Show More
@@ -1,310 +1,310 b'' | |||
|
1 | 1 | # -*- coding: utf-8 -*- |
|
2 | 2 | """ |
|
3 | 3 | Class and program to colorize python source code for ANSI terminals. |
|
4 | 4 | |
|
5 | 5 | Based on an HTML code highlighter by Jurgen Hermann found at: |
|
6 | 6 | http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/52298 |
|
7 | 7 | |
|
8 | 8 | Modifications by Fernando Perez (fperez@colorado.edu). |
|
9 | 9 | |
|
10 | 10 | Information on the original HTML highlighter follows: |
|
11 | 11 | |
|
12 | 12 | MoinMoin - Python Source Parser |
|
13 | 13 | |
|
14 | 14 | Title: Colorize Python source using the built-in tokenizer |
|
15 | 15 | |
|
16 | 16 | Submitter: Jurgen Hermann |
|
17 | 17 | Last Updated:2001/04/06 |
|
18 | 18 | |
|
19 | 19 | Version no:1.2 |
|
20 | 20 | |
|
21 | 21 | Description: |
|
22 | 22 | |
|
23 | 23 | This code is part of MoinMoin (http://moin.sourceforge.net/) and converts |
|
24 | 24 | Python source code to HTML markup, rendering comments, keywords, |
|
25 | 25 | operators, numeric and string literals in different colors. |
|
26 | 26 | |
|
27 | 27 | It shows how to use the built-in keyword, token and tokenize modules to |
|
28 | 28 | scan Python source code and re-emit it with no changes to its original |
|
29 | 29 | formatting (which is the hard part). |
|
30 | 30 | """ |
|
31 | 31 | from __future__ import print_function |
|
32 | 32 | |
|
33 | 33 | from __future__ import unicode_literals |
|
34 | 34 | |
|
35 | 35 | __all__ = ['ANSICodeColors','Parser'] |
|
36 | 36 | |
|
37 | 37 | _scheme_default = 'Linux' |
|
38 | 38 | |
|
39 | 39 | |
|
40 | 40 | # Imports |
|
41 | 41 | import StringIO |
|
42 | 42 | import keyword |
|
43 | 43 | import os |
|
44 | import optparse | |
|
45 | 44 | import sys |
|
46 | 45 | import token |
|
47 | 46 | import tokenize |
|
48 | 47 | |
|
49 | 48 | try: |
|
50 | 49 | generate_tokens = tokenize.generate_tokens |
|
51 | 50 | except AttributeError: |
|
52 | 51 | # Python 3. Note that we use the undocumented _tokenize because it expects |
|
53 | 52 | # strings, not bytes. See also Python issue #9969. |
|
54 | 53 | generate_tokens = tokenize._tokenize |
|
55 | 54 | |
|
56 | 55 | from IPython.utils.coloransi import * |
|
57 | 56 | |
|
58 | 57 | ############################################################################# |
|
59 | 58 | ### Python Source Parser (does Hilighting) |
|
60 | 59 | ############################################################################# |
|
61 | 60 | |
|
62 | 61 | _KEYWORD = token.NT_OFFSET + 1 |
|
63 | 62 | _TEXT = token.NT_OFFSET + 2 |
|
64 | 63 | |
|
65 | 64 | #**************************************************************************** |
|
66 | 65 | # Builtin color schemes |
|
67 | 66 | |
|
68 | 67 | Colors = TermColors # just a shorthand |
|
69 | 68 | |
|
70 | 69 | # Build a few color schemes |
|
71 | 70 | NoColor = ColorScheme( |
|
72 | 71 | 'NoColor',{ |
|
73 | 72 | token.NUMBER : Colors.NoColor, |
|
74 | 73 | token.OP : Colors.NoColor, |
|
75 | 74 | token.STRING : Colors.NoColor, |
|
76 | 75 | tokenize.COMMENT : Colors.NoColor, |
|
77 | 76 | token.NAME : Colors.NoColor, |
|
78 | 77 | token.ERRORTOKEN : Colors.NoColor, |
|
79 | 78 | |
|
80 | 79 | _KEYWORD : Colors.NoColor, |
