v4.1.0

Compare Source

   🚀 Features

• benchmarks:
  • Add keyed and nested-group datasets covering v4 tabular forms  -  by @​johannschopplich (ebe80)
  • Corrupt encoded text for structural-validation datasets  -  by @​johannschopplich (1de7b)
  • Add wilson 95% intervals to accuracy tables  -  by @​johannschopplich (ac638)
• encode:
  • rawString marker and escapeString export for verbatim replacer output  -  by @​yilmazhasan in #​321 (b36f8)

   🐞 Bug Fixes

• Accept the indentSize option named by the spec  -  by @​johannschopplich (b41d5)
• benchmarks:
  • Normalize non-ISO dates in local time to avoid timezone-dependent day shifts  -  by @​johannschopplich (c7b6f)
  • Pin faker reference date so regenerated datasets stay reproducible  -  by @​johannschopplich (d7d78)
  • Split accuracy report into csv-excluded and flat-only tracks  -  by @​johannschopplich (911ba)
  • Floor grok-4.5 reasoning at low  -  by @​johannschopplich (7c792)
• cli:
  • Error on ill-formed UTF-8 input in strict mode  -  by @​johannschopplich (5f04e)
• decode:
  • Strip a byte-order mark and trailing spaces, and stop truncating at a declared [N]  -  by @​johannschopplich (d8ca3)
  • Keep a quoted colon inside an unquoted key  -  by @​johannschopplich (41c92)
  • Omit leaf fields that have no cell instead of yielding undefined  -  by @​johannschopplich (c2c25)
  • Reject a bare token line outside root primitive position  -  by @​johannschopplich (ea050)
• encode:
  • Reject unpaired surrogates and invalid delimiters  -  by @​johannschopplich (db67c)

    View changes on GitHub

v4.0.0

Compare Source

   🚨 Breaking Changes

• Remove key folding and path expansion  -  by @​johannschopplich (46956)

   🚀 Features

• Decode-side comment lines with encoder quoting of #-leading strings  -  by @​johannschopplich (13561)
• Nested field groups in tabular headers  -  by @​johannschopplich (128fc)
• Keyed tabular form for objects of uniform objects  -  by @​johannschopplich (68464)
• decode: Reject indentation depth jumps in strict mode  -  by @​johannschopplich (6db88)

   🐞 Bug Fixes

• decode:
  • Prevent false array header detection for unquoted keys c…  -  by @​rayriffy in #​314 (13360)
  • Trim value tokens by U+0020 only  -  by @​johannschopplich (fea26)
  • Reject over-indented orphan lines and trailing root content in strict mode  -  by @​johannschopplich (4489d)
  • Treat a trailing carriage return as part of the line terminator  -  by @​johannschopplich (9562c)
  • Reject keyless headers outside root and list-item positions  -  by @​johannschopplich (1560d)
  • Reject inline content after fields-bearing headers  -  by @​johannschopplich (06bd0)
• encode:
  • Quote leading-plus numeric-like strings  -  by @​johannschopplich (b769d)

    View changes on GitHub

v2.3.1

Compare Source

   🐞 Bug Fixes

• Harden encode and decode against prototype pollution  -  by @​ckorhonen in #​316 (94a2b)
• decode:
  • Slice key-value rest at key token end offset  -  by @​johannschopplich (70230)
  • Treat quoted spans as opaque in line classification  -  by @​johannschopplich (9024a)

    View changes on GitHub