Show More
@@ -1,743 +1,741 b'' | |||
|
1 | 1 | """Input handling and transformation machinery. |
|
2 | 2 | |
|
3 | 3 | The first class in this module, :class:`InputSplitter`, is designed to tell when |
|
4 | 4 | input from a line-oriented frontend is complete and should be executed, and when |
|
5 | 5 | the user should be prompted for another line of code instead. The name 'input |
|
6 | 6 | splitter' is largely for historical reasons. |
|
7 | 7 | |
|
8 | 8 | A companion, :class:`IPythonInputSplitter`, provides the same functionality but |
|
9 | 9 | with full support for the extended IPython syntax (magics, system calls, etc). |
|
10 | 10 | The code to actually do these transformations is in :mod:`IPython.core.inputtransformer`. |
|
11 | 11 | :class:`IPythonInputSplitter` feeds the raw code to the transformers in order |
|
12 | 12 | and stores the results. |
|
13 | 13 | |
|
14 | 14 | For more details, see the class docstrings below. |
|
15 | 15 | """ |
|
16 | 16 | |
|
17 | 17 | # Copyright (c) IPython Development Team. |
|
18 | 18 | # Distributed under the terms of the Modified BSD License. |
|
19 | 19 | import ast |
|
20 | 20 | import codeop |
|
21 | 21 | import io |
|
22 | 22 | import re |
|
23 | 23 | import sys |
|
24 | 24 | import tokenize |
|
25 | 25 | import warnings |
|
26 | 26 | |
|
27 | from IPython.utils.py3compat import cast_unicode | |
|
28 | 27 | from IPython.core.inputtransformer import (leading_indent, |
|
29 | 28 | classic_prompt, |
|
30 | 29 | ipy_prompt, |
|
31 | 30 | cellmagic, |
|
32 | 31 | assemble_logical_lines, |
|
33 | 32 | help_end, |
|
34 | 33 | escaped_commands, |
|
35 | 34 | assign_from_magic, |
|
36 | 35 | assign_from_system, |
|
37 | 36 | assemble_python_lines, |
|
38 | 37 | ) |
|
39 | 38 | |
|
40 | 39 | # These are available in this module for backwards compatibility. |
|
41 | 40 | from IPython.core.inputtransformer import (ESC_SHELL, ESC_SH_CAP, ESC_HELP, |
|
42 | 41 | ESC_HELP2, ESC_MAGIC, ESC_MAGIC2, |
|
43 | 42 | ESC_QUOTE, ESC_QUOTE2, ESC_PAREN, ESC_SEQUENCES) |
|
44 | 43 | |
|
45 | 44 | #----------------------------------------------------------------------------- |
|
46 | 45 | # Utilities |
|
47 | 46 | #----------------------------------------------------------------------------- |
|
48 | 47 | |
|
49 | 48 | # FIXME: These are general-purpose utilities that later can be moved to the |
|
50 | 49 | # general ward. Kept here for now because we're being very strict about test |
|
51 | 50 | # coverage with this code, and this lets us ensure that we keep 100% coverage |
|
52 | 51 | # while developing. |
|
53 | 52 | |
|
54 | 53 | # compiled regexps for autoindent management |
|
55 | 54 | dedent_re = re.compile('|'.join([ |
|
56 | 55 | r'^\s+raise(\s.*)?$', # raise statement (+ space + other stuff, maybe) |
|
57 | 56 | r'^\s+raise\([^\)]*\).*$', # wacky raise with immediate open paren |
|
58 | 57 | r'^\s+return(\s.*)?$', # normal return (+ space + other stuff, maybe) |
|
59 | 58 | r'^\s+return\([^\)]*\).*$', # wacky return with immediate open paren |
|
60 | 59 | r'^\s+pass\s*$', # pass (optionally followed by trailing spaces) |
|
61 | 60 | r'^\s+break\s*$', # break (optionally followed by trailing spaces) |
|
62 | 61 | r'^\s+continue\s*$', # continue (optionally followed by trailing spaces) |
|
63 | 62 | ])) |
|
64 | 63 | ini_spaces_re = re.compile(r'^([ \t\r\f\v]+)') |
|
65 | 64 | |
|
66 | 65 | # regexp to match pure comment lines so we don't accidentally insert 'if 1:' |
|
67 | 66 | # before pure comments |
|
68 | 67 | comment_line_re = re.compile('^\s*\#') |
|
69 | 68 | |
|
70 | 69 | |
|
71 | 70 | def num_ini_spaces(s): |
|
72 | 71 | """Return the number of initial spaces in a string. |
|
73 | 72 | |
|
74 | 73 | Note that tabs are counted as a single space. For now, we do *not* support |
|
75 | 74 | mixing of tabs and spaces in the user's input. |
|
76 | 75 | |
|
77 | 76 | Parameters |
|
78 | 77 | ---------- |
|
79 | 78 | s : string |
|
80 | 79 | |
|
81 | 80 | Returns |
|
82 | 81 | ------- |
|
83 | 82 | n : int |
|
84 | 83 | """ |
|
85 | 84 | |
|
86 | 85 | ini_spaces = ini_spaces_re.match(s) |
|
87 | 86 | if ini_spaces: |
|
88 | 87 | return ini_spaces.end() |
|
89 | 88 | else: |
|
90 | 89 | return 0 |
|
91 | 90 | |
|
92 | 91 | # Fake token types for partial_tokenize: |
|
93 | 92 | INCOMPLETE_STRING = tokenize.N_TOKENS |
|
94 | 93 | IN_MULTILINE_STATEMENT = tokenize.N_TOKENS + 1 |
|
95 | 94 | |
|
96 | 95 | # The 2 classes below have the same API as TokenInfo, but don't try to look up |
|
97 | 96 | # a token type name that they won't find. |
|
98 | 97 | class IncompleteString: |
|
99 | 98 | type = exact_type = INCOMPLETE_STRING |
|
100 | 99 | def __init__(self, s, start, end, line): |
|
101 | 100 | self.s = s |
|
102 | 101 | self.start = start |
|
103 | 102 | self.end = end |
|
104 | 103 | self.line = line |
|
105 | 104 | |
|
106 | 105 | class InMultilineStatement: |
|
107 | 106 | type = exact_type = IN_MULTILINE_STATEMENT |
|
108 | 107 | def __init__(self, pos, line): |
|
109 | 108 | self.s = '' |
|
110 | 109 | self.start = self.end = pos |
|
111 | 110 | self.line = line |
|
112 | 111 | |
|
113 | 112 | def partial_tokens(s): |
|
114 | 113 | """Iterate over tokens from a possibly-incomplete string of code. |
