diff options
| -rw-r--r-- | pygments/lexers/_mapping.py | 1 | ||||
| -rw-r--r-- | pygments/lexers/snobol_lexer.py | 80 | ||||
| -rw-r--r-- | tests/examplefiles/example.snobol | 15 |
3 files changed, 96 insertions, 0 deletions
diff --git a/pygments/lexers/_mapping.py b/pygments/lexers/_mapping.py index 4772a9a0..7653e18b 100644 --- a/pygments/lexers/_mapping.py +++ b/pygments/lexers/_mapping.py @@ -204,6 +204,7 @@ LEXERS = { 'ScssLexer': ('pygments.lexers.web', 'SCSS', ('scss',), ('*.scss',), ('text/x-scss',)), 'SmalltalkLexer': ('pygments.lexers.other', 'Smalltalk', ('smalltalk', 'squeak'), ('*.st',), ('text/x-smalltalk',)), 'SmartyLexer': ('pygments.lexers.templates', 'Smarty', ('smarty',), ('*.tpl',), ('application/x-smarty',)), + 'SnobolLexer': ('pygments.lexers.snobol_lexer', 'Snobol', ('snobol',), ('*.snobol',), ()), 'SourcesListLexer': ('pygments.lexers.text', 'Debian Sourcelist', ('sourceslist', 'sources.list'), ('sources.list',), ()), 'SqlLexer': ('pygments.lexers.other', 'SQL', ('sql',), ('*.sql',), ('text/x-sql',)), 'SqliteConsoleLexer': ('pygments.lexers.other', 'sqlite3con', ('sqlite3',), ('*.sqlite3-console',), ('text/x-sqlite3-console',)), diff --git a/pygments/lexers/snobol_lexer.py b/pygments/lexers/snobol_lexer.py new file mode 100644 index 00000000..8565fc59 --- /dev/null +++ b/pygments/lexers/snobol_lexer.py @@ -0,0 +1,80 @@ +# -*- coding: utf-8 -*- +""" + pygments.lexers.snobol_lexer + ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + + Just one lone lexer (SNOBOL4) + + :copyright: Copyright 2011 by Martin Harriman + :license: BSD, see LICENSE for details. +""" + +from pygments.lexer import RegexLexer, bygroups +from pygments.token import * + +__all__ = ['SnobolLexer'] + +class SnobolLexer(RegexLexer): + """ + Lexer for the SNOBOL4 programming language + + Recognizes the common ASCII equivalents of the original SNOBOL4 operators. + Does not require spaces around binary operators. + """ + + name = "Snobol" + aliases = ["snobol"] + filenames = ['*.snobol'] + + tokens = { + # root state, start of line + # comments, continuation lines, and directives start in column 1 + # as do labels + 'root': [ + (r'\*.*\n', Comment), + (r'[\+\.] ', Punctuation, 'statement'), + (r'-.*\n', Comment), + (r'END\s*\n', Name.Label, 'heredoc'), + (r'[A-Za-z\$][\w$]*', Name.Label, 'statement'), + (r'\s+', Whitespace, 'statement'), + ], + # statement state, line after continuation or label + 'statement': [ + (r'\s*\n', Whitespace, '#pop'), + (r'\s+', Whitespace), + (r'(?<=[^\w.])(LT|LE|EQ|NE|GE|GT|INTEGER|IDENT|DIFFER|LGT|SIZE|' + r'REPLACE|TRIM|DUPL|REMDR|DATE|TIME|EVAL|APPLY|OPSYN|LOAD|UNLOAD|' + r'LEN|SPAN|BREAK|ANY|NOTANY|TAB|RTAB|REM|POS|RPOS|FAIL|FENCE|' + r'ABORT|ARB|ARBNO|BAL|SUCCEED|INPUT|OUTPUT|TERMINAL)(?=[^\w.])', + Name.Builtin), + (r'[A-Za-z][\w\.]*', Name), + # ASCII equivalents of original operators + # | for the EBCDIC equivalent, ! likewise + # \ for EBCDIC negation + (r'\*\*|[\?\$\.!%\*/#+\-@\|&\\!=]', Operator), + (r'"[^"]*"', String), + (r"'[^']*'", String), + # Accept SPITBOL syntax for real numbers + # as well as Macro SNOBOL4 + (r'[0-9]+(?=[^\.EeDd])', Number.Integer), + (r'[0-9]+(\.[0-9]*)?([EDed][-+]?[0-9]+)?', Number.Float), + # Goto + (r':', Punctuation, 'goto'), + (r'[\(\)<>,;]', Punctuation), + ], + # Goto block + 'goto': [ + (r'\s*\n', Whitespace, "#pop:2"), + (r'\s+', Whitespace), + (r'F|S', Keyword), + (r'(\()([A-Za-z][\w.]*)(\))', + bygroups(Punctuation, Name.Label, Punctuation)) + ], + # everything after the END statement is basically one + # big heredoc. + 'heredoc': [ + (r'.*\n', String.Heredoc) + ] + } + + diff --git a/tests/examplefiles/example.snobol b/tests/examplefiles/example.snobol new file mode 100644 index 00000000..26ca5cf4 --- /dev/null +++ b/tests/examplefiles/example.snobol @@ -0,0 +1,15 @@ +-SOME RANDOM DIRECTIVE WOULD GO HERE +* +* SNOBOL4 example file for lexer +* + SOME.THING_OR_OTHER32 = 1 + 1.0 - 1E3 * 1E-3 ** 2.718284590E0 ++ :F(END)S(IN_LOOP) + PATTERN = LEN(3) ("GAR" | "BAR") +IN_LOOP THING = INPUT :F(END) + THING LEN(3) ("GAR" | "BAR") :S(OK) + OUTPUT = THING " : Failure!" :(IN_LOOP) +OK OUTPUT = THING ' : "Success"!' :(IN_LOOP) +END +FOOBAR +FOOGAR +THiNIg |