|
1 # -*- coding: utf-8 -*- |
|
2 """ |
|
3 pygments.lexers.fortran |
|
4 ~~~~~~~~~~~~~~~~~~~~~~~ |
|
5 |
|
6 Lexers for Fortran languages. |
|
7 |
|
8 :copyright: Copyright 2006-2017 by the Pygments team, see AUTHORS. |
|
9 :license: BSD, see LICENSE for details. |
|
10 """ |
|
11 |
|
12 import re |
|
13 |
|
14 from pygments.lexer import RegexLexer, bygroups, include, words, using, default |
|
15 from pygments.token import Text, Comment, Operator, Keyword, Name, String, \ |
|
16 Number, Punctuation, Generic |
|
17 |
|
18 __all__ = ['FortranLexer', 'FortranFixedLexer'] |
|
19 |
|
20 |
|
21 class FortranLexer(RegexLexer): |
|
22 """ |
|
23 Lexer for FORTRAN 90 code. |
|
24 |
|
25 .. versionadded:: 0.10 |
|
26 """ |
|
27 name = 'Fortran' |
|
28 aliases = ['fortran'] |
|
29 filenames = ['*.f03', '*.f90', '*.F03', '*.F90'] |
|
30 mimetypes = ['text/x-fortran'] |
|
31 flags = re.IGNORECASE | re.MULTILINE |
|
32 |
|
33 # Data Types: INTEGER, REAL, COMPLEX, LOGICAL, CHARACTER and DOUBLE PRECISION |
|
34 # Operators: **, *, +, -, /, <, >, <=, >=, ==, /= |
|
35 # Logical (?): NOT, AND, OR, EQV, NEQV |
|
36 |
|
37 # Builtins: |
|
38 # http://gcc.gnu.org/onlinedocs/gcc-3.4.6/g77/Table-of-Intrinsic-Functions.html |
|
39 |
|
40 tokens = { |
|
41 'root': [ |
|
42 (r'^#.*\n', Comment.Preproc), |
|
43 (r'!.*\n', Comment), |
|
44 include('strings'), |
|
45 include('core'), |
|
46 (r'[a-z][\w$]*', Name), |
|
47 include('nums'), |
|
48 (r'[\s]+', Text), |
|
49 ], |
|
50 'core': [ |
|
51 # Statements |
|
52 (words(( |
|
53 'ABSTRACT', 'ACCEPT', 'ALL', 'ALLSTOP', 'ALLOCATABLE', 'ALLOCATE', |
|
54 'ARRAY', 'ASSIGN', 'ASSOCIATE', 'ASYNCHRONOUS', 'BACKSPACE', 'BIND', |
|
55 'BLOCK', 'BLOCKDATA', 'BYTE', 'CALL', 'CASE', 'CLASS', 'CLOSE', |
|
56 'CODIMENSION', 'COMMON', 'CONCURRRENT', 'CONTIGUOUS', 'CONTAINS', |
|
57 'CONTINUE', 'CRITICAL', 'CYCLE', 'DATA', 'DEALLOCATE', 'DECODE', |
|
58 'DEFERRED', 'DIMENSION', 'DO', 'ELEMENTAL', 'ELSE', 'ENCODE', 'END', |
|
59 'ENTRY', 'ENUM', 'ENUMERATOR', 'EQUIVALENCE', 'EXIT', 'EXTENDS', |
|
60 'EXTERNAL', 'EXTRINSIC', 'FILE', 'FINAL', 'FORALL', 'FORMAT', |
|
61 'FUNCTION', 'GENERIC', 'GOTO', 'IF', 'IMAGES', 'IMPLICIT', |
|
62 'IMPORT', 'IMPURE', 'INCLUDE', 'INQUIRE', 'INTENT', 'INTERFACE', |
|
63 'INTRINSIC', 'IS', 'LOCK', 'MEMORY', 'MODULE', 'NAMELIST', 'NULLIFY', |
|
