Add output validation against a defined JSON schema

This commit is contained in:
2026-04-27 21:27:19 -05:00
parent 8a95dba276
commit 6cb739be55
14 changed files with 638 additions and 30 deletions

View File

@@ -0,0 +1,68 @@
package artifact
import (
"gitea.maximumdirect.net/eric/seriatim/internal/config"
"gitea.maximumdirect.net/eric/seriatim/internal/model"
"gitea.maximumdirect.net/eric/seriatim/schema"
)
const (
ApplicationName = "seriatim"
Version = "dev"
)
// FromMerged converts the internal merged transcript model into the public
// serialized output contract.
func FromMerged(cfg config.Config, merged model.MergedTranscript) schema.Transcript {
segments := make([]schema.Segment, len(merged.Segments))
for index, segment := range merged.Segments {
segments[index] = schema.Segment{
ID: segment.ID,
Source: segment.Source,
SourceSegmentIndex: copyIntPtr(segment.SourceSegmentIndex),
SourceRef: segment.SourceRef,
DerivedFrom: append([]string(nil), segment.DerivedFrom...),
Speaker: segment.Speaker,
Start: segment.Start,
End: segment.End,
Text: segment.Text,
Categories: append([]string(nil), segment.Categories...),
OverlapGroupID: segment.OverlapGroupID,
}
}
overlapGroups := make([]schema.OverlapGroup, len(merged.OverlapGroups))
for index, group := range merged.OverlapGroups {
overlapGroups[index] = schema.OverlapGroup{
ID: group.ID,
Start: group.Start,
End: group.End,
Segments: append([]string(nil), group.Segments...),
Speakers: append([]string(nil), group.Speakers...),
Class: group.Class,
Resolution: group.Resolution,
}
}
return schema.Transcript{
Metadata: schema.Metadata{
Application: ApplicationName,
Version: Version,
InputReader: cfg.InputReader,
InputFiles: append([]string(nil), cfg.InputFiles...),
PreprocessingModules: append([]string(nil), cfg.PreprocessingModules...),
PostprocessingModules: append([]string(nil), cfg.PostprocessingModules...),
OutputModules: append([]string(nil), cfg.OutputModules...),
},
Segments: segments,
OverlapGroups: overlapGroups,
}
}
func copyIntPtr(value *int) *int {
if value == nil {
return nil
}
copied := *value
return &copied
}

View File

@@ -6,8 +6,8 @@ import (
"os"
"gitea.maximumdirect.net/eric/seriatim/internal/config"
"gitea.maximumdirect.net/eric/seriatim/internal/model"
"gitea.maximumdirect.net/eric/seriatim/internal/report"
"gitea.maximumdirect.net/eric/seriatim/schema"
)
type jsonOutputWriter struct{}
@@ -16,7 +16,7 @@ func (jsonOutputWriter) Name() string {
return "json"
}
func (jsonOutputWriter) Write(ctx context.Context, out model.FinalTranscript, rpt report.Report, cfg config.Config) ([]report.Event, error) {
func (jsonOutputWriter) Write(ctx context.Context, out schema.Transcript, rpt report.Report, cfg config.Config) ([]report.Event, error) {
if err := ctx.Err(); err != nil {
return nil, err
}
@@ -34,6 +34,6 @@ func (jsonOutputWriter) Write(ctx context.Context, out model.FinalTranscript, rp
}
return []report.Event{
report.Info("output", "json", "wrote placeholder transcript JSON"),
report.Info("output", "json", "wrote transcript JSON"),
}, nil
}

