partial reads never cached; extractor version 3 discards 0.0.7 caches

3 files changed, 34 insertions, 2 deletions

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 80355e1..93b0854 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -14,7 +14,8 @@
   krino could not read; an exclude holds on an unknown only when the text
   could change its answer; `explain` shows `?`. A document read only in part
   answers the keywords found in what was read, and leaves the others
-  unknown.
+  unknown; its answers are never cached, and keyword caches written by
+  0.0.7 are discarded.
 - `krino log` marks a run "(partly undone)" while some of its reversible
   steps are not reversed, and "(undone)" only once every one is.
 - Tests read real documents made by LibreOffice and pandoc (docx, odt, ods,
diff --git a/internal/engine/exclude_test.go b/internal/engine/exclude_test.go
index 8308a82..5a758bb 100644
--- a/internal/engine/exclude_test.go
+++ b/internal/engine/exclude_test.go
@@ -455,3 +455,31 @@ func TestExplainLeavesTheCacheAlone(t *testing.T) {
 		t.Errorf("explain removed the cache: %v", err)
 	}
 }
+
+// TestPartlyReadableDocumentIsNotCached: a partial read's answers never
+// enter the keyword cache, so a second run with the cache still sets the
+// file aside (plan 11 review M2).
+func TestPartlyReadableDocumentIsNotCached(t *testing.T) {
+	h, dl := excludeTree(t, map[string]string{})
+	os.MkdirAll(dl, 0o755)
+	partDocx(t, filepath.Join(dl, "part.docx"), "good body text")
+	main := writeConfig(t, h, `(include "dl")`, map[string]string{"dl": `
+(path "~/dl")
+(exclude (content "confidential"))
+(rule "all" (move "Out"))
+`})
+	for run := 1; run <= 2; run++ {
+		e, errs := Load(main)
+		if len(errs) > 0 {
+			t.Fatal(errs)
+		}
+		e.CacheDir = filepath.Join(h, "cache")
+		r, err := e.Match(context.Background(), e.Dirs[0])
+		if err != nil {
+			t.Fatal(err)
+		}
+		if len(r.Matched) != 1 || !strings.HasSuffix(r.Matched[0].Excluded, "(content unreadable)") {
+			t.Fatalf("run %d: matched = %+v; want part.docx set aside", run, r.Matched)
+		}
+	}
+}
diff --git a/internal/extract/extract.go b/internal/extract/extract.go
index 0843c42..c0f0262 100644
--- a/internal/extract/extract.go
+++ b/internal/extract/extract.go
@@ -17,7 +17,10 @@ import (
 // Version is the version of the text this package extracts. Bump it
 // whenever a change could make any format's text differ, so every keyword
 // cache built from the old text is discarded (Fingerprint).
-const Version = 2
+//
+// 3: a document read only in part (ErrPartial) and text that turns binary
+// answer differently from 0.0.7, whose caches may hold their answers.
+const Version = 3
 
 var (
 	// ErrUnsupported is returned when the format carries no text krino