Add a AfterFileSeen Collector method for exporting file stats from …

…plugins

PiperOrigin-RevId: 638675789

extractor/filesystem/extractor.go

@@ -57,6 +57,8 @@ type ScanInput struct {
 	// A reader for accessing contents of the file.
 	// Note that the file is closed by the core library, not the plugin.
 	Reader io.Reader
+	// Stats collector to export stats internally from a plugin.
+	Stats stats.Collector
 }
 
 // Config stores the config settings for an extraction run.
@@ -279,7 +281,13 @@ func (wc *walkContext) runExtractor(ex Extractor, path string, mode fs.FileMode)
 	wc.extractCalls++
 
 	start := time.Now()
-	results, err := ex.Extract(wc.ctx, &ScanInput{Path: path, ScanRoot: wc.scanRoot, Info: info, Reader: rc})
+	results, err := ex.Extract(wc.ctx, &ScanInput{
+		Path:     path,
+		ScanRoot: wc.scanRoot,
+		Info:     info,
+		Reader:   rc,
+		Stats:    wc.stats,
+	})
 	wc.stats.AfterExtractorRun(ex.Name(), time.Since(start), err)
 	if err != nil {
 		addErrToMap(wc.errors, ex.Name(), fmt.Errorf("%s: %w", path, err))

stats/collector.go

@@ -32,6 +32,19 @@ type Collector interface {
 	// AfterResultsExported is called after results have been exported. destination should merely be
 	// a category of where the result was written to (e.g. 'file', 'http'), not the precise location.
 	AfterResultsExported(destination string, bytes int, err error)
+	AfterFileSeen(pluginName string, filestats *FileStats)
+}
+
+// FileStats is a struct containing stats about a file that was extracted. If
+// the file was skipped due to an error during extraction, `Error` will be
+// populated.
+type FileStats struct {
+	Path          string
+	Error         error
+	FileSizeBytes int64
+	// For extractors that unarchive a compressed files, this reports the bytes
+	// that were opened during the unarchiving process.
+	UncompressedBytes int64
 }
 
 // NoopCollector implements Collector by doing nothing.
@@ -51,3 +64,6 @@ func (c NoopCollector) AfterScan(runtime time.Duration, status *plugin.ScanStatu
 
 // AfterResultsExported implements Collector by doing nothing.
 func (c NoopCollector) AfterResultsExported(destination string, bytes int, err error) {}
+
+// AfterFileSeen implements Collector by doing nothing.
+func (c NoopCollector) AfterFileSeen(name string, filestats *FileStats) {}