path: root/runtime/lua/vim/treesitter/query.lua

local a = vim.api
local language = require('vim.treesitter.language')

-- query: pattern matching on trees
-- predicate matching is implemented in lua
local Query = {}
Query.__index = Query

local M = {}

---@private
local function dedupe_files(files)
  local result = {}
  local seen = {}

  for _, path in ipairs(files) do
    if not seen[path] then
      table.insert(result, path)
      seen[path] = true
    end
  end

  return result
end

---@private
local function safe_read(filename, read_quantifier)
  local file, err = io.open(filename, 'r')
  if not file then
    error(err)
  end
  local content = file:read(read_quantifier)
  io.close(file)
  return content
end

--- Gets the list of files used to make up a query
---
---@param lang The language
---@param query_name The name of the query to load
---@param is_included Internal parameter, most of the time left as `nil`
function M.get_query_files(lang, query_name, is_included)
  local query_path = string.format('queries/%s/%s.scm', lang, query_name)
  local lang_files = dedupe_files(a.nvim_get_runtime_file(query_path, true))

  if #lang_files == 0 then
    return {}
  end

  local base_query = nil
  local extensions = {}

  local base_langs = {}

  -- Now get the base languages by looking at the first line of every file
  -- The syntax is the following :
  -- ;+ inherits: ({language},)*{language}
  --
  -- {language} ::= {lang} | ({lang})
  local MODELINE_FORMAT = '^;+%s*inherits%s*:?%s*([a-z_,()]+)%s*$'
  local EXTENDS_FORMAT = '^;+%s*extends%s*$'

  for _, filename in ipairs(lang_files) do
    local file, err = io.open(filename, 'r')
    if not file then
      error(err)
    end

    local extension = false

    for modeline in
      function()
        return file:read('*l')
      end
    do
      if not vim.startswith(modeline, ';') then
        break
      end

      local langlist = modeline:match(MODELINE_FORMAT)
      if langlist then
        for _, incllang in ipairs(vim.split(langlist, ',', true)) do
          local is_optional = incllang:match('%(.*%)')

          if is_optional then
            if not is_included then
              table.insert(base_langs, incllang:sub(2, #incllang - 1))
            end
          else
            table.insert(base_langs, incllang)
          end
        end
      elseif modeline:match(EXTENDS_FORMAT) then
        extension = true
      end
    end

    if extension then
      table.insert(extensions, filename)
    elseif base_query == nil then
      base_query = filename
    end
    io.close(file)
  end

  local query_files = { base_query }
  for _, base_lang in ipairs(base_langs) do
    local base_files = M.get_query_files(base_lang, query_name, true)
    vim.list_extend(query_files, base_files)
  end
  vim.list_extend(query_files, extensions)

  return query_files
end

---@private
local function read_query_files(filenames)
  local contents = {}

  for _, filename in ipairs(filenames) do
    table.insert(contents, safe_read(filename, '*a'))
  end

  return table.concat(contents, '')
end

--- The explicitly set queries from |vim.treesitter.query.set_query()|
local explicit_queries = setmetatable({}, {
  __index = function(t, k)
    local lang_queries = {}
    rawset(t, k, lang_queries)

    return lang_queries
  end,
})

--- Sets the runtime query {query_name} for {lang}
---
--- This allows users to override any runtime files and/or configuration
--- set by plugins.
---
---@param lang string: The language to use for the query
---@param query_name string: The name of the query (i.e. "highlights")
---@param text string: The query text (unparsed).
function M.set_query(lang, query_name, text)
  explicit_queries[lang][query_name] = M.parse_query(lang, text)
end

--- Returns the runtime query {query_name} for {lang}.
---
---@param lang The language to use for the query
---@param query_name The name of the query (i.e. "highlights")
---
---@return The corresponding query, parsed.
function M.get_query(lang, query_name)
  if explicit_queries[lang][query_name] then
    return explicit_queries[lang][query_name]
  end

  local query_files = M.get_query_files(lang, query_name)
  local query_string = read_query_files(query_files)

  if #query_string > 0 then
    return M.parse_query(lang, query_string)
  end
end

local query_cache = vim.defaulttable(function()
  return setmetatable({}, { __mode = 'v' })
end)

--- Parse {query} as a string. (If the query is in a file, the caller
--- should read the contents into a string before calling).
---
--- Returns a `Query` (see |lua-treesitter-query|) object which can be used to
--- search nodes in the syntax tree for the patterns defined in {query}
--- using `iter_*` methods below.
---
--- Exposes `info` and `captures` with additional context about {query}.
---   - `captures` contains the list of unique capture names defined in
---     {query}.
---   -` info.captures` also points to `captures`.
---   - `info.patterns` contains information about predicates.
---
---@param lang string The language
---@param query string A string containing the query (s-expr syntax)
---
---@returns The query
function M.parse_query(lang, query)
  language.require_language(lang)
  local cached = query_cache[lang][query]
  if cached then
    return cached
  else
    local self = setmetatable({}, Query)
    self.query = vim._ts_parse_query(lang, query)
    self.info = self.query:inspect()
    self.captures = self.info.captures
    query_cache[lang][query] = self
    return self
  end
end

