Reformatted Lua LPeg lexers and added new convenience functions and pattern.

`lexer.range()` replaces `lexer.delimited_range()` and `lexer.nested_pair()`.
`lexer.to_eol()` replaces `patt * lexer.nonnewline^0` constructs.
`lexer.number` replaces `lexer.float + lexer.integer`.
Also added unit tests for lexer functions.

1 files changed, 37 insertions, 37 deletions

diff --git a/lexlua/yaml.lua b/lexlua/yaml.lua
index 2cd54d210..fd70182fc 100644
--- a/lexlua/yaml.lua
+++ b/lexlua/yaml.lua
@@ -10,15 +10,16 @@ local M = {_NAME = 'yaml'}
 
 -- Whitespace.
 local indent = #lexer.starts_line(S(' \t')) *
-               (token(lexer.WHITESPACE, ' ') + token('indent_error', '\t'))^1
+  (token(lexer.WHITESPACE, ' ') + token('indent_error', '\t'))^1
 local ws = token(lexer.WHITESPACE, S(' \t')^1 + lexer.newline^1)
 
 -- Comments.
-local comment = token(lexer.COMMENT, '#' * lexer.nonnewline^0)
+local comment = token(lexer.COMMENT, lexer.to_eol('#'))
 
 -- Strings.
-local string = token(lexer.STRING, lexer.delimited_range("'") +
-                                   lexer.delimited_range('"'))
+local sq_str = lexer.range("'")
+local dq_str = lexer.range('"')
+local string = token(lexer.STRING, sq_str + dq_str)
 
 -- Numbers.
 local integer = lexer.dec_num + lexer.hex_num + '0' * S('oO') * R('07')^1
@@ -26,22 +27,23 @@ local special_num = '.' * word_match({'inf', 'nan'}, nil, true)
 local number = token(lexer.NUMBER, special_num + lexer.float + integer)
 
 -- Timestamps.
-local ts = token('timestamp',
-                 lexer.digit * lexer.digit * lexer.digit * lexer.digit * -- year
-                 '-' * lexer.digit * lexer.digit^-1 * -- month
-                 '-' * lexer.digit * lexer.digit^-1 * -- day
-                 ((S(' \t')^1 + S('tT'))^-1 * -- separator
-                  lexer.digit * lexer.digit^-1 * -- hour
-                  ':' * lexer.digit * lexer.digit * -- minute
-                  ':' * lexer.digit * lexer.digit * -- second
-                  ('.' * lexer.digit^0)^-1 * -- fraction
-                  ('Z' + -- timezone
-                   S(' \t')^0 * S('-+') * lexer.digit * lexer.digit^-1 *
-                   (':' * lexer.digit * lexer.digit)^-1)^-1)^-1)
+local year = lexer.digit * lexer.digit * lexer.digit * lexer.digit
+local month = lexer.digit * lexer.digit^-1
+local day = lexer.digit * lexer.digit^-1
+local date = year * '-' * month * '-' * day
+local hours = lexer.digit * lexer.digit^-1
+local minutes = lexer.digit * lexer.digit
+local seconds = lexer.digit * lexer.digit
+local fraction = '.' * lexer.digit^0
+local time = hours * ':' * minutes * ':' * seconds * fraction^-1
+local T = S(' \t')^1 + S('tT')
+local zone = 'Z' + S(' \t')^0 * S('-+') * hours * (':' * minutes)^-1
+local ts = token('timestamp', date * (T * time * zone^-1))
 
 -- Constants.
-local constant = token(lexer.CONSTANT,
-                       word_match({'null', 'true', 'false'}, nil, true))
+local constant = token(lexer.CONSTANT, word_match({
+  'null', 'true', 'false'
+}, nil, true))
 
 -- Types.
 local type = token(lexer.TYPE, '!!' * word_match({
@@ -50,38 +52,36 @@ local type = token(lexer.TYPE, '!!' * word_match({
   -- Scalar types.
   'binary', 'bool', 'float', 'int', 'merge', 'null', 'str', 'timestamp',
   'value', 'yaml'
-}, nil, true) + '!' * lexer.delimited_range('<>'))
+}, nil, true) + '!' * lexer.range('<', '>', true))
 
 -- Document boundaries.
 local doc_bounds = token('document', lexer.starts_line(P('---') + '...'))
 
 -- Directives
 local directive = token('directive', lexer.starts_line('%') *
-                                     lexer.nonnewline^1)
+  lexer.nonnewline^1)
 
 local word = (lexer.alpha + '-' * -lexer.space) * (lexer.alnum + '-')^0
 
 -- Keys and literals.
 local colon = S(' \t')^0 * ':' * (lexer.space + -1)
-local key = token(lexer.KEYWORD,
-                  #word * (lexer.nonnewline - colon)^1 * #colon *
-                  P(function(input, index)
-                    local line = input:sub(1, index - 1):match('[^\r\n]+$')
-                    return not line:find('[%w-]+:') and index
-                  end))
+local key = token(lexer.KEYWORD, #word * (lexer.nonnewline - colon)^1 * #colon *
+  P(function(input, index)
+    local line = input:sub(1, index - 1):match('[^\r\n]+$')
+    return not line:find('[%w-]+:') and index
+  end))
 local value = #word * (lexer.nonnewline - lexer.space^0 * S(',]}'))^1
 local block = S('|>') * S('+-')^-1 * (lexer.newline + -1) *
-              function(input, index)
-                local rest = input:sub(index)
-                local level = #rest:match('^( *)')
-                for pos, indent, line in rest:gmatch('() *()([^\r\n]+)') do
-                  if indent - pos < level and line ~= ' ' or
-                     level == 0 and pos > 1 then
-                    return index + pos - 1
-                  end
-                end
-                return #input + 1
-              end
+  function(input, index)
+    local rest = input:sub(index)
+    local level = #rest:match('^( *)')
+    for pos, indent, line in rest:gmatch('() *()([^\r\n]+)') do
+      if indent - pos < level and line ~= ' ' or level == 0 and pos > 1 then
+        return index + pos - 1
+      end
+    end
+    return #input + 1
+  end
 local literal = token('literal', value + block)
 
 -- Indicators.