|
115 | 114 | |
|
116 | 115 | This adds two special token types: INCOMPLETE_STRING and |
|
117 | 116 | IN_MULTILINE_STATEMENT. These can only occur as the last token yielded, and |
|
118 | 117 | represent the two main ways for code to be incomplete. |
|
119 | 118 | """ |
|
120 | 119 | readline = io.StringIO(s).readline |
|
121 | 120 | token = tokenize.TokenInfo(tokenize.NEWLINE, '', (1, 0), (1, 0), '') |
|
122 | 121 | try: |
|
123 | 122 | for token in tokenize.generate_tokens(readline): |
|
124 | 123 | yield token |
|
125 | 124 | except tokenize.TokenError as e: |
|
126 | 125 | # catch EOF error |
|
127 | 126 | lines = s.splitlines(keepends=True) |
|
128 | 127 | end = len(lines), len(lines[-1]) |
|
129 | 128 | if 'multi-line string' in e.args[0]: |
|
130 | 129 | l, c = start = token.end |
|
131 | 130 | s = lines[l-1][c:] + ''.join(lines[l:]) |
|
132 | 131 | yield IncompleteString(s, start, end, lines[-1]) |
|
133 | 132 | elif 'multi-line statement' in e.args[0]: |
|
134 | 133 | yield InMultilineStatement(end, lines[-1]) |
|
135 | 134 | else: |
|
136 | 135 | raise |
|
137 | 136 | |
|
138 | 137 | def find_next_indent(code): |
|
139 | 138 | """Find the number of spaces for the next line of indentation""" |
|
140 | 139 | tokens = list(partial_tokens(code)) |
|
141 | 140 | if tokens[-1].type == tokenize.ENDMARKER: |
|
142 | 141 | tokens.pop() |
|
143 | 142 | if not tokens: |
|
144 | 143 | return 0 |
|
145 | 144 | while (tokens[-1].type in {tokenize.DEDENT, tokenize.NEWLINE, tokenize.COMMENT}): |
|
146 | 145 | tokens.pop() |
|
147 | 146 | |
|
148 | 147 | if tokens[-1].type == INCOMPLETE_STRING: |
|
149 | 148 | # Inside a multiline string |
|
150 | 149 | return 0 |
|
151 | 150 | |
|
152 | 151 | # Find the indents used before |
|
153 | 152 | prev_indents = [0] |
|
154 | 153 | def _add_indent(n): |
|
155 | 154 | if n != prev_indents[-1]: |
|
156 | 155 | prev_indents.append(n) |
|
157 | 156 | |
|
158 | 157 | tokiter = iter(tokens) |
|
159 | 158 | for tok in tokiter: |
|
160 | 159 | if tok.type in {tokenize.INDENT, tokenize.DEDENT}: |
|
161 | 160 | _add_indent(tok.end[1]) |
|
162 | 161 | elif (tok.type == tokenize.NL): |
|
163 | 162 | try: |
|
164 | 163 | _add_indent(next(tokiter).start[1]) |
|
165 | 164 | except StopIteration: |
|
166 | 165 | break |
|
167 | 166 | |
|
168 | 167 | last_indent = prev_indents.pop() |
|
169 | 168 | |
|
170 | 169 | # If we've just opened a multiline statement (e.g. 'a = ['), indent more |
|
171 | 170 | if tokens[-1].type == IN_MULTILINE_STATEMENT: |
|
172 | 171 | if tokens[-2].exact_type in {tokenize.LPAR, tokenize.LSQB, tokenize.LBRACE}: |
|
173 | 172 | return last_indent + 4 |
|
174 | 173 | return last_indent |
|
175 | 174 | |
|
176 | 175 | if tokens[-1].exact_type == tokenize.COLON: |
|
177 | 176 | # Line ends with colon - indent |
|
178 | 177 | return last_indent + 4 |
|
179 | 178 | |
|
180 | 179 | if last_indent: |
|
181 | 180 | # Examine the last line for dedent cues - statements like return or |
|
182 | 181 | # raise which normally end a block of code. |
|
183 | 182 | last_line_starts = 0 |
|
184 | 183 | for i, tok in enumerate(tokens): |
|
185 | 184 | if tok.type == tokenize.NEWLINE: |
|
186 | 185 | last_line_starts = i + 1 |
|
187 | 186 | |
|
188 | 187 | last_line_tokens = tokens[last_line_starts:] |
|
189 | 188 | names = [t.string for t in last_line_tokens if t.type == tokenize.NAME] |
|
190 | 189 | if names and names[0] in {'raise', 'return', 'pass', 'break', 'continue'}: |
|
191 | 190 | # Find the most recent indentation less than the current level |
|
192 | 191 | for indent in reversed(prev_indents): |
|
193 | 192 | if indent < last_indent: |
|
194 | 193 | return indent |
|
195 | 194 | |
|
196 | 195 | return last_indent |
|
197 | 196 | |
|
198 | 197 | |
|
199 | 198 | def last_blank(src): |
|
200 | 199 | """Determine if the input source ends in a blank. |
|
201 | 200 | |
|
202 | 201 | A blank is either a newline or a line consisting of whitespace. |
|
203 | 202 | |
|
204 | 203 | Parameters |
|
205 | 204 | ---------- |
|
206 | 205 | src : string |
|
207 | 206 | A single or multiline string. |
|
208 | 207 | """ |
|
209 | 208 | if not src: return False |
|
210 | 209 | ll = src.splitlines()[-1] |
|
211 | 210 | return (ll == '') or ll.isspace() |
|
212 | 211 | |
|
213 | 212 | |
|
214 | 213 | last_two_blanks_re = re.compile(r'\n\s*\n\s*$', re.MULTILINE) |
|
215 | 214 | last_two_blanks_re2 = re.compile(r'.+\n\s*\n\s+$', re.MULTILINE) |
|
216 | 215 | |
|
217 | 216 | def last_two_blanks(src): |
|
218 | 217 | """Determine if the input source ends in two blanks. |
|
219 | 218 | |
|
220 | 219 | A blank is either a newline or a line consisting of whitespace. |
|
221 | 220 | |
|
222 | 221 | Parameters |
|
223 | 222 | ---------- |
|
224 | 223 | src : string |
|
225 | 224 | A single or multiline string. |
|
226 | 225 | """ |
|
227 | 226 | if not src: return False |
|
228 | 227 | # The logic here is tricky: I couldn't get a regexp to work and pass all |
|
229 | 228 | # the tests, so I took a different approach: split the source by lines, |
|
230 | 229 | # grab the last two and prepend '###\n' as a stand-in for whatever was in |
|
231 | 230 | # the body before the last two lines. Then, with that structure, it's |
