Implemented the initial Go framework

internal/builtin/preprocess.go

@@ -0,0 +1,82 @@
+package builtin
+
+import (
+	"context"
+	"fmt"
+
+	"gitea.maximumdirect.net/eric/seriatim/internal/config"
+	"gitea.maximumdirect.net/eric/seriatim/internal/model"
+	"gitea.maximumdirect.net/eric/seriatim/internal/pipeline"
+	"gitea.maximumdirect.net/eric/seriatim/internal/report"
+)
+
+type noopPreprocessor struct {
+	name     string
+	requires pipeline.ModelState
+	produces pipeline.ModelState
+}
+
+func (p noopPreprocessor) Name() string {
+	return p.name
+}
+
+func (p noopPreprocessor) Requires() pipeline.ModelState {
+	return p.requires
+}
+
+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) {
+	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)
+	}
+
+	in.State = p.produces
+	return in, []report.Event{
+		report.Info("preprocessing", p.name, "completed no-op preprocessing module"),
+	}, nil
+}
+
+type normalizeSpeakers struct{}
+
+func (normalizeSpeakers) Name() string {
+	return "normalize-speakers"
+}
+
+func (normalizeSpeakers) Requires() pipeline.ModelState {
+	return pipeline.StateRaw
+}
+
+func (normalizeSpeakers) Produces() pipeline.ModelState {
+	return pipeline.StateCanonical
+}
+
+func (normalizeSpeakers) 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 != pipeline.StateRaw {
+		return pipeline.PreprocessState{}, nil, fmt.Errorf("preprocessing module %q requires state %q but received %q", "normalize-speakers", pipeline.StateRaw, in.State)
+	}
+
+	canonical := make([]model.CanonicalTranscript, 0, len(in.Raw))
+	for _, raw := range in.Raw {
+		canonical = append(canonical, model.CanonicalTranscript{
+			Source:   raw.Source,
+			Segments: nil,
+		})
+	}
+
+	return pipeline.PreprocessState{
+			State:     pipeline.StateCanonical,
+			Raw:       append([]model.RawTranscript(nil), in.Raw...),
+			Canonical: canonical,
+		}, []report.Event{
+			report.Info("preprocessing", "normalize-speakers", "created placeholder canonical transcript(s)"),
+		}, nil
+}