rana-cli/wakatime/packages/pygments/lexers/agile.py

2290 lines
98 KiB
Python

# -*- coding: utf-8 -*-
"""
pygments.lexers.agile
~~~~~~~~~~~~~~~~~~~~~
Lexers for agile languages.
:copyright: Copyright 2006-2013 by the Pygments team, see AUTHORS.
:license: BSD, see LICENSE for details.
"""
import re
from pygments.lexer import Lexer, RegexLexer, ExtendedRegexLexer, \
LexerContext, include, combined, do_insertions, bygroups, using, this
from pygments.token import Error, Text, Other, \
Comment, Operator, Keyword, Name, String, Number, Generic, Punctuation
from pygments.util import get_bool_opt, get_list_opt, shebang_matches
from pygments import unistring as uni
__all__ = ['PythonLexer', 'PythonConsoleLexer', 'PythonTracebackLexer',
'Python3Lexer', 'Python3TracebackLexer', 'RubyLexer',
'RubyConsoleLexer', 'PerlLexer', 'LuaLexer', 'MoonScriptLexer',
'CrocLexer', 'MiniDLexer', 'IoLexer', 'TclLexer', 'FactorLexer',
'FancyLexer', 'DgLexer', 'Perl6Lexer']
# b/w compatibility
from pygments.lexers.functional import SchemeLexer
from pygments.lexers.jvm import IokeLexer, ClojureLexer
line_re = re.compile('.*?\n')
class PythonLexer(RegexLexer):
"""
For `Python <http://www.python.org>`_ source code.
"""
name = 'Python'
aliases = ['python', 'py', 'sage']
filenames = ['*.py', '*.pyw', '*.sc', 'SConstruct', 'SConscript', '*.tac', '*.sage']
mimetypes = ['text/x-python', 'application/x-python']
tokens = {
'root': [
(r'\n', Text),
(r'^(\s*)([rRuU]{,2}"""(?:.|\n)*?""")', bygroups(Text, String.Doc)),
(r"^(\s*)([rRuU]{,2}'''(?:.|\n)*?''')", bygroups(Text, String.Doc)),
(r'[^\S\n]+', Text),
(r'#.*$', Comment),
(r'[]{}:(),;[]', Punctuation),
(r'\\\n', Text),
(r'\\', Text),
(r'(in|is|and|or|not)\b', Operator.Word),
(r'!=|==|<<|>>|[-~+/*%=<>&^|.]', Operator),
include('keywords'),
(r'(def)((?:\s|\\\s)+)', bygroups(Keyword, Text), 'funcname'),
(r'(class)((?:\s|\\\s)+)', bygroups(Keyword, Text), 'classname'),
(r'(from)((?:\s|\\\s)+)', bygroups(Keyword.Namespace, Text),
'fromimport'),
(r'(import)((?:\s|\\\s)+)', bygroups(Keyword.Namespace, Text),
'import'),
include('builtins'),
include('backtick'),
('(?:[rR]|[uU][rR]|[rR][uU])"""', String, 'tdqs'),
("(?:[rR]|[uU][rR]|[rR][uU])'''", String, 'tsqs'),
('(?:[rR]|[uU][rR]|[rR][uU])"', String, 'dqs'),
("(?:[rR]|[uU][rR]|[rR][uU])'", String, 'sqs'),
('[uU]?"""', String, combined('stringescape', 'tdqs')),
("[uU]?'''", String, combined('stringescape', 'tsqs')),
('[uU]?"', String, combined('stringescape', 'dqs')),
("[uU]?'", String, combined('stringescape', 'sqs')),
include('name'),
include('numbers'),
],
'keywords': [
(r'(assert|break|continue|del|elif|else|except|exec|'
r'finally|for|global|if|lambda|pass|print|raise|'
r'return|try|while|yield(\s+from)?|as|with)\b', Keyword),
],
'builtins': [
(r'(?<!\.)(__import__|abs|all|any|apply|basestring|bin|bool|buffer|'
r'bytearray|bytes|callable|chr|classmethod|cmp|coerce|compile|'
r'complex|delattr|dict|dir|divmod|enumerate|eval|execfile|exit|'
r'file|filter|float|frozenset|getattr|globals|hasattr|hash|hex|id|'
r'input|int|intern|isinstance|issubclass|iter|len|list|locals|'
r'long|map|max|min|next|object|oct|open|ord|pow|property|range|'
r'raw_input|reduce|reload|repr|reversed|round|set|setattr|slice|'
r'sorted|staticmethod|str|sum|super|tuple|type|unichr|unicode|'
r'vars|xrange|zip)\b', Name.Builtin),
(r'(?<!\.)(self|None|Ellipsis|NotImplemented|False|True'
r')\b', Name.Builtin.Pseudo),
(r'(?<!\.)(ArithmeticError|AssertionError|AttributeError|'
r'BaseException|DeprecationWarning|EOFError|EnvironmentError|'
r'Exception|FloatingPointError|FutureWarning|GeneratorExit|IOError|'
r'ImportError|ImportWarning|IndentationError|IndexError|KeyError|'
r'KeyboardInterrupt|LookupError|MemoryError|NameError|'
r'NotImplemented|NotImplementedError|OSError|OverflowError|'
r'OverflowWarning|PendingDeprecationWarning|ReferenceError|'
r'RuntimeError|RuntimeWarning|StandardError|StopIteration|'
r'SyntaxError|SyntaxWarning|SystemError|SystemExit|TabError|'
r'TypeError|UnboundLocalError|UnicodeDecodeError|'
r'UnicodeEncodeError|UnicodeError|UnicodeTranslateError|'
r'UnicodeWarning|UserWarning|ValueError|VMSError|Warning|'
r'WindowsError|ZeroDivisionError)\b', Name.Exception),
],
'numbers': [
(r'(\d+\.\d*|\d*\.\d+)([eE][+-]?[0-9]+)?j?', Number.Float),
(r'\d+[eE][+-]?[0-9]+j?', Number.Float),
(r'0[0-7]+j?', Number.Oct),
(r'0[xX][a-fA-F0-9]+', Number.Hex),
(r'\d+L', Number.Integer.Long),
(r'\d+j?', Number.Integer)
],
'backtick': [
('`.*?`', String.Backtick),
],
'name': [
(r'@[a-zA-Z0-9_.]+', Name.Decorator),
('[a-zA-Z_][a-zA-Z0-9_]*', Name),
],
'funcname': [
('[a-zA-Z_][a-zA-Z0-9_]*', Name.Function, '#pop')
],
'classname': [
('[a-zA-Z_][a-zA-Z0-9_]*', Name.Class, '#pop')
],
'import': [
(r'(?:[ \t]|\\\n)+', Text),
(r'as\b', Keyword.Namespace),
(r',', Operator),
(r'[a-zA-Z_][a-zA-Z0-9_.]*', Name.Namespace),
(r'', Text, '#pop') # all else: go back
],
'fromimport': [
(r'(?:[ \t]|\\\n)+', Text),
(r'import\b', Keyword.Namespace, '#pop'),
# if None occurs here, it's "raise x from None", since None can
# never be a module name
(r'None\b', Name.Builtin.Pseudo, '#pop'),
# sadly, in "raise x from y" y will be highlighted as namespace too
(r'[a-zA-Z_.][a-zA-Z0-9_.]*', Name.Namespace),
# anything else here also means "raise x from y" and is therefore
# not an error
(r'', Text, '#pop'),
],
'stringescape': [
(r'\\([\\abfnrtv"\']|\n|N{.*?}|u[a-fA-F0-9]{4}|'
r'U[a-fA-F0-9]{8}|x[a-fA-F0-9]{2}|[0-7]{1,3})', String.Escape)
],
'strings': [
(r'%(\([a-zA-Z0-9_]+\))?[-#0 +]*([0-9]+|[*])?(\.([0-9]+|[*]))?'
'[hlL]?[diouxXeEfFgGcrs%]', String.Interpol),
(r'[^\\\'"%\n]+', String),
# quotes, percents and backslashes must be parsed one at a time
(r'[\'"\\]', String),
# unhandled string formatting sign
(r'%', String)
# newlines are an error (use "nl" state)
],
'nl': [
(r'\n', String)
],
'dqs': [
(r'"', String, '#pop'),
(r'\\\\|\\"|\\\n', String.Escape), # included here for raw strings
include('strings')
],
'sqs': [
(r"'", String, '#pop'),
(r"\\\\|\\'|\\\n", String.Escape), # included here for raw strings
include('strings')
],
'tdqs': [
(r'"""', String, '#pop'),
include('strings'),
include('nl')
],
'tsqs': [
(r"'''", String, '#pop'),
include('strings'),
include('nl')
],
}
def analyse_text(text):
return shebang_matches(text, r'pythonw?(2(\.\d)?)?') or \
'import ' in text[:1000]
class Python3Lexer(RegexLexer):
"""
For `Python <http://www.python.org>`_ source code (version 3.0).
*New in Pygments 0.10.*
"""
name = 'Python 3'
aliases = ['python3', 'py3']
filenames = [] # Nothing until Python 3 gets widespread
mimetypes = ['text/x-python3', 'application/x-python3']
flags = re.MULTILINE | re.UNICODE
uni_name = "[%s][%s]*" % (uni.xid_start, uni.xid_continue)
tokens = PythonLexer.tokens.copy()
tokens['keywords'] = [
(r'(assert|break|continue|del|elif|else|except|'
r'finally|for|global|if|lambda|pass|raise|nonlocal|'
r'return|try|while|yield(\s+from)?|as|with|True|False|None)\b',
Keyword),
]
tokens['builtins'] = [
(r'(?<!\.)(__import__|abs|all|any|bin|bool|bytearray|bytes|'
r'chr|classmethod|cmp|compile|complex|delattr|dict|dir|'
r'divmod|enumerate|eval|filter|float|format|frozenset|getattr|'
r'globals|hasattr|hash|hex|id|input|int|isinstance|issubclass|'
r'iter|len|list|locals|map|max|memoryview|min|next|object|oct|'
r'open|ord|pow|print|property|range|repr|reversed|round|'
r'set|setattr|slice|sorted|staticmethod|str|sum|super|tuple|type|'
r'vars|zip)\b', Name.Builtin),
(r'(?<!\.)(self|Ellipsis|NotImplemented)\b', Name.Builtin.Pseudo),
(r'(?<!\.)(ArithmeticError|AssertionError|AttributeError|'
r'BaseException|BufferError|BytesWarning|DeprecationWarning|'
r'EOFError|EnvironmentError|Exception|FloatingPointError|'
r'FutureWarning|GeneratorExit|IOError|ImportError|'
r'ImportWarning|IndentationError|IndexError|KeyError|'
r'KeyboardInterrupt|LookupError|MemoryError|NameError|'
r'NotImplementedError|OSError|OverflowError|'
r'PendingDeprecationWarning|ReferenceError|'
r'RuntimeError|RuntimeWarning|StopIteration|'
r'SyntaxError|SyntaxWarning|SystemError|SystemExit|TabError|'
r'TypeError|UnboundLocalError|UnicodeDecodeError|'
r'UnicodeEncodeError|UnicodeError|UnicodeTranslateError|'
r'UnicodeWarning|UserWarning|ValueError|VMSError|Warning|'
r'WindowsError|ZeroDivisionError|'
# new builtin exceptions from PEP 3151
r'BlockingIOError|ChildProcessError|ConnectionError|'
r'BrokenPipeError|ConnectionAbortedError|ConnectionRefusedError|'
r'ConnectionResetError|FileExistsError|FileNotFoundError|'
r'InterruptedError|IsADirectoryError|NotADirectoryError|'
r'PermissionError|ProcessLookupError|TimeoutError)\b',
Name.Exception),
]
tokens['numbers'] = [
(r'(\d+\.\d*|\d*\.\d+)([eE][+-]?[0-9]+)?', Number.Float),
(r'0[oO][0-7]+', Number.Oct),
(r'0[bB][01]+', Number.Bin),
(r'0[xX][a-fA-F0-9]+', Number.Hex),
(r'\d+', Number.Integer)
]
tokens['backtick'] = []
tokens['name'] = [
(r'@[a-zA-Z0-9_]+', Name.Decorator),
(uni_name, Name),
]
tokens['funcname'] = [
(uni_name, Name.Function, '#pop')
]
tokens['classname'] = [
(uni_name, Name.Class, '#pop')
]
tokens['import'] = [
(r'(\s+)(as)(\s+)', bygroups(Text, Keyword, Text)),
(r'\.', Name.Namespace),
(uni_name, Name.Namespace),
(r'(\s*)(,)(\s*)', bygroups(Text, Operator, Text)),
(r'', Text, '#pop') # all else: go back
]
tokens['fromimport'] = [
(r'(\s+)(import)\b', bygroups(Text, Keyword), '#pop'),
(r'\.', Name.Namespace),
(uni_name, Name.Namespace),
(r'', Text, '#pop'),
]
# don't highlight "%s" substitutions
tokens['strings'] = [
(r'[^\\\'"%\n]+', String),
# quotes, percents and backslashes must be parsed one at a time
(r'[\'"\\]', String),
# unhandled string formatting sign
(r'%', String)
# newlines are an error (use "nl" state)
]
def analyse_text(text):
return shebang_matches(text, r'pythonw?3(\.\d)?')
class PythonConsoleLexer(Lexer):
"""
For Python console output or doctests, such as:
.. sourcecode:: pycon
>>> a = 'foo'
>>> print a
foo
>>> 1 / 0
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ZeroDivisionError: integer division or modulo by zero
Additional options:
`python3`
Use Python 3 lexer for code. Default is ``False``.
*New in Pygments 1.0.*
"""
name = 'Python console session'
aliases = ['pycon']
mimetypes = ['text/x-python-doctest']
def __init__(self, **options):
self.python3 = get_bool_opt(options, 'python3', False)
Lexer.__init__(self, **options)
def get_tokens_unprocessed(self, text):
if self.python3:
pylexer = Python3Lexer(**self.options)
tblexer = Python3TracebackLexer(**self.options)
else:
pylexer = PythonLexer(**self.options)
tblexer = PythonTracebackLexer(**self.options)
curcode = ''
insertions = []
curtb = ''
tbindex = 0
tb = 0
for match in line_re.finditer(text):
line = match.group()
if line.startswith(u'>>> ') or line.startswith(u'... '):
tb = 0
insertions.append((len(curcode),
[(0, Generic.Prompt, line[:4])]))
curcode += line[4:]
elif line.rstrip() == u'...' and not tb:
# only a new >>> prompt can end an exception block
# otherwise an ellipsis in place of the traceback frames
# will be mishandled
insertions.append((len(curcode),
[(0, Generic.Prompt, u'...')]))
curcode += line[3:]
else:
if curcode:
for item in do_insertions(insertions,
pylexer.get_tokens_unprocessed(curcode)):
yield item
curcode = ''
insertions = []
if (line.startswith(u'Traceback (most recent call last):') or
re.match(ur' File "[^"]+", line \d+\n$', line)):
tb = 1
curtb = line
tbindex = match.start()
elif line == 'KeyboardInterrupt\n':
yield match.start(), Name.Class, line
elif tb:
curtb += line
if not (line.startswith(' ') or line.strip() == u'...'):
tb = 0
for i, t, v in tblexer.get_tokens_unprocessed(curtb):
yield tbindex+i, t, v
else:
yield match.start(), Generic.Output, line
if curcode:
for item in do_insertions(insertions,
pylexer.get_tokens_unprocessed(curcode)):
yield item
class PythonTracebackLexer(RegexLexer):
"""
For Python tracebacks.
*New in Pygments 0.7.*
"""
name = 'Python Traceback'
aliases = ['pytb']
filenames = ['*.pytb']
mimetypes = ['text/x-python-traceback']
tokens = {
'root': [
(r'^Traceback \(most recent call last\):\n',
Generic.Traceback, 'intb'),
# SyntaxError starts with this.
(r'^(?= File "[^"]+", line \d+)', Generic.Traceback, 'intb'),
(r'^.*\n', Other),
],
'intb': [
(r'^( File )("[^"]+")(, line )(\d+)(, in )(.+)(\n)',
bygroups(Text, Name.Builtin, Text, Number, Text, Name, Text)),
(r'^( File )("[^"]+")(, line )(\d+)(\n)',
bygroups(Text, Name.Builtin, Text, Number, Text)),
(r'^( )(.+)(\n)',
bygroups(Text, using(PythonLexer), Text)),
(r'^([ \t]*)(\.\.\.)(\n)',
bygroups(Text, Comment, Text)), # for doctests...
(r'^([^:]+)(: )(.+)(\n)',
bygroups(Generic.Error, Text, Name, Text), '#pop'),
(r'^([a-zA-Z_][a-zA-Z0-9_]*)(:?\n)',
bygroups(Generic.Error, Text), '#pop')
],
}
class Python3TracebackLexer(RegexLexer):
"""
For Python 3.0 tracebacks, with support for chained exceptions.
*New in Pygments 1.0.*
"""
name = 'Python 3.0 Traceback'
aliases = ['py3tb']
filenames = ['*.py3tb']
mimetypes = ['text/x-python3-traceback']
tokens = {
'root': [
(r'\n', Text),
(r'^Traceback \(most recent call last\):\n', Generic.Traceback, 'intb'),
(r'^During handling of the above exception, another '
r'exception occurred:\n\n', Generic.Traceback),
(r'^The above exception was the direct cause of the '
r'following exception:\n\n', Generic.Traceback),
(r'^(?= File "[^"]+", line \d+)', Generic.Traceback, 'intb'),
],
'intb': [
(r'^( File )("[^"]+")(, line )(\d+)(, in )(.+)(\n)',
bygroups(Text, Name.Builtin, Text, Number, Text, Name, Text)),
(r'^( File )("[^"]+")(, line )(\d+)(\n)',
bygroups(Text, Name.Builtin, Text, Number, Text)),
(r'^( )(.+)(\n)',
bygroups(Text, using(Python3Lexer), Text)),
(r'^([ \t]*)(\.\.\.)(\n)',
bygroups(Text, Comment, Text)), # for doctests...
(r'^([^:]+)(: )(.+)(\n)',
bygroups(Generic.Error, Text, Name, Text), '#pop'),
(r'^([a-zA-Z_][a-zA-Z0-9_]*)(:?\n)',
bygroups(Generic.Error, Text), '#pop')
],
}
class RubyLexer(ExtendedRegexLexer):
"""
For `Ruby <http://www.ruby-lang.org>`_ source code.
"""
name = 'Ruby'
aliases = ['rb', 'ruby', 'duby']
filenames = ['*.rb', '*.rbw', 'Rakefile', '*.rake', '*.gemspec',
'*.rbx', '*.duby']
mimetypes = ['text/x-ruby', 'application/x-ruby']
flags = re.DOTALL | re.MULTILINE
def heredoc_callback(self, match, ctx):
# okay, this is the hardest part of parsing Ruby...
# match: 1 = <<-?, 2 = quote? 3 = name 4 = quote? 5 = rest of line
start = match.start(1)
yield start, Operator, match.group(1) # <<-?
yield match.start(2), String.Heredoc, match.group(2) # quote ", ', `
yield match.start(3), Name.Constant, match.group(3) # heredoc name
yield match.start(4), String.Heredoc, match.group(4) # quote again
heredocstack = ctx.__dict__.setdefault('heredocstack', [])
outermost = not bool(heredocstack)
heredocstack.append((match.group(1) == '<<-', match.group(3)))
ctx.pos = match.start(5)
ctx.end = match.end(5)
# this may find other heredocs
for i, t, v in self.get_tokens_unprocessed(context=ctx):
yield i, t, v
ctx.pos = match.end()
if outermost:
# this is the outer heredoc again, now we can process them all
for tolerant, hdname in heredocstack:
lines = []
for match in line_re.finditer(ctx.text, ctx.pos):
if tolerant:
check = match.group().strip()
else:
check = match.group().rstrip()
if check == hdname:
for amatch in lines:
yield amatch.start(), String.Heredoc, amatch.group()
yield match.start(), Name.Constant, match.group()
ctx.pos = match.end()
break
else:
lines.append(match)
else:
# end of heredoc not found -- error!
for amatch in lines:
yield amatch.start(), Error, amatch.group()
ctx.end = len(ctx.text)
del heredocstack[:]
def gen_rubystrings_rules():
def intp_regex_callback(self, match, ctx):
yield match.start(1), String.Regex, match.group(1) # begin
nctx = LexerContext(match.group(3), 0, ['interpolated-regex'])
for i, t, v in self.get_tokens_unprocessed(context=nctx):
yield match.start(3)+i, t, v
yield match.start(4), String.Regex, match.group(4) # end[mixounse]*
ctx.pos = match.end()
def intp_string_callback(self, match, ctx):
yield match.start(1), String.Other, match.group(1)
nctx = LexerContext(match.group(3), 0, ['interpolated-string'])
for i, t, v in self.get_tokens_unprocessed(context=nctx):
yield match.start(3)+i, t, v
yield match.start(4), String.Other, match.group(4) # end
ctx.pos = match.end()
states = {}
states['strings'] = [
# easy ones
(r'\:@{0,2}([a-zA-Z_]\w*[\!\?]?|\*\*?|[-+]@?|'
r'[/%&|^`~]|\[\]=?|<<|>>|<=?>|>=?|===?)', String.Symbol),
(r":'(\\\\|\\'|[^'])*'", String.Symbol),
(r"'(\\\\|\\'|[^'])*'", String.Single),
(r':"', String.Symbol, 'simple-sym'),
(r'([a-zA-Z_][a-zA-Z0-9]*)(:)(?!:)',
bygroups(String.Symbol, Punctuation)), # Since Ruby 1.9
(r'"', String.Double, 'simple-string'),
(r'(?<!\.)`', String.Backtick, 'simple-backtick'),
]
# double-quoted string and symbol
for name, ttype, end in ('string', String.Double, '"'), \
('sym', String.Symbol, '"'), \
('backtick', String.Backtick, '`'):
states['simple-'+name] = [
include('string-intp-escaped'),
(r'[^\\%s#]+' % end, ttype),
(r'[\\#]', ttype),
(end, ttype, '#pop'),
]
# braced quoted strings
for lbrace, rbrace, name in ('\\{', '\\}', 'cb'), \
('\\[', '\\]', 'sb'), \
('\\(', '\\)', 'pa'), \
('<', '>', 'ab'):
states[name+'-intp-string'] = [
(r'\\[\\' + lbrace + rbrace + ']', String.Other),
(r'(?<!\\)' + lbrace, String.Other, '#push'),
(r'(?<!\\)' + rbrace, String.Other, '#pop'),
include('string-intp-escaped'),
(r'[\\#' + lbrace + rbrace + ']', String.Other),
(r'[^\\#' + lbrace + rbrace + ']+', String.Other),
]
states['strings'].append((r'%[QWx]?' + lbrace, String.Other,
name+'-intp-string'))
states[name+'-string'] = [
(r'\\[\\' + lbrace + rbrace + ']', String.Other),
(r'(?<!\\)' + lbrace, String.Other, '#push'),
(r'(?<!\\)' + rbrace, String.Other, '#pop'),
(r'[\\#' + lbrace + rbrace + ']', String.Other),
(r'[^\\#' + lbrace + rbrace + ']+', String.Other),
]
states['strings'].append((r'%[qsw]' + lbrace, String.Other,
name+'-string'))
states[name+'-regex'] = [
(r'\\[\\' + lbrace + rbrace + ']', String.Regex),
(r'(?<!\\)' + lbrace, String.Regex, '#push'),
(r'(?<!\\)' + rbrace + '[mixounse]*', String.Regex, '#pop'),
include('string-intp'),
(r'[\\#' + lbrace + rbrace + ']', String.Regex),
(r'[^\\#' + lbrace + rbrace + ']+', String.Regex),
]
states['strings'].append((r'%r' + lbrace, String.Regex,
name+'-regex'))
# these must come after %<brace>!
states['strings'] += [
# %r regex
(r'(%r([^a-zA-Z0-9]))((?:\\\2|(?!\2).)*)(\2[mixounse]*)',
intp_regex_callback),
# regular fancy strings with qsw
(r'%[qsw]([^a-zA-Z0-9])((?:\\\1|(?!\1).)*)\1', String.Other),
(r'(%[QWx]([^a-zA-Z0-9]))((?:\\\2|(?!\2).)*)(\2)',
intp_string_callback),
# special forms of fancy strings after operators or
# in method calls with braces
(r'(?<=[-+/*%=<>&!^|~,(])(\s*)(%([\t ])(?:(?:\\\3|(?!\3).)*)\3)',
bygroups(Text, String.Other, None)),
# and because of fixed width lookbehinds the whole thing a
# second time for line startings...
(r'^(\s*)(%([\t ])(?:(?:\\\3|(?!\3).)*)\3)',
bygroups(Text, String.Other, None)),
# all regular fancy strings without qsw
(r'(%([^a-zA-Z0-9\s]))((?:\\\2|(?!\2).)*)(\2)',
intp_string_callback),
]
return states
tokens = {
'root': [
(r'#.*?$', Comment.Single),
(r'=begin\s.*?\n=end.*?$', Comment.Multiline),
# keywords
(r'(BEGIN|END|alias|begin|break|case|defined\?|'
r'do|else|elsif|end|ensure|for|if|in|next|redo|'
r'rescue|raise|retry|return|super|then|undef|unless|until|when|'
r'while|yield)\b', Keyword),
# start of function, class and module names
(r'(module)(\s+)([a-zA-Z_][a-zA-Z0-9_]*'
r'(?:::[a-zA-Z_][a-zA-Z0-9_]*)*)',
bygroups(Keyword, Text, Name.Namespace)),
(r'(def)(\s+)', bygroups(Keyword, Text), 'funcname'),
(r'def(?=[*%&^`~+-/\[<>=])', Keyword, 'funcname'),
(r'(class)(\s+)', bygroups(Keyword, Text), 'classname'),
# special methods
(r'(initialize|new|loop|include|extend|raise|attr_reader|'
r'attr_writer|attr_accessor|attr|catch|throw|private|'
r'module_function|public|protected|true|false|nil)\b',
Keyword.Pseudo),
(r'(not|and|or)\b', Operator.Word),
(r'(autoload|block_given|const_defined|eql|equal|frozen|include|'
r'instance_of|is_a|iterator|kind_of|method_defined|nil|'
r'private_method_defined|protected_method_defined|'
r'public_method_defined|respond_to|tainted)\?', Name.Builtin),
(r'(chomp|chop|exit|gsub|sub)!', Name.Builtin),
(r'(?<!\.)(Array|Float|Integer|String|__id__|__send__|abort|'
r'ancestors|at_exit|autoload|binding|callcc|caller|'
r'catch|chomp|chop|class_eval|class_variables|'
r'clone|const_defined\?|const_get|const_missing|const_set|'
r'constants|display|dup|eval|exec|exit|extend|fail|fork|'
r'format|freeze|getc|gets|global_variables|gsub|'
r'hash|id|included_modules|inspect|instance_eval|'
r'instance_method|instance_methods|'
r'instance_variable_get|instance_variable_set|instance_variables|'
r'lambda|load|local_variables|loop|'
r'method|method_missing|methods|module_eval|name|'
r'object_id|open|p|print|printf|private_class_method|'
r'private_instance_methods|'
r'private_methods|proc|protected_instance_methods|'
r'protected_methods|public_class_method|'
r'public_instance_methods|public_methods|'
r'putc|puts|raise|rand|readline|readlines|require|'
r'scan|select|self|send|set_trace_func|singleton_methods|sleep|'
r'split|sprintf|srand|sub|syscall|system|taint|'
r'test|throw|to_a|to_s|trace_var|trap|untaint|untrace_var|'
r'warn)\b', Name.Builtin),
(r'__(FILE|LINE)__\b', Name.Builtin.Pseudo),
# normal heredocs
(r'(?<!\w)(<<-?)(["`\']?)([a-zA-Z_]\w*)(\2)(.*?\n)',
heredoc_callback),
# empty string heredocs
(r'(<<-?)("|\')()(\2)(.*?\n)', heredoc_callback),
(r'__END__', Comment.Preproc, 'end-part'),
# multiline regex (after keywords or assignments)
(r'(?:^|(?<=[=<>~!:])|'
r'(?<=(?:\s|;)when\s)|'
r'(?<=(?:\s|;)or\s)|'
r'(?<=(?:\s|;)and\s)|'
r'(?<=(?:\s|;|\.)index\s)|'
r'(?<=(?:\s|;|\.)scan\s)|'
r'(?<=(?:\s|;|\.)sub\s)|'
r'(?<=(?:\s|;|\.)sub!\s)|'
r'(?<=(?:\s|;|\.)gsub\s)|'
r'(?<=(?:\s|;|\.)gsub!\s)|'
r'(?<=(?:\s|;|\.)match\s)|'
r'(?<=(?:\s|;)if\s)|'
r'(?<=(?:\s|;)elsif\s)|'
r'(?<=^when\s)|'
r'(?<=^index\s)|'
r'(?<=^scan\s)|'
r'(?<=^sub\s)|'
r'(?<=^gsub\s)|'
r'(?<=^sub!\s)|'
r'(?<=^gsub!\s)|'
r'(?<=^match\s)|'
r'(?<=^if\s)|'
r'(?<=^elsif\s)'
r')(\s*)(/)', bygroups(Text, String.Regex), 'multiline-regex'),
# multiline regex (in method calls or subscripts)
(r'(?<=\(|,|\[)/', String.Regex, 'multiline-regex'),
# multiline regex (this time the funny no whitespace rule)
(r'(\s+)(/)(?![\s=])', bygroups(Text, String.Regex),
'multiline-regex'),
# lex numbers and ignore following regular expressions which
# are division operators in fact (grrrr. i hate that. any
# better ideas?)
# since pygments 0.7 we also eat a "?" operator after numbers
# so that the char operator does not work. Chars are not allowed
# there so that you can use the ternary operator.
# stupid example:
# x>=0?n[x]:""
(r'(0_?[0-7]+(?:_[0-7]+)*)(\s*)([/?])?',
bygroups(Number.Oct, Text, Operator)),
(r'(0x[0-9A-Fa-f]+(?:_[0-9A-Fa-f]+)*)(\s*)([/?])?',
bygroups(Number.Hex, Text, Operator)),
(r'(0b[01]+(?:_[01]+)*)(\s*)([/?])?',
bygroups(Number.Bin, Text, Operator)),
(r'([\d]+(?:_\d+)*)(\s*)([/?])?',
bygroups(Number.Integer, Text, Operator)),
# Names
(r'@@[a-zA-Z_][a-zA-Z0-9_]*', Name.Variable.Class),
(r'@[a-zA-Z_][a-zA-Z0-9_]*', Name.Variable.Instance),
(r'\$[a-zA-Z0-9_]+', Name.Variable.Global),
(r'\$[!@&`\'+~=/\\,;.<>_*$?:"]', Name.Variable.Global),
(r'\$-[0adFiIlpvw]', Name.Variable.Global),
(r'::', Operator),
include('strings'),
# chars
(r'\?(\\[MC]-)*' # modifiers
r'(\\([\\abefnrstv#"\']|x[a-fA-F0-9]{1,2}|[0-7]{1,3})|\S)'
r'(?!\w)',
String.Char),
(r'[A-Z][a-zA-Z0-9_]+', Name.Constant),
# this is needed because ruby attributes can look
# like keywords (class) or like this: ` ?!?
(r'(\.|::)([a-zA-Z_]\w*[\!\?]?|[*%&^`~+-/\[<>=])',
bygroups(Operator, Name)),
(r'[a-zA-Z_]\w*[\!\?]?', Name),
(r'(\[|\]|\*\*|<<?|>>?|>=|<=|<=>|=~|={3}|'
r'!~|&&?|\|\||\.{1,3})', Operator),
(r'[-+/*%=<>&!^|~]=?', Operator),
(r'[(){};,/?:\\]', Punctuation),
(r'\s+', Text)
],
'funcname': [
(r'\(', Punctuation, 'defexpr'),
(r'(?:([a-zA-Z_][a-zA-Z0-9_]*)(\.))?'
r'([a-zA-Z_]\w*[\!\?]?|\*\*?|[-+]@?|'
r'[/%&|^`~]|\[\]=?|<<|>>|<=?>|>=?|===?)',
bygroups(Name.Class, Operator, Name.Function), '#pop'),
(r'', Text, '#pop')
],
'classname': [
(r'\(', Punctuation, 'defexpr'),
(r'<<', Operator, '#pop'),
(r'[A-Z_]\w*', Name.Class, '#pop'),
(r'', Text, '#pop')
],
'defexpr': [
(r'(\))(\.|::)?', bygroups(Punctuation, Operator), '#pop'),
(r'\(', Operator, '#push'),
include('root')
],
'in-intp': [
('}', String.Interpol, '#pop'),
include('root'),
],
'string-intp': [
(r'#{', String.Interpol, 'in-intp'),
(r'#@@?[a-zA-Z_][a-zA-Z0-9_]*', String.Interpol),
(r'#\$[a-zA-Z_][a-zA-Z0-9_]*', String.Interpol)
],
'string-intp-escaped': [
include('string-intp'),
(r'\\([\\abefnrstv#"\']|x[a-fA-F0-9]{1,2}|[0-7]{1,3})',
String.Escape)
],
'interpolated-regex': [
include('string-intp'),
(r'[\\#]', String.Regex),
(r'[^\\#]+', String.Regex),
],
'interpolated-string': [
include('string-intp'),
(r'[\\#]', String.Other),
(r'[^\\#]+', String.Other),
],
'multiline-regex': [
include('string-intp'),
(r'\\\\', String.Regex),
(r'\\/', String.Regex),
(r'[\\#]', String.Regex),
(r'[^\\/#]+', String.Regex),
(r'/[mixounse]*', String.Regex, '#pop'),
],
'end-part': [
(r'.+', Comment.Preproc, '#pop')
]
}
tokens.update(gen_rubystrings_rules())
def analyse_text(text):
return shebang_matches(text, r'ruby(1\.\d)?')
class RubyConsoleLexer(Lexer):
"""
For Ruby interactive console (**irb**) output like:
.. sourcecode:: rbcon
irb(main):001:0> a = 1
=> 1
irb(main):002:0> puts a
1
=> nil
"""
name = 'Ruby irb session'
aliases = ['rbcon', 'irb']
mimetypes = ['text/x-ruby-shellsession']
_prompt_re = re.compile('irb\([a-zA-Z_][a-zA-Z0-9_]*\):\d{3}:\d+[>*"\'] '
'|>> |\?> ')
def get_tokens_unprocessed(self, text):
rblexer = RubyLexer(**self.options)
curcode = ''
insertions = []
for match in line_re.finditer(text):
line = match.group()
m = self._prompt_re.match(line)
if m is not None:
end = m.end()
insertions.append((len(curcode),
[(0, Generic.Prompt, line[:end])]))
curcode += line[end:]
else:
if curcode:
for item in do_insertions(insertions,
rblexer.get_tokens_unprocessed(curcode)):
yield item
curcode = ''
insertions = []
yield match.start(), Generic.Output, line
if curcode:
for item in do_insertions(insertions,
rblexer.get_tokens_unprocessed(curcode)):
yield item
class PerlLexer(RegexLexer):
"""
For `Perl <http://www.perl.org>`_ source code.
"""
name = 'Perl'
aliases = ['perl', 'pl']
filenames = ['*.pl', '*.pm']
mimetypes = ['text/x-perl', 'application/x-perl']
flags = re.DOTALL | re.MULTILINE
# TODO: give this to a perl guy who knows how to parse perl...
tokens = {
'balanced-regex': [
(r'/(\\\\|\\[^\\]|[^\\/])*/[egimosx]*', String.Regex, '#pop'),
(r'!(\\\\|\\[^\\]|[^\\!])*![egimosx]*', String.Regex, '#pop'),
(r'\\(\\\\|[^\\])*\\[egimosx]*', String.Regex, '#pop'),
(r'{(\\\\|\\[^\\]|[^\\}])*}[egimosx]*', String.Regex, '#pop'),
(r'<(\\\\|\\[^\\]|[^\\>])*>[egimosx]*', String.Regex, '#pop'),
(r'\[(\\\\|\\[^\\]|[^\\\]])*\][egimosx]*', String.Regex, '#pop'),
(r'\((\\\\|\\[^\\]|[^\\\)])*\)[egimosx]*', String.Regex, '#pop'),
(r'@(\\\\|\\[^\\]|[^\\\@])*@[egimosx]*', String.Regex, '#pop'),
(r'%(\\\\|\\[^\\]|[^\\\%])*%[egimosx]*', String.Regex, '#pop'),
(r'\$(\\\\|\\[^\\]|[^\\\$])*\$[egimosx]*', String.Regex, '#pop'),
],
'root': [
(r'\#.*?$', Comment.Single),
(r'^=[a-zA-Z0-9]+\s+.*?\n=cut', Comment.Multiline),
(r'(case|continue|do|else|elsif|for|foreach|if|last|my|'
r'next|our|redo|reset|then|unless|until|while|use|'
r'print|new|BEGIN|CHECK|INIT|END|return)\b', Keyword),
(r'(format)(\s+)([a-zA-Z0-9_]+)(\s*)(=)(\s*\n)',
bygroups(Keyword, Text, Name, Text, Punctuation, Text), 'format'),
(r'(eq|lt|gt|le|ge|ne|not|and|or|cmp)\b', Operator.Word),
# common delimiters
(r's/(\\\\|\\[^\\]|[^\\/])*/(\\\\|\\[^\\]|[^\\/])*/[egimosx]*',
String.Regex),
(r's!(\\\\|\\!|[^!])*!(\\\\|\\!|[^!])*![egimosx]*', String.Regex),
(r's\\(\\\\|[^\\])*\\(\\\\|[^\\])*\\[egimosx]*', String.Regex),
(r's@(\\\\|\\[^\\]|[^\\@])*@(\\\\|\\[^\\]|[^\\@])*@[egimosx]*',
String.Regex),
(r's%(\\\\|\\[^\\]|[^\\%])*%(\\\\|\\[^\\]|[^\\%])*%[egimosx]*',
String.Regex),
# balanced delimiters
(r's{(\\\\|\\[^\\]|[^\\}])*}\s*', String.Regex, 'balanced-regex'),
(r's<(\\\\|\\[^\\]|[^\\>])*>\s*', String.Regex, 'balanced-regex'),
(r's\[(\\\\|\\[^\\]|[^\\\]])*\]\s*', String.Regex,
'balanced-regex'),
(r's\((\\\\|\\[^\\]|[^\\\)])*\)\s*', String.Regex,
'balanced-regex'),
(r'm?/(\\\\|\\[^\\]|[^\\/\n])*/[gcimosx]*', String.Regex),
(r'm(?=[/!\\{<\[\(@%\$])', String.Regex, 'balanced-regex'),
(r'((?<==~)|(?<=\())\s*/(\\\\|\\[^\\]|[^\\/])*/[gcimosx]*',
String.Regex),
(r'\s+', Text),
(r'(abs|accept|alarm|atan2|bind|binmode|bless|caller|chdir|'
r'chmod|chomp|chop|chown|chr|chroot|close|closedir|connect|'
r'continue|cos|crypt|dbmclose|dbmopen|defined|delete|die|'
r'dump|each|endgrent|endhostent|endnetent|endprotoent|'
r'endpwent|endservent|eof|eval|exec|exists|exit|exp|fcntl|'
r'fileno|flock|fork|format|formline|getc|getgrent|getgrgid|'
r'getgrnam|gethostbyaddr|gethostbyname|gethostent|getlogin|'
r'getnetbyaddr|getnetbyname|getnetent|getpeername|getpgrp|'
r'getppid|getpriority|getprotobyname|getprotobynumber|'
r'getprotoent|getpwent|getpwnam|getpwuid|getservbyname|'
r'getservbyport|getservent|getsockname|getsockopt|glob|gmtime|'
r'goto|grep|hex|import|index|int|ioctl|join|keys|kill|last|'
r'lc|lcfirst|length|link|listen|local|localtime|log|lstat|'
r'map|mkdir|msgctl|msgget|msgrcv|msgsnd|my|next|no|oct|open|'
r'opendir|ord|our|pack|package|pipe|pop|pos|printf|'
r'prototype|push|quotemeta|rand|read|readdir|'
r'readline|readlink|readpipe|recv|redo|ref|rename|require|'
r'reverse|rewinddir|rindex|rmdir|scalar|seek|seekdir|'
r'select|semctl|semget|semop|send|setgrent|sethostent|setnetent|'
r'setpgrp|setpriority|setprotoent|setpwent|setservent|'
r'setsockopt|shift|shmctl|shmget|shmread|shmwrite|shutdown|'
r'sin|sleep|socket|socketpair|sort|splice|split|sprintf|sqrt|'
r'srand|stat|study|substr|symlink|syscall|sysopen|sysread|'
r'sysseek|system|syswrite|tell|telldir|tie|tied|time|times|tr|'
r'truncate|uc|ucfirst|umask|undef|unlink|unpack|unshift|untie|'
r'utime|values|vec|wait|waitpid|wantarray|warn|write'
r')\b', Name.Builtin),
(r'((__(DATA|DIE|WARN)__)|(STD(IN|OUT|ERR)))\b', Name.Builtin.Pseudo),
(r'<<([\'"]?)([a-zA-Z_][a-zA-Z0-9_]*)\1;?\n.*?\n\2\n', String),
(r'__END__', Comment.Preproc, 'end-part'),
(r'\$\^[ADEFHILMOPSTWX]', Name.Variable.Global),
(r"\$[\\\"\[\]'&`+*.,;=%~?@$!<>(^|/-](?!\w)", Name.Variable.Global),
(r'[$@%#]+', Name.Variable, 'varname'),
(r'0_?[0-7]+(_[0-7]+)*', Number.Oct),
(r'0x[0-9A-Fa-f]+(_[0-9A-Fa-f]+)*', Number.Hex),
(r'0b[01]+(_[01]+)*', Number.Bin),
(r'(?i)(\d*(_\d*)*\.\d+(_\d*)*|\d+(_\d*)*\.\d+(_\d*)*)(e[+-]?\d+)?',
Number.Float),
(r'(?i)\d+(_\d*)*e[+-]?\d+(_\d*)*', Number.Float),
(r'\d+(_\d+)*', Number.Integer),
(r"'(\\\\|\\[^\\]|[^'\\])*'", String),
(r'"(\\\\|\\[^\\]|[^"\\])*"', String),
(r'`(\\\\|\\[^\\]|[^`\\])*`', String.Backtick),
(r'<([^\s>]+)>', String.Regex),
(r'(q|qq|qw|qr|qx)\{', String.Other, 'cb-string'),
(r'(q|qq|qw|qr|qx)\(', String.Other, 'rb-string'),
(r'(q|qq|qw|qr|qx)\[', String.Other, 'sb-string'),
(r'(q|qq|qw|qr|qx)\<', String.Other, 'lt-string'),
(r'(q|qq|qw|qr|qx)([^a-zA-Z0-9])(.|\n)*?\2', String.Other),
(r'package\s+', Keyword, 'modulename'),
(r'sub\s+', Keyword, 'funcname'),
(r'(\[\]|\*\*|::|<<|>>|>=|<=>|<=|={3}|!=|=~|'
r'!~|&&?|\|\||\.{1,3})', Operator),
(r'[-+/*%=<>&^|!\\~]=?', Operator),
(r'[\(\)\[\]:;,<>/\?\{\}]', Punctuation), # yes, there's no shortage
# of punctuation in Perl!
(r'(?=\w)', Name, 'name'),
],
'format': [
(r'\.\n', String.Interpol, '#pop'),
(r'[^\n]*\n', String.Interpol),
],
'varname': [
(r'\s+', Text),
(r'\{', Punctuation, '#pop'), # hash syntax?
(r'\)|,', Punctuation, '#pop'), # argument specifier
(r'[a-zA-Z0-9_]+::', Name.Namespace),
(r'[a-zA-Z0-9_:]+', Name.Variable, '#pop'),
],
'name': [
(r'[a-zA-Z0-9_]+::', Name.Namespace),
(r'[a-zA-Z0-9_:]+', Name, '#pop'),
(r'[A-Z_]+(?=[^a-zA-Z0-9_])', Name.Constant, '#pop'),
(r'(?=[^a-zA-Z0-9_])', Text, '#pop'),
],
'modulename': [
(r'[a-zA-Z_]\w*', Name.Namespace, '#pop')
],
'funcname': [
(r'[a-zA-Z_]\w*[\!\?]?', Name.Function),
(r'\s+', Text),
# argument declaration
(r'(\([$@%]*\))(\s*)', bygroups(Punctuation, Text)),
(r'.*?{', Punctuation, '#pop'),
(r';', Punctuation, '#pop'),
],
'cb-string': [
(r'\\[\{\}\\]', String.Other),
(r'\\', String.Other),
(r'\{', String.Other, 'cb-string'),
(r'\}', String.Other, '#pop'),
(r'[^\{\}\\]+', String.Other)
],
'rb-string': [
(r'\\[\(\)\\]', String.Other),
(r'\\', String.Other),
(r'\(', String.Other, 'rb-string'),
(r'\)', String.Other, '#pop'),
(r'[^\(\)]+', String.Other)
],
'sb-string': [
(r'\\[\[\]\\]', String.Other),
(r'\\', String.Other),
(r'\[', String.Other, 'sb-string'),
(r'\]', String.Other, '#pop'),
(r'[^\[\]]+', String.Other)
],
'lt-string': [
(r'\\[\<\>\\]', String.Other),
(r'\\', String.Other),
(r'\<', String.Other, 'lt-string'),
(r'\>', String.Other, '#pop'),
(r'[^\<\>]+', String.Other)
],
'end-part': [
(r'.+', Comment.Preproc, '#pop')
]
}
def analyse_text(text):
if shebang_matches(text, r'perl'):
return True
if 'my $' in text:
return 0.9
return 0.1 # who knows, might still be perl!
class LuaLexer(RegexLexer):
"""
For `Lua <http://www.lua.org>`_ source code.
Additional options accepted:
`func_name_highlighting`
If given and ``True``, highlight builtin function names
(default: ``True``).
`disabled_modules`
If given, must be a list of module names whose function names
should not be highlighted. By default all modules are highlighted.
To get a list of allowed modules have a look into the
`_luabuiltins` module:
.. sourcecode:: pycon
>>> from pygments.lexers._luabuiltins import MODULES
>>> MODULES.keys()
['string', 'coroutine', 'modules', 'io', 'basic', ...]
"""
name = 'Lua'
aliases = ['lua']
filenames = ['*.lua', '*.wlua']
mimetypes = ['text/x-lua', 'application/x-lua']
tokens = {
'root': [
# lua allows a file to start with a shebang
(r'#!(.*?)$', Comment.Preproc),
(r'', Text, 'base'),
],
'base': [
(r'(?s)--\[(=*)\[.*?\]\1\]', Comment.Multiline),
('--.*$', Comment.Single),
(r'(?i)(\d*\.\d+|\d+\.\d*)(e[+-]?\d+)?', Number.Float),
(r'(?i)\d+e[+-]?\d+', Number.Float),
('(?i)0x[0-9a-f]*', Number.Hex),
(r'\d+', Number.Integer),
(r'\n', Text),
(r'[^\S\n]', Text),
# multiline strings
(r'(?s)\[(=*)\[.*?\]\1\]', String),
(r'(==|~=|<=|>=|\.\.\.|\.\.|[=+\-*/%^<>#])', Operator),
(r'[\[\]\{\}\(\)\.,:;]', Punctuation),
(r'(and|or|not)\b', Operator.Word),
('(break|do|else|elseif|end|for|if|in|repeat|return|then|until|'
r'while)\b', Keyword),
(r'(local)\b', Keyword.Declaration),
(r'(true|false|nil)\b', Keyword.Constant),
(r'(function)\b', Keyword, 'funcname'),
(r'[A-Za-z_][A-Za-z0-9_]*(\.[A-Za-z_][A-Za-z0-9_]*)?', Name),
("'", String.Single, combined('stringescape', 'sqs')),
('"', String.Double, combined('stringescape', 'dqs'))
],
'funcname': [
(r'\s+', Text),
('(?:([A-Za-z_][A-Za-z0-9_]*)(\.))?([A-Za-z_][A-Za-z0-9_]*)',
bygroups(Name.Class, Punctuation, Name.Function), '#pop'),
# inline function
('\(', Punctuation, '#pop'),
],
# if I understand correctly, every character is valid in a lua string,
# so this state is only for later corrections
'string': [
('.', String)
],
'stringescape': [
(r'''\\([abfnrtv\\"']|\d{1,3})''', String.Escape)
],
'sqs': [
("'", String, '#pop'),
include('string')
],
'dqs': [
('"', String, '#pop'),
include('string')
]
}
def __init__(self, **options):
self.func_name_highlighting = get_bool_opt(
options, 'func_name_highlighting', True)
self.disabled_modules = get_list_opt(options, 'disabled_modules', [])
self._functions = set()
if self.func_name_highlighting:
from pygments.lexers._luabuiltins import MODULES
for mod, func in MODULES.iteritems():
if mod not in self.disabled_modules:
self._functions.update(func)
RegexLexer.__init__(self, **options)
def get_tokens_unprocessed(self, text):
for index, token, value in \
RegexLexer.get_tokens_unprocessed(self, text):
if token is Name:
if value in self._functions:
yield index, Name.Builtin, value
continue
elif '.' in value:
a, b = value.split('.')
yield index, Name, a
yield index + len(a), Punctuation, u'.'
yield index + len(a) + 1, Name, b
continue
yield index, token, value
class MoonScriptLexer(LuaLexer):
"""
For `MoonScript <http://moonscript.org.org>`_ source code.
*New in Pygments 1.5.*
"""
name = "MoonScript"
aliases = ["moon", "moonscript"]
filenames = ["*.moon"]
mimetypes = ['text/x-moonscript', 'application/x-moonscript']
tokens = {
'root': [
(r'#!(.*?)$', Comment.Preproc),
(r'', Text, 'base'),
],
'base': [
('--.*$', Comment.Single),
(r'(?i)(\d*\.\d+|\d+\.\d*)(e[+-]?\d+)?', Number.Float),
(r'(?i)\d+e[+-]?\d+', Number.Float),
(r'(?i)0x[0-9a-f]*', Number.Hex),
(r'\d+', Number.Integer),
(r'\n', Text),
(r'[^\S\n]+', Text),
(r'(?s)\[(=*)\[.*?\]\1\]', String),
(r'(->|=>)', Name.Function),
(r':[a-zA-Z_][a-zA-Z0-9_]*', Name.Variable),
(r'(==|!=|~=|<=|>=|\.\.\.|\.\.|[=+\-*/%^<>#!.\\:])', Operator),
(r'[;,]', Punctuation),
(r'[\[\]\{\}\(\)]', Keyword.Type),
(r'[a-zA-Z_][a-zA-Z0-9_]*:', Name.Variable),
(r"(class|extends|if|then|super|do|with|import|export|"
r"while|elseif|return|for|in|from|when|using|else|"
r"and|or|not|switch|break)\b", Keyword),
(r'(true|false|nil)\b', Keyword.Constant),
(r'(and|or|not)\b', Operator.Word),
(r'(self)\b', Name.Builtin.Pseudo),
(r'@@?([a-zA-Z_][a-zA-Z0-9_]*)?', Name.Variable.Class),
(r'[A-Z]\w*', Name.Class), # proper name
(r'[A-Za-z_][A-Za-z0-9_]*(\.[A-Za-z_][A-Za-z0-9_]*)?', Name),
("'", String.Single, combined('stringescape', 'sqs')),
('"', String.Double, combined('stringescape', 'dqs'))
],
'stringescape': [
(r'''\\([abfnrtv\\"']|\d{1,3})''', String.Escape)
],
'sqs': [
("'", String.Single, '#pop'),
(".", String)
],
'dqs': [
('"', String.Double, '#pop'),
(".", String)
]
}
def get_tokens_unprocessed(self, text):
# set . as Operator instead of Punctuation
for index, token, value in \
LuaLexer.get_tokens_unprocessed(self, text):
if token == Punctuation and value == ".":
token = Operator
yield index, token, value
class CrocLexer(RegexLexer):
"""
For `Croc <http://jfbillingsley.com/croc>`_ source.
"""
name = 'Croc'
filenames = ['*.croc']
aliases = ['croc']
mimetypes = ['text/x-crocsrc']
tokens = {
'root': [
(r'\n', Text),
(r'\s+', Text),
# Comments
(r'//(.*?)\n', Comment.Single),
(r'/\*', Comment.Multiline, 'nestedcomment'),
# Keywords
(r'(as|assert|break|case|catch|class|continue|default'
r'|do|else|finally|for|foreach|function|global|namespace'
r'|if|import|in|is|local|module|return|scope|super|switch'
r'|this|throw|try|vararg|while|with|yield)\b', Keyword),
(r'(false|true|null)\b', Keyword.Constant),
# FloatLiteral
(r'([0-9][0-9_]*)(?=[.eE])(\.[0-9][0-9_]*)?([eE][+\-]?[0-9_]+)?',
Number.Float),
# IntegerLiteral
# -- Binary
(r'0[bB][01][01_]*', Number),
# -- Hexadecimal
(r'0[xX][0-9a-fA-F][0-9a-fA-F_]*', Number.Hex),
# -- Decimal
(r'([0-9][0-9_]*)(?![.eE])', Number.Integer),
# CharacterLiteral
(r"""'(\\['"\\nrt]|\\x[0-9a-fA-F]{2}|\\[0-9]{1,3}"""
r"""|\\u[0-9a-fA-F]{4}|\\U[0-9a-fA-F]{8}|.)'""",
String.Char
),
# StringLiteral
# -- WysiwygString
(r'@"(""|[^"])*"', String),
(r'@`(``|[^`])*`', String),
(r"@'(''|[^'])*'", String),
# -- DoubleQuotedString
(r'"(\\\\|\\"|[^"])*"', String),
# Tokens
(
r'(~=|\^=|%=|\*=|==|!=|>>>=|>>>|>>=|>>|>=|<=>|\?=|-\>'
r'|<<=|<<|<=|\+\+|\+=|--|-=|\|\||\|=|&&|&=|\.\.|/=)'
r'|[-/.&$@|\+<>!()\[\]{}?,;:=*%^~#\\]', Punctuation
),
# Identifier
(r'[a-zA-Z_]\w*', Name),
],
'nestedcomment': [
(r'[^*/]+', Comment.Multiline),
(r'/\*', Comment.Multiline, '#push'),
(r'\*/', Comment.Multiline, '#pop'),
(r'[*/]', Comment.Multiline),
],
}
class MiniDLexer(CrocLexer):
"""
For MiniD source. MiniD is now known as Croc.
"""
name = 'MiniD'
filenames = ['*.md']
aliases = ['minid']
mimetypes = ['text/x-minidsrc']
class IoLexer(RegexLexer):
"""
For `Io <http://iolanguage.com/>`_ (a small, prototype-based
programming language) source.
*New in Pygments 0.10.*
"""
name = 'Io'
filenames = ['*.io']
aliases = ['io']
mimetypes = ['text/x-iosrc']
tokens = {
'root': [
(r'\n', Text),
(r'\s+', Text),
# Comments
(r'//(.*?)\n', Comment.Single),
(r'#(.*?)\n', Comment.Single),
(r'/(\\\n)?[*](.|\n)*?[*](\\\n)?/', Comment.Multiline),
(r'/\+', Comment.Multiline, 'nestedcomment'),
# DoubleQuotedString
(r'"(\\\\|\\"|[^"])*"', String),
# Operators
(r'::=|:=|=|\(|\)|;|,|\*|-|\+|>|<|@|!|/|\||\^|\.|%|&|\[|\]|\{|\}',
Operator),
# keywords
(r'(clone|do|doFile|doString|method|for|if|else|elseif|then)\b',
Keyword),
# constants
(r'(nil|false|true)\b', Name.Constant),
# names
(r'(Object|list|List|Map|args|Sequence|Coroutine|File)\b',
Name.Builtin),
('[a-zA-Z_][a-zA-Z0-9_]*', Name),
# numbers
(r'(\d+\.?\d*|\d*\.\d+)([eE][+-]?[0-9]+)?', Number.Float),
(r'\d+', Number.Integer)
],
'nestedcomment': [
(r'[^+/]+', Comment.Multiline),
(r'/\+', Comment.Multiline, '#push'),
(r'\+/', Comment.Multiline, '#pop'),
(r'[+/]', Comment.Multiline),
]
}
class TclLexer(RegexLexer):
"""
For Tcl source code.
*New in Pygments 0.10.*
"""
keyword_cmds_re = (
r'\b(after|apply|array|break|catch|continue|elseif|else|error|'
r'eval|expr|for|foreach|global|if|namespace|proc|rename|return|'
r'set|switch|then|trace|unset|update|uplevel|upvar|variable|'
r'vwait|while)\b'
)
builtin_cmds_re = (
r'\b(append|bgerror|binary|cd|chan|clock|close|concat|dde|dict|'
r'encoding|eof|exec|exit|fblocked|fconfigure|fcopy|file|'
r'fileevent|flush|format|gets|glob|history|http|incr|info|interp|'
r'join|lappend|lassign|lindex|linsert|list|llength|load|loadTk|'
r'lrange|lrepeat|lreplace|lreverse|lsearch|lset|lsort|mathfunc|'
r'mathop|memory|msgcat|open|package|pid|pkg::create|pkg_mkIndex|'
r'platform|platform::shell|puts|pwd|re_syntax|read|refchan|'
r'regexp|registry|regsub|scan|seek|socket|source|split|string|'
r'subst|tell|time|tm|unknown|unload)\b'
)
name = 'Tcl'
aliases = ['tcl']
filenames = ['*.tcl']
mimetypes = ['text/x-tcl', 'text/x-script.tcl', 'application/x-tcl']
def _gen_command_rules(keyword_cmds_re, builtin_cmds_re, context=""):
return [
(keyword_cmds_re, Keyword, 'params' + context),
(builtin_cmds_re, Name.Builtin, 'params' + context),
(r'([\w\.\-]+)', Name.Variable, 'params' + context),
(r'#', Comment, 'comment'),
]
tokens = {
'root': [
include('command'),
include('basic'),
include('data'),
(r'}', Keyword), # HACK: somehow we miscounted our braces
],
'command': _gen_command_rules(keyword_cmds_re, builtin_cmds_re),
'command-in-brace': _gen_command_rules(keyword_cmds_re,
builtin_cmds_re,
"-in-brace"),
'command-in-bracket': _gen_command_rules(keyword_cmds_re,
builtin_cmds_re,
"-in-bracket"),
'command-in-paren': _gen_command_rules(keyword_cmds_re,
builtin_cmds_re,
"-in-paren"),
'basic': [
(r'\(', Keyword, 'paren'),
(r'\[', Keyword, 'bracket'),
(r'\{', Keyword, 'brace'),
(r'"', String.Double, 'string'),
(r'(eq|ne|in|ni)\b', Operator.Word),
(r'!=|==|<<|>>|<=|>=|&&|\|\||\*\*|[-+~!*/%<>&^|?:]', Operator),
],
'data': [
(r'\s+', Text),
(r'0x[a-fA-F0-9]+', Number.Hex),
(r'0[0-7]+', Number.Oct),
(r'\d+\.\d+', Number.Float),
(r'\d+', Number.Integer),
(r'\$([\w\.\-\:]+)', Name.Variable),
(r'([\w\.\-\:]+)', Text),
],
'params': [
(r';', Keyword, '#pop'),
(r'\n', Text, '#pop'),
(r'(else|elseif|then)\b', Keyword),
include('basic'),
include('data'),
],
'params-in-brace': [
(r'}', Keyword, ('#pop', '#pop')),
include('params')
],
'params-in-paren': [
(r'\)', Keyword, ('#pop', '#pop')),
include('params')
],
'params-in-bracket': [
(r'\]', Keyword, ('#pop', '#pop')),
include('params')
],
'string': [
(r'\[', String.Double, 'string-square'),
(r'(?s)(\\\\|\\[0-7]+|\\.|[^"\\])', String.Double),
(r'"', String.Double, '#pop')
],
'string-square': [
(r'\[', String.Double, 'string-square'),
(r'(?s)(\\\\|\\[0-7]+|\\.|\\\n|[^\]\\])', String.Double),
(r'\]', String.Double, '#pop')
],
'brace': [
(r'}', Keyword, '#pop'),
include('command-in-brace'),
include('basic'),
include('data'),
],
'paren': [
(r'\)', Keyword, '#pop'),
include('command-in-paren'),
include('basic'),
include('data'),
],
'bracket': [
(r'\]', Keyword, '#pop'),
include('command-in-bracket'),
include('basic'),
include('data'),
],
'comment': [
(r'.*[^\\]\n', Comment, '#pop'),
(r'.*\\\n', Comment),
],
}
def analyse_text(text):
return shebang_matches(text, r'(tcl)')
class FactorLexer(RegexLexer):
"""
Lexer for the `Factor <http://factorcode.org>`_ language.
*New in Pygments 1.4.*
"""
name = 'Factor'
aliases = ['factor']
filenames = ['*.factor']
mimetypes = ['text/x-factor']
flags = re.MULTILINE | re.UNICODE
builtin_kernel = (
r'(?:or|2bi|2tri|while|wrapper|nip|4dip|wrapper\\?|bi\\*|'
r'callstack>array|both\\?|hashcode|die|dupd|callstack|'
r'callstack\\?|3dup|tri@|pick|curry|build|\\?execute|3bi|'
r'prepose|>boolean|\\?if|clone|eq\\?|tri\\*|\\?|=|swapd|'
r'2over|2keep|3keep|clear|2dup|when|not|tuple\\?|dup|2bi\\*|'
r'2tri\\*|call|tri-curry|object|bi@|do|unless\\*|if\\*|loop|'
r'bi-curry\\*|drop|when\\*|assert=|retainstack|assert\\?|-rot|'
r'execute|2bi@|2tri@|boa|with|either\\?|3drop|bi|curry\\?|'
r'datastack|until|3dip|over|3curry|tri-curry\\*|tri-curry@|swap|'
r'and|2nip|throw|bi-curry|\\(clone\\)|hashcode\\*|compose|2dip|if|3tri|'
r'unless|compose\\?|tuple|keep|2curry|equal\\?|assert|tri|2drop|'
r'most|<wrapper>|boolean\\?|identity-hashcode|identity-tuple\\?|'
r'null|new|dip|bi-curry@|rot|xor|identity-tuple|boolean)\s'
)
builtin_assocs = (
r'(?:\\?at|assoc\\?|assoc-clone-like|assoc=|delete-at\\*|'
r'assoc-partition|extract-keys|new-assoc|value\\?|assoc-size|'
r'map>assoc|push-at|assoc-like|key\\?|assoc-intersect|'
r'assoc-refine|update|assoc-union|assoc-combine|at\\*|'
r'assoc-empty\\?|at\\+|set-at|assoc-all\\?|assoc-subset\\?|'
r'assoc-hashcode|change-at|assoc-each|assoc-diff|zip|values|'
r'value-at|rename-at|inc-at|enum\\?|at|cache|assoc>map|<enum>|'
r'assoc|assoc-map|enum|value-at\\*|assoc-map-as|>alist|'
r'assoc-filter-as|clear-assoc|assoc-stack|maybe-set-at|'
r'substitute|assoc-filter|2cache|delete-at|assoc-find|keys|'
r'assoc-any\\?|unzip)\s'
)
builtin_combinators = (
r'(?:case|execute-effect|no-cond|no-case\\?|3cleave>quot|2cleave|'
r'cond>quot|wrong-values\\?|no-cond\\?|cleave>quot|no-case|'
r'case>quot|3cleave|wrong-values|to-fixed-point|alist>quot|'
r'case-find|cond|cleave|call-effect|2cleave>quot|recursive-hashcode|'
r'linear-case-quot|spread|spread>quot)\s'
)
builtin_math = (
r'(?:number=|if-zero|next-power-of-2|each-integer|\\?1\\+|'
r'fp-special\\?|imaginary-part|unless-zero|float>bits|number\\?|'
r'fp-infinity\\?|bignum\\?|fp-snan\\?|denominator|fp-bitwise=|\\*|'
r'\\+|power-of-2\\?|-|u>=|/|>=|bitand|log2-expects-positive|<|'
r'log2|>|integer\\?|number|bits>double|2/|zero\\?|(find-integer)|'
r'bits>float|float\\?|shift|ratio\\?|even\\?|ratio|fp-sign|bitnot|'
r'>fixnum|complex\\?|/i|/f|byte-array>bignum|when-zero|sgn|>bignum|'
r'next-float|u<|u>|mod|recip|rational|find-last-integer|>float|'
r'(all-integers\\?)|2^|times|integer|fixnum\\?|neg|fixnum|sq|'
r'bignum|(each-integer)|bit\\?|fp-qnan\\?|find-integer|complex|'
r'<fp-nan>|real|double>bits|bitor|rem|fp-nan-payload|all-integers\\?|'
r'real-part|log2-expects-positive\\?|prev-float|align|unordered\\?|'
r'float|fp-nan\\?|abs|bitxor|u<=|odd\\?|<=|/mod|rational\\?|>integer|'
r'real\\?|numerator)\s'
)
builtin_sequences = (
r'(?:member-eq\\?|append|assert-sequence=|find-last-from|trim-head-slice|'
r'clone-like|3sequence|assert-sequence\\?|map-as|last-index-from|'
r'reversed|index-from|cut\\*|pad-tail|remove-eq!|concat-as|'
r'but-last|snip|trim-tail|nths|nth|2selector|sequence|slice\\?|'
r'<slice>|partition|remove-nth|tail-slice|empty\\?|tail\\*|'
r'if-empty|find-from|virtual-sequence\\?|member\\?|set-length|'
r'drop-prefix|unclip|unclip-last-slice|iota|map-sum|'
r'bounds-error\\?|sequence-hashcode-step|selector-for|'
r'accumulate-as|map|start|midpoint@|\\(accumulate\\)|rest-slice|'
r'prepend|fourth|sift|accumulate!|new-sequence|follow|map!|'
r'like|first4|1sequence|reverse|slice|unless-empty|padding|'
r'virtual@|repetition\\?|set-last|index|4sequence|max-length|'
r'set-second|immutable-sequence|first2|first3|replicate-as|'
r'reduce-index|unclip-slice|supremum|suffix!|insert-nth|'
r'trim-tail-slice|tail|3append|short|count|suffix|concat|'
r'flip|filter|sum|immutable\\?|reverse!|2sequence|map-integers|'
r'delete-all|start\\*|indices|snip-slice|check-slice|sequence\\?|'
r'head|map-find|filter!|append-as|reduce|sequence=|halves|'
r'collapse-slice|interleave|2map|filter-as|binary-reduce|'
r'slice-error\\?|product|bounds-check\\?|bounds-check|harvest|'
r'immutable|virtual-exemplar|find|produce|remove|pad-head|last|'
r'replicate|set-fourth|remove-eq|shorten|reversed\\?|'
r'map-find-last|3map-as|2unclip-slice|shorter\\?|3map|find-last|'
r'head-slice|pop\\*|2map-as|tail-slice\\*|but-last-slice|'
r'2map-reduce|iota\\?|collector-for|accumulate|each|selector|'
r'append!|new-resizable|cut-slice|each-index|head-slice\\*|'
r'2reverse-each|sequence-hashcode|pop|set-nth|\\?nth|'
r'<flat-slice>|second|join|when-empty|collector|'
r'immutable-sequence\\?|<reversed>|all\\?|3append-as|'
r'virtual-sequence|subseq\\?|remove-nth!|push-either|new-like|'
r'length|last-index|push-if|2all\\?|lengthen|assert-sequence|'
r'copy|map-reduce|move|third|first|3each|tail\\?|set-first|'
r'prefix|bounds-error|any\\?|<repetition>|trim-slice|exchange|'
r'surround|2reduce|cut|change-nth|min-length|set-third|produce-as|'
r'push-all|head\\?|delete-slice|rest|sum-lengths|2each|head\\*|'
r'infimum|remove!|glue|slice-error|subseq|trim|replace-slice|'
r'push|repetition|map-index|trim-head|unclip-last|mismatch)\s'
)
builtin_namespaces = (
r'(?:global|\\+@|change|set-namestack|change-global|init-namespaces|'
r'on|off|set-global|namespace|set|with-scope|bind|with-variable|'
r'inc|dec|counter|initialize|namestack|get|get-global|make-assoc)\s'
)
builtin_arrays = (
r'(?:<array>|2array|3array|pair|>array|1array|4array|pair\\?|'
r'array|resize-array|array\\?)\s'
)
builtin_io = (
r'(?:\\+character\\+|bad-seek-type\\?|readln|each-morsel|stream-seek|'
r'read|print|with-output-stream|contents|write1|stream-write1|'
r'stream-copy|stream-element-type|with-input-stream|'
r'stream-print|stream-read|stream-contents|stream-tell|'
r'tell-output|bl|seek-output|bad-seek-type|nl|stream-nl|write|'
r'flush|stream-lines|\\+byte\\+|stream-flush|read1|'
r'seek-absolute\\?|stream-read1|lines|stream-readln|'
r'stream-read-until|each-line|seek-end|with-output-stream\\*|'
r'seek-absolute|with-streams|seek-input|seek-relative\\?|'
r'input-stream|stream-write|read-partial|seek-end\\?|'
r'seek-relative|error-stream|read-until|with-input-stream\\*|'
r'with-streams\\*|tell-input|each-block|output-stream|'
r'stream-read-partial|each-stream-block|each-stream-line)\s'
)
builtin_strings = (
r'(?:resize-string|>string|<string>|1string|string|string\\?)\s'
)
builtin_vectors = (
r'(?:vector\\?|<vector>|\\?push|vector|>vector|1vector)\s'
)
builtin_continuations = (
r'(?:with-return|restarts|return-continuation|with-datastack|'
r'recover|rethrow-restarts|<restart>|ifcc|set-catchstack|'
r'>continuation<|cleanup|ignore-errors|restart\\?|'
r'compute-restarts|attempt-all-error|error-thread|continue|'
r'<continuation>|attempt-all-error\\?|condition\\?|'
r'<condition>|throw-restarts|error|catchstack|continue-with|'
r'thread-error-hook|continuation|rethrow|callcc1|'
r'error-continuation|callcc0|attempt-all|condition|'
r'continuation\\?|restart|return)\s'
)
tokens = {
'root': [
# TODO: (( inputs -- outputs ))
# TODO: << ... >>
# defining words
(r'(\s*)(:|::|MACRO:|MEMO:)(\s+)(\S+)',
bygroups(Text, Keyword, Text, Name.Function)),
(r'(\s*)(M:)(\s+)(\S+)(\s+)(\S+)',
bygroups(Text, Keyword, Text, Name.Class, Text, Name.Function)),
(r'(\s*)(GENERIC:)(\s+)(\S+)',
bygroups(Text, Keyword, Text, Name.Function)),
(r'(\s*)(HOOK:|GENERIC#)(\s+)(\S+)(\s+)(\S+)',
bygroups(Text, Keyword, Text, Name.Function, Text, Name.Function)),
(r'(\()(\s+)', bygroups(Name.Function, Text), 'stackeffect'),
(r'\;\s', Keyword),
# imports and namespaces
(r'(USING:)((?:\s|\\\s)+)',
bygroups(Keyword.Namespace, Text), 'import'),
(r'(USE:)(\s+)(\S+)',
bygroups(Keyword.Namespace, Text, Name.Namespace)),
(r'(UNUSE:)(\s+)(\S+)',
bygroups(Keyword.Namespace, Text, Name.Namespace)),
(r'(QUALIFIED:)(\s+)(\S+)',
bygroups(Keyword.Namespace, Text, Name.Namespace)),
(r'(QUALIFIED-WITH:)(\s+)(\S+)',
bygroups(Keyword.Namespace, Text, Name.Namespace)),
(r'(FROM:|EXCLUDE:)(\s+)(\S+)(\s+)(=>)',
bygroups(Keyword.Namespace, Text, Name.Namespace, Text, Text)),
(r'(IN:)(\s+)(\S+)',
bygroups(Keyword.Namespace, Text, Name.Namespace)),
(r'(?:ALIAS|DEFER|FORGET|POSTPONE):', Keyword.Namespace),
# tuples and classes
(r'(TUPLE:)(\s+)(\S+)(\s+<\s+)(\S+)',
bygroups(Keyword, Text, Name.Class, Text, Name.Class), 'slots'),
(r'(TUPLE:)(\s+)(\S+)',
bygroups(Keyword, Text, Name.Class), 'slots'),
(r'(UNION:)(\s+)(\S+)', bygroups(Keyword, Text, Name.Class)),
(r'(INTERSECTION:)(\s+)(\S+)', bygroups(Keyword, Text, Name.Class)),
(r'(PREDICATE:)(\s+)(\S+)(\s+<\s+)(\S+)',
bygroups(Keyword, Text, Name.Class, Text, Name.Class)),
(r'(C:)(\s+)(\S+)(\s+)(\S+)',
bygroups(Keyword, Text, Name.Function, Text, Name.Class)),
(r'INSTANCE:', Keyword),
(r'SLOT:', Keyword),
(r'MIXIN:', Keyword),
(r'(?:SINGLETON|SINGLETONS):', Keyword),
# other syntax
(r'CONSTANT:', Keyword),
(r'(?:SYMBOL|SYMBOLS):', Keyword),
(r'ERROR:', Keyword),
(r'SYNTAX:', Keyword),
(r'(HELP:)(\s+)(\S+)', bygroups(Keyword, Text, Name.Function)),
(r'(MAIN:)(\s+)(\S+)',
bygroups(Keyword.Namespace, Text, Name.Function)),
(r'(?:ALIEN|TYPEDEF|FUNCTION|STRUCT):', Keyword),
# vocab.private
# TODO: words inside vocab.private should have red names?
(r'(?:<PRIVATE|PRIVATE>)', Keyword.Namespace),
# strings
(r'"""\s+(?:.|\n)*?\s+"""', String),
(r'"(?:\\\\|\\"|[^"])*"', String),
(r'CHAR:\s+(\\[\\abfnrstv]*|\S)\s', String.Char),
# comments
(r'\!\s+.*$', Comment),
(r'#\!\s+.*$', Comment),
# boolean constants
(r'(t|f)\s', Name.Constant),
# numbers
(r'-?\d+\.\d+\s', Number.Float),
(r'-?\d+\s', Number.Integer),
(r'HEX:\s+[a-fA-F\d]+\s', Number.Hex),
(r'BIN:\s+[01]+\s', Number.Integer),
(r'OCT:\s+[0-7]+\s', Number.Oct),
# operators
(r'[-+/*=<>^]\s', Operator),
# keywords
(r'(?:deprecated|final|foldable|flushable|inline|recursive)\s',
Keyword),
# builtins
(builtin_kernel, Name.Builtin),
(builtin_assocs, Name.Builtin),
(builtin_combinators, Name.Builtin),
(builtin_math, Name.Builtin),
(builtin_sequences, Name.Builtin),
(builtin_namespaces, Name.Builtin),
(builtin_arrays, Name.Builtin),
(builtin_io, Name.Builtin),
(builtin_strings, Name.Builtin),
(builtin_vectors, Name.Builtin),
(builtin_continuations, Name.Builtin),
# whitespaces - usually not relevant
(r'\s+', Text),
# everything else is text
(r'\S+', Text),
],
'stackeffect': [
(r'\s*\(', Name.Function, 'stackeffect'),
(r'\)', Name.Function, '#pop'),
(r'\-\-', Name.Function),
(r'\s+', Text),
(r'\S+', Name.Variable),
],
'slots': [
(r'\s+', Text),
(r';\s', Keyword, '#pop'),
(r'\S+', Name.Variable),
],
'import': [
(r';', Keyword, '#pop'),
(r'\S+', Name.Namespace),
(r'\s+', Text),
],
}
class FancyLexer(RegexLexer):
"""
Pygments Lexer For `Fancy <http://www.fancy-lang.org/>`_.
Fancy is a self-hosted, pure object-oriented, dynamic,
class-based, concurrent general-purpose programming language
running on Rubinius, the Ruby VM.
*New in Pygments 1.5.*
"""
name = 'Fancy'
filenames = ['*.fy', '*.fancypack']
aliases = ['fancy', 'fy']
mimetypes = ['text/x-fancysrc']
tokens = {
# copied from PerlLexer:
'balanced-regex': [
(r'/(\\\\|\\/|[^/])*/[egimosx]*', String.Regex, '#pop'),
(r'!(\\\\|\\!|[^!])*![egimosx]*', String.Regex, '#pop'),
(r'\\(\\\\|[^\\])*\\[egimosx]*', String.Regex, '#pop'),
(r'{(\\\\|\\}|[^}])*}[egimosx]*', String.Regex, '#pop'),
(r'<(\\\\|\\>|[^>])*>[egimosx]*', String.Regex, '#pop'),
(r'\[(\\\\|\\\]|[^\]])*\][egimosx]*', String.Regex, '#pop'),
(r'\((\\\\|\\\)|[^\)])*\)[egimosx]*', String.Regex, '#pop'),
(r'@(\\\\|\\\@|[^\@])*@[egimosx]*', String.Regex, '#pop'),
(r'%(\\\\|\\\%|[^\%])*%[egimosx]*', String.Regex, '#pop'),
(r'\$(\\\\|\\\$|[^\$])*\$[egimosx]*', String.Regex, '#pop'),
],
'root': [
(r'\s+', Text),
# balanced delimiters (copied from PerlLexer):
(r's{(\\\\|\\}|[^}])*}\s*', String.Regex, 'balanced-regex'),
(r's<(\\\\|\\>|[^>])*>\s*', String.Regex, 'balanced-regex'),
(r's\[(\\\\|\\\]|[^\]])*\]\s*', String.Regex, 'balanced-regex'),
(r's\((\\\\|\\\)|[^\)])*\)\s*', String.Regex, 'balanced-regex'),
(r'm?/(\\\\|\\/|[^/\n])*/[gcimosx]*', String.Regex),
(r'm(?=[/!\\{<\[\(@%\$])', String.Regex, 'balanced-regex'),
# Comments
(r'#(.*?)\n', Comment.Single),
# Symbols
(r'\'([^\'\s\[\]\(\)\{\}]+|\[\])', String.Symbol),
# Multi-line DoubleQuotedString
(r'"""(\\\\|\\"|[^"])*"""', String),
# DoubleQuotedString
(r'"(\\\\|\\"|[^"])*"', String),
# keywords
(r'(def|class|try|catch|finally|retry|return|return_local|match|'
r'case|->|=>)\b', Keyword),
# constants
(r'(self|super|nil|false|true)\b', Name.Constant),
(r'[(){};,/?\|:\\]', Punctuation),
# names
(r'(Object|Array|Hash|Directory|File|Class|String|Number|'
r'Enumerable|FancyEnumerable|Block|TrueClass|NilClass|'
r'FalseClass|Tuple|Symbol|Stack|Set|FancySpec|Method|Package|'
r'Range)\b', Name.Builtin),
# functions
(r'[a-zA-Z]([a-zA-Z0-9_]|[-+?!=*/^><%])*:', Name.Function),
# operators, must be below functions
(r'[-+*/~,<>=&!?%^\[\]\.$]+', Operator),
('[A-Z][a-zA-Z0-9_]*', Name.Constant),
('@[a-zA-Z_][a-zA-Z0-9_]*', Name.Variable.Instance),
('@@[a-zA-Z_][a-zA-Z0-9_]*', Name.Variable.Class),
('@@?', Operator),
('[a-zA-Z_][a-zA-Z0-9_]*', Name),
# numbers - / checks are necessary to avoid mismarking regexes,
# see comment in RubyLexer
(r'(0[oO]?[0-7]+(?:_[0-7]+)*)(\s*)([/?])?',
bygroups(Number.Oct, Text, Operator)),
(r'(0[xX][0-9A-Fa-f]+(?:_[0-9A-Fa-f]+)*)(\s*)([/?])?',
bygroups(Number.Hex, Text, Operator)),
(r'(0[bB][01]+(?:_[01]+)*)(\s*)([/?])?',
bygroups(Number.Bin, Text, Operator)),
(r'([\d]+(?:_\d+)*)(\s*)([/?])?',
bygroups(Number.Integer, Text, Operator)),
(r'\d+([eE][+-]?[0-9]+)|\d+\.\d+([eE][+-]?[0-9]+)?', Number.Float),
(r'\d+', Number.Integer)
]
}
class DgLexer(RegexLexer):
"""
Lexer for `dg <http://pyos.github.com/dg>`_,
a functional and object-oriented programming language
running on the CPython 3 VM.
*New in Pygments 1.6.*
"""
name = 'dg'
aliases = ['dg']
filenames = ['*.dg']
mimetypes = ['text/x-dg']
tokens = {
'root': [
# Whitespace:
(r'\s+', Text),
(r'#.*?$', Comment.Single),
# Lexemes:
# Numbers
(r'0[bB][01]+', Number.Bin),
(r'0[oO][0-7]+', Number.Oct),
(r'0[xX][\da-fA-F]+', Number.Hex),
(r'[+-]?\d+\.\d+([eE][+-]?\d+)?[jJ]?', Number.Float),
(r'[+-]?\d+[eE][+-]?\d+[jJ]?', Number.Float),
(r'[+-]?\d+[jJ]?', Number.Integer),
# Character/String Literals
(r"[br]*'''", String, combined('stringescape', 'tsqs', 'string')),
(r'[br]*"""', String, combined('stringescape', 'tdqs', 'string')),
(r"[br]*'", String, combined('stringescape', 'sqs', 'string')),
(r'[br]*"', String, combined('stringescape', 'dqs', 'string')),
# Operators
(r"`\w+'*`", Operator), # Infix links
# Reserved infix links
(r'\b(or|and|if|else|where|is|in)\b', Operator.Word),
(r'[!$%&*+\-./:<-@\\^|~;,]+', Operator),
# Identifiers
# Python 3 types
(r"(?<!\.)(bool|bytearray|bytes|classmethod|complex|dict'?|"
r"float|frozenset|int|list'?|memoryview|object|property|range|"
r"set'?|slice|staticmethod|str|super|tuple'?|type)"
r"(?!['\w])", Name.Builtin),
# Python 3 builtins + some more
(r'(?<!\.)(__import__|abs|all|any|bin|bind|chr|cmp|compile|complex|'
r'delattr|dir|divmod|drop|dropwhile|enumerate|eval|filter|flip|'
r'foldl1?|format|fst|getattr|globals|hasattr|hash|head|hex|id|'
r'init|input|isinstance|issubclass|iter|iterate|last|len|locals|'
r'map|max|min|next|oct|open|ord|pow|print|repr|reversed|round|'
r'setattr|scanl1?|snd|sorted|sum|tail|take|takewhile|vars|zip)'
r"(?!['\w])", Name.Builtin),
(r"(?<!\.)(self|Ellipsis|NotImplemented|None|True|False)(?!['\w])",
Name.Builtin.Pseudo),
(r"(?<!\.)[A-Z]\w*(Error|Exception|Warning)'*(?!['\w])",
Name.Exception),
(r"(?<!\.)(KeyboardInterrupt|SystemExit|StopIteration|"
r"GeneratorExit)(?!['\w])", Name.Exception),
# Compiler-defined identifiers
(r"(?<![\.\w])(import|inherit|for|while|switch|not|raise|unsafe|"
r"yield|with)(?!['\w])", Keyword.Reserved),
# Other links
(r"[A-Z_']+\b", Name),
(r"[A-Z][\w']*\b", Keyword.Type),
(r"\w+'*", Name),
# Blocks
(r'[()]', Punctuation),
],
'stringescape': [
(r'\\([\\abfnrtv"\']|\n|N{.*?}|u[a-fA-F0-9]{4}|'
r'U[a-fA-F0-9]{8}|x[a-fA-F0-9]{2}|[0-7]{1,3})', String.Escape)
],
'string': [
(r'%(\([a-zA-Z0-9_]+\))?[-#0 +]*([0-9]+|[*])?(\.([0-9]+|[*]))?'
'[hlL]?[diouxXeEfFgGcrs%]', String.Interpol),
(r'[^\\\'"%\n]+', String),
# quotes, percents and backslashes must be parsed one at a time
(r'[\'"\\]', String),
# unhandled string formatting sign
(r'%', String),
(r'\n', String)
],
'dqs': [
(r'"', String, '#pop')
],
'sqs': [
(r"'", String, '#pop')
],
'tdqs': [
(r'"""', String, '#pop')
],
'tsqs': [
(r"'''", String, '#pop')
],
}
class Perl6Lexer(ExtendedRegexLexer):
"""
For `Perl 6 <http://www.perl6.org>`_ source code.
*New in Pygments 1.7.*
"""
name = 'Perl6'
aliases = ['perl6', 'pl6']
filenames = ['*.pl', '*.pm', '*.nqp', '*.p6', '*.6pl', '*.p6l', '*.pl6',
'*.6pm', '*.p6m', '*.pm6']
mimetypes = ['text/x-perl6', 'application/x-perl6']
flags = re.MULTILINE | re.DOTALL | re.UNICODE
PERL6_IDENTIFIER_RANGE = "['a-zA-Z0-9_:-]"
PERL6_KEYWORDS = (
'BEGIN', 'CATCH', 'CHECK', 'CONTROL', 'END', 'ENTER', 'FIRST', 'INIT',
'KEEP', 'LAST', 'LEAVE', 'NEXT', 'POST', 'PRE', 'START', 'TEMP',
'UNDO', 'as', 'assoc', 'async', 'augment', 'binary', 'break', 'but',
'cached', 'category', 'class', 'constant', 'contend', 'continue',
'copy', 'deep', 'default', 'defequiv', 'defer', 'die', 'do', 'else',
'elsif', 'enum', 'equiv', 'exit', 'export', 'fail', 'fatal', 'for',
'gather', 'given', 'goto', 'grammar', 'handles', 'has', 'if', 'inline',
'irs', 'is', 'last', 'leave', 'let', 'lift', 'loop', 'looser', 'macro',
'make', 'maybe', 'method', 'module', 'multi', 'my', 'next', 'of',
'ofs', 'only', 'oo', 'ors', 'our', 'package', 'parsed', 'prec',
'proto', 'readonly', 'redo', 'ref', 'regex', 'reparsed', 'repeat',
'require', 'required', 'return', 'returns', 'role', 'rule', 'rw',
'self', 'slang', 'state', 'sub', 'submethod', 'subset', 'supersede',
'take', 'temp', 'tighter', 'token', 'trusts', 'try', 'unary',
'unless', 'until', 'use', 'warn', 'when', 'where', 'while', 'will',
)
PERL6_BUILTINS = (
'ACCEPTS', 'HOW', 'REJECTS', 'VAR', 'WHAT', 'WHENCE', 'WHERE', 'WHICH',
'WHO', 'abs', 'acos', 'acosec', 'acosech', 'acosh', 'acotan', 'acotanh',
'all', 'any', 'approx', 'arity', 'asec', 'asech', 'asin', 'asinh'
'assuming', 'atan', 'atan2', 'atanh', 'attr', 'bless', 'body', 'by'
'bytes', 'caller', 'callsame', 'callwith', 'can', 'capitalize', 'cat',
'ceiling', 'chars', 'chmod', 'chomp', 'chop', 'chr', 'chroot',
'circumfix', 'cis', 'classify', 'clone', 'close', 'cmp_ok', 'codes',
'comb', 'connect', 'contains', 'context', 'cos', 'cosec', 'cosech',
'cosh', 'cotan', 'cotanh', 'count', 'defined', 'delete', 'diag',
'dies_ok', 'does', 'e', 'each', 'eager', 'elems', 'end', 'eof', 'eval',
'eval_dies_ok', 'eval_elsewhere', 'eval_lives_ok', 'evalfile', 'exists',
'exp', 'first', 'flip', 'floor', 'flunk', 'flush', 'fmt', 'force_todo',
'fork', 'from', 'getc', 'gethost', 'getlogin', 'getpeername', 'getpw',
'gmtime', 'graphs', 'grep', 'hints', 'hyper', 'im', 'index', 'infix',
'invert', 'is_approx', 'is_deeply', 'isa', 'isa_ok', 'isnt', 'iterator',
'join', 'key', 'keys', 'kill', 'kv', 'lastcall', 'lazy', 'lc', 'lcfirst',
'like', 'lines', 'link', 'lives_ok', 'localtime', 'log', 'log10', 'map',
'max', 'min', 'minmax', 'name', 'new', 'nextsame', 'nextwith', 'nfc',
'nfd', 'nfkc', 'nfkd', 'nok_error', 'nonce', 'none', 'normalize', 'not',
'nothing', 'ok', 'once', 'one', 'open', 'opendir', 'operator', 'ord',
'p5chomp', 'p5chop', 'pack', 'pair', 'pairs', 'pass', 'perl', 'pi',
'pick', 'plan', 'plan_ok', 'polar', 'pop', 'pos', 'postcircumfix',
'postfix', 'pred', 'prefix', 'print', 'printf', 'push', 'quasi',
'quotemeta', 'rand', 're', 'read', 'readdir', 'readline', 'reduce',
'reverse', 'rewind', 'rewinddir', 'rindex', 'roots', 'round',
'roundrobin', 'run', 'runinstead', 'sameaccent', 'samecase', 'say',
'sec', 'sech', 'sech', 'seek', 'shape', 'shift', 'sign', 'signature',
'sin', 'sinh', 'skip', 'skip_rest', 'sleep', 'slurp', 'sort', 'splice',
'split', 'sprintf', 'sqrt', 'srand', 'strand', 'subst', 'substr', 'succ',
'sum', 'symlink', 'tan', 'tanh', 'throws_ok', 'time', 'times', 'to',
'todo', 'trim', 'trim_end', 'trim_start', 'true', 'truncate', 'uc',
'ucfirst', 'undef', 'undefine', 'uniq', 'unlike', 'unlink', 'unpack',
'unpolar', 'unshift', 'unwrap', 'use_ok', 'value', 'values', 'vec',
'version_lt', 'void', 'wait', 'want', 'wrap', 'write', 'zip',
)
PERL6_BUILTIN_CLASSES = (
'Abstraction', 'Any', 'AnyChar', 'Array', 'Associative', 'Bag', 'Bit',
'Blob', 'Block', 'Bool', 'Buf', 'Byte', 'Callable', 'Capture', 'Char', 'Class',
'Code', 'Codepoint', 'Comparator', 'Complex', 'Decreasing', 'Exception',
'Failure', 'False', 'Grammar', 'Grapheme', 'Hash', 'IO', 'Increasing',
'Int', 'Junction', 'KeyBag', 'KeyExtractor', 'KeyHash', 'KeySet',
'KitchenSink', 'List', 'Macro', 'Mapping', 'Match', 'Matcher', 'Method',
'Module', 'Num', 'Object', 'Ordered', 'Ordering', 'OrderingPair',
'Package', 'Pair', 'Positional', 'Proxy', 'Range', 'Rat', 'Regex',
'Role', 'Routine', 'Scalar', 'Seq', 'Set', 'Signature', 'Str', 'StrLen',
'StrPos', 'Sub', 'Submethod', 'True', 'UInt', 'Undef', 'Version', 'Void',
'Whatever', 'bit', 'bool', 'buf', 'buf1', 'buf16', 'buf2', 'buf32',
'buf4', 'buf64', 'buf8', 'complex', 'int', 'int1', 'int16', 'int2',
'int32', 'int4', 'int64', 'int8', 'num', 'rat', 'rat1', 'rat16', 'rat2',
'rat32', 'rat4', 'rat64', 'rat8', 'uint', 'uint1', 'uint16', 'uint2',
'uint32', 'uint4', 'uint64', 'uint8', 'utf16', 'utf32', 'utf8',
)
PERL6_OPERATORS = (
'X', 'Z', 'after', 'also', 'and', 'andthen', 'before', 'cmp', 'div',
'eq', 'eqv', 'extra', 'ff', 'fff', 'ge', 'gt', 'le', 'leg', 'lt', 'm',
'mm', 'mod', 'ne', 'or', 'orelse', 'rx', 's', 'tr', 'x', 'xor', 'xx',
'++', '--', '**', '!', '+', '-', '~', '?', '|', '||', '+^', '~^', '?^',
'^', '*', '/', '%', '%%', '+&', '+<', '+>', '~&', '~<', '~>', '?&',
'gcd', 'lcm', '+', '-', '+|', '+^', '~|', '~^', '?|', '?^',
'~', '&', '^', 'but', 'does', '<=>', '..', '..^', '^..', '^..^',
'!=', '==', '<', '<=', '>', '>=', '~~', '===', '!eqv',
'&&', '||', '^^', '//', 'min', 'max', '??', '!!', 'ff', 'fff', 'so',
'not', '<==', '==>', '<<==', '==>>',
)
# Perl 6 has a *lot* of possible bracketing characters
# this list was lifted from STD.pm6 (https://github.com/perl6/std)
PERL6_BRACKETS = {
u'\u0028' : u'\u0029', u'\u003c' : u'\u003e', u'\u005b' : u'\u005d', u'\u007b' : u'\u007d',
u'\u00ab' : u'\u00bb', u'\u0f3a' : u'\u0f3b', u'\u0f3c' : u'\u0f3d', u'\u169b' : u'\u169c',
u'\u2018' : u'\u2019', u'\u201a' : u'\u2019', u'\u201b' : u'\u2019', u'\u201c' : u'\u201d',
u'\u201e' : u'\u201d', u'\u201f' : u'\u201d', u'\u2039' : u'\u203a', u'\u2045' : u'\u2046',
u'\u207d' : u'\u207e', u'\u208d' : u'\u208e', u'\u2208' : u'\u220b', u'\u2209' : u'\u220c',
u'\u220a' : u'\u220d', u'\u2215' : u'\u29f5', u'\u223c' : u'\u223d', u'\u2243' : u'\u22cd',
u'\u2252' : u'\u2253', u'\u2254' : u'\u2255', u'\u2264' : u'\u2265', u'\u2266' : u'\u2267',
u'\u2268' : u'\u2269', u'\u226a' : u'\u226b', u'\u226e' : u'\u226f', u'\u2270' : u'\u2271',
u'\u2272' : u'\u2273', u'\u2274' : u'\u2275', u'\u2276' : u'\u2277', u'\u2278' : u'\u2279',
u'\u227a' : u'\u227b', u'\u227c' : u'\u227d', u'\u227e' : u'\u227f', u'\u2280' : u'\u2281',
u'\u2282' : u'\u2283', u'\u2284' : u'\u2285', u'\u2286' : u'\u2287', u'\u2288' : u'\u2289',
u'\u228a' : u'\u228b', u'\u228f' : u'\u2290', u'\u2291' : u'\u2292', u'\u2298' : u'\u29b8',
u'\u22a2' : u'\u22a3', u'\u22a6' : u'\u2ade', u'\u22a8' : u'\u2ae4', u'\u22a9' : u'\u2ae3',
u'\u22ab' : u'\u2ae5', u'\u22b0' : u'\u22b1', u'\u22b2' : u'\u22b3', u'\u22b4' : u'\u22b5',
u'\u22b6' : u'\u22b7', u'\u22c9' : u'\u22ca', u'\u22cb' : u'\u22cc', u'\u22d0' : u'\u22d1',
u'\u22d6' : u'\u22d7', u'\u22d8' : u'\u22d9', u'\u22da' : u'\u22db', u'\u22dc' : u'\u22dd',
u'\u22de' : u'\u22df', u'\u22e0' : u'\u22e1', u'\u22e2' : u'\u22e3', u'\u22e4' : u'\u22e5',
u'\u22e6' : u'\u22e7', u'\u22e8' : u'\u22e9', u'\u22ea' : u'\u22eb', u'\u22ec' : u'\u22ed',
u'\u22f0' : u'\u22f1', u'\u22f2' : u'\u22fa', u'\u22f3' : u'\u22fb', u'\u22f4' : u'\u22fc',
u'\u22f6' : u'\u22fd', u'\u22f7' : u'\u22fe', u'\u2308' : u'\u2309', u'\u230a' : u'\u230b',
u'\u2329' : u'\u232a', u'\u23b4' : u'\u23b5', u'\u2768' : u'\u2769', u'\u276a' : u'\u276b',
u'\u276c' : u'\u276d', u'\u276e' : u'\u276f', u'\u2770' : u'\u2771', u'\u2772' : u'\u2773',
u'\u2774' : u'\u2775', u'\u27c3' : u'\u27c4', u'\u27c5' : u'\u27c6', u'\u27d5' : u'\u27d6',
u'\u27dd' : u'\u27de', u'\u27e2' : u'\u27e3', u'\u27e4' : u'\u27e5', u'\u27e6' : u'\u27e7',
u'\u27e8' : u'\u27e9', u'\u27ea' : u'\u27eb', u'\u2983' : u'\u2984', u'\u2985' : u'\u2986',
u'\u2987' : u'\u2988', u'\u2989' : u'\u298a', u'\u298b' : u'\u298c', u'\u298d' : u'\u298e',
u'\u298f' : u'\u2990', u'\u2991' : u'\u2992', u'\u2993' : u'\u2994', u'\u2995' : u'\u2996',
u'\u2997' : u'\u2998', u'\u29c0' : u'\u29c1', u'\u29c4' : u'\u29c5', u'\u29cf' : u'\u29d0',
u'\u29d1' : u'\u29d2', u'\u29d4' : u'\u29d5', u'\u29d8' : u'\u29d9', u'\u29da' : u'\u29db',
u'\u29f8' : u'\u29f9', u'\u29fc' : u'\u29fd', u'\u2a2b' : u'\u2a2c', u'\u2a2d' : u'\u2a2e',
u'\u2a34' : u'\u2a35', u'\u2a3c' : u'\u2a3d', u'\u2a64' : u'\u2a65', u'\u2a79' : u'\u2a7a',
u'\u2a7d' : u'\u2a7e', u'\u2a7f' : u'\u2a80', u'\u2a81' : u'\u2a82', u'\u2a83' : u'\u2a84',
u'\u2a8b' : u'\u2a8c', u'\u2a91' : u'\u2a92', u'\u2a93' : u'\u2a94', u'\u2a95' : u'\u2a96',
u'\u2a97' : u'\u2a98', u'\u2a99' : u'\u2a9a', u'\u2a9b' : u'\u2a9c', u'\u2aa1' : u'\u2aa2',
u'\u2aa6' : u'\u2aa7', u'\u2aa8' : u'\u2aa9', u'\u2aaa' : u'\u2aab', u'\u2aac' : u'\u2aad',
u'\u2aaf' : u'\u2ab0', u'\u2ab3' : u'\u2ab4', u'\u2abb' : u'\u2abc', u'\u2abd' : u'\u2abe',
u'\u2abf' : u'\u2ac0', u'\u2ac1' : u'\u2ac2', u'\u2ac3' : u'\u2ac4', u'\u2ac5' : u'\u2ac6',
u'\u2acd' : u'\u2ace', u'\u2acf' : u'\u2ad0', u'\u2ad1' : u'\u2ad2', u'\u2ad3' : u'\u2ad4',
u'\u2ad5' : u'\u2ad6', u'\u2aec' : u'\u2aed', u'\u2af7' : u'\u2af8', u'\u2af9' : u'\u2afa',
u'\u2e02' : u'\u2e03', u'\u2e04' : u'\u2e05', u'\u2e09' : u'\u2e0a', u'\u2e0c' : u'\u2e0d',
u'\u2e1c' : u'\u2e1d', u'\u2e20' : u'\u2e21', u'\u3008' : u'\u3009', u'\u300a' : u'\u300b',
u'\u300c' : u'\u300d', u'\u300e' : u'\u300f', u'\u3010' : u'\u3011', u'\u3014' : u'\u3015',
u'\u3016' : u'\u3017', u'\u3018' : u'\u3019', u'\u301a' : u'\u301b', u'\u301d' : u'\u301e',
u'\ufd3e' : u'\ufd3f', u'\ufe17' : u'\ufe18', u'\ufe35' : u'\ufe36', u'\ufe37' : u'\ufe38',
u'\ufe39' : u'\ufe3a', u'\ufe3b' : u'\ufe3c', u'\ufe3d' : u'\ufe3e', u'\ufe3f' : u'\ufe40',
u'\ufe41' : u'\ufe42', u'\ufe43' : u'\ufe44', u'\ufe47' : u'\ufe48', u'\ufe59' : u'\ufe5a',
u'\ufe5b' : u'\ufe5c', u'\ufe5d' : u'\ufe5e', u'\uff08' : u'\uff09', u'\uff1c' : u'\uff1e',
u'\uff3b' : u'\uff3d', u'\uff5b' : u'\uff5d', u'\uff5f' : u'\uff60', u'\uff62' : u'\uff63',
}
def _build_word_match(words, boundary_regex_fragment = None, prefix = '', suffix = ''):
if boundary_regex_fragment is None:
return r'\b(' + prefix + r'|'.join([ re.escape(x) for x in words]) + suffix + r')\b'
else:
return r'(?<!' + boundary_regex_fragment + ')' + prefix + '(' + \
r'|'.join([ re.escape(x) for x in words]) + r')' + suffix + '(?!' + boundary_regex_fragment + ')'
def brackets_callback(token_class):
def callback(lexer, match, context):
groups = match.groupdict()
opening_chars = groups['delimiter']
n_chars = len(opening_chars)
adverbs = groups.get('adverbs')
closer = Perl6Lexer.PERL6_BRACKETS.get(opening_chars[0])
text = context.text
if closer is None: # it's not a mirrored character, which means we
# just need to look for the next occurrence
end_pos = text.find(opening_chars, match.start('delimiter') + n_chars)
else: # we need to look for the corresponding closing character,
# keep nesting in mind
closing_chars = closer * n_chars
nesting_level = 1
search_pos = match.start('delimiter')
while nesting_level > 0:
next_open_pos = text.find(opening_chars, search_pos + n_chars)
next_close_pos = text.find(closing_chars, search_pos + n_chars)
if next_close_pos == -1:
next_close_pos = len(text)
nesting_level = 0
elif next_open_pos != -1 and next_open_pos < next_close_pos:
nesting_level += 1
search_pos = next_open_pos
else: # next_close_pos < next_open_pos
nesting_level -= 1
search_pos = next_close_pos
end_pos = next_close_pos
if adverbs is not None and re.search(r':to\b', adverbs):
heredoc_terminator = text[match.start('delimiter') + n_chars : end_pos]
end_heredoc = re.search(r'^\s*' + re.escape(heredoc_terminator) + r'\s*$', text[ match.end('delimiter') : ], re.MULTILINE)
if end_heredoc:
end_pos = match.end('delimiter') + end_heredoc.end()
else:
end_pos = len(text)
yield match.start(), token_class, text[match.start() : end_pos + n_chars]
context.pos = end_pos + n_chars
return callback
def opening_brace_callback(lexer, match, context):
stack = context.stack
yield match.start(), Text, context.text[match.start() : match.end()]
context.pos = match.end()
# if we encounter an opening brace and we're one level
# below a token state, it means we need to increment
# the nesting level for braces so we know later when
# we should return to the token rules.
if len(stack) > 2 and stack[-2] == 'token':
context.perl6_token_nesting_level += 1
def closing_brace_callback(lexer, match, context):
stack = context.stack
yield match.start(), Text, context.text[match.start() : match.end()]
context.pos = match.end()
# if we encounter a free closing brace and we're one level
# below a token state, it means we need to check the nesting
# level to see if we need to return to the token state.
if len(stack) > 2 and stack[-2] == 'token':
context.perl6_token_nesting_level -= 1
if context.perl6_token_nesting_level == 0:
stack.pop()
def embedded_perl6_callback(lexer, match, context):
context.perl6_token_nesting_level = 1
yield match.start(), Text, context.text[match.start() : match.end()]
context.pos = match.end()
context.stack.append('root')
# If you're modifying these rules, be careful if you need to process '{' or '}' characters.
# We have special logic for processing these characters (due to the fact that you can nest
# Perl 6 code in regex blocks), so if you need to process one of them, make sure you also
# process the corresponding one!
tokens = {
'common' : [
(r'#[`|=](?P<delimiter>(?P<first_char>[' + ''.join(PERL6_BRACKETS.keys()) + r'])(?P=first_char)*)', brackets_callback(Comment.Multiline)),
(r'#[^\n]*$', Comment.Singleline),
(r'^(\s*)=begin\s+(\w+)\b.*?^\1=end\s+\2', Comment.Multiline),
(r'^(\s*)=for.*?\n\s*?\n', Comment.Multiline),
(r'^=.*?\n\s*?\n', Comment.Multiline),
(r'(regex|token|rule)(\s*' + PERL6_IDENTIFIER_RANGE + '+:sym)', bygroups(Keyword, Name), 'token-sym-brackets'),
(r'(regex|token|rule)(?!' + PERL6_IDENTIFIER_RANGE + ')(\s*' + PERL6_IDENTIFIER_RANGE + '+)?', bygroups(Keyword, Name), 'pre-token'),
# deal with a special case in the Perl 6 grammar (role q { ... })
(r'(role)(\s+)(q)(\s*)', bygroups(Keyword, Text, Name, Text)),
(_build_word_match(PERL6_KEYWORDS, PERL6_IDENTIFIER_RANGE), Keyword),
(_build_word_match(PERL6_BUILTIN_CLASSES, PERL6_IDENTIFIER_RANGE, suffix = '(?::[UD])?'), Name.Builtin),
(_build_word_match(PERL6_BUILTINS, PERL6_IDENTIFIER_RANGE), Name.Builtin),
# copied from PerlLexer
(r'[$@%&][.^:?=!~]?' + PERL6_IDENTIFIER_RANGE + u'+(?:<<.*?>>|<.*?>|«.*?»)*', Name.Variable),
(r'\$[!/](?:<<.*?>>|<.*?>|«.*?»)*', Name.Variable.Global),
(r'::\?\w+', Name.Variable.Global),
(r'[$@%&]\*' + PERL6_IDENTIFIER_RANGE + u'+(?:<<.*?>>|<.*?>|«.*?»)*', Name.Variable.Global),
(r'\$(?:<.*?>)+', Name.Variable),
(r'(?:q|qq|Q)[a-zA-Z]?\s*(?P<adverbs>:[\w\s:]+)?\s*(?P<delimiter>(?P<first_char>[^0-9a-zA-Z:\s])(?P=first_char)*)', brackets_callback(String)),
# copied from PerlLexer
(r'0_?[0-7]+(_[0-7]+)*', Number.Oct),
(r'0x[0-9A-Fa-f]+(_[0-9A-Fa-f]+)*', Number.Hex),
(r'0b[01]+(_[01]+)*', Number.Bin),
(r'(?i)(\d*(_\d*)*\.\d+(_\d*)*|\d+(_\d*)*\.\d+(_\d*)*)(e[+-]?\d+)?', Number.Float),
(r'(?i)\d+(_\d*)*e[+-]?\d+(_\d*)*', Number.Float),
(r'\d+(_\d+)*', Number.Integer),
(r'(?<=~~)\s*/(?:\\\\|\\/|.)*?/', String.Regex),
(r'(?<=[=(,])\s*/(?:\\\\|\\/|.)*?/', String.Regex),
(r'm\w+(?=\()', Name),
(r'(?:m|ms|rx)\s*(?P<adverbs>:[\w\s:]+)?\s*(?P<delimiter>(?P<first_char>[^0-9a-zA-Z:\s])(?P=first_char)*)', brackets_callback(String.Regex)),
(r'(?:s|ss|tr)\s*(?::[\w\s:]+)?\s*/(?:\\\\|\\/|.)*?/(?:\\\\|\\/|.)*?/', String.Regex),
(r'<[^\s=].*?\S>', String),
(_build_word_match(PERL6_OPERATORS), Operator),
(r'[0-9a-zA-Z_]' + PERL6_IDENTIFIER_RANGE + '*', Name),
(r"'(\\\\|\\[^\\]|[^'\\])*'", String),
(r'"(\\\\|\\[^\\]|[^"\\])*"', String),
],
'root' : [
include('common'),
(r'\{', opening_brace_callback),
(r'\}', closing_brace_callback),
(r'.+?', Text),
],
'pre-token' : [
include('common'),
(r'\{', Text, ('#pop', 'token')),
(r'.+?', Text),
],
'token-sym-brackets' : [
(r'(?P<delimiter>(?P<first_char>[' + ''.join(PERL6_BRACKETS.keys()) + '])(?P=first_char)*)', brackets_callback(Name), ('#pop', 'pre-token')),
(r'', Name, ('#pop', 'pre-token')),
],
'token': [
(r'}', Text, '#pop'),
(r'(?<=:)(?:my|our|state|constant|temp|let).*?;', using(this)),
# make sure that quotes in character classes aren't treated as strings
(r'<(?:[-!?+.]\s*)?\[.*?\]>', String.Regex),
# make sure that '#' characters in quotes aren't treated as comments
(r"(?<!\\)'(\\\\|\\[^\\]|[^'\\])*'", String.Regex),
(r'(?<!\\)"(\\\\|\\[^\\]|[^"\\])*"', String.Regex),
(r'#.*?$', Comment.Singleline),
(r'\{', embedded_perl6_callback),
('.+?', String.Regex),
],
}
def analyse_text(text):
def strip_pod(lines):
in_pod = False
stripped_lines = []
for line in lines:
if re.match(r'^=(?:end|cut)', line):
in_pod = False
elif re.match(r'^=\w+', line):
in_pod = True
elif not in_pod:
stripped_lines.append(line)
return stripped_lines
lines = text.splitlines()
lines = strip_pod(lines)
text = '\n'.join(lines)
if shebang_matches(text, r'perl6|rakudo|niecza'):
return True
if 'use v6' in text:
return 0.91 # 0.01 greater than Perl says for 'my $'
if re.search(r'[$@%]\*[A-Z]+', text): # Perl 6-style globals ($*OS)
return 0.91
if re.search(r'[$@%]\?[A-Z]+', text): # Perl 6 compiler variables ($?PACKAGE)
return 0.91
if re.search(r'[$@%][!.][A-Za-z0-9_-]+', text): # Perl 6 member variables
return 0.91
for line in text.splitlines():
if re.match(r'\s*(?:my|our)?\s*module', line): # module declarations
return 0.91
if re.match(r'\s*(?:my|our)?\s*role', line): # role declarations
return 0.91
if re.match(r'\s*(?:my|our)?\s*class\b', line): # class declarations
return 0.91
return False
def __init__(self, **options):
super(Perl6Lexer, self).__init__(**options)
self.encoding = options.get('encoding', 'utf-8')