|
232 | 231 | # possible to analyze with two regexps. Not the most elegant solution, but |
|
233 | 232 | # it works. If anyone tries to change this logic, make sure to validate |
|
234 | 233 | # the whole test suite first! |
|
235 | 234 | new_src = '\n'.join(['###\n'] + src.splitlines()[-2:]) |
|
236 | 235 | return (bool(last_two_blanks_re.match(new_src)) or |
|
237 | 236 | bool(last_two_blanks_re2.match(new_src)) ) |
|
238 | 237 | |
|
239 | 238 | |
|
240 | 239 | def remove_comments(src): |
|
241 | 240 | """Remove all comments from input source. |
|
242 | 241 | |
|
243 | 242 | Note: comments are NOT recognized inside of strings! |
|
244 | 243 | |
|
245 | 244 | Parameters |
|
246 | 245 | ---------- |
|
247 | 246 | src : string |
|
248 | 247 | A single or multiline input string. |
|
249 | 248 | |
|
250 | 249 | Returns |
|
251 | 250 | ------- |
|
252 | 251 | String with all Python comments removed. |
|
253 | 252 | """ |
|
254 | 253 | |
|
255 | 254 | return re.sub('#.*', '', src) |
|
256 | 255 | |
|
257 | 256 | |
|
258 | 257 | def get_input_encoding(): |
|
259 | 258 | """Return the default standard input encoding. |
|
260 | 259 | |
|
261 | 260 | If sys.stdin has no encoding, 'ascii' is returned.""" |
|
262 | 261 | # There are strange environments for which sys.stdin.encoding is None. We |
|
263 | 262 | # ensure that a valid encoding is returned. |
|
264 | 263 | encoding = getattr(sys.stdin, 'encoding', None) |
|
265 | 264 | if encoding is None: |
|
266 | 265 | encoding = 'ascii' |
|
267 | 266 | return encoding |
|
268 | 267 | |
|
269 | 268 | #----------------------------------------------------------------------------- |
|
270 | 269 | # Classes and functions for normal Python syntax handling |
|
271 | 270 | #----------------------------------------------------------------------------- |
|
272 | 271 | |
|
273 | 272 | class InputSplitter(object): |
|
274 | 273 | r"""An object that can accumulate lines of Python source before execution. |
|
275 | 274 | |
|
276 | 275 | This object is designed to be fed python source line-by-line, using |
|
277 | 276 | :meth:`push`. It will return on each push whether the currently pushed |
|
278 | 277 | code could be executed already. In addition, it provides a method called |
|
279 | 278 | :meth:`push_accepts_more` that can be used to query whether more input |
|
280 | 279 | can be pushed into a single interactive block. |
|
281 | 280 | |
|
282 | 281 | This is a simple example of how an interactive terminal-based client can use |
|
283 | 282 | this tool:: |
|
284 | 283 | |
|
285 | 284 | isp = InputSplitter() |
|
286 | 285 | while isp.push_accepts_more(): |
|
287 | 286 | indent = ' '*isp.indent_spaces |
|
288 | 287 | prompt = '>>> ' + indent |
|
289 | 288 | line = indent + raw_input(prompt) |
|
290 | 289 | isp.push(line) |
|
291 | 290 | print 'Input source was:\n', isp.source_reset(), |
|
292 | 291 | """ |
|
293 | 292 | # Number of spaces of indentation computed from input that has been pushed |
|
294 | 293 | # so far. This is the attributes callers should query to get the current |
|
295 | 294 | # indentation level, in order to provide auto-indent facilities. |
|
296 | 295 | indent_spaces = 0 |
|
297 | 296 | # String, indicating the default input encoding. It is computed by default |
|
298 | 297 | # at initialization time via get_input_encoding(), but it can be reset by a |
|
299 | 298 | # client with specific knowledge of the encoding. |
|
300 | 299 | encoding = '' |
|
301 | 300 | # String where the current full source input is stored, properly encoded. |
|
302 | 301 | # Reading this attribute is the normal way of querying the currently pushed |
|
303 | 302 | # source code, that has been properly encoded. |
|
304 | 303 | source = '' |
|
305 | 304 | # Code object corresponding to the current source. It is automatically |
|
306 | 305 | # synced to the source, so it can be queried at any time to obtain the code |
|
307 | 306 | # object; it will be None if the source doesn't compile to valid Python. |
|
308 | 307 | code = None |
|
309 | 308 | |
|
310 | 309 | # Private attributes |
|
311 | 310 | |
|
312 | 311 | # List with lines of input accumulated so far |
|
313 | 312 | _buffer = None |
|
314 | 313 | # Command compiler |
|
315 | 314 | _compile = None |
|
316 | 315 | # Mark when input has changed indentation all the way back to flush-left |
|
317 | 316 | _full_dedent = False |
|
318 | 317 | # Boolean indicating whether the current block is complete |
|
319 | 318 | _is_complete = None |
|
320 | 319 | # Boolean indicating whether the current block has an unrecoverable syntax error |
|
321 | 320 | _is_invalid = False |
|
322 | 321 | |
|
323 | 322 | def __init__(self): |
|
324 | 323 | """Create a new InputSplitter instance. |
|
325 | 324 | """ |
|
326 | 325 | self._buffer = [] |
|
327 | 326 | self._compile = codeop.CommandCompiler() |
|
328 | 327 | self.encoding = get_input_encoding() |
|
329 | 328 | |
|
330 | 329 | def reset(self): |
|
331 | 330 | """Reset the input buffer and associated state.""" |
|
332 | 331 | self.indent_spaces = 0 |
|
333 | 332 | self._buffer[:] = [] |
|
334 | 333 | self.source = '' |
|
335 | 334 | self.code = None |
|
336 | 335 | self._is_complete = False |
|
337 | 336 | self._is_invalid = False |
