| 1 | #!/usr/bin/env python |
|---|
| 2 | # -*- coding: utf-8 -*- |
|---|
| 3 | |
|---|
| 4 | """ |
|---|
| 5 | | This file is part of the web2py Web Framework |
|---|
| 6 | | Copyrighted by Massimo Di Pierro <mdipierro@cs.depaul.edu> |
|---|
| 7 | | License: LGPLv3 (http://www.gnu.org/licenses/lgpl.html) |
|---|
| 8 | """ |
|---|
| 9 | |
|---|
| 10 | from pydal._compat import xrange |
|---|
| 11 | from yatl.sanitizer import xmlescape |
|---|
| 12 | import re |
|---|
| 13 | |
|---|
| 14 | __all__ = ['highlight'] |
|---|
| 15 | |
|---|
| 16 | |
|---|
| 17 | class all_styles(object): |
|---|
| 18 | """ |
|---|
| 19 | Custom non-data descriptor for lazy initialization of |
|---|
| 20 | Highlighter.all_styles class attribute. |
|---|
| 21 | |
|---|
| 22 | see: |
|---|
| 23 | https://docs.python.org/2/reference/datamodel.html#implementing-descriptors |
|---|
| 24 | or |
|---|
| 25 | https://docs.python.org/3/reference/datamodel.html#implementing-descriptors |
|---|
| 26 | """ |
|---|
| 27 | |
|---|
| 28 | def __get__(self, instance, owner): |
|---|
| 29 | val = _get_all_styles(owner) |
|---|
| 30 | setattr(owner, 'all_styles', val) |
|---|
| 31 | return val |
|---|
| 32 | |
|---|
| 33 | |
|---|
| 34 | class Highlighter(object): |
|---|
| 35 | |
|---|
| 36 | """Does syntax highlighting. |
|---|
| 37 | """ |
|---|
| 38 | |
|---|
| 39 | def __init__( |
|---|
| 40 | self, |
|---|
| 41 | mode, |
|---|
| 42 | link=None, |
|---|
| 43 | styles=None, |
|---|
| 44 | ): |
|---|
| 45 | """ |
|---|
| 46 | Initialize highlighter: |
|---|
| 47 | mode = language (PYTHON, WEB2PY, C, CPP, HTML, HTML_PLAIN) |
|---|
| 48 | """ |
|---|
| 49 | styles = styles or {} |
|---|
| 50 | mode = mode.upper() |
|---|
| 51 | if link and link[-1] != '/': |
|---|
| 52 | link = link + '/' |
|---|
| 53 | self.link = link |
|---|
| 54 | self.styles = styles |
|---|
| 55 | self.output = [] |
|---|
| 56 | self.span_style = None |
|---|
| 57 | if mode == 'WEB2PY': |
|---|
| 58 | (mode, self.suppress_tokens) = ('PYTHON', []) |
|---|
| 59 | elif mode == 'PYTHON': |
|---|
| 60 | self.suppress_tokens = ['GOTOHTML'] |
|---|
| 61 | elif mode == 'CPP': |
|---|
| 62 | (mode, self.suppress_tokens) = ('C', []) |
|---|
| 63 | elif mode == 'C': |
|---|
| 64 | self.suppress_tokens = ['CPPKEYWORD'] |
|---|
| 65 | elif mode == 'HTML_PLAIN': |
|---|
| 66 | (mode, self.suppress_tokens) = ('HTML', ['GOTOPYTHON']) |
|---|
| 67 | elif mode == 'HTML': |
|---|
| 68 | self.suppress_tokens = [] |
|---|
| 69 | else: |
|---|
| 70 | raise SyntaxError('Unknown mode: %s' % mode) |
|---|
| 71 | self.mode = mode |
|---|
| 72 | |
|---|
| 73 | def c_tokenizer( |
|---|
| 74 | self, |
|---|
| 75 | token, |
|---|
| 76 | match, |
|---|
| 77 | style, |
|---|
| 78 | ): |
|---|
| 79 | """ |
|---|
| 80 | Callback for C specific highlighting. |
|---|
| 81 | """ |
|---|
| 82 | |
|---|
| 83 | value = xmlescape(match.group(), quote=False) |
|---|
| 84 | self.change_style(token, style) |
|---|
| 85 | self.output.append(value) |
|---|
| 86 | |
|---|
| 87 | def python_tokenizer( |
|---|
| 88 | self, |
|---|
| 89 | token, |
|---|
| 90 | match, |
|---|
| 91 | style, |
|---|
| 92 | ): |
|---|