64 'NONE', 'NON_INTRINSIC', 'NON_OVERRIDABLE', 'NOPASS', 'OPEN', 'OPTIONAL', |
|
65 'OPTIONS', 'PARAMETER', 'PASS', 'PAUSE', 'POINTER', 'PRINT', 'PRIVATE', |
|
66 'PROGRAM', 'PROCEDURE', 'PROTECTED', 'PUBLIC', 'PURE', 'READ', |
|
67 'RECURSIVE', 'RESULT', 'RETURN', 'REWIND', 'SAVE', 'SELECT', 'SEQUENCE', |
|
68 'STOP', 'SUBMODULE', 'SUBROUTINE', 'SYNC', 'SYNCALL', 'SYNCIMAGES', |
|
69 'SYNCMEMORY', 'TARGET', 'THEN', 'TYPE', 'UNLOCK', 'USE', 'VALUE', |
|
70 'VOLATILE', 'WHERE', 'WRITE', 'WHILE'), prefix=r'\b', suffix=r'\s*\b'), |
|
71 Keyword), |
|
72 |
|
73 # Data Types |
|
74 (words(( |
|
75 'CHARACTER', 'COMPLEX', 'DOUBLE PRECISION', 'DOUBLE COMPLEX', 'INTEGER', |
|
76 'LOGICAL', 'REAL', 'C_INT', 'C_SHORT', 'C_LONG', 'C_LONG_LONG', |
|
77 'C_SIGNED_CHAR', 'C_SIZE_T', 'C_INT8_T', 'C_INT16_T', 'C_INT32_T', |
|
78 'C_INT64_T', 'C_INT_LEAST8_T', 'C_INT_LEAST16_T', 'C_INT_LEAST32_T', |
|
79 'C_INT_LEAST64_T', 'C_INT_FAST8_T', 'C_INT_FAST16_T', 'C_INT_FAST32_T', |
|
80 'C_INT_FAST64_T', 'C_INTMAX_T', 'C_INTPTR_T', 'C_FLOAT', 'C_DOUBLE', |
|
81 'C_LONG_DOUBLE', 'C_FLOAT_COMPLEX', 'C_DOUBLE_COMPLEX', |
|
82 'C_LONG_DOUBLE_COMPLEX', 'C_BOOL', 'C_CHAR', 'C_PTR', 'C_FUNPTR'), |
|
83 prefix=r'\b', suffix=r'\s*\b'), |
|
84 Keyword.Type), |
|
85 |
|
86 # Operators |
|
87 (r'(\*\*|\*|\+|-|\/|<|>|<=|>=|==|\/=|=)', Operator), |
|
88 |
|
89 (r'(::)', Keyword.Declaration), |
|
90 |
|
91 (r'[()\[\],:&%;.]', Punctuation), |
|
92 # Intrinsics |
|
93 (words(( |
|
94 'Abort', 'Abs', 'Access', 'AChar', 'ACos', 'ACosH', 'AdjustL', |
|
95 'AdjustR', 'AImag', 'AInt', 'Alarm', 'All', 'Allocated', 'ALog', |
|
96 'AMax', 'AMin', 'AMod', 'And', 'ANInt', 'Any', 'ASin', 'ASinH', |
|
97 'Associated', 'ATan', 'ATanH', 'Atomic_Define', 'Atomic_Ref', |
|
98 'BesJ', 'BesJN', 'Bessel_J0', 'Bessel_J1', 'Bessel_JN', 'Bessel_Y0', |
|
99 'Bessel_Y1', 'Bessel_YN', 'BesY', 'BesYN', 'BGE', 'BGT', 'BLE', |
|
100 'BLT', 'Bit_Size', 'BTest', 'CAbs', 'CCos', 'Ceiling', 'CExp', |
|
101 'Char', 'ChDir', 'ChMod', 'CLog', 'Cmplx', 'Command_Argument_Count', |
|
102 'Complex', 'Conjg', 'Cos', 'CosH', 'Count', 'CPU_Time', 'CShift', |
|
103 'CSin', 'CSqRt', 'CTime', 'C_Loc', 'C_Associated', |
|
104 'C_Null_Ptr', 'C_Null_Funptr', 'C_F_Pointer', 'C_F_ProcPointer', |
|
105 'C_Null_Char', 'C_Alert', 'C_Backspace', 'C_Form_Feed', 'C_FunLoc', |
