7267
|
1 <html>
|
|
2 <head>
|
|
3 <title>PLY (Python Lex-Yacc)</title>
|
|
4 </head>
|
|
5 <body bgcolor="#ffffff">
|
|
6
|
|
7 <h1>PLY (Python Lex-Yacc)</h1>
|
|
8
|
|
9 <b>
|
|
10 David M. Beazley <br>
|
|
11 dave@dabeaz.com<br>
|
|
12 </b>
|
|
13
|
|
14 <p>
|
|
15 <b>PLY Version: 3.6</b>
|
|
16 <p>
|
|
17
|
|
18 <!-- INDEX -->
|
|
19 <div class="sectiontoc">
|
|
20 <ul>
|
|
21 <li><a href="#ply_nn1">Preface and Requirements</a>
|
|
22 <li><a href="#ply_nn1">Introduction</a>
|
|
23 <li><a href="#ply_nn2">PLY Overview</a>
|
|
24 <li><a href="#ply_nn3">Lex</a>
|
|
25 <ul>
|
|
26 <li><a href="#ply_nn4">Lex Example</a>
|
|
27 <li><a href="#ply_nn5">The tokens list</a>
|
|
28 <li><a href="#ply_nn6">Specification of tokens</a>
|
|
29 <li><a href="#ply_nn7">Token values</a>
|
|
30 <li><a href="#ply_nn8">Discarded tokens</a>
|
|
31 <li><a href="#ply_nn9">Line numbers and positional information</a>
|
|
32 <li><a href="#ply_nn10">Ignored characters</a>
|
|
33 <li><a href="#ply_nn11">Literal characters</a>
|
|
34 <li><a href="#ply_nn12">Error handling</a>
|
|
35 <li><a href="#ply_nn14">EOF Handling</a>
|
|
36 <li><a href="#ply_nn13">Building and using the lexer</a>
|
|
37 <li><a href="#ply_nn14">The @TOKEN decorator</a>
|
|
38 <li><a href="#ply_nn15">Optimized mode</a>
|
|
39 <li><a href="#ply_nn16">Debugging</a>
|
|
40 <li><a href="#ply_nn17">Alternative specification of lexers</a>
|
|
41 <li><a href="#ply_nn18">Maintaining state</a>
|
|
42 <li><a href="#ply_nn19">Lexer cloning</a>
|
|
43 <li><a href="#ply_nn20">Internal lexer state</a>
|
|
44 <li><a href="#ply_nn21">Conditional lexing and start conditions</a>
|
|
45 <li><a href="#ply_nn21">Miscellaneous Issues</a>
|
|
46 </ul>
|
|
47 <li><a href="#ply_nn22">Parsing basics</a>
|
|
48 <li><a href="#ply_nn23">Yacc</a>
|
|
49 <ul>
|
|
50 <li><a href="#ply_nn24">An example</a>
|
|
51 <li><a href="#ply_nn25">Combining Grammar Rule Functions</a>
|
|
52 <li><a href="#ply_nn26">Character Literals</a>
|
|
53 <li><a href="#ply_nn26">Empty Productions</a>
|
|
54 <li><a href="#ply_nn28">Changing the starting symbol</a>
|
|
55 <li><a href="#ply_nn27">Dealing With Ambiguous Grammars</a>
|
|
56 <li><a href="#ply_nn28">The parser.out file</a>
|
|
57 <li><a href="#ply_nn29">Syntax Error Handling</a>
|
|
58 <ul>
|
|
59 <li><a href="#ply_nn30">Recovery and resynchronization with error rules</a>
|
|
60 <li><a href="#ply_nn31">Panic mode recovery</a>
|
|
61 <li><a href="#ply_nn35">Signalling an error from a production</a>
|
|
62 <li><a href="#ply_nn38">When Do Syntax Errors Get Reported</a>
|
|
63 <li><a href="#ply_nn32">General comments on error handling</a>
|
|
64 </ul>
|
|
65 <li><a href="#ply_nn33">Line Number and Position Tracking</a>
|
|
66 <li><a href="#ply_nn34">AST Construction</a>
|
|
67 <li><a href="#ply_nn35">Embedded Actions</a>
|
|
68 <li><a href="#ply_nn36">Miscellaneous Yacc Notes</a>
|
|
69 </ul>
|
|
70 <li><a href="#ply_nn37">Multiple Parsers and Lexers</a>
|
|
71 <li><a href="#ply_nn38">Using Python's Optimized Mode</a>
|
|
72 <li><a href="#ply_nn44">Advanced Debugging</a>
|
|
73 <ul>
|
|
74 <li><a href="#ply_nn45">Debugging the lex() and yacc() commands</a>
|
|
75 <li><a href="#ply_nn46">Run-time Debugging</a>
|
|
76 </ul>
|
|
77 <li><a href="#ply_nn49">Packaging Advice</a>
|
|
78 <li><a href="#ply_nn39">Where to go from here?</a>
|
|
79 </ul>
|
|
80 </div>
|
|
81 <!-- INDEX -->
|
|
82
|
|
83
|
|
84
|
|
85
|
|
86
|
|
87 <H2><a name="ply_nn1"></a>1. Preface and Requirements</H2>
|
|
88
|
|
89
|
|
90 <p>
|
|
91 This document provides an overview of lexing and parsing with PLY.
|
|
92 Given the intrinsic complexity of parsing, I would strongly advise
|
|
93 that you read (or at least skim) this entire document before jumping
|
|
94 into a big development project with PLY.