| 93 | """ |
|---|
| 94 | Callback for python specific highlighting. |
|---|
| 95 | """ |
|---|
| 96 | |
|---|
| 97 | value = xmlescape(match.group(), quote=False) |
|---|
| 98 | if token == 'MULTILINESTRING': |
|---|
| 99 | self.change_style(token, style) |
|---|
| 100 | self.output.append(value) |
|---|
| 101 | self.strMultilineString = match.group(1) |
|---|
| 102 | return 'PYTHONMultilineString' |
|---|
| 103 | elif token == 'ENDMULTILINESTRING': |
|---|
| 104 | if match.group(1) == self.strMultilineString: |
|---|
| 105 | self.output.append(value) |
|---|
| 106 | self.strMultilineString = '' |
|---|
| 107 | return 'PYTHON' |
|---|
| 108 | if style and style[:5] == 'link:': |
|---|
| 109 | self.change_style(None, None) |
|---|
| 110 | (url, style) = style[5:].split(';', 1) |
|---|
| 111 | if url == 'None' or url == '': |
|---|
| 112 | self.output.append('<span style="%s">%s</span>' |
|---|
| 113 | % (style, value)) |
|---|
| 114 | else: |
|---|
| 115 | self.output.append('<a href="%s%s" style="%s">%s</a>' |
|---|
| 116 | % (url, value, style, value)) |
|---|
| 117 | else: |
|---|
| 118 | self.change_style(token, style) |
|---|
| 119 | self.output.append(value) |
|---|
| 120 | if token == 'GOTOHTML': |
|---|
| 121 | return 'HTML' |
|---|
| 122 | return None |
|---|
| 123 | |
|---|
| 124 | def html_tokenizer( |
|---|
| 125 | self, |
|---|
| 126 | token, |
|---|
| 127 | match, |
|---|
| 128 | style, |
|---|
| 129 | ): |
|---|
| 130 | """ |
|---|
| 131 | Callback for HTML specific highlighting. |
|---|
| 132 | """ |
|---|
| 133 | |
|---|
| 134 | value = xmlescape(match.group(), quote=False) |
|---|
| 135 | self.change_style(token, style) |
|---|
| 136 | self.output.append(value) |
|---|
| 137 | if token == 'GOTOPYTHON': |
|---|
| 138 | return 'PYTHON' |
|---|
| 139 | return None |
|---|
| 140 | |
|---|
| 141 | all_styles = all_styles() |
|---|
| 142 | |
|---|
| 143 | def highlight(self, data): |
|---|
| 144 | """ |
|---|
| 145 | Syntax highlight some python code. |
|---|
| 146 | Returns html version of code. |
|---|
| 147 | """ |
|---|
| 148 | |
|---|
| 149 | i = 0 |
|---|
| 150 | mode = self.mode |
|---|
| 151 | while i < len(data): |
|---|
| 152 | for (token, o_re, style) in Highlighter.all_styles[mode][1]: |
|---|
| 153 | if token not in self.suppress_tokens: |
|---|
| 154 | match = o_re.match(data, i) |
|---|
| 155 | if match: |
|---|
| 156 | if style: |
|---|
| 157 | new_mode = \ |
|---|
| 158 | Highlighter.all_styles[mode][0](self, |
|---|
| 159 | token, match, style |
|---|
| 160 | % dict(link=self.link)) |
|---|
| 161 | else: |
|---|
| 162 | new_mode = \ |
|---|
| 163 | Highlighter.all_styles[mode][0](self, |
|---|
| 164 | token, match, style) |
|---|
| 165 | if new_mode is not None: |
|---|
| 166 | mode = new_mode |
|---|
| 167 | i += max(1, len(match.group())) |
|---|
| 168 | break |
|---|
| 169 | else: |
|---|
| 170 | self.change_style(None, None) |
|---|
| 171 | self.output.append(data[i]) |
|---|
| 172 | i += 1 |
|---|
| 173 | self.change_style(None, None) |
|---|
| 174 | return ''.join(self.output).expandtabs(4) |
|---|
| 175 | |
|---|
| 176 | def change_style(self, token, style): |
|---|