|
106 'C_Sizeof', 'C_New_Line', 'C_Carriage_Return', |
|
107 'C_Horizontal_Tab', 'C_Vertical_Tab', 'DAbs', 'DACos', 'DASin', |
|
108 'DATan', 'Date_and_Time', 'DbesJ', 'DbesJN', 'DbesY', |
|
109 'DbesYN', 'Dble', 'DCos', 'DCosH', 'DDiM', 'DErF', |
|
110 'DErFC', 'DExp', 'Digits', 'DiM', 'DInt', 'DLog', 'DMax', |
|
111 'DMin', 'DMod', 'DNInt', 'Dot_Product', 'DProd', 'DSign', 'DSinH', |
|
112 'DShiftL', 'DShiftR', 'DSin', 'DSqRt', 'DTanH', 'DTan', 'DTime', |
|
113 'EOShift', 'Epsilon', 'ErF', 'ErFC', 'ErFC_Scaled', 'ETime', |
|
114 'Execute_Command_Line', 'Exit', 'Exp', 'Exponent', 'Extends_Type_Of', |
|
115 'FDate', 'FGet', 'FGetC', 'FindLoc', 'Float', 'Floor', 'Flush', |
|
116 'FNum', 'FPutC', 'FPut', 'Fraction', 'FSeek', 'FStat', 'FTell', |
|
117 'Gamma', 'GError', 'GetArg', 'Get_Command', 'Get_Command_Argument', |
|
118 'Get_Environment_Variable', 'GetCWD', 'GetEnv', 'GetGId', 'GetLog', |
|
119 'GetPId', 'GetUId', 'GMTime', 'HostNm', 'Huge', 'Hypot', 'IAbs', |
|
120 'IAChar', 'IAll', 'IAnd', 'IAny', 'IArgC', 'IBClr', 'IBits', |
|
121 'IBSet', 'IChar', 'IDate', 'IDiM', 'IDInt', 'IDNInt', 'IEOr', |
|
122 'IErrNo', 'IFix', 'Imag', 'ImagPart', 'Image_Index', 'Index', |
|
123 'Int', 'IOr', 'IParity', 'IRand', 'IsaTty', 'IShft', 'IShftC', |
|
124 'ISign', 'Iso_C_Binding', 'Is_Contiguous', 'Is_Iostat_End', |
|
125 'Is_Iostat_Eor', 'ITime', 'Kill', 'Kind', 'LBound', 'LCoBound', |
|
126 'Len', 'Len_Trim', 'LGe', 'LGt', 'Link', 'LLe', 'LLt', 'LnBlnk', |
|
127 'Loc', 'Log', 'Log_Gamma', 'Logical', 'Long', 'LShift', 'LStat', |
|
128 'LTime', 'MaskL', 'MaskR', 'MatMul', 'Max', 'MaxExponent', |
|
129 'MaxLoc', 'MaxVal', 'MClock', 'Merge', 'Merge_Bits', 'Move_Alloc', |
|
130 'Min', 'MinExponent', 'MinLoc', 'MinVal', 'Mod', 'Modulo', 'MvBits', |
|
131 'Nearest', 'New_Line', 'NInt', 'Norm2', 'Not', 'Null', 'Num_Images', |
|
132 'Or', 'Pack', 'Parity', 'PError', 'Precision', 'Present', 'Product', |
|
133 'Radix', 'Rand', 'Random_Number', 'Random_Seed', 'Range', 'Real', |
|
134 'RealPart', 'Rename', 'Repeat', 'Reshape', 'RRSpacing', 'RShift', |
|
135 'Same_Type_As', 'Scale', 'Scan', 'Second', 'Selected_Char_Kind', |
|
136 'Selected_Int_Kind', 'Selected_Real_Kind', 'Set_Exponent', 'Shape', |
|
137 'ShiftA', 'ShiftL', 'ShiftR', 'Short', 'Sign', 'Signal', 'SinH', |
|
138 'Sin', 'Sleep', 'Sngl', 'Spacing', 'Spread', 'SqRt', 'SRand', |
