Add WhisperX configuration contract

This commit is contained in:
2026-05-03 15:23:47 -05:00
parent 598584cf2d
commit c2a715d3f6
8 changed files with 227 additions and 10 deletions

View File

@@ -14,13 +14,14 @@ func TestLoadAndValidate(t *testing.T) {
sessionYAML string
wantLoadErr string
wantValidate string
checkDefault bool
}{
{
name: "valid minimal config",
pipelineYAML: `workspace:
root: /tmp/narratio
whisperx:
timeout: 10m
transcribe_url: https://transcription.ai.rakestrawhome.com/transcribe
seriatim:
timeout: 30s
audita:
@@ -37,6 +38,7 @@ inputs:
autocorrect_file: ./autocorrect.yml
glossary_file: ./glossary.yml
`,
checkDefault: true,
},
{
name: "unknown pipeline field fails",
@@ -53,6 +55,23 @@ inputs:
`,
wantLoadErr: "pipeline file",
},
{
name: "unknown whisperx field fails",
pipelineYAML: `workspace:
root: /tmp/narratio
whisperx:
transcribe_url: https://transcription.ai.rakestrawhome.com/transcribe
bogus: true
`,
sessionYAML: `session_id: 2026-05-03
inputs:
audio_dir: ./audio
speakers_file: ./speakers.yml
autocorrect_file: ./autocorrect.yml
glossary_file: ./glossary.yml
`,
wantLoadErr: "strict decode failed",
},
{
name: "unknown session field fails",
pipelineYAML: `workspace:
@@ -72,6 +91,8 @@ unknown_field: true
name: "missing required field fails",
pipelineYAML: `workspace:
root: /tmp/narratio
whisperx:
transcribe_url: https://transcription.ai.rakestrawhome.com/transcribe
`,
sessionYAML: `session_id: ""
inputs:
@@ -82,11 +103,43 @@ inputs:
`,
wantValidate: "session config \"session.yml\" invalid: session.session_id is required",
},
{
name: "missing transcribe_url fails",
pipelineYAML: `workspace:
root: /tmp/narratio
whisperx:
`,
sessionYAML: `session_id: 2026-05-03
inputs:
audio_dir: ./audio
speakers_file: ./speakers.yml
autocorrect_file: ./autocorrect.yml
glossary_file: ./glossary.yml
`,
wantValidate: "pipeline config \"pipeline.yml\" invalid: pipeline.whisperx.transcribe_url is required",
},
{
name: "invalid transcribe_url fails",
pipelineYAML: `workspace:
root: /tmp/narratio
whisperx:
transcribe_url: ://
`,
sessionYAML: `session_id: 2026-05-03
inputs:
audio_dir: ./audio
speakers_file: ./speakers.yml
autocorrect_file: ./autocorrect.yml
glossary_file: ./glossary.yml
`,
wantValidate: "pipeline config \"pipeline.yml\" invalid: pipeline.whisperx.transcribe_url must be a valid URL",
},
{
name: "invalid timeout fails",
pipelineYAML: `workspace:
root: /tmp/narratio
whisperx:
transcribe_url: https://transcription.ai.rakestrawhome.com/transcribe
timeout: definitely-not-a-duration
`,
sessionYAML: `session_id: 2026-05-03
@@ -98,6 +151,57 @@ inputs:
`,
wantValidate: "pipeline config \"pipeline.yml\" invalid: pipeline.whisperx.timeout must be a valid duration",
},
{
name: "invalid retry_delay fails",
pipelineYAML: `workspace:
root: /tmp/narratio
whisperx:
transcribe_url: https://transcription.ai.rakestrawhome.com/transcribe
retry_delay: not-a-duration
`,
sessionYAML: `session_id: 2026-05-03
inputs:
audio_dir: ./audio
speakers_file: ./speakers.yml
autocorrect_file: ./autocorrect.yml
glossary_file: ./glossary.yml
`,
wantValidate: "pipeline config \"pipeline.yml\" invalid: pipeline.whisperx.retry_delay must be a valid duration",
},
{
name: "negative retries fails",
pipelineYAML: `workspace:
root: /tmp/narratio
whisperx:
transcribe_url: https://transcription.ai.rakestrawhome.com/transcribe
retries: -1
`,
sessionYAML: `session_id: 2026-05-03
inputs:
audio_dir: ./audio
speakers_file: ./speakers.yml
autocorrect_file: ./autocorrect.yml
glossary_file: ./glossary.yml
`,
wantValidate: "pipeline config \"pipeline.yml\" invalid: pipeline.whisperx.retries must be >= 0",
},
{
name: "invalid concurrency fails",
pipelineYAML: `workspace:
root: /tmp/narratio
whisperx:
transcribe_url: https://transcription.ai.rakestrawhome.com/transcribe
concurrency: 0
`,
sessionYAML: `session_id: 2026-05-03
inputs:
audio_dir: ./audio
speakers_file: ./speakers.yml
autocorrect_file: ./autocorrect.yml
glossary_file: ./glossary.yml
`,
wantValidate: "pipeline config \"pipeline.yml\" invalid: pipeline.whisperx.concurrency must be > 0",
},
}
for _, tt := range tests {
@@ -127,6 +231,23 @@ inputs:
if cfg.SessionPath != sessionPath {
t.Fatalf("SessionPath = %q, want %q", cfg.SessionPath, sessionPath)
}
if tt.checkDefault {
if cfg.Pipeline.WhisperX.Language != "en" {
t.Fatalf("whisperx.language = %q, want %q", cfg.Pipeline.WhisperX.Language, "en")
}
if cfg.Pipeline.WhisperX.Timeout != "30m" {
t.Fatalf("whisperx.timeout = %q, want %q", cfg.Pipeline.WhisperX.Timeout, "30m")
}
if cfg.Pipeline.WhisperX.RetryDelay != "2s" {
t.Fatalf("whisperx.retry_delay = %q, want %q", cfg.Pipeline.WhisperX.RetryDelay, "2s")
}
if cfg.Pipeline.WhisperX.Retries == nil || *cfg.Pipeline.WhisperX.Retries != 3 {
t.Fatalf("whisperx.retries = %v, want 3", cfg.Pipeline.WhisperX.Retries)
}
if cfg.Pipeline.WhisperX.Concurrency == nil || *cfg.Pipeline.WhisperX.Concurrency != 2 {
t.Fatalf("whisperx.concurrency = %v, want 2", cfg.Pipeline.WhisperX.Concurrency)
}
}
err = Validate(cfg)
if tt.wantValidate != "" {
@@ -146,8 +267,20 @@ inputs:
}
func TestValidateMissingAudioSource(t *testing.T) {
retries := 3
concurrency := 2
cfg := &Config{
Pipeline: &PipelineConfig{Workspace: WorkspaceConfig{Root: "/tmp/narratio"}},
Pipeline: &PipelineConfig{
Workspace: WorkspaceConfig{Root: "/tmp/narratio"},
WhisperX: WhisperXConfig{
TranscribeURL: "https://transcription.ai.rakestrawhome.com/transcribe",
Language: "en",
Timeout: "30m",
Retries: &retries,
RetryDelay: "2s",
Concurrency: &concurrency,
},
},
Session: &SessionConfig{
SessionID: "2026-05-03",
Inputs: SessionInputsConfig{