|
81 | 80 | _TEXT : Colors.NoColor, |
|
82 | 81 | |
|
83 | 82 | 'normal' : Colors.NoColor # color off (usu. Colors.Normal) |
|
84 | 83 | } ) |
|
85 | 84 | |
|
86 | 85 | LinuxColors = ColorScheme( |
|
87 | 86 | 'Linux',{ |
|
88 | 87 | token.NUMBER : Colors.LightCyan, |
|
89 | 88 | token.OP : Colors.Yellow, |
|
90 | 89 | token.STRING : Colors.LightBlue, |
|
91 | 90 | tokenize.COMMENT : Colors.LightRed, |
|
92 | 91 | token.NAME : Colors.Normal, |
|
93 | 92 | token.ERRORTOKEN : Colors.Red, |
|
94 | 93 | |
|
95 | 94 | _KEYWORD : Colors.LightGreen, |
|
96 | 95 | _TEXT : Colors.Yellow, |
|
97 | 96 | |
|
98 | 97 | 'normal' : Colors.Normal # color off (usu. Colors.Normal) |
|
99 | 98 | } ) |
|
100 | 99 | |
|
101 | 100 | LightBGColors = ColorScheme( |
|
102 | 101 | 'LightBG',{ |
|
103 | 102 | token.NUMBER : Colors.Cyan, |
|
104 | 103 | token.OP : Colors.Blue, |
|
105 | 104 | token.STRING : Colors.Blue, |
|
106 | 105 | tokenize.COMMENT : Colors.Red, |
|
107 | 106 | token.NAME : Colors.Normal, |
|
108 | 107 | token.ERRORTOKEN : Colors.Red, |
|
109 | 108 | |
|
110 | 109 | _KEYWORD : Colors.Green, |
|
111 | 110 | _TEXT : Colors.Blue, |
|
112 | 111 | |
|
113 | 112 | 'normal' : Colors.Normal # color off (usu. Colors.Normal) |
|
114 | 113 | } ) |
|
115 | 114 | |
|
116 | 115 | # Build table of color schemes (needed by the parser) |
|
117 | 116 | ANSICodeColors = ColorSchemeTable([NoColor,LinuxColors,LightBGColors], |
|
118 | 117 | _scheme_default) |
|
119 | 118 | |
|
120 | 119 | class Parser: |
|
121 | 120 | """ Format colored Python source. |
|
122 | 121 | """ |
|
123 | 122 | |
|
124 | 123 | def __init__(self, color_table=None,out = sys.stdout): |
|
125 | 124 | """ Create a parser with a specified color table and output channel. |
|
126 | 125 | |
|
127 | 126 | Call format() to process code. |
|
128 | 127 | """ |
|
129 | 128 | self.color_table = color_table and color_table or ANSICodeColors |
|
130 | 129 | self.out = out |
|
131 | 130 | |
|
132 | 131 | def format(self, raw, out = None, scheme = ''): |
|
133 | 132 | return self.format2(raw, out, scheme)[0] |
|
134 | 133 | |
|
135 | 134 | def format2(self, raw, out = None, scheme = ''): |
|
136 | 135 | """ Parse and send the colored source. |
|
137 | 136 | |
|
138 | 137 | If out and scheme are not specified, the defaults (given to |
|
139 | 138 | constructor) are used. |
|
140 | 139 | |
|
141 | 140 | out should be a file-type object. Optionally, out can be given as the |
|
142 | 141 | string 'str' and the parser will automatically return the output in a |
|
143 | 142 | string.""" |
|
144 | 143 | |
|
145 | 144 | string_output = 0 |
|
146 | 145 | if out == 'str' or self.out == 'str' or \ |
|
147 | 146 | isinstance(self.out,StringIO.StringIO): |
|
148 | 147 | # XXX - I don't really like this state handling logic, but at this |
|
149 | 148 | # point I don't want to make major changes, so adding the |
|
150 | 149 | # isinstance() check is the simplest I can do to ensure correct |
|
151 | 150 | # behavior. |
|
152 | 151 | out_old = self.out |
|
153 | 152 | self.out = StringIO.StringIO() |
|
154 | 153 | string_output = 1 |
|
155 | 154 | elif out is not None: |
|
156 | 155 | self.out = out |
|
157 | 156 | |
|
158 | 157 | # Fast return of the unmodified input for NoColor scheme |
|
159 | 158 | if scheme == 'NoColor': |
|
160 | 159 | error = False |
|
161 | 160 | self.out.write(raw) |
|
162 | 161 | if string_output: |
|
163 | 162 | return raw,error |
|
164 | 163 | else: |
|
165 | 164 | return None,error |
|
166 | 165 | |
|
167 | 166 | # local shorthands |
|
168 | 167 | colors = self.color_table[scheme].colors |
|
169 | 168 | self.colors = colors # put in object so __call__ sees it |
|
170 | 169 | |
|
171 | 170 | # Remove trailing whitespace and normalize tabs |
|
172 | 171 | self.raw = raw.expandtabs().rstrip() |
|
173 | 172 | |
|
174 | 173 | # store line offsets in self.lines |
|
175 | 174 | self.lines = [0, 0] |
|
176 | 175 | pos = 0 |
|
177 | 176 | raw_find = self.raw.find |
|
178 | 177 | lines_append = self.lines.append |
|
179 | 178 | while 1: |
|
180 | 179 | pos = raw_find('\n', pos) + 1 |
|
181 | 180 | if not pos: break |
|
182 | 181 | lines_append(pos) |
|
183 | 182 | lines_append(len(self.raw)) |
|
184 | 183 | |
|
185 | 184 | # parse the source and write it |
|
186 | 185 | self.pos = 0 |
|
187 | 186 | text = StringIO.StringIO(self.raw) |
|
188 | 187 | |
|
189 | 188 | error = False |
|
190 | 189 | try: |
|
191 | 190 | for atoken in generate_tokens(text.readline): |
|
192 | 191 | self(*atoken) |
|
193 | 192 | except tokenize.TokenError as ex: |
|
194 | 193 | msg = ex.args[0] |
|
195 | 194 | line = ex.args[1][0] |
|
196 | 195 | self.out.write("%s\n\n*** ERROR: %s%s%s\n" % |
|
197 | 196 | (colors[token.ERRORTOKEN], |
|
198 | 197 | msg, self.raw[self.lines[line]:], |
|
199 | 198 | colors.normal) |
|
200 | 199 | ) |
|
201 | 200 | error = True |
|
202 | 201 | self.out.write(colors.normal+'\n') |
|
203 | 202 | if string_output: |
|
204 | 203 | output = self.out.getvalue() |
|
205 | 204 | self.out = out_old |
|
206 | 205 | return (output, error) |
|
207 | 206 | return (None, error) |
|
208 | 207 | |
|
209 | 208 | def __call__(self, toktype, toktext, start_pos, end_pos, line): |
|
210 | 209 | """ Token handler, with syntax highlighting.""" |
|
211 | 210 | (srow,scol) = start_pos |
|
212 | 211 | (erow,ecol) = end_pos |
|
213 | 212 | colors = self.colors |
|
214 | 213 | owrite = self.out.write |
|
215 | 214 | |
|
216 | 215 | # line separator, so this works across platforms |
|
217 | 216 | linesep = os.linesep |
|
218 | 217 | |
|
219 | 218 | # calculate new positions |
|
220 | 219 | oldpos = self.pos |
|
221 | 220 | newpos = self.lines[srow] + scol |
|
222 | 221 | self.pos = newpos + len(toktext) |
|
223 | 222 | |
|
224 | 223 | # send the original whitespace, if needed |
|
225 | 224 | if newpos > oldpos: |
|
226 | 225 | owrite(self.raw[oldpos:newpos]) |
|
227 | 226 | |
|
228 | 227 | # skip indenting tokens |
|
229 | 228 | if toktype in [token.INDENT, token.DEDENT]: |
|
230 | 229 | self.pos = newpos |
|
231 | 230 | return |
|
232 | 231 | |
|
233 | 232 | # map token type to a color group |
|
234 | 233 | if token.LPAR <= toktype and toktype <= token.OP: |
|
235 | 234 | toktype = token.OP |
|
236 | 235 | elif toktype == token.NAME and keyword.iskeyword(toktext): |
|
237 | 236 | toktype = _KEYWORD |
|
238 | 237 | color = colors.get(toktype, colors[_TEXT]) |
