commit 05f1afb7ad40eccea874e9cd37b10f74bbb72969
author Markus Kusano <kusano@google.com> Thu Jun 27 16:00:16 2024 -0700
committer Copybara-Service <copybara-worker@google.com> Tue Jul 02 06:38:38 2024 -0700
tree ce566c80a1d385edc0d5091e96b9b0720f8f9c32
parent aadda567e858f7d424a5d647b5190df292720c40

#Centipede Add support for directly passing a corpus shard in a SeedCorpusSource

PiperOrigin-RevId: 647474909

centipede/BUILD

diff --git a/centipede/BUILD b/centipede/BUILD
index e851c68..8092c65 100644
--- a/centipede/BUILD
+++ b/centipede/BUILD
@@ -101,6 +101,7 @@
 
 cc_proto_library(
     name = "seed_corpus_config_cc_proto",
+    visibility = EXTENDED_API_VISIBILITY,
     deps = [":seed_corpus_config_proto"],
 )
 
@@ -1057,6 +1058,7 @@
     name = "seed_corpus_maker_lib",
     srcs = ["seed_corpus_maker_lib.cc"],
     hdrs = ["seed_corpus_maker_lib.h"],
+    visibility = EXTENDED_API_VISIBILITY,
     deps = [
         ":corpus_io",
         ":feature",

centipede/seed_corpus_config.proto

diff --git a/centipede/seed_corpus_config.proto b/centipede/seed_corpus_config.proto
index 8abc201..f71799d 100644
--- a/centipede/seed_corpus_config.proto
+++ b/centipede/seed_corpus_config.proto
@@ -34,6 +34,9 @@
   // A relative glob used to find corpus shards in each `dir_glob`-matching dir.
   // Required.
   string shard_rel_glob = 3;
+  // Path directly to a corpus shard. Will be added to the shards specified by
+  // `dir_glob`.
+  string corpus_shard_path = 6;
   // An absolute or relative number of corpus elements to sample from all the
   // corpus shards in all `dir_glob` matching dirs of this source. Optional: the
   // default is to take all the elements.

centipede/seed_corpus_maker_lib.cc

diff --git a/centipede/seed_corpus_maker_lib.cc b/centipede/seed_corpus_maker_lib.cc
index b43d7e9..2239a2b 100644
--- a/centipede/seed_corpus_maker_lib.cc
+++ b/centipede/seed_corpus_maker_lib.cc
@@ -182,6 +182,12 @@
   }
   LOG(INFO) << "Found " << corpus_shard_fnames.size()
             << " shard(s) total in source " << source.dir_glob();
+  if (!source.corpus_shard_path().empty()) {
+    LOG(INFO) << "Adding corpus shard path: " << source.corpus_shard_path();
+    corpus_shard_fnames.push_back(source.corpus_shard_path());
+  }
+  LOG(INFO) << "Total number of corpus shards found: "
+            << corpus_shard_fnames.size();
 
   if (corpus_shard_fnames.empty()) {
     LOG(WARNING) << "Skipping empty source " << source.dir_glob();