View File

@@ -4,6 +4,7 @@ import (
"context"
"fmt"
"gitea.maximumdirect.net/eric/seriatim/internal/artifact"
"gitea.maximumdirect.net/eric/seriatim/internal/autocorrect"
"gitea.maximumdirect.net/eric/seriatim/internal/backchannel"
"gitea.maximumdirect.net/eric/seriatim/internal/coalesce"
@@ -12,6 +13,7 @@ import (
"gitea.maximumdirect.net/eric/seriatim/internal/model"
"gitea.maximumdirect.net/eric/seriatim/internal/overlap"
"gitea.maximumdirect.net/eric/seriatim/internal/report"
"gitea.maximumdirect.net/eric/seriatim/schema"
)
type noopPostprocessor struct {
@@ -52,6 +54,27 @@ func (assignIDs) Process(ctx context.Context, in model.MergedTranscript, cfg con
}, nil
}
type validateOutput struct{}
func (validateOutput) Name() string {
return "validate-output"
}
func (validateOutput) Process(ctx context.Context, in model.MergedTranscript, cfg config.Config) (model.MergedTranscript, []report.Event, error) {
if err := ctx.Err(); err != nil {
return model.MergedTranscript{}, nil, err
}
transcript := artifact.FromMerged(cfg, in)
if err := schema.ValidateTranscript(transcript); err != nil {
return model.MergedTranscript{}, nil, fmt.Errorf("validate-output: %w", err)
}
return in, []report.Event{
report.Info("postprocessing", "validate-output", fmt.Sprintf("validated %d output segment(s)", len(in.Segments))),
}, nil
}
type detectOverlaps struct{}
func (detectOverlaps) Name() string {

View File

@@ -0,0 +1,60 @@
package builtin
import (
"context"
"strings"
"testing"
"gitea.maximumdirect.net/eric/seriatim/internal/config"
"gitea.maximumdirect.net/eric/seriatim/internal/model"
)
func TestValidateOutputSucceedsAfterAssignIDs(t *testing.T) {
merged := model.MergedTranscript{
Segments: []model.Segment{
{Source: "input.json", Speaker: "Alice", Start: 1, End: 2, Text: "hello"},
{Source: "input.json", Speaker: "Alice", Start: 3, End: 4, Text: "again"},
},
}
withIDs, _, err := assignIDs{}.Process(context.Background(), merged, testConfig())
if err != nil {
t.Fatalf("assign IDs: %v", err)
}
got, events, err := validateOutput{}.Process(context.Background(), withIDs, testConfig())
if err != nil {
t.Fatalf("validate output: %v", err)
}
if len(got.Segments) != 2 {
t.Fatalf("segment count = %d, want 2", len(got.Segments))
}
if len(events) != 1 || !strings.Contains(events[0].Message, "validated 2 output segment(s)") {
t.Fatalf("events = %#v", events)
}
}
func TestValidateOutputFailsBeforeAssignIDs(t *testing.T) {
merged := model.MergedTranscript{
Segments: []model.Segment{
{Source: "input.json", Speaker: "Alice", Start: 1, End: 2, Text: "hello"},
},
}
_, _, err := validateOutput{}.Process(context.Background(), merged, testConfig())
if err == nil {
t.Fatal("expected validation error")
}
if !strings.Contains(err.Error(), "segment 0 has id 0; want 1") {
t.Fatalf("unexpected error: %v", err)
}
}
func testConfig() config.Config {
return config.Config{
InputReader: config.DefaultInputReader,
InputFiles: []string{"input.json"},
PreprocessingModules: []string{"validate-raw", "normalize-speakers", "trim-text"},
PostprocessingModules: []string{"assign-ids", "validate-output"},
OutputModules: []string{"json"},
}
}

View File

@@ -17,7 +17,7 @@ func NewRegistry() *pipeline.Registry {
registry.RegisterPostprocessor(fillerPostprocessor{})
registry.RegisterPostprocessor(coalescePostprocessor{})
registry.RegisterPostprocessor(assignIDs{})
registry.RegisterPostprocessor(noopPostprocessor{name: "validate-output"})
registry.RegisterPostprocessor(validateOutput{})
registry.RegisterPostprocessor(autocorrectPostprocessor{})
registry.RegisterOutputWriter(jsonOutputWriter{})