|
338 | 337 | self._full_dedent = False |
|
339 | 338 | |
|
340 | 339 | def source_reset(self): |
|
341 | 340 | """Return the input source and perform a full reset. |
|
342 | 341 | """ |
|
343 | 342 | out = self.source |
|
344 | 343 | self.reset() |
|
345 | 344 | return out |
|
346 | 345 | |
|
347 | 346 | def check_complete(self, source): |
|
348 | 347 | """Return whether a block of code is ready to execute, or should be continued |
|
349 | 348 | |
|
350 | 349 | This is a non-stateful API, and will reset the state of this InputSplitter. |
|
351 | 350 | |
|
352 | 351 | Parameters |
|
353 | 352 | ---------- |
|
354 | 353 | source : string |
|
355 | 354 | Python input code, which can be multiline. |
|
356 | 355 | |
|
357 | 356 | Returns |
|
358 | 357 | ------- |
|
359 | 358 | status : str |
|
360 | 359 | One of 'complete', 'incomplete', or 'invalid' if source is not a |
|
361 | 360 | prefix of valid code. |
|
362 | 361 | indent_spaces : int or None |
|
363 | 362 | The number of spaces by which to indent the next line of code. If |
|
364 | 363 | status is not 'incomplete', this is None. |
|
365 | 364 | """ |
|
366 | 365 | self.reset() |
|
367 | 366 | try: |
|
368 | 367 | self.push(source) |
|
369 | 368 | except SyntaxError: |
|
370 | 369 | # Transformers in IPythonInputSplitter can raise SyntaxError, |
|
371 | 370 | # which push() will not catch. |
|
372 | 371 | return 'invalid', None |
|
373 | 372 | else: |
|
374 | 373 | if self._is_invalid: |
|
375 | 374 | return 'invalid', None |
|
376 | 375 | elif self.push_accepts_more(): |
|
377 | 376 | return 'incomplete', self.indent_spaces |
|
378 | 377 | else: |
|
379 | 378 | return 'complete', None |
|
380 | 379 | finally: |
|
381 | 380 | self.reset() |
|
382 | 381 | |
|
383 | 382 | def push(self, lines): |
|
384 | 383 | """Push one or more lines of input. |
|
385 | 384 | |
|
386 | 385 | This stores the given lines and returns a status code indicating |
|
387 | 386 | whether the code forms a complete Python block or not. |
|
388 | 387 | |
|
389 | 388 | Any exceptions generated in compilation are swallowed, but if an |
|
390 | 389 | exception was produced, the method returns True. |
|
391 | 390 | |
|
392 | 391 | Parameters |
|
393 | 392 | ---------- |
|
394 | 393 | lines : string |
|
395 | 394 | One or more lines of Python input. |
|
396 | 395 | |
|
397 | 396 | Returns |
|
398 | 397 | ------- |
|
399 | 398 | is_complete : boolean |
|
400 | 399 | True if the current input source (the result of the current input |
|
401 | 400 | plus prior inputs) forms a complete Python execution block. Note that |
|
402 | 401 | this value is also stored as a private attribute (``_is_complete``), so it |
|
403 | 402 | can be queried at any time. |
|
404 | 403 | """ |
|
405 | 404 | self._store(lines) |
|
406 | 405 | source = self.source |
|
407 | 406 | |
|
408 | 407 | # Before calling _compile(), reset the code object to None so that if an |
|
409 | 408 | # exception is raised in compilation, we don't mislead by having |
|
410 | 409 | # inconsistent code/source attributes. |
|
411 | 410 | self.code, self._is_complete = None, None |
|
412 | 411 | self._is_invalid = False |
|
413 | 412 | |
|
414 | 413 | # Honor termination lines properly |
|
415 | 414 | if source.endswith('\\\n'): |
|
416 | 415 | return False |
|
417 | 416 | |
|
418 | 417 | self._update_indent() |
|
419 | 418 | try: |
|
420 | 419 | with warnings.catch_warnings(): |
|
421 | 420 | warnings.simplefilter('error', SyntaxWarning) |
|
422 | 421 | self.code = self._compile(source, symbol="exec") |
|
423 | 422 | # Invalid syntax can produce any of a number of different errors from |
|
424 | 423 | # inside the compiler, so we have to catch them all. Syntax errors |
|
425 | 424 | # immediately produce a 'ready' block, so the invalid Python can be |
|
426 | 425 | # sent to the kernel for evaluation with possible ipython |
|
427 | 426 | # special-syntax conversion. |
|
428 | 427 | except (SyntaxError, OverflowError, ValueError, TypeError, |
|
429 | 428 | MemoryError, SyntaxWarning): |
|
430 | 429 | self._is_complete = True |
|
431 | 430 | self._is_invalid = True |
|
432 | 431 | else: |
|
433 | 432 | # Compilation didn't produce any exceptions (though it may not have |
|
434 | 433 | # given a complete code object) |
|
435 | 434 | self._is_complete = self.code is not None |
|
436 | 435 | |
|
437 | 436 | return self._is_complete |
|
438 | 437 | |
|
439 | 438 | def push_accepts_more(self): |
|
440 | 439 | """Return whether a block of interactive input can accept more input. |
|
441 | 440 | |
|
442 | 441 | This method is meant to be used by line-oriented frontends, who need to |
|
443 | 442 | guess whether a block is complete or not based solely on prior and |
|
444 | 443 | current input lines. The InputSplitter considers it has a complete |
|
445 | 444 | interactive block and will not accept more input when either: |
|
446 | 445 | |
|
447 | 446 | * A SyntaxError is raised |
|
448 | 447 | |
|
449 | 448 | * The code is complete and consists of a single line or a single |
|
450 | 449 | non-compound statement |
|
451 | 450 | |
|
452 | 451 | * The code is complete and has a blank line at the end |
|
453 | 452 | |
|
