.transcript.final_trimmed
reason: reviewed
`), nil)
if err != nil {
t.Fatalf("LoadArchiveLockStoreBytes() error = %v", err)
}
if len(store.Locks) != 1 || store.Locks[0].Source != "narratio.transcript.trimmed" || store.Locks[0].Reason != "reviewed" {
if len(store.Locks) != 1 || store.Locks[0].Source != "narratio.transcript.final_trimmed" || store.Locks[0].Reason != "reviewed" {
t.Fatalf("locks = %#v", store.Locks)
}
_, err = LoadArchiveLockStoreBytes("locks.yml", []byte(`locks:
- source: narratio.transcript.trimmed
dest: transcripts/trimmed.json
- source: narratio.transcript.final_trimmed
dest: transcripts/final.trimmed.json
`), nil)
if err == nil || !strings.Contains(err.Error(), "strict decode failed") {
t.Fatalf("unknown field error = %v, want strict decode failed", err)
}
_, err = LoadArchiveLockStoreBytes("locks.yml", []byte(`locks:
- source: narratio.transcript.trimmed
- source: narratio.transcript.trimmed
- source: narratio.transcript.final_trimmed
- source: narratio.transcript.final_trimmed
`), nil)
if err == nil || !strings.Contains(err.Error(), "duplicates another archive lock source") {
t.Fatalf("duplicate error = %v", err)
@@ -469,19 +526,19 @@ func TestArchiveLockStoreBytesStrictDecodeAndValidation(t *testing.T) {
func TestMergeArchiveLockRulesStaticWins(t *testing.T) {
merged := MergeArchiveLockRules(
[]ArchiveLockRule{{Source: "narratio.transcript.trimmed", Reason: "static"}},
[]ArchiveLockRule{{Source: "narratio.transcript.final_trimmed", Reason: "static"}},
[]ArchiveLockRule{
{Source: "narratio.transcript.trimmed", Reason: "remote"},
{Source: "narratio.transcript.full", Reason: "remote full"},
{Source: "narratio.transcript.final_trimmed", Reason: "remote"},
{Source: "narratio.transcript.final", Reason: "remote full"},
},
)
if len(merged) != 2 {
t.Fatalf("merged len = %d, want 2: %#v", len(merged), merged)
}
if merged[0].Source != "narratio.transcript.trimmed" || merged[0].Reason != "static" {
if merged[0].Source != "narratio.transcript.final_trimmed" || merged[0].Reason != "static" {
t.Fatalf("merged[0] = %#v, want static lock", merged[0])
}
if merged[1].Source != "narratio.transcript.full" {
if merged[1].Source != "narratio.transcript.final" {
t.Fatalf("merged[1] = %#v, want remote full lock", merged[1])
}
}