| 177 | """ |
|---|
| 178 | Generate output to change from existing style to another style only. |
|---|
| 179 | """ |
|---|
| 180 | |
|---|
| 181 | if token in self.styles: |
|---|
| 182 | style = self.styles[token] |
|---|
| 183 | if self.span_style != style: |
|---|
| 184 | if style != 'Keep': |
|---|
| 185 | if self.span_style is not None: |
|---|
| 186 | self.output.append('</span>') |
|---|
| 187 | if style is not None: |
|---|
| 188 | self.output.append('<span style="%s">' % style) |
|---|
| 189 | self.span_style = style |
|---|
| 190 | |
|---|
| 191 | |
|---|
| 192 | def _get_all_styles(cls): |
|---|
| 193 | return { |
|---|
| 194 | 'C': (cls.c_tokenizer, ( |
|---|
| 195 | ('COMMENT', re.compile(r'//.*\r?\n'), |
|---|
| 196 | 'color: green; font-style: italic'), |
|---|
| 197 | ('MULTILINECOMMENT', re.compile(r'/\*.*?\*/', re.DOTALL), |
|---|
| 198 | 'color: green; font-style: italic'), |
|---|
| 199 | ('PREPROCESSOR', re.compile(r'\s*#.*?[^\\]\s*\n', |
|---|
| 200 | re.DOTALL), 'color: magenta; font-style: italic'), |
|---|
| 201 | ('PUNC', re.compile(r'[-+*!&|^~/%\=<>\[\]{}(),.:]'), |
|---|
| 202 | 'font-weight: bold'), |
|---|
| 203 | ('NUMBER', |
|---|
| 204 | re.compile(r'0x[0-9a-fA-F]+|[+-]?\d+(\.\d+)?([eE][+-]\d+)?|\d+'), |
|---|
| 205 | 'color: red'), |
|---|
| 206 | ('KEYWORD', re.compile(r'(sizeof|int|long|short|char|void|' |
|---|
| 207 | + r'signed|unsigned|float|double|' |
|---|
| 208 | + r'goto|break|return|continue|asm|' |
|---|
| 209 | + r'case|default|if|else|switch|while|for|do|' |
|---|
| 210 | + r'struct|union|enum|typedef|' |
|---|
| 211 | + r'static|register|auto|volatile|extern|const)(?![a-zA-Z0-9_])'), |
|---|
| 212 | 'color:#185369; font-weight: bold'), |
|---|
| 213 | ('CPPKEYWORD', |
|---|
| 214 | re.compile(r'(class|private|protected|public|template|new|delete|' |
|---|
| 215 | + r'this|friend|using|inline|export|bool|throw|try|catch|' |
|---|
| 216 | + r'operator|typeid|virtual)(?![a-zA-Z0-9_])'), |
|---|
| 217 | 'color: blue; font-weight: bold'), |
|---|
| 218 | ('STRING', re.compile(r'r?u?\'(.*?)(?<!\\)\'|"(.*?)(?<!\\)"'), |
|---|
| 219 | 'color: #FF9966'), |
|---|
| 220 | ('IDENTIFIER', re.compile(r'[a-zA-Z_][a-zA-Z0-9_]*'), |
|---|
| 221 | None), |
|---|
| 222 | ('WHITESPACE', re.compile(r'[ \r\n]+'), 'Keep'), |
|---|
| 223 | )), |
|---|
| 224 | 'PYTHON': (cls.python_tokenizer, ( |
|---|
| 225 | ('GOTOHTML', re.compile(r'\}\}'), 'color: red'), |
|---|
| 226 | ('PUNC', re.compile(r'[-+*!|&^~/%\=<>\[\]{}(),.:]'), |
|---|
| 227 | 'font-weight: bold'), |
|---|
| 228 | ('NUMBER', |
|---|
| 229 | re.compile(r'0x[0-9a-fA-F]+|[+-]?\d+(\.\d+)?([eE][+-]\d+)?|\d+' |
|---|
| 230 | ), 'color: red'), |
|---|
| 231 | ('KEYWORD', |
|---|
| 232 | re.compile(r'(def|class|break|continue|del|exec|finally|pass|' |
|---|
| 233 | + r'print|raise|return|try|except|global|assert|lambda|' |
|---|
| 234 | + r'yield|for|while|if|elif|else|and|in|is|not|or|import|' |
|---|
| 235 | + r'from|True|False)(?![a-zA-Z0-9_])'), |
|---|
| 236 | 'color:#185369; font-weight: bold'), |
|---|
| 237 | ('WEB2PY', |
|---|