|
239 | 238 | |
|
240 | 239 | #print '<%s>' % toktext, # dbg |
|
241 | 240 | |
|
242 | 241 | # Triple quoted strings must be handled carefully so that backtracking |
|
243 | 242 | # in pagers works correctly. We need color terminators on _each_ line. |
|
244 | 243 | if linesep in toktext: |
|
245 | 244 | toktext = toktext.replace(linesep, '%s%s%s' % |
|
246 | 245 | (colors.normal,linesep,color)) |
|
247 | 246 | |
|
248 | 247 | # send text |
|
249 | 248 | owrite('%s%s%s' % (color,toktext,colors.normal)) |
|
250 | 249 | |
|
251 | 250 | def main(argv=None): |
|
252 | 251 | """Run as a command-line script: colorize a python file or stdin using ANSI |
|
253 | 252 | color escapes and print to stdout. |
|
254 | 253 | |
|
255 | 254 | Inputs: |
|
256 | 255 | |
|
257 | 256 | - argv(None): a list of strings like sys.argv[1:] giving the command-line |
|
258 | 257 | arguments. If None, use sys.argv[1:]. |
|
259 | 258 | """ |
|
260 | 259 | |
|
261 | 260 | usage_msg = """%prog [options] [filename] |
|
262 | 261 | |
|
263 | 262 | Colorize a python file or stdin using ANSI color escapes and print to stdout. |
|
264 | 263 | If no filename is given, or if filename is -, read standard input.""" |
|
265 | 264 | |
|
265 | import optparse | |
|
266 | 266 | parser = optparse.OptionParser(usage=usage_msg) |
|
267 | 267 | newopt = parser.add_option |
|
268 | 268 | newopt('-s','--scheme',metavar='NAME',dest='scheme_name',action='store', |
|
269 | 269 | choices=['Linux','LightBG','NoColor'],default=_scheme_default, |
|
270 | 270 | help="give the color scheme to use. Currently only 'Linux'\ |
|
271 | 271 | (default) and 'LightBG' and 'NoColor' are implemented (give without\ |
|
272 | 272 | quotes)") |
|
273 | 273 | |
|
274 | 274 | opts,args = parser.parse_args(argv) |
|
275 | 275 | |
|
276 | 276 | if len(args) > 1: |
|
277 | 277 | parser.error("you must give at most one filename.") |
|
278 | 278 | |
|
279 | 279 | if len(args) == 0: |
|
280 | 280 | fname = '-' # no filename given; setup to read from stdin |
|
281 | 281 | else: |
|
282 | 282 | fname = args[0] |
|
283 | 283 | |
|
284 | 284 | if fname == '-': |
|
285 | 285 | stream = sys.stdin |
|
286 | 286 | else: |
|
287 | 287 | try: |
|
288 | 288 | stream = open(fname) |
|
289 | 289 | except IOError as msg: |
|
290 | 290 | print(msg, file=sys.stderr) |
|
291 | 291 | sys.exit(1) |
|
292 | 292 | |
|
293 | 293 | parser = Parser() |
|
294 | 294 | |
|
295 | 295 | # we need nested try blocks because pre-2.5 python doesn't support unified |
|
296 | 296 | # try-except-finally |
|
297 | 297 | try: |
|
298 | 298 | try: |
|
299 | 299 | # write colorized version to stdout |
|
300 | 300 | parser.format(stream.read(),scheme=opts.scheme_name) |
|
301 | 301 | except IOError as msg: |
|
302 | 302 | # if user reads through a pager and quits, don't print traceback |
|
303 | 303 | if msg.args != (32,'Broken pipe'): |
|
304 | 304 | raise |
|
305 | 305 | finally: |
|
306 | 306 | if stream is not sys.stdin: |
|
307 | 307 | stream.close() # in case a non-handled exception happened above |
|
308 | 308 | |
|
309 | 309 | if __name__ == "__main__": |
|
310 | 310 | main() |
General Comments 0
You need to be logged in to leave comments.
Login now