diff options
Diffstat (limited to 'pygments/lexers/stata.py')
| -rw-r--r-- | pygments/lexers/stata.py | 145 |
1 files changed, 104 insertions, 41 deletions
diff --git a/pygments/lexers/stata.py b/pygments/lexers/stata.py index a015a23e..b7b22edb 100644 --- a/pygments/lexers/stata.py +++ b/pygments/lexers/stata.py @@ -5,10 +5,11 @@ Lexer for Stata - :copyright: Copyright 2006-2017 by the Pygments team, see AUTHORS. + :copyright: Copyright 2006-2019 by the Pygments team, see AUTHORS. :license: BSD, see LICENSE for details. """ +import re from pygments.lexer import RegexLexer, include, words from pygments.token import Comment, Keyword, Name, Number, \ String, Text, Operator @@ -33,56 +34,118 @@ class StataLexer(RegexLexer): aliases = ['stata', 'do'] filenames = ['*.do', '*.ado'] mimetypes = ['text/x-stata', 'text/stata', 'application/x-stata'] + flags = re.MULTILINE | re.DOTALL tokens = { 'root': [ include('comments'), - include('vars-strings'), + include('strings'), + include('macros'), include('numbers'), include('keywords'), + include('operators'), + include('format'), (r'.', Text), ], - # Global and local macros; regular and special strings - 'vars-strings': [ - (r'\$[\w{]', Name.Variable.Global, 'var_validglobal'), - (r'`\w{0,31}\'', Name.Variable), - (r'"', String, 'string_dquote'), - (r'`"', String, 'string_mquote'), - ], - # For either string type, highlight macros as macros - 'string_dquote': [ - (r'"', String, '#pop'), - (r'\\\\|\\"|\\\n', String.Escape), - (r'\$', Name.Variable.Global, 'var_validglobal'), - (r'`', Name.Variable, 'var_validlocal'), - (r'[^$`"\\]+', String), - (r'[$"\\]', String), - ], - 'string_mquote': [ + # Comments are a complicated beast in Stata because they can be + # nested and there are a few corner cases with that. See: + # - github.com/kylebarron/language-stata/issues/90 + # - statalist.org/forums/forum/general-stata-discussion/general/1448244 + 'comments': [ + (r'(^//|(?<=\s)//)(?!/)', Comment.Single, 'comments-double-slash'), + (r'^\s*\*', Comment.Single, 'comments-star'), + (r'/\*', Comment.Multiline, 'comments-block'), + (r'(^///|(?<=\s)///)', Comment.Special, 'comments-triple-slash') + ], + 'comments-block': [ + (r'/\*', Comment.Multiline, '#push'), + # this ends and restarts a comment block. but need to catch this so + # that it doesn\'t start _another_ level of comment blocks + (r'\*/\*', Comment.Multiline), + (r'(\*/\s+\*(?!/)[^\n]*)|(\*/)', Comment.Multiline, '#pop'), + # Match anything else as a character inside the comment + (r'.', Comment.Multiline), + ], + 'comments-star': [ + (r'///.*?\n', Comment.Single, + ('#pop', 'comments-triple-slash')), + (r'(^//|(?<=\s)//)(?!/)', Comment.Single, + ('#pop', 'comments-double-slash')), + (r'/\*', Comment.Multiline, 'comments-block'), + (r'.(?=\n)', Comment.Single, '#pop'), + (r'.', Comment.Single), + ], + 'comments-triple-slash': [ + (r'\n', Comment.Special, '#pop'), + # A // breaks out of a comment for the rest of the line + (r'//.*?(?=\n)', Comment.Single, '#pop'), + (r'.', Comment.Special), + ], + 'comments-double-slash': [ + (r'\n', Text, '#pop'), + (r'.', Comment.Single), + ], + # `"compound string"' and regular "string"; note the former are + # nested. + 'strings': [ + (r'`"', String, 'string-compound'), + (r'(?