|
|
95 </p>
|
|
96
|
|
97 <p>
|
|
98 PLY-3.5 is compatible with both Python 2 and Python 3. If you are using
|
|
99 Python 2, you have to use Python 2.6 or newer.
|
|
100 </p>
|
|
101
|
|
102 <H2><a name="ply_nn1"></a>2. Introduction</H2>
|
|
103
|
|
104
|
|
105 PLY is a pure-Python implementation of the popular compiler
|
|
106 construction tools lex and yacc. The main goal of PLY is to stay
|
|
107 fairly faithful to the way in which traditional lex/yacc tools work.
|
|
108 This includes supporting LALR(1) parsing as well as providing
|
|
109 extensive input validation, error reporting, and diagnostics. Thus,
|
|
110 if you've used yacc in another programming language, it should be
|
|
111 relatively straightforward to use PLY.
|
|
112
|
|
113 <p>
|
|
114 Early versions of PLY were developed to support an Introduction to
|
|
115 Compilers Course I taught in 2001 at the University of Chicago.
|
|
116 Since PLY was primarily developed as an instructional tool, you will
|
|
117 find it to be fairly picky about token and grammar rule
|
|
118 specification. In part, this
|
|
119 added formality is meant to catch common programming mistakes made by
|
|
120 novice users. However, advanced users will also find such features to
|
|
121 be useful when building complicated grammars for real programming
|
|
122 languages. It should also be noted that PLY does not provide much in
|
|
123 the way of bells and whistles (e.g., automatic construction of
|
|
124 abstract syntax trees, tree traversal, etc.). Nor would I consider it
|
|
125 to be a parsing framework. Instead, you will find a bare-bones, yet
|
|
126 fully capable lex/yacc implementation written entirely in Python.
|
|
127
|
|
128 <p>
|
|
129 The rest of this document assumes that you are somewhat familiar with
|
|
130 parsing theory, syntax directed translation, and the use of compiler
|
|
131 construction tools such as lex and yacc in other programming
|
|
132 languages. If you are unfamiliar with these topics, you will probably
|
|
133 want to consult an introductory text such as "Compilers: Principles,
|
|
134 Techniques, and Tools", by Aho, Sethi, and Ullman. O'Reilly's "Lex
|
|
135 and Yacc" by John Levine may also be handy. In fact, the O'Reilly book can be
|
|
136 used as a reference for PLY as the concepts are virtually identical.
|
|
137
|
|
138 <H2><a name="ply_nn2"></a>3. PLY Overview</H2>
|
|
139
|
|
140
|
|
141 <p>
|
|
142 PLY consists of two separate modules; <tt>lex.py</tt> and
|
|
143 <tt>yacc.py</tt>, both of which are found in a Python package
|
|
144 called <tt>ply</tt>. The <tt>lex.py</tt> module is used to break input text into a
|
|
145 collection of tokens specified by a collection of regular expression
|
|
146 rules. <tt>yacc.py</tt> is used to recognize language syntax that has
|
|
147 been specified in the form of a context free grammar.
|
|
148 </p>
|
|
149
|
|
150 <p>
|
|
151 The two tools are meant to work together. Specifically,
|
|
152 <tt>lex.py</tt> provides an external interface in the form of a
|
|
153 <tt>token()</tt> function that returns the next valid token on the
|
|
154 input stream. <tt>yacc.py</tt> calls this repeatedly to retrieve
|
|
155 tokens and invoke grammar rules. The output of <tt>yacc.py</tt> is
|
|
156 often an Abstract Syntax Tree (AST). However, this is entirely up to
|
|
157 the user. If desired, <tt>yacc.py</tt> can also be used to implement
|
|
158 simple one-pass compilers.
|
|
159
|
|
160 <p>
|
|
161 Like its Unix counterpart, <tt>yacc.py</tt> provides most of the
|
|
162 features you expect including extensive error checking, grammar
|
|
163 validation, support for empty productions, error tokens, and ambiguity
|
|
164 resolution via precedence rules. In fact, almost everything that is possible in traditional yacc
|
|
165 should be supported in PLY.
|
|
166
|
|
167 <p>
|
|
168 The primary difference between
|
|
169 <tt>yacc.py</tt> and Unix <tt>yacc</tt> is that <tt>yacc.py</tt>
|
|
170 doesn't involve a separate code-generation process.