|
139 'Stat', 'Storage_Size', 'Sum', 'SymLnk', 'System', 'System_Clock', |
|
140 'Tan', 'TanH', 'Time', 'This_Image', 'Tiny', 'TrailZ', 'Transfer', |
|
141 'Transpose', 'Trim', 'TtyNam', 'UBound', 'UCoBound', 'UMask', |
|
142 'Unlink', 'Unpack', 'Verify', 'XOr', 'ZAbs', 'ZCos', 'ZExp', |
|
143 'ZLog', 'ZSin', 'ZSqRt'), prefix=r'\b', suffix=r'\s*\b'), |
|
144 Name.Builtin), |
|
145 |
|
146 # Booleans |
|
147 (r'\.(true|false)\.', Name.Builtin), |
|
148 # Comparing Operators |
|
149 (r'\.(eq|ne|lt|le|gt|ge|not|and|or|eqv|neqv)\.', Operator.Word), |
|
150 ], |
|
151 |
|
152 'strings': [ |
|
153 (r'(?s)"(\\\\|\\[0-7]+|\\.|[^"\\])*"', String.Double), |
|
154 (r"(?s)'(\\\\|\\[0-7]+|\\.|[^'\\])*'", String.Single), |
|
155 ], |
|
156 |
|
157 'nums': [ |
|
158 (r'\d+(?![.e])(_[a-z]\w+)?', Number.Integer), |
|
159 (r'[+-]?\d*\.\d+([ed][-+]?\d+)?(_[a-z]\w+)?', Number.Float), |
|
160 (r'[+-]?\d+\.\d*([ed][-+]?\d+)?(_[a-z]\w+)?', Number.Float), |
|
161 (r'[+-]?\d+(\.\d*)?[ed][-+]?\d+(_[a-z]\w+)?', Number.Float), |
|
162 ], |
|
163 } |
|
164 |
|
165 |
|
166 class FortranFixedLexer(RegexLexer): |
|
167 """ |
|
168 Lexer for fixed format Fortran. |
|
169 |
|
170 .. versionadded:: 2.1 |
|
171 """ |
|
172 name = 'FortranFixed' |
|
173 aliases = ['fortranfixed'] |
|
174 filenames = ['*.f', '*.F'] |
|
175 |
|
176 flags = re.IGNORECASE |
|
177 |
|
178 def _lex_fortran(self, match, ctx=None): |
|
179 """Lex a line just as free form fortran without line break.""" |
|
180 lexer = FortranLexer() |
|
181 text = match.group(0) + "\n" |
|
182 for index, token, value in lexer.get_tokens_unprocessed(text): |
|
183 value = value.replace('\n', '') |
|
184 if value != '': |
|
185 yield index, token, value |
|
186 |
|
187 tokens = { |
|
188 'root': [ |
|
189 (r'[C*].*\n', Comment), |
|
190 (r'#.*\n', Comment.Preproc), |
|
191 (r' {0,4}!.*\n', Comment), |
|
192 (r'(.{5})', Name.Label, 'cont-char'), |
|
193 (r'.*\n', using(FortranLexer)), |
|
194 ], |
|
195 'cont-char': [ |
|
196 (' ', Text, 'code'), |
|
197 ('0', Comment, 'code'), |
|
198 ('.', Generic.Strong, 'code'), |
|
199 ], |
|
200 'code': [ |
|
201 (r'(.{66})(.*)(\n)', |
|
202 bygroups(_lex_fortran, Comment, Text), 'root'), |
|
203 (r'(.*)(\n)', bygroups(_lex_fortran, Text), 'root'), |
|
204 default('root'), |
|
205 ] |
|
206 } |