Cleaned up documentation and development artifcats in advance of release

internal/builtin/preprocess.go

@@ -13,35 +13,58 @@ import (
 	"gitea.maximumdirect.net/eric/seriatim/internal/speaker"
 )
 
-type noopPreprocessor struct {
-	name     string
-	requires pipeline.ModelState
-	produces pipeline.ModelState
+type validateRaw struct{}
+
+func (validateRaw) Name() string {
+	return "validate-raw"
 }
 
-func (p noopPreprocessor) Name() string {
-	return p.name
+func (validateRaw) Requires() pipeline.ModelState {
+	return pipeline.StateRaw
 }
 
-func (p noopPreprocessor) Requires() pipeline.ModelState {
-	return p.requires
+func (validateRaw) Produces() pipeline.ModelState {
+	return pipeline.StateRaw
 }
 
-func (p noopPreprocessor) Produces() pipeline.ModelState {
-	return p.produces
-}
-
-func (p noopPreprocessor) Process(ctx context.Context, in pipeline.PreprocessState, cfg config.Config) (pipeline.PreprocessState, []report.Event, error) {
+func (validateRaw) Process(ctx context.Context, in pipeline.PreprocessState, cfg config.Config) (pipeline.PreprocessState, []report.Event, error) {
 	if err := ctx.Err(); err != nil {
 		return pipeline.PreprocessState{}, nil, err
 	}
-	if in.State != p.requires {
-		return pipeline.PreprocessState{}, nil, fmt.Errorf("preprocessing module %q requires state %q but received %q", p.name, p.requires, in.State)
+	if in.State != pipeline.StateRaw {
+		return pipeline.PreprocessState{}, nil, fmt.Errorf("preprocessing module %q requires state %q but received %q", "validate-raw", pipeline.StateRaw, in.State)
+	}
+	if len(in.Raw) == 0 {
+		return pipeline.PreprocessState{}, nil, fmt.Errorf("validate-raw: no raw transcript(s) to validate")
+	}
+
+	for transcriptIndex, transcript := range in.Raw {
+		if strings.TrimSpace(transcript.Source) == "" {
+			return pipeline.PreprocessState{}, nil, fmt.Errorf("validate-raw: raw transcript %d has empty source", transcriptIndex)
+		}
+		for segmentIndex, segment := range transcript.Segments {
+			if segment.Start < 0 {
+				return pipeline.PreprocessState{}, nil, fmt.Errorf("validate-raw: raw transcript %q segment %d has negative start", transcript.Source, segmentIndex)
+			}
+			if segment.End < segment.Start {
+				return pipeline.PreprocessState{}, nil, fmt.Errorf("validate-raw: raw transcript %q segment %d has end before start", transcript.Source, segmentIndex)
+			}
+			for wordIndex, word := range segment.Words {
+				if !word.Timed {
+					continue
+				}
+				if word.Start < 0 {
+					return pipeline.PreprocessState{}, nil, fmt.Errorf("validate-raw: raw transcript %q segment %d word %d has negative start", transcript.Source, segmentIndex, wordIndex)
+				}
+				if word.End < word.Start {
+					return pipeline.PreprocessState{}, nil, fmt.Errorf("validate-raw: raw transcript %q segment %d word %d has end before start", transcript.Source, segmentIndex, wordIndex)
+				}
+			}
+		}
 	}
 
-	in.State = p.produces
 	return in, []report.Event{
-		report.Info("preprocessing", p.name, "completed no-op preprocessing module"),
+		report.Info("preprocessing", "validate-raw", fmt.Sprintf("validated %d raw transcript(s)", len(in.Raw))),
 	}, nil
 }