Dup names

.gitlab-ci.yml

@@ -195,6 +195,9 @@ build-job:
     - docker buildx build --cache-from type=local,src=${HOME}/docker-cache/base --cache-from type=local,src=${HOME}/docker-cache/build --cache-from type=local,src=${HOME}/docker-cache/run --cache-to type=inline --platform="${PLATFORMS}" --build-arg THREADS=${THREADS} --target run -t ${CI_REPO}:${DOCKER_TAG} -f Dockerfile --push .
   variables:
     GIT_SUBMODULE_STRATEGY: recursive
+    # Work around <https://gitlab.com/gitlab-org/gitlab-runner/-/work_items/39384>
+    # See <https://gitlab.com/gitlab-org/gitlab-runner/-/work_items/38908>
+    GIT_STRATEGY: clone
 
 # The arm container build takes like 90 minutes, so we don't want to run it
 # before the main test phase where the other long tests live.
@@ -243,6 +246,9 @@ production-build-job:
     # manage to hit a 6 hour timeout on our extremely slow emulators.
   variables:
     GIT_SUBMODULE_STRATEGY: recursive
+    # Work around <https://gitlab.com/gitlab-org/gitlab-runner/-/work_items/39384>
+    # See <https://gitlab.com/gitlab-org/gitlab-runner/-/work_items/38908>
+    GIT_STRATEGY: clone
 
 # We also run the toil-vg/pytest-based tests
 # Note that WE ONLY RUN TESTS LISTED IN vgci/test-list.txt

.gitmodules

@@ -1,6 +1,6 @@
 [submodule "vcflib"]
 	path = deps/vcflib
-	url = https://github.com/vcflib/vcflib.git
+	url = https://github.com/vgteam/vcflib.git
 [submodule "gssw"]
 	path = deps/gssw
 	url = https://github.com/vgteam/gssw.git

src/constructor.cpp

@@ -2444,6 +2444,11 @@ namespace vg {
                 exit(1);
             }
             for (auto& kv : *(reference->index)) {
+                if (reference_for.count(kv.first)) {
+                    #pragma omp critical (cerr)
+                    cerr << "error:[vg::Constructor] Contig " << kv.first << " appears multiple times" << endl;
+                    exit(1);
+                }
                 // For every sequence name and index entry, point to this reference
                 reference_for[kv.first] = reference;
 #ifdef debug

test/t/02_vg_construct.t

@@ -7,7 +7,7 @@ PATH=../bin:$PATH # for vg
 
 export LC_ALL="C" # force a consistent sort order 
 
-plan tests 30
+plan tests 33
 
 is $(vg construct -m 1000 -r small/x.fa -v small/x.vcf.gz | vg stats -z - | grep nodes | cut -f 2) 210 "construction produces the right number of nodes"
 
@@ -120,3 +120,16 @@ is $? 0 "Reference with ambiguity codes has them coerced to Ns"
 is "$(vg view -j tiny.vg | jq -r '.node[].sequence' | tr -d 'ACGT\n' | wc -c)" "10" "Expected number of Ns are created"
 rm -f tiny.vg
 
+vg construct -r small/x.fa -r small/x.fa > /dev/null
+is $? 1 "Names may not be duplicated across files"
+sed "s/y/x/" small/xy.fa > xx.fa
+samtools faidx xx.fa
+# Samtools drops duplicate entries so we need to manually make an index with duplicates.
+cat xx.fa.fai xx.fa.fai >xx.fa.fai2
+mv xx.fa.fai2 xx.fa.fai
+vg construct -r xx.fa > /dev/null
+is $? 1 "Names may not be duplicated within an index"
+rm -f xx.fa.fai
+vg construct -r xx.fa > /dev/null
+is $? 1 "Names may not be duplicated within an unindexed file"
+rm -f xx.fa xx.fa.fai