|
|
171 Instead, PLY relies on reflection (introspection)
|
|
172 to build its lexers and parsers. Unlike traditional lex/yacc which
|
|
173 require a special input file that is converted into a separate source
|
|
174 file, the specifications given to PLY <em>are</em> valid Python
|
|
175 programs. This means that there are no extra source files nor is
|
|
176 there a special compiler construction step (e.g., running yacc to
|
|
177 generate Python code for the compiler). Since the generation of the
|
|
178 parsing tables is relatively expensive, PLY caches the results and
|
|
179 saves them to a file. If no changes are detected in the input source,
|
|
180 the tables are read from the cache. Otherwise, they are regenerated.
|
|
181
|
|
182 <H2><a name="ply_nn3"></a>4. Lex</H2>
|
|
183
|
|
184
|
|
185 <tt>lex.py</tt> is used to tokenize an input string. For example, suppose
|
|
186 you're writing a programming language and a user supplied the following input string:
|
|
187
|
|
188 <blockquote>
|
|
189 <pre>
|
|
190 x = 3 + 42 * (s - t)
|
|
191 </pre>
|
|
192 </blockquote>
|
|
193
|
|
194 A tokenizer splits the string into individual tokens
|
|
195
|
|
196 <blockquote>
|
|
197 <pre>
|
|
198 'x','=', '3', '+', '42', '*', '(', 's', '-', 't', ')'
|
|
199 </pre>
|
|
200 </blockquote>
|
|
201
|
|
202 Tokens are usually given names to indicate what they are. For example:
|
|
203
|
|
204 <blockquote>
|
|
205 <pre>
|
|
206 'ID','EQUALS','NUMBER','PLUS','NUMBER','TIMES',
|
|
207 'LPAREN','ID','MINUS','ID','RPAREN'
|
|
208 </pre>
|
|
209 </blockquote>
|
|
210
|
|
211 More specifically, the input is broken into pairs of token types and values. For example:
|
|
212
|
|
213 <blockquote>
|
|
214 <pre>
|
|
215 ('ID','x'), ('EQUALS','='), ('NUMBER','3'),
|
|
216 ('PLUS','+'), ('NUMBER','42), ('TIMES','*'),
|
|
217 ('LPAREN','('), ('ID','s'), ('MINUS','-'),
|
|
218 ('ID','t'), ('RPAREN',')'
|
|
219 </pre>
|
|
220 </blockquote>
|
|
221
|
|
222 The identification of tokens is typically done by writing a series of regular expression
|
|
223 rules. The next section shows how this is done using <tt>lex.py</tt>.
|
|
224
|
|
225 <H3><a name="ply_nn4"></a>4.1 Lex Example</H3>
|
|
226
|
|
227
|
|
228 The following example shows how <tt>lex.py</tt> is used to write a simple tokenizer.
|
|
229
|
|
230 <blockquote>
|
|
231 <pre>
|
|
232 # ------------------------------------------------------------
|
|
233 # calclex.py
|
|
234 #
|
|
235 # tokenizer for a simple expression evaluator for
|
|
236 # numbers and +,-,*,/
|
|
237 # ------------------------------------------------------------
|
|
238 import ply.lex as lex
|
|
239
|
|
240 # List of token names. This is always required
|
|
241 tokens = (
|
|
242 'NUMBER',
|
|
243 'PLUS',
|
|
244 'MINUS',
|
|
245 'TIMES',
|
|
246 'DIVIDE',
|
|
247 'LPAREN',
|
|
248 'RPAREN',
|
|
249 )
|
|
250
|
|
251 # Regular expression rules for simple tokens
|
|
252 t_PLUS = r'\+'
|
|
253 t_MINUS = r'-'
|
|
254 t_TIMES = r'\*'
|
|
255 t_DIVIDE = r'/'
|
|
256 t_LPAREN = r'\('
|
|
257 t_RPAREN = r'\)'
|
|
258
|
|
259 # A regular expression rule with some action code
|
|
260 def t_NUMBER(t):
|
|
261 r'\d+'
|
|
262 t.value = int(t.value)
|
|
263 return t
|
|
264
|
|
265 # Define a rule so we can track line numbers
|
|
266 def t_newline(t):
|
|
267 r'\n+'
|
|
268 t.lexer.lineno += len(t.value)
|
|
269
|
|
270 # A string containing ignored characters (spaces and tabs)
|
|
271 t_ignore = ' \t'
|
|
272
|
|
273 # Error handling rule
|
|
274 def t_error(t):
|
|
275 print("Illegal character '%s'" % t.value[0])
|
|
276 t.lexer.skip(1)
|
|
277
|
|
278 # Build the lexer
|
|
279 lexer = lex.lex()
|
|
280
|
|
281 </pre>
|
|
282 </blockquote>
|
|
283 To use the lexer, you first need to feed it some input text using
|
|
284 its <tt>input()</tt> method. After that, repeated calls
|
|