<!`)"', String, 'string-regular'), + ], + 'string-compound': [ + (r'`"', String, '#push'), (r'"\'', String, '#pop'), - (r'\\\\|\\"|\\\n', String.Escape), - (r'\$', Name.Variable.Global, 'var_validglobal'), - (r'`', Name.Variable, 'var_validlocal'), - (r'[^$`"\\]+', String), - (r'[$"\\]', String), - ], - 'var_validglobal': [ - (r'\{\w{0,32}\}', Name.Variable.Global, '#pop'), - (r'\w{1,32}', Name.Variable.Global, '#pop'), + (r'\\\\|\\"|\\\$|\\`|\\\n', String.Escape), + include('macros'), + (r'.', String) ], - 'var_validlocal': [ - (r'\w{0,31}\'', Name.Variable, '#pop'), + 'string-regular': [ + (r'(")(?!\')|(?=\n)', String, '#pop'), + (r'\\\\|\\"|\\\$|\\`|\\\n', String.Escape), + include('macros'), + (r'.', String) ], - # * only OK at line start, // OK anywhere - 'comments': [ - (r'^\s*\*.*$', Comment), - (r'//.*', Comment.Single), - (r'/\*.*?\*/', Comment.Multiline), - (r'/[*](.|\n)*?[*]/', Comment.Multiline), + # A local is usually + # `\w{0,31}' + # `:extended macro' + # `=expression' + # `[rsen](results)' + # `(++--)scalar(++--)' + # + # However, there are all sorts of weird rules wrt edge + # cases. Instead of writing 27 exceptions, anything inside + # `' is a local. + # + # A global is more restricted, so we do follow rules. Note only + # locals explicitly enclosed ${} can be nested. + 'macros': [ + (r'\$(\{|(?=[\$`]))', Name.Variable.Global, 'macro-global-nested'), + (r'\$', Name.Variable.Global, 'macro-global-name'), + (r'`', Name.Variable, 'macro-local'), + ], + 'macro-local': [ + (r'`', Name.Variable, '#push'), + (r"'", Name.Variable, '#pop'), + (r'\$(\{|(?=[\$`]))', Name.Variable.Global, 'macro-global-nested'), + (r'\$', Name.Variable.Global, 'macro-global-name'), + (r'.', Name.Variable), # fallback + ], + 'macro-global-nested': [ + (r'\$(\{|(?=[\$`]))', Name.Variable.Global, '#push'), + (r'\}', Name.Variable.Global, '#pop'), + (r'\$', Name.Variable.Global, 'macro-global-name'), + (r'`', Name.Variable, 'macro-local'), + (r'\w', Name.Variable.Global), # fallback + (r'(?!\w)', Name.Variable.Global, '#pop'), + ], + 'macro-global-name': [ + (r'\$(\{|(?=[\$`]))', Name.Variable.Global, 'macro-global-nested', '#pop'), + (r'\$', Name.Variable.Global, 'macro-global-name', '#pop'), + (r'`', Name.Variable, 'macro-local', '#pop'), + (r'\w{1,32}', Name.Variable.Global, '#pop'), ], # Built in functions and statements 'keywords': [ - (words(builtins_functions, prefix = r'\b', suffix = r'\('), + (words(builtins_functions, prefix = r'\b', suffix = r'(?=\()'), Name.Function), (words(builtins_base, prefix = r'(^\s*|\s)', suffix = r'\b'), Keyword), @@ -100,9 +163,9 @@ class StataLexer(RegexLexer): ], # Stata formats 'format': [ - (r'%-?\d{1,2}(\.\d{1,2})?[gfe]c?', Name.Variable), - (r'%(21x|16H|16L|8H|8L)', Name.Variable), - (r'%-?(tc|tC|td|tw|tm|tq|th|ty|tg).{0,32}', Name.Variable), - (r'%[-~]?\d{1,4}s', Name.Variable), + (r'%-?\d{1,2}(\.\d{1,2})?[gfe]c?', Name.Other), + (r'%(21x|16H|16L|8H|8L)', Name.Other), + (r'%-?(tc|tC|td|tw|tm|tq|th|ty|tg)\S{0,32}', Name.Other), + (r'%[-~]?\d{1,4}s', Name.Other), ] } |