454 | 453 | If the current input produces a syntax error, this method immediately |
|
455 | 454 | returns False but does *not* raise the syntax error exception, as |
|
456 | 455 | typically clients will want to send invalid syntax to an execution |
|
457 | 456 | backend which might convert the invalid syntax into valid Python via |
|
458 | 457 | one of the dynamic IPython mechanisms. |
|
459 | 458 | """ |
|
460 | 459 | |
|
461 | 460 | # With incomplete input, unconditionally accept more |
|
462 | 461 | # A syntax error also sets _is_complete to True - see push() |
|
463 | 462 | if not self._is_complete: |
|
464 | 463 | #print("Not complete") # debug |
|
465 | 464 | return True |
|
466 | 465 | |
|
467 | 466 | # The user can make any (complete) input execute by leaving a blank line |
|
468 | 467 | last_line = self.source.splitlines()[-1] |
|
469 | 468 | if (not last_line) or last_line.isspace(): |
|
470 | 469 | #print("Blank line") # debug |
|
471 | 470 | return False |
|
472 | 471 | |
|
473 | 472 | # If there's just a single line or AST node, and we're flush left, as is |
|
474 | 473 | # the case after a simple statement such as 'a=1', we want to execute it |
|
475 | 474 | # straight away. |
|
476 | 475 | if self.indent_spaces==0: |
|
477 | 476 | if len(self.source.splitlines()) <= 1: |
|
478 | 477 | return False |
|
479 | 478 | |
|
480 | 479 | try: |
|
481 | 480 | code_ast = ast.parse(u''.join(self._buffer)) |
|
482 | 481 | except Exception: |
|
483 | 482 | #print("Can't parse AST") # debug |
|
484 | 483 | return False |
|
485 | 484 | else: |
|
486 | 485 | if len(code_ast.body) == 1 and \ |
|
487 | 486 | not hasattr(code_ast.body[0], 'body'): |
|
488 | 487 | #print("Simple statement") # debug |
|
489 | 488 | return False |
|
490 | 489 | |
|
491 | 490 | # General fallback - accept more code |
|
492 | 491 | return True |
|
493 | 492 | |
|
494 | 493 | def _update_indent(self): |
|
495 | 494 | # self.source always has a trailing newline |
|
496 | 495 | self.indent_spaces = find_next_indent(self.source[:-1]) |
|
497 | 496 | self._full_dedent = (self.indent_spaces == 0) |
|
498 | 497 | |
|
499 | 498 | def _store(self, lines, buffer=None, store='source'): |
|
500 | 499 | """Store one or more lines of input. |
|
501 | 500 | |
|
502 | 501 | If input lines are not newline-terminated, a newline is automatically |
|
503 | 502 | appended.""" |
|
504 | 503 | |
|
505 | 504 | if buffer is None: |
|
506 | 505 | buffer = self._buffer |
|
507 | 506 | |
|
508 | 507 | if lines.endswith('\n'): |
|
509 | 508 | buffer.append(lines) |
|
510 | 509 | else: |
|
511 | 510 | buffer.append(lines+'\n') |
|
512 | 511 | setattr(self, store, self._set_source(buffer)) |
|
513 | 512 | |
|
514 | 513 | def _set_source(self, buffer): |
|
515 | 514 | return u''.join(buffer) |
|
516 | 515 | |
|
517 | 516 | |
|
518 | 517 | class IPythonInputSplitter(InputSplitter): |
|
519 | 518 | """An input splitter that recognizes all of IPython's special syntax.""" |
|
520 | 519 | |
|
521 | 520 | # String with raw, untransformed input. |
|
522 | 521 | source_raw = '' |
|
523 | 522 | |
|
524 | 523 | # Flag to track when a transformer has stored input that it hasn't given |
|
525 | 524 | # back yet. |
|
526 | 525 | transformer_accumulating = False |
|
527 | 526 | |
|
528 | 527 | # Flag to track when assemble_python_lines has stored input that it hasn't |
|
529 | 528 | # given back yet. |
|
530 | 529 | within_python_line = False |
|
531 | 530 | |
|
532 | 531 | # Private attributes |
|
533 | 532 | |
|
534 | 533 | # List with lines of raw input accumulated so far. |
|
535 | 534 | _buffer_raw = None |
|
536 | 535 | |
|
537 | 536 | def __init__(self, line_input_checker=True, physical_line_transforms=None, |
|
538 | 537 | logical_line_transforms=None, python_line_transforms=None): |
|
539 | 538 | super(IPythonInputSplitter, self).__init__() |
|
540 | 539 | self._buffer_raw = [] |
|
541 | 540 | self._validate = True |
|
542 | 541 | |
|
543 | 542 | if physical_line_transforms is not None: |
|
544 | 543 | self.physical_line_transforms = physical_line_transforms |
|
545 | 544 | else: |
|
546 | 545 | self.physical_line_transforms = [ |
|
547 | 546 | leading_indent(), |
|
548 | 547 | classic_prompt(), |
|
549 | 548 | ipy_prompt(), |
|
550 | 549 | cellmagic(end_on_blank_line=line_input_checker), |
|
551 | 550 | ] |
|
552 | 551 | |
|
553 | 552 | self.assemble_logical_lines = assemble_logical_lines() |
|
554 | 553 | if logical_line_transforms is not None: |
|
555 | 554 | self.logical_line_transforms = logical_line_transforms |
|
556 | 555 | else: |
|
557 | 556 | self.logical_line_transforms = [ |
|
558 | 557 | help_end(), |
|
559 | 558 | escaped_commands(), |
|
560 | 559 | assign_from_magic(), |
|
561 | 560 | assign_from_system(), |
|
562 | 561 | ] |
|
563 | 562 | |
|
564 | 563 | self.assemble_python_lines = assemble_python_lines() |
|
565 | 564 | if python_line_transforms is not None: |
|
566 | 565 | self.python_line_transforms = python_line_transforms |
|
567 | 566 | else: |
|
568 | 567 | # We don't use any of these at present |
|
569 | 568 | self.python_line_transforms = [] |
|
570 | 569 | |
|
571 | 570 | @property |
|
572 | 571 | def transforms(self): |
|
573 | 572 | "Quick access to all transformers." |
|
574 | 573 | return self.physical_line_transforms + \ |