285 to <tt>token()</tt> produce tokens. The following code shows how this
|
|
286 works:
|
|
287
|
|
288 <blockquote>
|
|
289 <pre>
|
|
290
|
|
291 # Test it out
|
|
292 data = '''
|
|
293 3 + 4 * 10
|
|
294 + -20 *2
|
|
295 '''
|
|
296
|
|
297 # Give the lexer some input
|
|
298 lexer.input(data)
|
|
299
|
|
300 # Tokenize
|
|
301 while True:
|
|
302 tok = lexer.token()
|
|
303 if not tok:
|
|
304 break # No more input
|
|
305 print(tok)
|
|
306 </pre>
|
|
307 </blockquote>
|
|
308
|
|
309 When executed, the example will produce the following output:
|
|
310
|
|
311 <blockquote>
|
|
312 <pre>
|
|
313 $ python example.py
|
|
314 LexToken(NUMBER,3,2,1)
|
|
315 LexToken(PLUS,'+',2,3)
|
|
316 LexToken(NUMBER,4,2,5)
|
|
317 LexToken(TIMES,'*',2,7)
|
|
318 LexToken(NUMBER,10,2,10)
|
|
319 LexToken(PLUS,'+',3,14)
|
|
320 LexToken(MINUS,'-',3,16)
|
|
321 LexToken(NUMBER,20,3,18)
|
|
322 LexToken(TIMES,'*',3,20)
|
|
323 LexToken(NUMBER,2,3,21)
|
|
324 </pre>
|
|
325 </blockquote>
|
|
326
|
|
327 Lexers also support the iteration protocol. So, you can write the above loop as follows:
|
|
328
|
|
329 <blockquote>
|
|
330 <pre>
|
|
331 for tok in lexer:
|
|
332 print(tok)
|
|
333 </pre>
|
|
334 </blockquote>
|
|
335
|
|
336 The tokens returned by <tt>lexer.token()</tt> are instances
|
|
337 of <tt>LexToken</tt>. This object has
|
|
338 attributes <tt>tok.type</tt>, <tt>tok.value</tt>,
|
|
339 <tt>tok.lineno</tt>, and <tt>tok.lexpos</tt>. The following code shows an example of
|
|
340 accessing these attributes:
|
|
341
|
|
342 <blockquote>
|
|
343 <pre>
|
|
344 # Tokenize
|
|
345 while True:
|
|
346 tok = lexer.token()
|
|
347 if not tok:
|
|
348 break # No more input
|
|
349 print(tok.type, tok.value, tok.lineno, tok.lexpos)
|
|
350 </pre>
|
|
351 </blockquote>
|
|
352
|
|
353 The <tt>tok.type</tt> and <tt>tok.value</tt> attributes contain the
|
|
354 type and value of the token itself.
|
|
355 <tt>tok.line</tt> and <tt>tok.lexpos</tt> contain information about
|
|
356 the location of the token. <tt>tok.lexpos</tt> is the index of the
|
|
357 token relative to the start of the input text.
|
|
358
|
|
359 <H3><a name="ply_nn5"></a>4.2 The tokens list</H3>
|
|
360
|
|
361
|
|
362 <p>
|
|
363 All lexers must provide a list <tt>tokens</tt> that defines all of the possible token
|
|
364 names that can be produced by the lexer. This list is always required
|
|
365 and is used to perform a variety of validation checks. The tokens list is also used by the
|
|
366 <tt>yacc.py</tt> module to identify terminals.
|
|
367 </p>
|
|
368
|
|
369 <p>
|
|
370 In the example, the following code specified the token names:
|
|
371
|
|
372 <blockquote>
|
|
373 <pre>
|
|
374 tokens = (
|
|
375 'NUMBER',
|
|
376 'PLUS',
|
|
377 'MINUS',
|
|
378 'TIMES',
|
|
379 'DIVIDE',
|
|
380 'LPAREN',
|
|
381 'RPAREN',
|
|
382 )
|
|
383 </pre>
|
|
384 </blockquote>
|
|
385
|
|
386 <H3><a name="ply_nn6"></a>4.3 Specification of tokens</H3>
|
|
387
|
|
388
|
|
389 Each token is specified by writing a regular expression rule compatible with Python's <tt>re</tt> module. Each of these rules
|
|
390 are defined by making declarations with a special prefix <tt>t_</tt> to indicate that it
|
|
391 defines a token. For simple tokens, the regular expression can
|
|
392 be specified as strings such as this (note: Python raw strings are used since they are the
|
|
393 most convenient way to write regular expression strings):
|
|
394
|
|
395 <blockquote>
|
|
396 <pre>
|
|
397 t_PLUS = r'\+'
|
|
398 </pre>
|
|
399 </blockquote>
|
|
400
|
|
401 In this case, the name following the <tt>t_</tt> must exactly match one of the
|
|
402 names supplied in <tt>tokens</tt>. If some kind of action needs to be performed,
|
|
403 a token rule can be specified as a function. For example, this rule matches numbers and
|
|
404 converts the string into a Python integer.