--- Gets the text corresponding to a given node
---
---@param node table The node
---@param source table The buffer or string from which the node is extracted
---@param opts table Optional parameters.
---          - concat: (boolean default true) Concatenate result in a string
function M.get_node_text(node, source, opts)
  opts = opts or {}
  local concat = vim.F.if_nil(opts.concat, true)

  local start_row, start_col, start_byte = node:start()
  local end_row, end_col, end_byte = node:end_()

  if type(source) == 'number' then
    local lines
    local eof_row = a.nvim_buf_line_count(source)
    if start_row >= eof_row then
      return nil
    end

    if end_col == 0 then
      lines = a.nvim_buf_get_lines(source, start_row, end_row, true)
      end_col = -1
    else
      lines = a.nvim_buf_get_lines(source, start_row, end_row + 1, true)
    end

    if #lines > 0 then
      if #lines == 1 then
        lines[1] = string.sub(lines[1], start_col + 1, end_col)
      else
        lines[1] = string.sub(lines[1], start_col + 1)
        lines[#lines] = string.sub(lines[#lines], 1, end_col)
      end
    end

    return concat and table.concat(lines, '\n') or lines
  elseif type(source) == 'string' then
    return source:sub(start_byte + 1, end_byte)
  end
end

-- Predicate handler receive the following arguments
-- (match, pattern, bufnr, predicate)
local predicate_handlers = {
  ['eq?'] = function(match, _, source, predicate)
    local node = match[predicate[2]]
    if not node then
      return true
    end
    local node_text = M.get_node_text(node, source)

    local str
    if type(predicate[3]) == 'string' then
      -- (#eq? @aa "foo")
      str = predicate[3]
    else
      -- (#eq? @aa @bb)
      str = M.get_node_text(match[predicate[3]], source)
    end

    if node_text ~= str or str == nil then
      return false
    end

    return true
  end,

  ['lua-match?'] = function(match, _, source, predicate)
    local node = match[predicate[2]]
    if not node then
      return true
    end
    local regex = predicate[3]
    return string.find(M.get_node_text(node, source), regex)
  end,

  ['match?'] = (function()
    local magic_prefixes = { ['\\v'] = true, ['\\m'] = true, ['\\M'] = true, ['\\V'] = true }
    ---@private
    local function check_magic(str)
      if string.len(str) < 2 or magic_prefixes[string.sub(str, 1, 2)] then
        return str
      end
      return '\\v' .. str
    end

    local compiled_vim_regexes = setmetatable({}, {
      __index = function(t, pattern)
        local res = vim.regex(check_magic(pattern))
        rawset(t, pattern, res)
        return res
      end,
    })

    return function(match, _, source, pred)
      local node = match[pred[2]]
      if not node then
        return true
      end
      local regex = compiled_vim_regexes[pred[3]]
      return regex:match_str(M.get_node_text(node, source))
    end
  end)(),

  ['contains?'] = function(match, _, source, predicate)
    local node = match[predicate[2]]
    if not node then
      return true
    end
    local node_text = M.get_node_text(node, source)

    for i = 3, #predicate do
      if string.find(node_text, predicate[i], 1, true) then
        return true
      end
    end

    return false
  end,

  ['any-of?'] = function(match, _, source, predicate)
    local node = match[predicate[2]]
    if not node then
      return true
    end
    local node_text = M.get_node_text(node, source)

    -- Since 'predicate' will not be used by callers of this function, use it
    -- to store a string set built from the list of words to check against.
    local string_set = predicate['string_set']
    if not string_set then
      string_set = {}
      for i = 3, #predicate do
        string_set[predicate[i]] = true
      end
      predicate['string_set'] = string_set
    end

    return string_set[node_text]
  end,
}

-- As we provide lua-match? also expose vim-match?
predicate_handlers['vim-match?'] = predicate_handlers['match?']

-- Directives store metadata or perform side effects against a match.
-- Directives should always end with a `!`.
-- Directive handler receive the following arguments
-- (match, pattern, bufnr, predicate, metadata)
local directive_handlers = {
  ['set!'] = function(_, _, _, pred, metadata)
    if #pred == 4 then
      -- (#set! @capture "key" "value")
      local _, capture_id, key, value = unpack(pred)
      if not metadata[capture_id] then
        metadata[capture_id] = {}
      end
      metadata[capture_id][key] = value