feat(corpus checker): allow specifying an allowlist of corpora (#5033)

kythe · Aug 11, 2021 · 1458594 · 1458594
1 parent d1ccce0
commit 1458594
Show file tree

Hide file tree

Showing 4 changed files with 25 additions and 3 deletions.
diff --git a/kythe/cxx/indexer/cxx/testdata/BUILD b/kythe/cxx/indexer/cxx/testdata/BUILD
@@ -884,6 +884,10 @@ cc_indexer_test(
 
 empty_corpus_test(
     name = "empty_corpus_test",
+    allowed_corpora = [
+        "file",
+        "unit",
+    ],
     entries = ":vname_corpus_entries.entries.gz",
 )
 

diff --git a/kythe/go/test/tools/empty_corpus_checker/empty_corpus_checker.go b/kythe/go/test/tools/empty_corpus_checker/empty_corpus_checker.go
@@ -36,8 +36,12 @@ import (
 	spb "kythe.io/kythe/proto/storage_go_proto"
 )
 
+var allowedCorpora flagutil.StringSet
+
 func init() {
 	flag.Usage = flagutil.SimpleUsage("Checks a stream of Entry protos via stdin for empty vname.corpus")
+
+	flag.Var(&allowedCorpora, "allowed_corpora", "Comma-separated list of corpora allowed in the input entrystream")
 }
 
 func main() {
@@ -77,5 +81,13 @@ func main() {
 	if emptyCorpusCount != 0 {
 		log.Fatalf("FAILURE: found %d sources with empty corpus", emptyCorpusCount)
 	}
+
+	if len(allowedCorpora) > 0 {
+		diff := allCorpora.Diff(stringset.Set(allowedCorpora))
+		if !diff.Empty() {
+			log.Fatalf("FAILURE: found entries with disallowed corpora: %v", diff)
+		}
+	}
+
 	log.Printf("Success! All vnames have a non-empty corpus.")
 }
diff --git a/kythe/go/test/tools/empty_corpus_checker/empty_corpus_test.bzl b/kythe/go/test/tools/empty_corpus_checker/empty_corpus_test.bzl
@@ -1,9 +1,14 @@
-def empty_corpus_test(name, entries):
+def empty_corpus_test(name, entries, allowed_corpora = []):
     """Instantiates a test that ensures all vnames in `entries` have a non-empty corpus."""
+
+    args = ["$(location %s)" % entries]
+    if len(allowed_corpora) > 0:
+        args.append("--allowed_corpora=" + ",".join(allowed_corpora))
+
     native.sh_test(
         name = name,
         srcs = ["//kythe/go/test/tools/empty_corpus_checker:empty_corpus_test.sh"],
-        args = ["$(location %s)" % entries],
+        args = args,
         data = [
             entries,
             "//kythe/go/test/tools/empty_corpus_checker",

diff --git a/kythe/go/test/tools/empty_corpus_checker/empty_corpus_test.sh b/kythe/go/test/tools/empty_corpus_checker/empty_corpus_test.sh
@@ -19,5 +19,6 @@ set -euo pipefail
 # verify that all vnames have a non-empty corpus.
 
 ENTRIES="$1"
+shift
 
-gunzip -c "$ENTRIES" | "$EMPTY_CORPUS_CHECKER"
+gunzip -c "$ENTRIES" | "$EMPTY_CORPUS_CHECKER" "$@"