| 238 | re.compile(r'(request|response|session|cache|redirect|local_import|HTTP|TR|XML|URL|BEAUTIFY|A|BODY|BR|B|CAT|CENTER|CODE|COL|COLGROUP|DIV|EM|EMBED|FIELDSET|LEGEND|FORM|H1|H2|H3|H4|H5|H6|IFRAME|HEAD|HR|HTML|I|IMG|INPUT|LABEL|LI|LINK|MARKMIN|MENU|META|OBJECT|OL|ON|OPTION|P|PRE|SCRIPT|SELECT|SPAN|STYLE|TABLE|THEAD|TBODY|TFOOT|TAG|TD|TEXTAREA|TH|TITLE|TT|T|UL|XHTML|IS_SLUG|IS_STRONG|IS_LOWER|IS_UPPER|IS_ALPHANUMERIC|IS_DATETIME|IS_DATETIME_IN_RANGE|IS_DATE|IS_DATE_IN_RANGE|IS_DECIMAL_IN_RANGE|IS_EMAIL|IS_EXPR|IS_FILE|IS_FLOAT_IN_RANGE|IS_IMAGE|IS_INT_IN_RANGE|IS_IN_SET|IS_IPV4|IS_LIST_OF|IS_LENGTH|IS_MATCH|IS_EQUAL_TO|IS_EMPTY_OR|IS_NULL_OR|IS_NOT_EMPTY|IS_TIME|IS_UPLOAD_FILENAME|IS_URL|CLEANUP|CRYPT|IS_IN_DB|IS_NOT_IN_DB|DAL|Field|SQLFORM|SQLTABLE|xmlescape|embed64)(?![a-zA-Z0-9_])' |
|---|
| 239 | ), 'link:%(link)s;text-decoration:None;color:#FF5C1F;'), |
|---|
| 240 | ('MAGIC', re.compile(r'self|None'), |
|---|
| 241 | 'color:#185369; font-weight: bold'), |
|---|
| 242 | ('MULTILINESTRING', re.compile(r'r?u?(\'\'\'|""")'), |
|---|
| 243 | 'color: #FF9966'), |
|---|
| 244 | ('STRING', re.compile(r'r?u?\'(.*?)(?<!\\)\'|"(.*?)(?<!\\)"' |
|---|
| 245 | ), 'color: #FF9966'), |
|---|
| 246 | ('IDENTIFIER', re.compile(r'[a-zA-Z_][a-zA-Z0-9_]*'), |
|---|
| 247 | None), |
|---|
| 248 | ('COMMENT', re.compile(r'\#.*\r?\n'), |
|---|
| 249 | 'color: green; font-style: italic'), |
|---|
| 250 | ('WHITESPACE', re.compile(r'[ \r\n]+'), 'Keep'), |
|---|
| 251 | )), |
|---|
| 252 | 'PYTHONMultilineString': (cls.python_tokenizer, |
|---|
| 253 | (('ENDMULTILINESTRING', |
|---|
| 254 | re.compile(r'.*?("""|\'\'\')', |
|---|
| 255 | re.DOTALL), 'color: darkred'), )), |
|---|
| 256 | 'HTML': (cls.html_tokenizer, ( |
|---|
| 257 | ('GOTOPYTHON', re.compile(r'\{\{'), 'color: red'), |
|---|
| 258 | ('COMMENT', re.compile(r'<!--[^>]*-->|<!>'), |
|---|
| 259 | 'color: green; font-style: italic'), |
|---|
| 260 | ('XMLCRAP', re.compile(r'<![^>]*>'), |
|---|
| 261 | 'color: blue; font-style: italic'), |
|---|
| 262 | ('SCRIPT', re.compile(r'<script .*?</script>', re.IGNORECASE |
|---|
| 263 | + re.DOTALL), 'color: black'), |
|---|
| 264 | ('TAG', re.compile(r'</?\s*[a-zA-Z0-9]+'), |
|---|
| 265 | 'color: darkred; font-weight: bold'), |
|---|
| 266 | ('ENDTAG', re.compile(r'/?>'), |
|---|
| 267 | 'color: darkred; font-weight: bold'), |
|---|
| 268 | )), |
|---|
| 269 | } |
|---|
| 270 | |
|---|
| 271 | |
|---|
| 272 | def highlight( |
|---|
| 273 | code, |
|---|
| 274 | language, |
|---|
| 275 | link='/examples/globals/vars/', |
|---|
| 276 | counter=1, |
|---|
| 277 | styles=None, |
|---|
| 278 | highlight_line=None, |
|---|
| 279 | context_lines=None, |
|---|
| 280 | attributes=None, |
|---|
| 281 | ): |
|---|
| 282 | styles = styles or {} |
|---|
| 283 | attributes = attributes or {} |
|---|
| 284 | code_style = styles.get('CODE', None) or ''' |
|---|
| 285 | font-size: 11px; |
|---|
| 286 | font-family: Bitstream Vera Sans Mono,monospace; |
|---|
| 287 | background-color: transparent; |
|---|
| 288 | margin: 0; |
|---|
| 289 | padding: 5px; |
|---|
| 290 | border: none; |
|---|
| 291 | overflow: auto; |
|---|
| 292 | white-space: pre !important; |
|---|
| 293 | ''' |
|---|
