From e2c77bab89f9ac7d55b52697589feaf9afcf6c07 Mon Sep 17 00:00:00 2001
From: Jay Berkenbilt <ejb@ql.org>
Date: Sat, 15 Jun 2019 11:20:55 -0400
Subject: [PATCH] Populate seed corpus

---
 .gitignore    |  1 +
 fuzz/build.mk | 52 ++++++++++++++++++++++++++++++++++++++++++++++++++-
 fuzz/get_sha1 |  9 +++++++++
 3 files changed, 61 insertions(+), 1 deletion(-)
 create mode 100644 fuzz/get_sha1

diff --git a/.gitignore b/.gitignore
index 720eadcf..5c7f7fed 100644
--- a/.gitignore
+++ b/.gitignore
@@ -25,3 +25,4 @@ manual/html.xsl
 manual/print.xsl
 qpdf/build/
 zlib-flate/build/
+fuzz/qpdf_fuzzer_seed_corpus/
diff --git a/fuzz/build.mk b/fuzz/build.mk
index 1ecdac9b..5e95ada6 100644
--- a/fuzz/build.mk
+++ b/fuzz/build.mk
@@ -9,7 +9,7 @@ DEFAULT_FUZZ_RUNNER := standalone_fuzz_target_runner
 OBJ_DEFAULT_FUZZ := fuzz/$(OUTPUT_DIR)/$(DEFAULT_FUZZ_RUNNER).$(OBJ)
 
 BINS_fuzz = $(foreach B,$(FUZZERS),fuzz/$(OUTPUT_DIR)/$(call binname,$(B)))
-TARGETS_fuzz = $(OBJ_DEFAULT_FUZZ) $(BINS_fuzz)
+TARGETS_fuzz = $(OBJ_DEFAULT_FUZZ) $(BINS_fuzz) fuzz_corpus
 
 INCLUDES_fuzz = include
 
@@ -21,8 +21,58 @@ LIB_FUZZING_ENGINE ?= $(OBJ_DEFAULT_FUZZ)
 # by oss-fuzz, it will be there.
 $(BINS_fuzz): $(TARGETS_libqpdf) $(OBJ_DEFAULT_FUZZ)
 
+# Files from the test suite that are good for seeding the fuzzer.
+# Update $n_test_files in qtest/fuzz.test if you change this list.
+SEED_CORPUS_FILES = \
+	field-types.pdf \
+	image-streams.pdf \
+	need-appearances.pdf \
+	outlines-with-actions.pdf \
+	outlines-with-old-root-dests.pdf \
+	page-labels-and-outlines.pdf \
+	page-labels-num-tree.pdf \
+	issue-99b.pdf \
+	issue-99.pdf \
+	issue-100.pdf \
+	issue-101.pdf \
+	issue-106.pdf \
+	issue-117.pdf \
+	issue-119.pdf \
+	issue-120.pdf \
+	issue-141a.pdf \
+	issue-141b.pdf \
+	issue-143.pdf \
+	issue-146.pdf \
+	issue-147.pdf \
+	issue-148.pdf \
+	issue-149.pdf \
+	issue-150.pdf \
+	issue-202.pdf \
+	issue-263.pdf \
+	issue-335a.pdf \
+	issue-335b.pdf
+
 # -----
 
+CORPUS_FROM_TEST := $(foreach F,$(SEED_CORPUS_FILES),qpdf/qtest/qpdf/$F)
+CORPUS_DIR := fuzz/qpdf_fuzzer_seed_corpus
+
+.PHONY: fuzz_corpus
+fuzz_corpus:: fuzz/$(OUTPUT_DIR)/fuzz_corpus.stamp
+$(foreach F,$(CORPUS_FROM_TEST),$(eval \
+  SHA1_$(notdir $(F)) := $(shell perl fuzz/get_sha1 < $F)))
+$(foreach F,$(CORPUS_FROM_TEST),$(eval \
+  fuzz_corpus:: $(CORPUS_DIR)/$(SHA1_$(notdir $(F)))))
+$(foreach F,$(CORPUS_FROM_TEST),$(eval \
+  $(CORPUS_DIR)/$(SHA1_$(notdir $(F))): $(F) ; \
+	mkdir -p $(CORPUS_DIR); \
+	cp $(F) $(CORPUS_DIR)/$(SHA1_$(notdir $(F)))))
+
+fuzz/$(OUTPUT_DIR)/fuzz_corpus.stamp: fuzz/original-corpus.tar.gz $(CORPUS_FROM_TEST)
+	mkdir -p $(CORPUS_DIR)
+	(cd $(CORPUS_DIR); tar xzf ../original-corpus.tar.gz)
+	touch $@
+
 $(foreach B,$(FUZZERS),$(eval \
   OBJS_$(B) = $(call src_to_obj,fuzz/$(B).cc)))
 
diff --git a/fuzz/get_sha1 b/fuzz/get_sha1
new file mode 100644
index 00000000..4bc34fb6
--- /dev/null
+++ b/fuzz/get_sha1
@@ -0,0 +1,9 @@
+#!/usr/bin/env perl
+require 5.008;
+use warnings;
+use strict;
+use Digest::SHA;
+
+binmode STDIN;
+my $digest = Digest::SHA->new('sha1')->addfile(*STDIN)->hexdigest;
+print "$digest\n";