|
|
405
|
|
406 <blockquote>
|
|
407 <pre>
|
|
408 def t_NUMBER(t):
|
|
409 r'\d+'
|
|
410 t.value = int(t.value)
|
|
411 return t
|
|
412 </pre>
|
|
413 </blockquote>
|
|
414
|
|
415 When a function is used, the regular expression rule is specified in the function documentation string.
|
|
416 The function always takes a single argument which is an instance of
|
|
417 <tt>LexToken</tt>. This object has attributes of <tt>t.type</tt> which is the token type (as a string),
|
|
418 <tt>t.value</tt> which is the lexeme (the actual text matched), <tt>t.lineno</tt> which is the current line number, and <tt>t.lexpos</tt> which
|
|
419 is the position of the token relative to the beginning of the input text.
|
|
420 By default, <tt>t.type</tt> is set to the name following the <tt>t_</tt> prefix. The action
|
|
421 function can modify the contents of the <tt>LexToken</tt> object as appropriate. However,
|
|
422 when it is done, the resulting token should be returned. If no value is returned by the action
|
|
423 function, the token is simply discarded and the next token read.
|
|
424
|
|
425 <p>
|
|
426 Internally, <tt>lex.py</tt> uses the <tt>re</tt> module to do its pattern matching. Patterns are compiled
|
|
427 using the <tt>re.VERBOSE</tt> flag which can be used to help readability. However, be aware that unescaped
|
|
428 whitespace is ignored and comments are allowed in this mode. If your pattern involves whitespace, make sure you
|
|
429 use <tt>\s</tt>. If you need to match the <tt>#</tt> character, use <tt>[#]</tt>.
|
|
430 </p>
|
|
431
|
|
432 <p>
|
|
433 When building the master regular expression,
|
|
434 rules are added in the following order:
|
|
435 </p>
|
|
436
|
|
437 <p>
|
|
438 <ol>
|
|
439 <li>All tokens defined by functions are added in the same order as they appear in the lexer file.
|
|
440 <li>Tokens defined by strings are added next by sorting them in order of decreasing regular expression length (longer expressions
|
|
441 are added first).
|
|
442 </ol>
|
|
443 <p>
|
|
444 Without this ordering, it can be difficult to correctly match certain types of tokens. For example, if you
|
|
445 wanted to have separate tokens for "=" and "==", you need to make sure that "==" is checked first. By sorting regular
|
|
446 expressions in order of decreasing length, this problem is solved for rules defined as strings. For functions,
|
|
447 the order can be explicitly controlled since rules appearing first are checked first.
|
|
448
|
|
449 <p>
|
|
450 To handle reserved words, you should write a single rule to match an
|
|
451 identifier and do a special name lookup in a function like this:
|
|
452
|
|
453 <blockquote>
|
|
454 <pre>
|
|
455 reserved = {
|
|
456 'if' : 'IF',
|
|
457 'then' : 'THEN',
|
|
458 'else' : 'ELSE',
|
|
459 'while' : 'WHILE',
|
|
460 ...
|
|
461 }
|
|
462
|
|
463 tokens = ['LPAREN','RPAREN',...,'ID'] + list(reserved.values())
|
|
464
|
|
465 def t_ID(t):
|
|
466 r'[a-zA-Z_][a-zA-Z_0-9]*'
|
|
467 t.type = reserved.get(t.value,'ID') # Check for reserved words
|
|
468 return t
|
|
469 </pre>
|
|
470 </blockquote>
|
|
471
|
|
472 This approach greatly reduces the number of regular expression rules and is likely to make things a little faster.
|
|
473
|
|
474 <p>
|
|
475 <b>Note:</b> You should avoid writing individual rules for reserved words. For example, if you write rules like this,
|
|
476
|
|
477 <blockquote>
|
|
478 <pre>
|
|
479 t_FOR = r'for'
|
|
480 t_PRINT = r'print'
|
|
481 </pre>
|
|
482 </blockquote>
|
|
483
|
|
484 those rules will be triggered for identifiers that include those words as a prefix such as "forget" or "printed". This is probably not
|
|
485 what you want.
|
|
486
|
|
487 <H3><a name="ply_nn7"></a>4.4 Token values</H3>
|
|
488
|
|
489
|
|
490 When tokens are returned by lex, they have a value that is stored in the <tt>value</tt> attribute. Normally, the value is the text
|
|
491 that was matched. However, the value can be assigned to any Python object. For instance, when lexing identifiers, you may
|
|
492 want to return both the identifier name and information from some sort of symbol table. To do this, you might write a rule like this:
|
|
493
|
|
494 <blockquote>
|
|
495 <pre>
|
|
496 def t_ID(t):
|
|
497 ...
|
|
498 # Look up symbol table information and return a tuple
|
|
499 t.value = (t.value, symbol_lookup(t.value))
|
|
500 ...
