ii14 / autocmd-pattern-parser

autocmd-pattern-parser

A work-in-progress autocmd pattern parser, for converting legacy filetype detection autocommands to lua.

Build and run

make
./auparser /usr/share/nvim/runtime/filetype.vim

Options

• -u to unroll branches
• -t to exclude tree from output
• -p to parse raw patterns (one pattern per line)
• - for stdin

Output

[
  {
    "pattern": "...",       // original pattern
    "lnum": 1,              // source line number
    "cmd": "...",           // autocmd command, eg. "setf json"
    "tree": [               // tree
      [                     // branch, as in "abc" in "abc,def"
        {                   // token
          "type": "...",    // token type
          "value": "..."    // raw string
        },
        {
          "type": "Branch", // nested branch
          "value": [        // alternatives, same format as in branch ie. arrays of tokens
            [               // alternative
              {"type": "...", "value": "..."},
              {"type": "...", "value": "..."}
            ],
            []              // empty alternative, as in "a{b,}c"
          ]
        }
      ]
    ],
    "result": [             // results from branch unrolling
      {
        "pattern": "...",   // raw pattern for this branch
        "tokens": [         // array of tokens
          {"type": "...", "value": "..."},
          {"type": "...", "value": "..."}
        ]
      }
    ]
  },
  {
    "pattern": "...",
    "lnum": 1,
    "cmd": "...",
    "error": "..." // optional error, although it parses nvim runtime and polyglot just fine
  }
]

Token types

Defined as type_t in auparser.h.

• Literal: matches string literals
• AnyChar: matches ? (translates to . in vim/lua regex)
• AnyChars: matches * (translates to .* in vim/lua regex)
• Set: matches character sets, eg. [^2-3abc]
• Cls: matches character class, eg. \d \s \X
• Opts: matches vim regex settings, eg. \c for turning on ignorecase
• ZeroOrMore: matches \*
• ZeroOrOne: matches \= (basically \?)
• OneOrMore: matches \+
• Count: previous atom repeated N times, eg. \\\{6\}