Merge branch 'oopsie' into 'dev'

Fix nonstandard RG tag leading to segfault [CW-5364]

See merge request epi2melabs/fastcat!79

CHANGELOG.md

@@ -8,6 +8,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ## [v0.19.1]
 ### Fixed
 - Compilation on macOS with clang.
+- Segmentation fault with bad read group information.
 
 ## [v0.19.0]
 ### Added

Makefile

@@ -113,20 +113,22 @@ mem_check_fastcat_demultiplex: fastcat
 	$(VALGRIND) --error-exitcode=1 --tool=memcheck --leak-check=full --show-leak-kinds=all -s \
 		./fastcat test/data/*.fastq.gz --demultiplex demultiplex > /dev/null
 
-.PHONY: mem_check_fastcat_demultiplex
-mem_check_fastcat_demultiplex: fastcat
-	rm -rf demultiplex
-	$(VALGRIND) --error-exitcode=1 --tool=memcheck --leak-check=full --show-leak-kinds=all -s \
-		./fastcat test/data/*.fastq.gz --demultiplex demultiplex > /dev/null
-
 .PHONY: mem_check_bamstats
 mem_check_bamstats: bamstats
+	@echo "Memcheck bamstats with good data"
+	rm -rf bamstats-histograms
+	$(VALGRIND) --error-exitcode=1 --tool=memcheck --leak-check=full --show-leak-kinds=all -s \
+		./bamstats test/parse_rg/[email protected] > /dev/null
+	@echo "Memcheck bamstats with bad data"
+	@echo ""
 	rm -rf bamstats-histograms
 	$(VALGRIND) --error-exitcode=1 --tool=memcheck --leak-check=full --show-leak-kinds=all -s \
-		./bamstats test/bamstats/400ecoli-with-qcfail.bam
+		./bamstats test/parse_rg/bad-ones.bam > /dev/null
+	@echo ""
+	@echo "Memcheck bamstats with qcfails"
 	rm -rf bamstats-histograms
 	$(VALGRIND) --error-exitcode=1 --tool=memcheck --leak-check=full --show-leak-kinds=all -s \
-		./bamstats test/parse_rg/[email protected].bam
+		./bamstats test/bamstats/400ecoli-with-qcfail.bam > /dev/null
 
 .PHONY: mem_check_bamstats_duplex
 mem_check_bamstats_duplex: bamstats

src/bamstats/readstats.c

@@ -279,8 +279,12 @@ void process_bams(
         start_time = "";
         if (tags.RG != NULL) {
             rg_info = create_rg_info(tags.RG);
-            runid = rg_info->runid;
-            basecaller = rg_info->basecaller;
+            if (rg_info->runid != NULL) {
+                runid = rg_info->runid;
+            }
+            if (rg_info->basecaller != NULL) {
+                basecaller = rg_info->basecaller;
+            }
         } else if (tags.RD != NULL) {
             runid = tags.RD;
         }

src/kh_counter.c

@@ -30,6 +30,7 @@ int kh_counter_val(kh_counter_t *hash, char *key) {
 }
 
 size_t kh_counter_add(kh_counter_t *hash, char *key, int val) {
+    if (key == NULL) {return -1;}
     // note: key is copied so no need for caller to hold on to it
     int ret;
     khiter_t k = kh_put(KH_COUNTER, hash, key, &ret);
@@ -47,6 +48,7 @@ size_t kh_counter_add(kh_counter_t *hash, char *key, int val) {
 }
 
 size_t kh_counter_sub(kh_counter_t *hash, char *key, int val) {
+    if (key == NULL) {return -1;}
     // note: key is copied so no need for caller to hold on to it
     int ret;
     khiter_t k = kh_put(KH_COUNTER, hash, key, &ret);