biojppm/rapidyaml

Parser regression in 0.7.0

Closed this issue · 2 comments

marcalff commented

The following yaml parses fine with 0.6.0:

file_format: 0.0
tracer_provider:
  processors:
    - simple:
        exporter:
          console:
  sampler:
    trace_id_ratio_based:
      ratio:

With 0.7.0, I get a parse error on sampler::

8:3: (102B): ERROR: parse error
8:3:   sampler:  (size=10)
       ^~~~~~~~  (cols 3-11)
marcalff commented

Investigated a bit, turning ryml debug and looking at the parser debug trace.

Some observations.

  1. stdout and stderr can be mixed

When capturing stdout and stderr to the same file, messages can mixed, as seen here:

/data/malff/CODE/MY_GITHUB/rapidyaml/src/c4/yml/parse_engine.def.hpp:1757: searching for state with indentation 2. current: ind=8,level=5,node=6,flags=RMAP|BLCK|RNXT
/data/malff/CODE/MY_GITHUB/rapidyaml/src/c4/yml/parse_engine.def.hpp:1757: searching for state with indentation 2. current: ind=6,level=4,node=5,flags=RMAP|BLCK|RNXT
/data/malff/CODE/MY_GITHUB/rapidyaml/src/c4/yml/parse_engine.def.hpp:1757: searching for state with indentation 2. current: ind=4,level=3,node=4,flags=RSEQ|BLCK|RNXT
/data/malff/CODE/MY_GITHUB/rapidyaml/src/c4/yml/parse_engine.def.hpp:1757: searching for 8:3: (102B): ERROR:
/data/malff/CODE/MY_GITHUB/rapidyaml/src/c4/yml/parse_engine.def.hpp:6752: parse error
8:3:   sampler:  (size=10)
       ^~~~~~~~  (cols 3-11)
top state: RMAP|BLCK|RNXT

state with indentation 2. current: ind=2,level=2,node=3,flags=RMAP|BLCK|RNXT
/data/malff/CODE/MY_GITHUB/rapidyaml/src/c4/yml/parse_engine.def.hpp:1764: same indentation!!! level=2 node=3

Please consider to flush stdout before writing to stderr, when raising a parse error.

  1. parse error location

The error is raised at src/c4/yml/parse_engine.def.hpp:6752 (all line numbers using label v0.7.0).

From what I can tell, when seeing sampler::

  • the previous nodes are poped, closing maps and sequences
/data/malff/CODE/MY_GITHUB/rapidyaml/src/c4/yml/event_handler_tree.hpp:298: node[3]: end_seq_val
/data/malff/CODE/MY_GITHUB/rapidyaml/src/c4/yml/parse_engine.def.hpp:1699: current level is 2 (indentation=2)
/data/malff/CODE/MY_GITHUB/rapidyaml/src/c4/yml/parse_engine.def.hpp:6323: mapblck[RVAL]: still mapblck!
  • the code loops to find a mapblock
/data/malff/CODE/MY_GITHUB/rapidyaml/src/c4/yml/parse_engine.def.hpp:1420: line[8] (11 cols) progressed by 2:  col 1-->3   offset 100-->102
/data/malff/CODE/MY_GITHUB/rapidyaml/src/c4/yml/parse_engine.def.hpp:7044: mapblck: again
  • and then it chokes on finding sampler, not knowing what to do with a token which is neither a : nor a space.
/data/malff/CODE/MY_GITHUB/rapidyaml/src/c4/yml/parse_engine.def.hpp:5943: handle2_map_block: map_id=2 node_id=3 level=2 indref=2
/data/malff/CODE/MY_GITHUB/rapidyaml/src/c4/yml/parse_engine.def.hpp:6727: mapblck[RNXT]: 's'
marcalff commented

@biojppm

I hope this debugging helps.
Not sure how to get further into the code, as I don't know the parser states well enough.

  • 🚀1