View File

@@ -102,6 +102,9 @@ func TestMergeWritesMergedOutputAndReport(t *testing.T) {
if !equalStrings(gotModules, wantModules) {
t.Fatalf("report event order mismatch:\ngot %v\nwant %v", gotModules, wantModules)
}
if !hasReportEvent(rpt, "postprocessing", "validate-output", "validated 3 output segment(s)") {
t.Fatal("expected validate-output report event")
}
}
func TestMergeTieBreakOrder(t *testing.T) {
@@ -153,6 +156,28 @@ func TestMergeTieBreakOrder(t *testing.T) {
}
}
func TestMergeValidateOutputBeforeAssignIDsFails(t *testing.T) {
dir := t.TempDir()
input := writeJSONFile(t, dir, "input.json", `{
"segments": [
{"start": 1, "end": 2, "text": "hello"}
]
}`)
output := filepath.Join(dir, "merged.json")
err := executeMerge(
"--input-file", input,
"--output-file", output,
"--postprocessing-modules", "validate-output,assign-ids",
)
if err == nil {
t.Fatal("expected validation error")
}
if !strings.Contains(err.Error(), "validate-output: segment 0 has id 0; want 1") {
t.Fatalf("unexpected error: %v", err)
}
}
func TestMergeDetectsOverlapGroups(t *testing.T) {
dir := t.TempDir()
inputA := writeJSONFile(t, dir, "a.json", `{

View File

@@ -6,6 +6,7 @@ import (
"gitea.maximumdirect.net/eric/seriatim/internal/config"
"gitea.maximumdirect.net/eric/seriatim/internal/model"
"gitea.maximumdirect.net/eric/seriatim/internal/report"
"gitea.maximumdirect.net/eric/seriatim/schema"
)
// ModelState identifies which representation a preprocessing module consumes.
@@ -52,5 +53,5 @@ type Postprocessor interface {
// OutputWriter emits final artifacts.
type OutputWriter interface {
Name() string
Write(ctx context.Context, out model.FinalTranscript, rpt report.Report, cfg config.Config) ([]report.Event, error)
Write(ctx context.Context, out schema.Transcript, rpt report.Report, cfg config.Config) ([]report.Event, error)
}

View File

@@ -4,14 +4,16 @@ import (
"context"
"fmt"
"gitea.maximumdirect.net/eric/seriatim/internal/artifact"
"gitea.maximumdirect.net/eric/seriatim/internal/config"
"gitea.maximumdirect.net/eric/seriatim/internal/model"
"gitea.maximumdirect.net/eric/seriatim/internal/report"
"gitea.maximumdirect.net/eric/seriatim/schema"
)
const (
applicationName = "seriatim"
version = "dev"
applicationName = artifact.ApplicationName
version = artifact.Version
)
// Run validates module composition, executes the pipeline, and emits outputs.
@@ -139,29 +141,8 @@ func validatePreprocessors(modules []Preprocessor) error {
return nil
}
func finalizeTranscript(cfg config.Config, merged model.MergedTranscript) model.FinalTranscript {
segments := make([]model.Segment, len(merged.Segments))
copy(segments, merged.Segments)
for index := range segments {
segments[index].Words = nil
segments[index].DerivedFrom = append([]string(nil), segments[index].DerivedFrom...)
}
overlapGroups := make([]model.OverlapGroup, len(merged.OverlapGroups))
copy(overlapGroups, merged.OverlapGroups)
return model.FinalTranscript{
Metadata: model.OutputMetadata{
Application: applicationName,
Version: version,
InputReader: cfg.InputReader,
InputFiles: append([]string(nil), cfg.InputFiles...),
PreprocessingModules: append([]string(nil), cfg.PreprocessingModules...),
PostprocessingModules: append([]string(nil), cfg.PostprocessingModules...),
OutputModules: append([]string(nil), cfg.OutputModules...),
},
Segments: segments,
OverlapGroups: overlapGroups,
}
func finalizeTranscript(cfg config.Config, merged model.MergedTranscript) schema.Transcript {
return artifact.FromMerged(cfg, merged)
}
func finalizeReport(cfg config.Config, events []report.Event) report.Report {