|
575 | 574 | [self.assemble_logical_lines] + self.logical_line_transforms + \ |
|
576 | 575 | [self.assemble_python_lines] + self.python_line_transforms |
|
577 | 576 | |
|
578 | 577 | @property |
|
579 | 578 | def transforms_in_use(self): |
|
580 | 579 | """Transformers, excluding logical line transformers if we're in a |
|
581 | 580 | Python line.""" |
|
582 | 581 | t = self.physical_line_transforms[:] |
|
583 | 582 | if not self.within_python_line: |
|
584 | 583 | t += [self.assemble_logical_lines] + self.logical_line_transforms |
|
585 | 584 | return t + [self.assemble_python_lines] + self.python_line_transforms |
|
586 | 585 | |
|
587 | 586 | def reset(self): |
|
588 | 587 | """Reset the input buffer and associated state.""" |
|
589 | 588 | super(IPythonInputSplitter, self).reset() |
|
590 | 589 | self._buffer_raw[:] = [] |
|
591 | 590 | self.source_raw = '' |
|
592 | 591 | self.transformer_accumulating = False |
|
593 | 592 | self.within_python_line = False |
|
594 | 593 | |
|
595 | 594 | for t in self.transforms: |
|
596 | 595 | try: |
|
597 | 596 | t.reset() |
|
598 | 597 | except SyntaxError: |
|
599 | 598 | # Nothing that calls reset() expects to handle transformer |
|
600 | 599 | # errors |
|
601 | 600 | pass |
|
602 | 601 | |
|
603 | 602 | def flush_transformers(self): |
|
604 | 603 | def _flush(transform, outs): |
|
605 | 604 | """yield transformed lines |
|
606 | 605 | |
|
607 | 606 | always strings, never None |
|
608 | 607 | |
|
609 | 608 | transform: the current transform |
|
610 | 609 | outs: an iterable of previously transformed inputs. |
|
611 | 610 | Each may be multiline, which will be passed |
|
612 | 611 | one line at a time to transform. |
|
613 | 612 | """ |
|
614 | 613 | for out in outs: |
|
615 | 614 | for line in out.splitlines(): |
|
616 | 615 | # push one line at a time |
|
617 | 616 | tmp = transform.push(line) |
|
618 | 617 | if tmp is not None: |
|
619 | 618 | yield tmp |
|
620 | 619 | |
|
621 | 620 | # reset the transform |
|
622 | 621 | tmp = transform.reset() |
|
623 | 622 | if tmp is not None: |
|
624 | 623 | yield tmp |
|
625 | 624 | |
|
626 | 625 | out = [] |
|
627 | 626 | for t in self.transforms_in_use: |
|
628 | 627 | out = _flush(t, out) |
|
629 | 628 | |
|
630 | 629 | out = list(out) |
|
631 | 630 | if out: |
|
632 | 631 | self._store('\n'.join(out)) |
|
633 | 632 | |
|
634 | 633 | def raw_reset(self): |
|
635 | 634 | """Return raw input only and perform a full reset. |
|
636 | 635 | """ |
|
637 | 636 | out = self.source_raw |
|
638 | 637 | self.reset() |
|
639 | 638 | return out |
|
640 | 639 | |
|
641 | 640 | def source_reset(self): |
|
642 | 641 | try: |
|
643 | 642 | self.flush_transformers() |
|
644 | 643 | return self.source |
|
645 | 644 | finally: |
|
646 | 645 | self.reset() |
|
647 | 646 | |
|
648 | 647 | def push_accepts_more(self): |
|
649 | 648 | if self.transformer_accumulating: |
|
650 | 649 | return True |
|
651 | 650 | else: |
|
652 | 651 | return super(IPythonInputSplitter, self).push_accepts_more() |
|
653 | 652 | |
|
654 | 653 | def transform_cell(self, cell): |
|
655 | 654 | """Process and translate a cell of input. |
|
656 | 655 | """ |
|
657 | 656 | self.reset() |
|
658 | 657 | try: |
|
659 | 658 | self.push(cell) |
|
660 | 659 | self.flush_transformers() |
|
661 | 660 | return self.source |
|
662 | 661 | finally: |
|
663 | 662 | self.reset() |
|
664 | 663 | |
|
665 | 664 | def push(self, lines): |
|
666 | 665 | """Push one or more lines of IPython input. |
|
667 | 666 | |
|
668 | 667 | This stores the given lines and returns a status code indicating |
|
669 | 668 | whether the code forms a complete Python block or not, after processing |
|
670 | 669 | all input lines for special IPython syntax. |
|
671 | 670 | |
|
672 | 671 | Any exceptions generated in compilation are swallowed, but if an |
|
673 | 672 | exception was produced, the method returns True. |
|
674 | 673 | |
|
675 | 674 | Parameters |
|
676 | 675 | ---------- |
|
677 | 676 | lines : string |
|
678 | 677 | One or more lines of Python input. |
|
679 | 678 | |
|
680 | 679 | Returns |
|
681 | 680 | ------- |
|
682 | 681 | is_complete : boolean |
|
683 | 682 | True if the current input source (the result of the current input |
|
684 | 683 | plus prior inputs) forms a complete Python execution block. Note that |
|
685 | 684 | this value is also stored as a private attribute (_is_complete), so it |
|
686 | 685 | can be queried at any time. |
|
687 | 686 | """ |
|
688 | 687 | |
|
689 | 688 | # We must ensure all input is pure unicode |
|
690 | lines = cast_unicode(lines, self.encoding) | |
|
691 | 689 | # ''.splitlines() --> [], but we need to push the empty line to transformers |
|
692 | 690 | lines_list = lines.splitlines() |
|
693 | 691 | if not lines_list: |
|
694 | 692 | lines_list = [''] |
|
695 | 693 | |
|
696 | 694 | # Store raw source before applying any transformations to it. Note |
|
697 | 695 | # that this must be done *after* the reset() call that would otherwise |
|
698 | 696 | # flush the buffer. |
|
699 | 697 | self._store(lines, self._buffer_raw, 'source_raw') |
|
700 | 698 | |
|
701 | 699 | for line in lines_list: |
|
702 | 700 | out = self.push_line(line) |
|
