comunica / json-event-parser.js / 32377107465

100%

master: 100%

Pull #22

github

Speeds up string parsing by copying plain runs in bulk

Strings are the bulk of a typical JSON-LD document, and each character
was appended individually: a call and a bounds check per byte. Once
inside a string, scan ahead for the whole run of plain characters and
copy it in one go.

Also drops the Buffer.byteLength call on every string token. It only fed
this.offset, and re-measured a string that had just been built, so the
byte count is now accumulated as the string is appended instead. The
semantics are unchanged, including the existing inaccuracy for escaped
strings that #12 tracks.

The tokenizer is now faster than @bergos/jsonparse on every workload
except the 0.4KiB one, where per-parse setup dominates:

                       before    after   vs jsonparse
  person                0.025    0.028          1.27x
  sparql-init           0.049    0.045          0.88x
  toRdf-manifest        1.660    1.430          0.62x
  dbpedia-expanded    110.808   85.349          0.53x

Plain characters no longer reach appendStringChar, which left its buffer
flush reachable only through escape sequences, so a large token built
entirely from escapes was added to cover it.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EDHYSxPkD1cp78Qzea6DmQ

323 of 323 branches covered (100.0%)

Branch coverage included in aggregate %.

17 of 17 new or added lines in 1 file covered. (100.0%)

500 of 500 relevant lines covered (100.0%)

5176817.85 hits per line