From e0dddff176a01d410904b6750b3395de4f7e54db Mon Sep 17 00:00:00 2001
From: Lukasz Kasprzak <lukas@labunix.xyz>
Date: Mon, 14 Sep 2026 21:39:00 +0200
Subject: plan 9: keyword cache keys on extension, max-read and Unicode tables,
 trims removed keywords

---
 internal/engine/match.go | 29 +++++++++++++++++++++--------
 1 file changed, 21 insertions(+), 8 deletions(-)

(limited to 'internal/engine/match.go')

diff --git a/internal/engine/match.go b/internal/engine/match.go
index b922829..08d75af 100644
--- a/internal/engine/match.go
+++ b/internal/engine/match.go
@@ -114,7 +114,11 @@ func (e *Engine) Match(ctx context.Context, d *Dir) (*Result, error) {
 				ids = append(ids, fileCacheID(f))
 			}
 		}
-		if err := run.cache.Save(e.cacheFile(d), ids); err != nil {
+		keys := make([]string, len(d.ContentKeywords))
+		for i, k := range d.ContentKeywords {
+			keys[i] = k.Key()
+		}
+		if err := run.cache.Save(e.cacheFile(d), ids, keys); err != nil {
 			warnings = append(warnings, "cache: "+err.Error())
 		}
 	}
@@ -297,11 +301,12 @@ 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)
+// cacheFingerprint identifies what a cached keyword answer of d depends on
+// besides the file: the extractor (its version and tools), normalisation
+// and its Unicode tables, the Go release, and d's max-read, which caps what
+// is read. A cache written under any other is discarded.
+func (e *Engine) cacheFingerprint(d *Dir) string {
+	return fmt.Sprintf("%s %s %s max-read=%d", e.Extract.Fingerprint(), norm.Fingerprint(), runtime.Version(), d.Settings.MaxRead)
 }
 
 // cacheFile is d's keyword cache file.
@@ -313,10 +318,18 @@ func (e *Engine) cacheFile(d *Dir) string {
 // engine has a CacheDir and the directory has content tests at all. A cache
 // that cannot be read is replaced by an empty one, reported as a warning.
 func (e *Engine) openCache(run *matchRun) []string {
-	if e.CacheDir == "" || len(run.d.ContentKeywords) == 0 {
+	if e.CacheDir == "" {
+		return nil
+	}
+	if len(run.d.ContentKeywords) == 0 {
+		// No content test left: a cache from an earlier configuration only
+		// holds keywords this directory no longer uses (review cache F6).
+		if err := os.Remove(e.cacheFile(run.d)); err != nil && !os.IsNotExist(err) {
+			return []string{"cache: " + err.Error()}
+		}
 		return nil
 	}
-	c, err := kwcache.Load(e.cacheFile(run.d), e.cacheFingerprint())
+	c, err := kwcache.Load(e.cacheFile(run.d), e.cacheFingerprint(run.d))
 	run.cache = c
 	if err != nil {
 		return []string{"cache: " + err.Error() + " (starting a new one)"}
-- 
cgit v1.3