703 | 701 | |
|
704 | 702 | return out |
|
705 | 703 | |
|
706 | 704 | def push_line(self, line): |
|
707 | 705 | buf = self._buffer |
|
708 | 706 | |
|
709 | 707 | def _accumulating(dbg): |
|
710 | 708 | #print(dbg) |
|
711 | 709 | self.transformer_accumulating = True |
|
712 | 710 | return False |
|
713 | 711 | |
|
714 | 712 | for transformer in self.physical_line_transforms: |
|
715 | 713 | line = transformer.push(line) |
|
716 | 714 | if line is None: |
|
717 | 715 | return _accumulating(transformer) |
|
718 | 716 | |
|
719 | 717 | if not self.within_python_line: |
|
720 | 718 | line = self.assemble_logical_lines.push(line) |
|
721 | 719 | if line is None: |
|
722 | 720 | return _accumulating('acc logical line') |
|
723 | 721 | |
|
724 | 722 | for transformer in self.logical_line_transforms: |
|
725 | 723 | line = transformer.push(line) |
|
726 | 724 | if line is None: |
|
727 | 725 | return _accumulating(transformer) |
|
728 | 726 | |
|
729 | 727 | line = self.assemble_python_lines.push(line) |
|
730 | 728 | if line is None: |
|
731 | 729 | self.within_python_line = True |
|
732 | 730 | return _accumulating('acc python line') |
|
733 | 731 | else: |
|
734 | 732 | self.within_python_line = False |
|
735 | 733 | |
|
736 | 734 | for transformer in self.python_line_transforms: |
|
737 | 735 | line = transformer.push(line) |
|
738 | 736 | if line is None: |
|
739 | 737 | return _accumulating(transformer) |
|
740 | 738 | |
|
741 | 739 | #print("transformers clear") #debug |
|
742 | 740 | self.transformer_accumulating = False |
|
743 | 741 | return super(IPythonInputSplitter, self).push(line) |
@@ -1,139 +1,184 b'' | |||
|
1 | 1 | |
|
2 | 2 | =========================== |
|
3 | 3 | Custom input transformation |
|
4 | 4 | =========================== |
|
5 | 5 | |
|
6 | 6 | IPython extends Python syntax to allow things like magic commands, and help with |
|
7 | 7 | the ``?`` syntax. There are several ways to customise how the user's input is |
|
8 | 8 | processed into Python code to be executed. |
|
9 | 9 | |
|
10 | 10 | These hooks are mainly for other projects using IPython as the core of their |
|
11 | 11 | interactive interface. Using them carelessly can easily break IPython! |
|
12 | 12 | |
|
13 | 13 | String based transformations |
|
14 | 14 | ============================ |
|
15 | 15 | |
|
16 | 16 | .. currentmodule:: IPython.core.inputtransforms |
|
17 | 17 | |
|
18 | 18 | When the user enters a line of code, it is first processed as a string. By the |
|
19 | 19 | end of this stage, it must be valid Python syntax. |
|
20 | 20 | |
|
21 | 21 | These transformers all subclass :class:`IPython.core.inputtransformer.InputTransformer`, |
|
22 | 22 | and are used by :class:`IPython.core.inputsplitter.IPythonInputSplitter`. |
|
23 | 23 | |
|
24 | 24 | These transformers act in three groups, stored separately as lists of instances |
|
25 | 25 | in attributes of :class:`~IPython.core.inputsplitter.IPythonInputSplitter`: |
|
26 | 26 | |
|
27 | 27 | * ``physical_line_transforms`` act on the lines as the user enters them. For |
|
28 | 28 | example, these strip Python prompts from examples pasted in. |
|
29 | 29 | * ``logical_line_transforms`` act on lines as connected by explicit line |
|
30 | 30 | continuations, i.e. ``\`` at the end of physical lines. They are skipped |
|
31 | 31 | inside multiline Python statements. This is the point where IPython recognises |
|
32 | 32 | ``%magic`` commands, for instance. |
|
33 | 33 | * ``python_line_transforms`` act on blocks containing complete Python statements. |
|
34 | 34 | Multi-line strings, lists and function calls are reassembled before being |
|
35 | 35 | passed to these, but note that function and class *definitions* are still a |
|
36 | 36 | series of separate statements. IPython does not use any of these by default. |
|
37 | 37 | |
|
38 | 38 | An InteractiveShell instance actually has two |
|
39 | 39 | :class:`~IPython.core.inputsplitter.IPythonInputSplitter` instances, as the |
|
40 | 40 | attributes :attr:`~IPython.core.interactiveshell.InteractiveShell.input_splitter`, |
|
41 | 41 | to tell when a block of input is complete, and |
|
42 | 42 | :attr:`~IPython.core.interactiveshell.InteractiveShell.input_transformer_manager`, |
|
43 | 43 | to transform complete cells. If you add a transformer, you should make sure that |
|
44 | 44 | it gets added to both, e.g.:: |
|
45 | 45 | |
|
46 | 46 | ip.input_splitter.logical_line_transforms.append(my_transformer()) |
|
47 | 47 | ip.input_transformer_manager.logical_line_transforms.append(my_transformer()) |
|
48 | 48 | |
|
49 | 49 | These transformers may raise :exc:`SyntaxError` if the input code is invalid, but |
|
50 | 50 | in most cases it is clearer to pass unrecognised code through unmodified and let |
|
51 | 51 | Python's own parser decide whether it is valid. |
|
52 | 52 | |
|
53 | 53 | .. versionchanged:: 2.0 |
|
54 | 54 | |
|
55 | 55 | Added the option to raise :exc:`SyntaxError`. |
|
56 | 56 | |
|
57 | 57 | Stateless transformations |
|
58 | 58 | ------------------------- |
|
59 | 59 | |
|
60 | 60 | The simplest kind of transformations work one line at a time. Write a function |
|
61 | 61 | which takes a line and returns a line, and decorate it with |
|
62 | 62 | :meth:`StatelessInputTransformer.wrap`:: |
|
63 | 63 | |
|
64 | 64 | @StatelessInputTransformer.wrap |
|
65 | 65 | def my_special_commands(line): |
|
66 | 66 | if line.startswith("Β¬"): |
|
67 | 67 | return "specialcommand(" + repr(line) + ")" |
|
68 | 68 | return line |
|
69 | 69 | |
|
70 | 70 | The decorator returns a factory function which will produce instances of |
|
71 | 71 | :class:`~IPython.core.inputtransformer.StatelessInputTransformer` using your |
|
72 | 72 | function. |
|
73 | 73 | |
|
74 | Transforming a full block | |
|
75 | ------------------------- | |
|
76 | ||
|
77 | Transforming a full block of python code is possible by implementing a | |
|
78 | :class:`~IPython.core.inputtransformer.Inputtransformer` and overwriting the | |
|
79 | ``push`` and ``reset`` methods. The reset method should send the full block of | |
|
80 | transformed text. As an example a transformer the reversed the lines from last | |
|
81 | to first. | |
|
82 | ||
|
83 | from IPython.core.inputtransformer import InputTransformer | |
|
84 | ||
|
85 | class ReverseLineTransformer(InputTransformer): | |
|
86 | ||
|
87 | def __init__(self): | |
|
88 | self.acc = [] | |
|
89 | ||
|
90 | def push(self, line): | |
|
91 | self.acc.append(line) | |
|
92 | return None | |
|
93 | ||
|
94 | def reset(self): | |
|
95 | ret = '\n'.join(self.acc[::-1]) | |
|
96 | self.acc = [] | |
|
97 | return ret | |
|
98 | ||
|
99 | ||
|
74 | 100 | Coroutine transformers |
|
75 | 101 | ---------------------- |
|
76 | 102 | |
|
77 | 103 | More advanced transformers can be written as coroutines. The coroutine will be |
|
78 | 104 | sent each line in turn, followed by ``None`` to reset it. It can yield lines, or |
|
79 | 105 | ``None`` if it is accumulating text to yield at a later point. When reset, it |
|
80 | 106 | should give up any code it has accumulated. |
|
81 | 107 | |
|
108 | You may use :meth:`CoroutineInputTransformer.wrap` to simplify the creation of | |
|
109 | such a transformer. | |
|
110 | ||
|
111 | Here is a simple :class:`CoroutineInputTransformer` that can be though of be | |
|
112 | being the identity:: | |
|
113 | ||
|
114 | @CoroutineInputTransformer.wrap | |
|
115 | def noop(): | |
|
116 | line = '' | |
|
117 | while True: | |
|
118 | line = (yield line) | |
|
119 | ||
|
120 | ip = get_ipython() | |
|
121 | ||
|
122 | ip.input_splitter.logical_line_transforms.append(noop()) | |
|
123 | ip.input_transformer_manager.logical_line_transforms.append(noop()) | |
|
124 | ||
|
82 | 125 | This code in IPython strips a constant amount of leading indentation from each |
|
83 | 126 | line in a cell:: |
|
84 | 127 | |
|
128 | from IPython.core.inputtransformer import CoroutineInputTransformer | |
|
129 | ||
|
85 | 130 | @CoroutineInputTransformer.wrap |
|
86 | 131 |
|
|
87 | 132 | """Remove leading indentation. |
|
88 | 133 | |
|
89 | 134 | If the first line starts with a spaces or tabs, the same whitespace will be |
|
90 | 135 | removed from each following line until it is reset. |
|
91 | 136 | """ |
|
92 | 137 | space_re = re.compile(r'^[ \t]+') |
|
93 | 138 | line = '' |
|
94 | 139 | while True: |
|
95 | 140 | line = (yield line) |
|
96 | 141 | |
|
97 | 142 | if line is None: |
|
98 | 143 | continue |
|
99 | 144 | |
|
100 | 145 | m = space_re.match(line) |
|
101 | 146 | if m: |
|
102 | 147 | space = m.group(0) |
|
103 | 148 | while line is not None: |
|
104 | 149 | if line.startswith(space): |
|
105 | 150 | line = line[len(space):] |
|
106 | 151 | line = (yield line) |
|
107 | 152 | else: |
|
108 | 153 | # No leading spaces - wait for reset |
|
109 | 154 | while line is not None: |
|
110 | 155 | line = (yield line) |
|
111 | 156 | |
|
112 | 157 | |
|
113 | 158 | Token-based transformers |
|
114 | 159 | ------------------------ |
|
115 | 160 | |
|
116 | 161 | There is an experimental framework that takes care of tokenizing and |
|
117 | 162 | untokenizing lines of code. Define a function that accepts a list of tokens, and |
|
118 | 163 | returns an iterable of output tokens, and decorate it with |
|
119 | 164 | :meth:`TokenInputTransformer.wrap`. These should only be used in |
|
120 | 165 | ``python_line_transforms``. |
|
121 | 166 | |
|
122 | 167 | AST transformations |
|
123 | 168 | =================== |
|
124 | 169 | |
|
125 | 170 | After the code has been parsed as Python syntax, you can use Python's powerful |
|
126 | 171 | *Abstract Syntax Tree* tools to modify it. Subclass :class:`ast.NodeTransformer`, |
|
127 | 172 | and add an instance to ``shell.ast_transformers``. |
|
128 | 173 | |
|
129 | 174 | This example wraps integer literals in an ``Integer`` class, which is useful for |
|
130 | 175 | mathematical frameworks that want to handle e.g. ``1/3`` as a precise fraction:: |
|
131 | 176 | |
|
132 | 177 | |
|
133 | 178 | class IntegerWrapper(ast.NodeTransformer): |
|
134 | 179 | """Wraps all integers in a call to Integer()""" |
|
135 | 180 | def visit_Num(self, node): |
|
136 | 181 | if isinstance(node.n, int): |
|
137 | 182 | return ast.Call(func=ast.Name(id='Integer', ctx=ast.Load()), |
|
138 | 183 | args=[node], keywords=[]) |
|
139 | 184 | return node |
General Comments 0
You need to be logged in to leave comments.
Login now