|
|
501 return t
|
|
502 </pre>
|
|
503 </blockquote>
|
|
504
|
|
505 It is important to note that storing data in other attribute names is <em>not</em> recommended. The <tt>yacc.py</tt> module only exposes the
|
|
506 contents of the <tt>value</tt> attribute. Thus, accessing other attributes may be unnecessarily awkward. If you
|
|
507 need to store multiple values on a token, assign a tuple, dictionary, or instance to <tt>value</tt>.
|
|
508
|
|
509 <H3><a name="ply_nn8"></a>4.5 Discarded tokens</H3>
|
|
510
|
|
511
|
|
512 To discard a token, such as a comment, simply define a token rule that returns no value. For example:
|
|
513
|
|
514 <blockquote>
|
|
515 <pre>
|
|
516 def t_COMMENT(t):
|
|
517 r'\#.*'
|
|
518 pass
|
|
519 # No return value. Token discarded
|
|
520 </pre>
|
|
521 </blockquote>
|
|
522
|
|
523 Alternatively, you can include the prefix "ignore_" in the token declaration to force a token to be ignored. For example:
|
|
524
|
|
525 <blockquote>
|
|
526 <pre>
|
|
527 t_ignore_COMMENT = r'\#.*'
|
|
528 </pre>
|
|
529 </blockquote>
|
|
530
|
|
531 Be advised that if you are ignoring many different kinds of text, you may still want to use functions since these provide more precise
|
|
532 control over the order in which regular expressions are matched (i.e., functions are matched in order of specification whereas strings are
|
|
533 sorted by regular expression length).
|
|
534
|
|
535 <H3><a name="ply_nn9"></a>4.6 Line numbers and positional information</H3>
|
|
536
|
|
537
|
|
538 <p>By default, <tt>lex.py</tt> knows nothing about line numbers. This is because <tt>lex.py</tt> doesn't know anything
|
|
539 about what constitutes a "line" of input (e.g., the newline character or even if the input is textual data).
|
|
540 To update this information, you need to write a special rule. In the example, the <tt>t_newline()</tt> rule shows how to do this.
|
|
541
|
|
542 <blockquote>
|
|
543 <pre>
|
|
544 # Define a rule so we can track line numbers
|
|
545 def t_newline(t):
|
|
546 r'\n+'
|
|
547 t.lexer.lineno += len(t.value)
|
|
548 </pre>
|
|
549 </blockquote>
|
|
550 Within the rule, the <tt>lineno</tt> attribute of the underlying lexer <tt>t.lexer</tt> is updated.
|
|
551 After the line number is updated, the token is simply discarded since nothing is returned.
|
|
552
|
|
553 <p>
|
|
554 <tt>lex.py</tt> does not perform and kind of automatic column tracking. However, it does record positional
|
|
555 information related to each token in the <tt>lexpos</tt> attribute. Using this, it is usually possible to compute
|
|
556 column information as a separate step. For instance, just count backwards until you reach a newline.
|
|
557
|
|
558 <blockquote>
|
|
559 <pre>
|
|
560 # Compute column.
|
|
561 # input is the input text string
|
|
562 # token is a token instance
|
|
563 def find_column(input,token):
|
|
564 last_cr = input.rfind('\n',0,token.lexpos)
|
|
565 if last_cr < 0:
|
|
566 last_cr = 0
|
|
567 column = (token.lexpos - last_cr) + 1
|
|
568 return column
|
|
569 </pre>
|
|
570 </blockquote>
|
|
571
|
|
572 Since column information is often only useful in the context of error handling, calculating the column
|
|
573 position can be performed when needed as opposed to doing it for each token.
|
|
574
|
|
575 <H3><a name="ply_nn10"></a>4.7 Ignored characters</H3>
|
|
576
|
|
577
|
|
578 <p>
|
|
579 The special <tt>t_ignore</tt> rule is reserved by <tt>lex.py</tt> for characters
|
|
580 that should be completely ignored in the input stream.
|
|
581 Usually this is used to skip over whitespace and other non-essential characters.
|
|
582 Although it is possible to define a regular expression rule for whitespace in a manner
|
|
583 similar to <tt>t_newline()</tt>, the use of <tt>t_ignore</tt> provides substantially better
|
|
584 lexing performance because it is handled as a special case and is checked in a much
|
|
585 more efficient manner than the normal regular expression rules.
|
|
586 </p>
|
|
587
|
|
588 <p>
|
|
589 The characters given in <tt>t_ignore</tt> are not ignored when such characters are part of
|
|
590 other regular expression patterns. For example, if you had a rule to capture quoted text,
|
|
591 that pattern can include the ignored characters (which will be captured in the normal way). The
|
|
592 main purpose of <tt>t_ignore</tt> is to ignore whitespace and other padding between the
|
|
593 tokens that you actually want to parse.