| 294 | linenumbers_style = styles.get('LINENUMBERS', None) or ''' |
|---|
| 295 | font-size: 11px; |
|---|
| 296 | font-family: Bitstream Vera Sans Mono,monospace; |
|---|
| 297 | background-color: transparent; |
|---|
| 298 | margin: 0; |
|---|
| 299 | padding: 5px; |
|---|
| 300 | border: none; |
|---|
| 301 | color: #A0A0A0; |
|---|
| 302 | ''' |
|---|
| 303 | linehighlight_style = styles.get('LINEHIGHLIGHT', None) or \ |
|---|
| 304 | 'background-color: #EBDDE2;' |
|---|
| 305 | |
|---|
| 306 | if language and language.upper() in ['PYTHON', 'C', 'CPP', 'HTML', |
|---|
| 307 | 'WEB2PY']: |
|---|
| 308 | code = Highlighter(language, link, styles).highlight(code) |
|---|
| 309 | else: |
|---|
| 310 | code = xmlescape(code, quote=False) |
|---|
| 311 | lines = code.split('\n') |
|---|
| 312 | |
|---|
| 313 | if counter is None: |
|---|
| 314 | linenumbers = [''] * len(lines) |
|---|
| 315 | elif isinstance(counter, str): |
|---|
| 316 | linenumbers = [xmlescape(counter, quote=False)] * len(lines) |
|---|
| 317 | else: |
|---|
| 318 | linenumbers = [str(i + counter) + '.' for i in |
|---|
| 319 | xrange(len(lines))] |
|---|
| 320 | |
|---|
| 321 | if highlight_line: |
|---|
| 322 | if counter and not isinstance(counter, str): |
|---|
| 323 | lineno = highlight_line - counter |
|---|
| 324 | else: |
|---|
| 325 | lineno = highlight_line |
|---|
| 326 | if lineno < len(lines): |
|---|
| 327 | lines[lineno] = '<span style="%s">%s</span>' % ( |
|---|
| 328 | linehighlight_style, lines[lineno]) |
|---|
| 329 | linenumbers[lineno] = '<span style="%s">%s</span>' % ( |
|---|
| 330 | linehighlight_style, linenumbers[lineno]) |
|---|
| 331 | |
|---|
| 332 | if context_lines: |
|---|
| 333 | if lineno + context_lines < len(lines): |
|---|
| 334 | delslice = slice(lineno + context_lines + 1, len(lines)) |
|---|
| 335 | del lines[delslice] |
|---|
| 336 | del linenumbers[delslice] |
|---|
| 337 | if lineno - context_lines > 0: |
|---|
| 338 | delslice = slice(0, lineno - context_lines) |
|---|
| 339 | del lines[delslice] |
|---|
| 340 | del linenumbers[delslice] |
|---|
| 341 | |
|---|
| 342 | code = '<br/>'.join(lines) |
|---|
| 343 | numbers = '<br/>'.join(linenumbers) |
|---|
| 344 | |
|---|
| 345 | items = attributes.items() |
|---|
| 346 | fa = ' '.join([key[1:].lower() for (key, value) in items if key[:1] |
|---|
| 347 | == '_' and value is None] + ['%s="%s"' |
|---|
| 348 | % (key[1:].lower(), str(value).replace('"', "'")) |
|---|
| 349 | for (key, value) in items if key[:1] |
|---|
| 350 | == '_' and value]) |
|---|
| 351 | if fa: |
|---|
| 352 | fa = ' ' + fa |
|---|
| 353 | return '<table%s><tr style="vertical-align:top;">' \ |
|---|
| 354 | '<td style="min-width:40px; text-align: right;"><pre style="%s">%s</pre></td>' \ |
|---|
| 355 | '<td><pre style="%s">%s</pre></td></tr></table>' % (fa, linenumbers_style, numbers, code_style, code) |
|---|
| 356 | |
|---|
| 357 | |
|---|
| 358 | if __name__ == '__main__': |
|---|
| 359 | import sys |
|---|
| 360 | argfp = open(sys.argv[1]) |
|---|
| 361 | data = argfp.read() |
|---|
| 362 | argfp.close() |
|---|
| 363 | print('<html><body>' + highlight(data, sys.argv[2]) + '</body></html>') |
|---|