16
internal/config/trim_test.go
View File

@@ -17,7 +17,7 @@ func TestTrimLoadAndValidate(t *testing.T) {
name: "valid trim config",
trimYAML: `trim:
enabled: true
output_path: transcripts/trimmed.json
output_path: transcripts/final.trimmed.json
bounds:
prompt_id: dnd_session.bounds
profile_id: ""
@@ -45,7 +45,7 @@ func TestTrimLoadAndValidate(t *testing.T) {
{
name: "enabled omitted defaults disabled",
trimYAML: `trim:
output_path: transcripts/trimmed.json
output_path: transcripts/final.trimmed.json
bounds:
prompt_id: dnd_session.bounds
transcript_input_name: transcript
@@ -65,7 +65,7 @@ func TestTrimLoadAndValidate(t *testing.T) {
name: "missing prompt id fails when enabled",
trimYAML: `trim:
enabled: true
output_path: transcripts/trimmed.json
output_path: transcripts/final.trimmed.json
bounds:
transcript_input_name: transcript
output_path: artifacts/session_bounds.json
@@ -76,7 +76,7 @@ func TestTrimLoadAndValidate(t *testing.T) {
name: "missing transcript input name fails when enabled",
trimYAML: `trim:
enabled: true
output_path: transcripts/trimmed.json
output_path: transcripts/final.trimmed.json
bounds:
prompt_id: dnd_session.bounds
output_path: artifacts/session_bounds.json
@@ -87,7 +87,7 @@ func TestTrimLoadAndValidate(t *testing.T) {
name: "missing bounds output path fails when enabled",
trimYAML: `trim:
enabled: true
output_path: transcripts/trimmed.json
output_path: transcripts/final.trimmed.json
bounds:
prompt_id: dnd_session.bounds
transcript_input_name: transcript
@@ -109,7 +109,7 @@ func TestTrimLoadAndValidate(t *testing.T) {
name: "invalid timeout fails",
trimYAML: `trim:
enabled: true
output_path: transcripts/trimmed.json
output_path: transcripts/final.trimmed.json
bounds:
prompt_id: dnd_session.bounds
transcript_input_name: transcript
@@ -122,7 +122,7 @@ func TestTrimLoadAndValidate(t *testing.T) {
name: "render debug true requires render output path",
trimYAML: `trim:
enabled: true
output_path: transcripts/trimmed.json
output_path: transcripts/final.trimmed.json
bounds:
prompt_id: dnd_session.bounds
transcript_input_name: transcript
@@ -135,7 +135,7 @@ func TestTrimLoadAndValidate(t *testing.T) {
name: "unknown trim field fails strict decoding",
trimYAML: `trim:
enabled: true
output_path: transcripts/trimmed.json
output_path: transcripts/final.trimmed.json
bounds:
prompt_id: dnd_session.bounds
transcript_input_name: transcript

33
internal/config/validate.go
View File

@@ -7,6 +7,8 @@ import (
"regexp"
"strings"
"time"
"gitea.maximumdirect.net/eric/narratio/internal/artifactmodel"
)
// Validate checks resolved configuration for required fields and parseable durations.
@@ -224,12 +226,11 @@ func MergeArchiveLockRules(staticLocks, remoteLocks []ArchiveLockRule) []Archive
func archiveSourceKnown(source string, scriptorium *ScriptoriumConfig) (string, error) {
trimmed := strings.TrimSpace(source)
if _, ok := artifactmodel.LookupRuntimeTranscriptArtifact(trimmed); ok {
return "", nil
}
switch trimmed {
case "narratio.transcript.merged",
"narratio.transcript.polished",
"narratio.transcript.full",
"narratio.transcript.trimmed",
"narratio.bounds.session":
case "narratio.bounds.session":
return "", nil
}
matches := narratioArtifactSourceRE.FindStringSubmatch(trimmed)
@@ -248,15 +249,10 @@ func archiveSourceKnown(source string, scriptorium *ScriptoriumConfig) (string,
func deriveArchivePromotionDest(source string, scriptorium *ScriptoriumConfig) (string, error) {
trimmed := strings.TrimSpace(source)
if spec, ok := artifactmodel.LookupRuntimeTranscriptArtifact(trimmed); ok {
return spec.CanonicalRelPath, nil
}
switch trimmed {
case "narratio.transcript.merged":
return PathTranscriptMerged, nil
case "narratio.transcript.polished":
return PathTranscriptProcessed, nil
case "narratio.transcript.full":
return PathTranscriptNormalized, nil
case "narratio.transcript.trimmed":
return PathTranscriptTrimmed, nil
case "narratio.bounds.session":
return filepath.ToSlash(filepath.Join(PathArtifactsDirSegment, "session_bounds.json")), nil
}
@@ -716,15 +712,10 @@ func validateScriptoriumInputSource(artifactName, inputName, source string, conf
}
func isStaticSupportedScriptoriumInputSource(source string) bool {
if _, ok := artifactmodel.LookupRuntimeTranscriptArtifact(source); ok {
return true
}
switch source {
case "narratio.transcript.merged":
return true
case "narratio.transcript.polished":
return true
case "narratio.transcript.full":
return true
case "narratio.transcript.trimmed":
return true
case "narratio.bounds.session":
return true
default: