hub/venv/lib/python3.7/site-packages/pygments/lexers/julia.py

336 lines
14 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

# -*- coding: utf-8 -*-
"""
pygments.lexers.julia
~~~~~~~~~~~~~~~~~~~~~
Lexers for the Julia language.
:copyright: Copyright 2006-2019 by the Pygments team, see AUTHORS.
:license: BSD, see LICENSE for details.
"""
import re
from pygments.lexer import Lexer, RegexLexer, bygroups, do_insertions, \
words, include
from pygments.token import Text, Comment, Operator, Keyword, Name, String, \
Number, Punctuation, Generic
from pygments.util import shebang_matches, unirange
__all__ = ['JuliaLexer', 'JuliaConsoleLexer']
allowed_variable = (
u'(?:[a-zA-Z_\u00A1-\uffff]|%s)(?:[a-zA-Z_0-9\u00A1-\uffff]|%s)*!*' %
((unirange(0x10000, 0x10ffff),) * 2))
class JuliaLexer(RegexLexer):
"""
For `Julia <http://julialang.org/>`_ source code.
.. versionadded:: 1.6
"""
name = 'Julia'
aliases = ['julia', 'jl']
filenames = ['*.jl']
mimetypes = ['text/x-julia', 'application/x-julia']
flags = re.MULTILINE | re.UNICODE
tokens = {
'root': [
(r'\n', Text),
(r'[^\S\n]+', Text),
(r'#=', Comment.Multiline, "blockcomment"),
(r'#.*$', Comment),
(r'[\[\]{}(),;]', Punctuation),
# keywords
(r'in\b', Keyword.Pseudo),
(r'isa\b', Keyword.Pseudo),
(r'(true|false)\b', Keyword.Constant),
(r'(local|global|const)\b', Keyword.Declaration),
(words([
'function', 'type', 'typealias', 'abstract', 'immutable',
'baremodule', 'begin', 'bitstype', 'break', 'catch', 'ccall',
'continue', 'do', 'else', 'elseif', 'end', 'export', 'finally',
'for', 'if', 'import', 'importall', 'let', 'macro', 'module',
'mutable', 'primitive', 'quote', 'return', 'struct', 'try',
'using', 'while'],
suffix=r'\b'), Keyword),
# NOTE
# Patterns below work only for definition sites and thus hardly reliable.
#
# functions
# (r'(function)(\s+)(' + allowed_variable + ')',
# bygroups(Keyword, Text, Name.Function)),
#
# types
# (r'(type|typealias|abstract|immutable)(\s+)(' + allowed_variable + ')',
# bygroups(Keyword, Text, Name.Class)),
# type names
(words([
'ANY', 'ASCIIString', 'AbstractArray', 'AbstractChannel',
'AbstractFloat', 'AbstractMatrix', 'AbstractRNG',
'AbstractSparseArray', 'AbstractSparseMatrix',
'AbstractSparseVector', 'AbstractString', 'AbstractVecOrMat',
'AbstractVector', 'Any', 'ArgumentError', 'Array',
'AssertionError', 'Associative', 'Base64DecodePipe',
'Base64EncodePipe', 'Bidiagonal', 'BigFloat', 'BigInt',
'BitArray', 'BitMatrix', 'BitVector', 'Bool', 'BoundsError',
'Box', 'BufferStream', 'CapturedException', 'CartesianIndex',
'CartesianRange', 'Cchar', 'Cdouble', 'Cfloat', 'Channel',
'Char', 'Cint', 'Cintmax_t', 'Clong', 'Clonglong',
'ClusterManager', 'Cmd', 'Coff_t', 'Colon', 'Complex',
'Complex128', 'Complex32', 'Complex64', 'CompositeException',
'Condition', 'Cptrdiff_t', 'Cshort', 'Csize_t', 'Cssize_t',
'Cstring', 'Cuchar', 'Cuint', 'Cuintmax_t', 'Culong',
'Culonglong', 'Cushort', 'Cwchar_t', 'Cwstring', 'DataType',
'Date', 'DateTime', 'DenseArray', 'DenseMatrix',
'DenseVecOrMat', 'DenseVector', 'Diagonal', 'Dict',
'DimensionMismatch', 'Dims', 'DirectIndexString', 'Display',
'DivideError', 'DomainError', 'EOFError', 'EachLine', 'Enum',
'Enumerate', 'ErrorException', 'Exception', 'Expr',
'Factorization', 'FileMonitor', 'FileOffset', 'Filter',
'Float16', 'Float32', 'Float64', 'FloatRange', 'Function',
'GenSym', 'GlobalRef', 'GotoNode', 'HTML', 'Hermitian', 'IO',
'IOBuffer', 'IOStream', 'IPv4', 'IPv6', 'InexactError',
'InitError', 'Int', 'Int128', 'Int16', 'Int32', 'Int64', 'Int8',
'IntSet', 'Integer', 'InterruptException', 'IntrinsicFunction',
'InvalidStateException', 'Irrational', 'KeyError', 'LabelNode',
'LambdaStaticData', 'LinSpace', 'LineNumberNode', 'LoadError',
'LocalProcess', 'LowerTriangular', 'MIME', 'Matrix',
'MersenneTwister', 'Method', 'MethodError', 'MethodTable',
'Module', 'NTuple', 'NewvarNode', 'NullException', 'Nullable',
'Number', 'ObjectIdDict', 'OrdinalRange', 'OutOfMemoryError',
'OverflowError', 'Pair', 'ParseError', 'PartialQuickSort',
'Pipe', 'PollingFileWatcher', 'ProcessExitedException',
'ProcessGroup', 'Ptr', 'QuoteNode', 'RandomDevice', 'Range',
'Rational', 'RawFD', 'ReadOnlyMemoryError', 'Real',
'ReentrantLock', 'Ref', 'Regex', 'RegexMatch',
'RemoteException', 'RemoteRef', 'RepString', 'RevString',
'RopeString', 'RoundingMode', 'SegmentationFault',
'SerializationState', 'Set', 'SharedArray', 'SharedMatrix',
'SharedVector', 'Signed', 'SimpleVector', 'SparseMatrixCSC',
'StackOverflowError', 'StatStruct', 'StepRange', 'StridedArray',
'StridedMatrix', 'StridedVecOrMat', 'StridedVector', 'SubArray',
'SubString', 'SymTridiagonal', 'Symbol', 'SymbolNode',
'Symmetric', 'SystemError', 'TCPSocket', 'Task', 'Text',
'TextDisplay', 'Timer', 'TopNode', 'Tridiagonal', 'Tuple',
'Type', 'TypeConstructor', 'TypeError', 'TypeName', 'TypeVar',
'UDPSocket', 'UInt', 'UInt128', 'UInt16', 'UInt32', 'UInt64',
'UInt8', 'UTF16String', 'UTF32String', 'UTF8String',
'UndefRefError', 'UndefVarError', 'UnicodeError', 'UniformScaling',
'Union', 'UnitRange', 'Unsigned', 'UpperTriangular', 'Val',
'Vararg', 'VecOrMat', 'Vector', 'VersionNumber', 'Void', 'WString',
'WeakKeyDict', 'WeakRef', 'WorkerConfig', 'Zip'], suffix=r'\b'),
Keyword.Type),
# builtins
(words([
u'ARGS', u'CPU_CORES', u'C_NULL', u'DevNull', u'ENDIAN_BOM',
u'ENV', u'I', u'Inf', u'Inf16', u'Inf32', u'Inf64',
u'InsertionSort', u'JULIA_HOME', u'LOAD_PATH', u'MergeSort',
u'NaN', u'NaN16', u'NaN32', u'NaN64', u'OS_NAME',
u'QuickSort', u'RoundDown', u'RoundFromZero', u'RoundNearest',
u'RoundNearestTiesAway', u'RoundNearestTiesUp',
u'RoundToZero', u'RoundUp', u'STDERR', u'STDIN', u'STDOUT',
u'VERSION', u'WORD_SIZE', u'catalan', u'e', u'eu',
u'eulergamma', u'golden', u'im', u'nothing', u'pi', u'γ',
u'π', u'φ'],
suffix=r'\b'), Name.Builtin),
# operators
# see: https://github.com/JuliaLang/julia/blob/master/src/julia-parser.scm
(words([
# prec-assignment
u'=', u':=', u'+=', u'-=', u'*=', u'/=', u'//=', u'.//=', u'.*=', u'./=',
u'\\=', u'.\\=', u'^=', u'.^=', u'÷=', u'.÷=', u'%=', u'.%=', u'|=', u'&=',
u'$=', u'=>', u'<<=', u'>>=', u'>>>=', u'~', u'.+=', u'.-=',
# prec-conditional
u'?',
# prec-arrow
u'--', u'-->',
# prec-lazy-or
u'||',
# prec-lazy-and
u'&&',
# prec-comparison
u'>', u'<', u'>=', u'', u'<=', u'', u'==', u'===', u'', u'!=', u'',
u'!==', u'', u'.>', u'.<', u'.>=', u'.≥', u'.<=', u'.≤', u'.==', u'.!=',
u'.≠', u'.=', u'.!', u'<:', u'>:', u'', u'', u'', u'', u'',
u'', u'',
u'', u'',
# prec-pipe
u'|>', u'<|',
# prec-colon
u':',
# prec-plus
u'+', u'-', u'.+', u'.-', u'|', u'', u'$',
# prec-bitshift
u'<<', u'>>', u'>>>', u'.<<', u'.>>', u'.>>>',
# prec-times
u'*', u'/', u'./', u'÷', u'', u'%', u'', u'.%', u'.*', u'\\', u'.\\', u'&', u'',
# prec-rational
u'//', u'.//',
# prec-power
u'^', u'.^',
# prec-decl
u'::',
# prec-dot
u'.',
# unary op
u'+', u'-', u'!', u'', u'', u''
]), Operator),
# chars
(r"'(\\.|\\[0-7]{1,3}|\\x[a-fA-F0-9]{1,3}|\\u[a-fA-F0-9]{1,4}|"
r"\\U[a-fA-F0-9]{1,6}|[^\\\'\n])'", String.Char),
# try to match trailing transpose
(r'(?<=[.\w)\]])\'+', Operator),
# strings
(r'"""', String, 'tqstring'),
(r'"', String, 'string'),
# regular expressions
(r'r"""', String.Regex, 'tqregex'),
(r'r"', String.Regex, 'regex'),
# backticks
(r'`', String.Backtick, 'command'),
# names
(allowed_variable, Name),
(r'@' + allowed_variable, Name.Decorator),
# numbers
(r'(\d+(_\d+)+\.\d*|\d*\.\d+(_\d+)+)([eEf][+-]?[0-9]+)?', Number.Float),
(r'(\d+\.\d*|\d*\.\d+)([eEf][+-]?[0-9]+)?', Number.Float),
(r'\d+(_\d+)+[eEf][+-]?[0-9]+', Number.Float),
(r'\d+[eEf][+-]?[0-9]+', Number.Float),
(r'0b[01]+(_[01]+)+', Number.Bin),
(r'0b[01]+', Number.Bin),
(r'0o[0-7]+(_[0-7]+)+', Number.Oct),
(r'0o[0-7]+', Number.Oct),
(r'0x[a-fA-F0-9]+(_[a-fA-F0-9]+)+', Number.Hex),
(r'0x[a-fA-F0-9]+', Number.Hex),
(r'\d+(_\d+)+', Number.Integer),
(r'\d+', Number.Integer)
],
"blockcomment": [
(r'[^=#]', Comment.Multiline),
(r'#=', Comment.Multiline, '#push'),
(r'=#', Comment.Multiline, '#pop'),
(r'[=#]', Comment.Multiline),
],
'string': [
(r'"', String, '#pop'),
# FIXME: This escape pattern is not perfect.
(r'\\([\\"\'$nrbtfav]|(x|u|U)[a-fA-F0-9]+|\d+)', String.Escape),
# Interpolation is defined as "$" followed by the shortest full
# expression, which is something we can't parse.
# Include the most common cases here: $word, and $(paren'd expr).
(r'\$' + allowed_variable, String.Interpol),
# (r'\$[a-zA-Z_]+', String.Interpol),
(r'(\$)(\()', bygroups(String.Interpol, Punctuation), 'in-intp'),
# @printf and @sprintf formats
(r'%[-#0 +]*([0-9]+|[*])?(\.([0-9]+|[*]))?[hlL]?[E-GXc-giorsux%]',
String.Interpol),
(r'.|\s', String),
],
'tqstring': [
(r'"""', String, '#pop'),
(r'\\([\\"\'$nrbtfav]|(x|u|U)[a-fA-F0-9]+|\d+)', String.Escape),
(r'\$' + allowed_variable, String.Interpol),
(r'(\$)(\()', bygroups(String.Interpol, Punctuation), 'in-intp'),
(r'.|\s', String),
],
'regex': [
(r'"', String.Regex, '#pop'),
(r'\\"', String.Regex),
(r'.|\s', String.Regex),
],
'tqregex': [
(r'"""', String.Regex, '#pop'),
(r'.|\s', String.Regex),
],
'command': [
(r'`', String.Backtick, '#pop'),
(r'\$' + allowed_variable, String.Interpol),
(r'(\$)(\()', bygroups(String.Interpol, Punctuation), 'in-intp'),
(r'.|\s', String.Backtick)
],
'in-intp': [
(r'\(', Punctuation, '#push'),
(r'\)', Punctuation, '#pop'),
include('root'),
]
}
def analyse_text(text):
return shebang_matches(text, r'julia')
class JuliaConsoleLexer(Lexer):
"""
For Julia console sessions. Modeled after MatlabSessionLexer.
.. versionadded:: 1.6
"""
name = 'Julia console'
aliases = ['jlcon']
def get_tokens_unprocessed(self, text):
jllexer = JuliaLexer(**self.options)
start = 0
curcode = ''
insertions = []
output = False
error = False
for line in text.splitlines(True):
if line.startswith('julia>'):
insertions.append((len(curcode), [(0, Generic.Prompt, line[:6])]))
curcode += line[6:]
output = False
error = False
elif line.startswith('help?>') or line.startswith('shell>'):
yield start, Generic.Prompt, line[:6]
yield start + 6, Text, line[6:]
output = False
error = False
elif line.startswith(' ') and not output:
insertions.append((len(curcode), [(0, Text, line[:6])]))
curcode += line[6:]
else:
if curcode:
for item in do_insertions(
insertions, jllexer.get_tokens_unprocessed(curcode)):
yield item
curcode = ''
insertions = []
if line.startswith('ERROR: ') or error:
yield start, Generic.Error, line
error = True
else:
yield start, Generic.Output, line
output = True
start += len(line)
if curcode:
for item in do_insertions(
insertions, jllexer.get_tokens_unprocessed(curcode)):
yield item