IndexBuilderMain dumps core on empty stream input without good error message

[lehmann-4178656ch]

When using IndexBuilderMain inside a pipeline I mistyped the name of a volume and was greeted with the following message (replaced real path with iDoNotExist):

/bin/bash: line 1:     7 Exit 1                  bzcat iDoNotExist
         8 Segmentation fault      (core dumped) | IndexBuilderMain -l -i /index/truthy -F nt -f -

As it was not obvious to me that I mistyped the volume directory I initially was confused on what the error could be.

In this case the problematic line is bzcat: Can't open input file iDoNotExist: No such file or directory. but it's hard to spot inside the normal output of the IndexBuilder

~$ wharfer run --rm --entrypoint /bin/bash -v /tmp:/index/ niklas88/qlever:latest "-c" "bzcat iDoNotExist | IndexBuilderMain -l -i /index/truthy -F nt -f -"

IndexBuilderMain, version Jun  7 2019 07:47:36

Set locale LC_CTYPE to: C.UTF-8
Mon Jul  1 08:19:53.879	- DEBUG: Configuring STXXL...
Mon Jul  1 08:19:53.879	- DEBUG: done.
Mon Jul  1 08:19:53.879	- INFO:  Parsing from stdin.
Mon Jul  1 08:19:53.879	- INFO:  Reading from uncompressed NTriples file/dev/stdin
bzcat: Can't open input file iDoNotExist: No such file or directory.
FOXXLL v1.4.99 (prerelease/Release)
Disk '/index/truthy-stxxl.disk' is allocated, space: 1000000 MiB, I/O implementation: syscall queue=0 devid=0
Mon Jul  1 08:19:53.881	- INFO:  Lines (from KB-file) processed: 0
Mon Jul  1 08:19:53.881	- INFO:  Actual number of Triples in this section (include langfilter triples): 0
Mon Jul  1 08:19:53.881	- INFO:  writing partial vocabulary to /index/truthy.partial-vocabulary0
Mon Jul  1 08:19:53.881	- INFO:  it contains 1 elements
Mon Jul  1 08:19:53.881	- INFO:  Creating partial vocabulary from set ...
Mon Jul  1 08:19:53.881	- INFO:  ... sorting ...
Mon Jul  1 08:19:53.881	- INFO:  Done creating vocabulary.
Mon Jul  1 08:19:53.881	- INFO:  Writing vocabulary to binary file /index/truthy.partial-vocabulary0
Mon Jul  1 08:19:53.881	- INFO:  Done writing vocabulary to file.
Mon Jul  1 08:19:53.881	- INFO:  Pass done.
Mon Jul  1 08:19:53.881	- INFO:  Merging vocabulary
Mon Jul  1 08:19:53.882	- INFO:  Finished writing batch of merged words
Mon Jul  1 08:19:53.882	- INFO:  Finished Merging Vocabulary.
Mon Jul  1 08:19:53.885	- INFO:  total size of vocabulary (internal and external) is 1
Mon Jul  1 08:19:53.885	- INFO:  Initialized external vocabulary. It contains 0 elements.
Mon Jul  1 08:19:53.886	- INFO:  Updating Ids in stxxl vector to global Ids.
Mon Jul  1 08:19:53.886	- INFO:  Lines processed: 0
Mon Jul  1 08:19:53.886	- INFO:  Corresponding number of statements in original knowledgeBase:0
Mon Jul  1 08:19:53.886	- INFO:  Reading IdMap from /index/truthy.partial-ids-mmap0 ...
Mon Jul  1 08:19:53.886	- INFO:  Done reading idMap
Mon Jul  1 08:19:53.889	- INFO:  Lines processed: 0
Mon Jul  1 08:19:53.889	- INFO:  Pass done.
Mon Jul  1 08:19:53.889	- INFO:  Sorting for PSO permutation...
stl_in_memory_sort, range: 0
Mon Jul  1 08:19:53.889	- INFO:  Sort done.
Mon Jul  1 08:19:53.889	- INFO:  Performing unique to ensure RDF semantics...
Mon Jul  1 08:19:53.889	- INFO:  Size before: 0
Mon Jul  1 08:19:53.889	- INFO:  Done: unique.
Mon Jul  1 08:19:53.889	- INFO:  Size after: 0
Mon Jul  1 08:19:53.889	- WARN:  Attempt to write an empty index!
Mon Jul  1 08:19:53.889	- INFO:  Sorting for SPO permutation...
stl_in_memory_sort, range: 0
Mon Jul  1 08:19:53.890	- INFO:  Sort done.
Mon Jul  1 08:19:53.890	- WARN:  Attempt to write an empty index!
Mon Jul  1 08:19:53.890	- INFO:  Vector already sorted for pattern creation.
Mon Jul  1 08:19:53.890	- INFO:  Creating patterns file...
/bin/bash: line 1:     7 Exit 1                  bzcat iDoNotExist
         8 Segmentation fault      (core dumped) | IndexBuilderMain -l -i /index/truthy -F nt -f -
2019/07/01 10:19:54 exit status 139

It looks like IndexBuilderMain handles empty strings:

$ wharfer run --rm --entrypoint /bin/bash -v /tmp:/index/ niklas88/qlever:latest "-c" "echo \"\" | IndexBuilderMain -l -i /index/truthy -F nt -f -"

IndexBuilderMain, version Jun  7 2019 07:47:36

Set locale LC_CTYPE to: C.UTF-8
Mon Jul  1 08:25:20.539	- DEBUG: Configuring STXXL...
Mon Jul  1 08:25:20.539	- DEBUG: done.
Mon Jul  1 08:25:20.539	- INFO:  Parsing from stdin.
Mon Jul  1 08:25:20.539	- INFO:  Reading from uncompressed NTriples file/dev/stdin
Mon Jul  1 08:25:20.540	- ERROR: BAD INPUT STRING (Illegal whitespace only line; in /app/src/parser/NTriplesParser.cpp, line 29, function bool NTriplesParser::getLine(std::array<std::__cxx11::basic_string<char>, 3>&))

This can be observed using the latest and previous images:
niklas88/qlever latest 5e18076b6ad5 3 days ago 198MB
niklas88/qlever latest ce623d98a552 3 weeks ago 199MB