|
|
594 </p>
|
|
595
|
|
596 <H3><a name="ply_nn11"></a>4.8 Literal characters</H3>
|
|
597
|
|
598
|
|
599 <p>
|
|
600 Literal characters can be specified by defining a variable <tt>literals</tt> in your lexing module. For example:
|
|
601
|
|
602 <blockquote>
|
|
603 <pre>
|
|
604 literals = [ '+','-','*','/' ]
|
|
605 </pre>
|
|
606 </blockquote>
|
|
607
|
|
608 or alternatively
|
|
609
|
|
610 <blockquote>
|
|
611 <pre>
|
|
612 literals = "+-*/"
|
|
613 </pre>
|
|
614 </blockquote>
|
|
615
|
|
616 A literal character is simply a single character that is returned "as is" when encountered by the lexer. Literals are checked
|
|
617 after all of the defined regular expression rules. Thus, if a rule starts with one of the literal characters, it will always
|
|
618 take precedence.
|
|
619
|
|
620 <p>
|
|
621 When a literal token is returned, both its <tt>type</tt> and <tt>value</tt> attributes are set to the character itself. For example, <tt>'+'</tt>.
|
|
622 </p>
|
|
623
|
|
624 <p>
|
|
625 It's possible to write token functions that perform additional actions
|
|
626 when literals are matched. However, you'll need to set the token type
|
|
627 appropriately. For example:
|
|
628 </p>
|
|
629
|
|
630 <blockquote>
|
|
631 <pre>
|
|
632 literals = [ '{', '}' ]
|
|
633
|
|
634 def t_lbrace(t):
|
|
635 r'\{'
|
|
636 t.type = '{' # Set token type to the expected literal
|
|
637 return t
|
|
638
|
|
639 def t_rbrace(t):
|
|
640 r'\}'
|
|
641 t.type = '}' # Set token type to the expected literal
|
|
642 return t
|
|
643 </pre>
|
|
644 </blockquote>
|
|
645
|
|
646 <H3><a name="ply_nn12"></a>4.9 Error handling</H3>
|
|
647
|
|
648
|
|
649 <p>
|
|
650 The <tt>t_error()</tt>
|
|
651 function is used to handle lexing errors that occur when illegal
|
|
652 characters are detected. In this case, the <tt>t.value</tt> attribute contains the
|
|
653 rest of the input string that has not been tokenized. In the example, the error function
|
|
654 was defined as follows:
|
|
655
|
|
656 <blockquote>
|
|
657 <pre>
|
|
658 # Error handling rule
|
|
659 def t_error(t):
|
|
660 print("Illegal character '%s'" % t.value[0])
|
|
661 t.lexer.skip(1)
|
|
662 </pre>
|
|
663 </blockquote>
|
|
664
|
|
665 In this case, we simply print the offending character and skip ahead one character by calling <tt>t.lexer.skip(1)</tt>.
|
|
666
|
|
667 <H3><a name="ply_nn14"></a>4.10 EOF Handling</H3>
|
|
668
|
|
669
|
|
670 <p>
|
|
671 The <tt>t_eof()</tt> function is used to handle an end-of-file (EOF) condition in the input. As input, it
|
|
672 receives a token type <tt>'eof'</tt> with the <tt>lineno</tt> and <tt>lexpos</tt> attributes set appropriately.
|
|
673 The main use of this function is provide more input to the lexer so that it can continue to parse. Here is an
|
|
674 example of how this works:
|
|
675 </p>
|
|
676
|
|
677 <blockquote>
|
|
678 <pre>
|
|
679 # EOF handling rule
|
|
680 def t_eof(t):
|
|
681 # Get more input (Example)
|
|
682 more = raw_input('... ')
|
|
683 if more:
|
|
684 self.lexer.input(more)
|
|
685 return self.lexer.token()
|
|
686 return None
|
|
687 </pre>
|
|
688 </blockquote>
|
|
689
|
|
690 <p>
|
|
691 The EOF function should return the next available token (by calling <tt>self.lexer.token())</tt> or <tt>None</tt> to
|
|
692 indicate no more data. Be aware that setting more input with the <tt>self.lexer.input()</tt> method does
|
|
693 NOT reset the lexer state or the <tt>lineno</tt> attribute used for position tracking. The <tt>lexpos</tt>
|
|
694 attribute is reset so be aware of that if you're using it in error reporting.
|
|
695 </p>
|
|
696
|
|
697 <H3><a name="ply_nn13"></a>4.11 Building and using the lexer</H3>
|
|
698
|
|
699
|
|
700 <p>
|
|
701 To build the lexer, the function <tt>lex.lex()</tt> is used. For example:</p>
|
|
702
|
|
703 <blockquote>
|
|
704 <pre>
|
|
705 lexer = lex.lex()
|
|
706 </pre>
|
|
707 </blockquote>
|
|
708
|
|
709 <p>This function
|
|
710 uses Python reflection (or introspection) to read the regular expression rules
|
|
711 out of the calling context and build the lexer. Once the lexer has been built, two methods can
|
|
712 be used to control the lexer.
|
|
713 </p>
|
|
714 <ul>
|
|
715 <li><tt>lexer.input(data)</tt>. Reset the lexer and store a new input string.
