Implemented an autocorrect module at the postprocessing stage

internal/builtin/postprocess.go

@@ -2,7 +2,9 @@ package builtin
 
 import (
 	"context"
+	"fmt"
 
+	"gitea.maximumdirect.net/eric/seriatim/internal/autocorrect"
 	"gitea.maximumdirect.net/eric/seriatim/internal/config"
 	"gitea.maximumdirect.net/eric/seriatim/internal/model"
 	"gitea.maximumdirect.net/eric/seriatim/internal/report"
@@ -45,3 +47,31 @@ func (assignIDs) Process(ctx context.Context, in model.MergedTranscript, cfg con
 		report.Info("postprocessing", "assign-ids", "assigned final segment IDs"),
 	}, nil
 }
+
+type autocorrectPostprocessor struct{}
+
+func (autocorrectPostprocessor) Name() string {
+	return "autocorrect"
+}
+
+func (autocorrectPostprocessor) 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
+	}
+
+	rules, err := autocorrect.Load(cfg.AutocorrectFile)
+	if err != nil {
+		return model.MergedTranscript{}, nil, err
+	}
+
+	replacements := 0
+	for index := range in.Segments {
+		var count int
+		in.Segments[index].Text, count = rules.Apply(in.Segments[index].Text)
+		replacements += count
+	}
+
+	return in, []report.Event{
+		report.Info("postprocessing", "autocorrect", fmt.Sprintf("applied %d autocorrect replacement(s)", replacements)),
+	}, nil
+}

internal/builtin/registry.go

@@ -10,13 +10,12 @@ func NewRegistry() *pipeline.Registry {
 	registry.RegisterPreprocessor(noopPreprocessor{name: "validate-raw", requires: pipeline.StateRaw, produces: pipeline.StateRaw})
 	registry.RegisterPreprocessor(normalizeSpeakers{})
 	registry.RegisterPreprocessor(trimText{})
-	registry.RegisterPreprocessor(noopPreprocessor{name: "autocorrect", requires: pipeline.StateCanonical, produces: pipeline.StateCanonical})
 	registry.RegisterMerger(placeholderMerger{})
 	registry.RegisterPostprocessor(noopPostprocessor{name: "detect-overlaps"})
 	registry.RegisterPostprocessor(noopPostprocessor{name: "resolve-overlaps"})
 	registry.RegisterPostprocessor(assignIDs{})
 	registry.RegisterPostprocessor(noopPostprocessor{name: "validate-output"})
-	registry.RegisterPostprocessor(noopPostprocessor{name: "autocorrect"})
+	registry.RegisterPostprocessor(autocorrectPostprocessor{})
 	registry.RegisterOutputWriter(jsonOutputWriter{})
 
 	return registry