plan 8: fuzz decoders; fold ẞ and invalid UTF-8 correctly, refuse non-UTF-8 paths in krino new

1 files changed, 9 insertions, 1 deletions

diff --git a/internal/engine/match.go b/internal/engine/match.go
index 1a8077a..b922829 100644
--- a/internal/engine/match.go
+++ b/internal/engine/match.go
@@ -17,6 +17,7 @@ import (
 	"krino/internal/cond"
 	"krino/internal/config"
 	"krino/internal/kwcache"
+	"krino/internal/norm"
 	"krino/internal/plan"
 	"krino/internal/scan"
 	"krino/internal/xdg"
@@ -296,6 +297,13 @@ func (e *Engine) Explain(ctx context.Context, path string) (*Explanation, error)
 	return &Explanation{Dir: d, File: sf, Skip: skip, Excludes: excludes, Excluded: excluded, Rules: rules}, nil
 }
 
+// cacheFingerprint identifies what a cached keyword answer depends on
+// besides the file: the extractor (its version and tools) and the
+// normalisation version. A cache written under any other is discarded.
+func (e *Engine) cacheFingerprint() string {
+	return fmt.Sprintf("%s norm%d", e.Extract.Fingerprint(), norm.Version)
+}
+
 // cacheFile is d's keyword cache file.
 func (e *Engine) cacheFile(d *Dir) string {
 	return filepath.Join(e.CacheDir, d.Name+".cache")
@@ -308,7 +316,7 @@ func (e *Engine) openCache(run *matchRun) []string {
 	if e.CacheDir == "" || len(run.d.ContentKeywords) == 0 {
 		return nil
 	}
-	c, err := kwcache.Load(e.cacheFile(run.d), e.Extract.Fingerprint())
+	c, err := kwcache.Load(e.cacheFile(run.d), e.cacheFingerprint())
 	run.cache = c
 	if err != nil {
 		return []string{"cache: " + err.Error() + " (starting a new one)"}