|
|
716 <li><tt>lexer.token()</tt>. Return the next token. Returns a special <tt>LexToken</tt> instance on success or
|
|
717 None if the end of the input text has been reached.
|
|
718 </ul>
|
|
719
|
|
720 <H3><a name="ply_nn14"></a>4.12 The @TOKEN decorator</H3>
|
|
721
|
|
722
|
|
723 In some applications, you may want to define build tokens from as a series of
|
|
724 more complex regular expression rules. For example:
|
|
725
|
|
726 <blockquote>
|
|
727 <pre>
|
|
728 digit = r'([0-9])'
|
|
729 nondigit = r'([_A-Za-z])'
|
|
730 identifier = r'(' + nondigit + r'(' + digit + r'|' + nondigit + r')*)'
|
|
731
|
|
732 def t_ID(t):
|
|
733 # want docstring to be identifier above. ?????
|
|
734 ...
|
|
735 </pre>
|
|
736 </blockquote>
|
|
737
|
|
738 In this case, we want the regular expression rule for <tt>ID</tt> to be one of the variables above. However, there is no
|
|
739 way to directly specify this using a normal documentation string. To solve this problem, you can use the <tt>@TOKEN</tt>
|
|
740 decorator. For example:
|
|
741
|
|
742 <blockquote>
|
|
743 <pre>
|
|
744 from ply.lex import TOKEN
|
|
745
|
|
746 @TOKEN(identifier)
|
|
747 def t_ID(t):
|
|
748 ...
|
|
749 </pre>
|
|
750 </blockquote>
|
|
751
|
|
752 <p>
|
|
753 This will attach <tt>identifier</tt> to the docstring for <tt>t_ID()</tt> allowing <tt>lex.py</tt> to work normally.
|
|
754 </p>
|
|
755
|
|
756 <H3><a name="ply_nn15"></a>4.13 Optimized mode</H3>
|
|
757
|
|
758
|
|
759 For improved performance, it may be desirable to use Python's
|
|
760 optimized mode (e.g., running Python with the <tt>-O</tt>
|
|
761 option). However, doing so causes Python to ignore documentation
|
|
762 strings. This presents special problems for <tt>lex.py</tt>. To
|
|
763 handle this case, you can create your lexer using
|
|
764 the <tt>optimize</tt> option as follows:
|
|
765
|
|
766 <blockquote>
|
|
767 <pre>
|
|
768 lexer = lex.lex(optimize=1)
|
|
769 </pre>
|
|
770 </blockquote>
|
|
771
|
|
772 Next, run Python in its normal operating mode. When you do
|
|
773 this, <tt>lex.py</tt> will write a file called <tt>lextab.py</tt> in
|
|
774 the same directory as the module containing the lexer specification.
|
|
775 This file contains all of the regular
|
|
776 expression rules and tables used during lexing. On subsequent
|
|
777 executions,
|
|
778 <tt>lextab.py</tt> will simply be imported to build the lexer. This
|
|
779 approach substantially improves the startup time of the lexer and it
|
|
780 works in Python's optimized mode.
|
|
781
|
|
782 <p>
|
|
783 To change the name of the lexer-generated module, use the <tt>lextab</tt> keyword argument. For example:
|
|
784 </p>
|
|
785
|
|
786 <blockquote>
|
|
787 <pre>
|
|
788 lexer = lex.lex(optimize=1,lextab="footab")
|
|
789 </pre>
|
|
790 </blockquote>
|
|
791
|
|
792 When running in optimized mode, it is important to note that lex disables most error checking. Thus, this is really only recommended
|
|
793 if you're sure everything is working correctly and you're ready to start releasing production code.
|
|
794
|
|
795 <H3><a name="ply_nn16"></a>4.14 Debugging</H3>
|
|
796
|
|
797
|
|
798 For the purpose of debugging, you can run <tt>lex()</tt> in a debugging mode as follows:
|
|
799
|
|
800 <blockquote>
|
|
801 <pre>
|
|
802 lexer = lex.lex(debug=1)
|
|
803 </pre>
|
|
804 </blockquote>
|
|
805
|
|
806 <p>
|
|
807 This will produce various sorts of debugging information including all of the added rules,
|
|
808 the master regular expressions used by the lexer, and tokens generating during lexing.
|
|
809 </p>
|
|
810
|
|
811 <p>
|
|
812 In addition, <tt>lex.py</tt> comes with a simple main function which
|
|
813 will either tokenize input read from standard input or from a file specified
|
|
814 on the command line. To use it, simply put this in your lexer:
|
|
815 </p>
|
|
816
|
|
817 <blockquote>
|
|
818 <pre>
|
|
819 if __name__ == '__main__':
|
|
820 lex.runmain()
|
|
821 </pre>
|
|
822 </blockquote>
|
|
823
|
|
824 Please refer to the "Debugging" section near the end for some more advanced details
|
|
825 of debugging.
|
|
826
|
|
827 <H3><a name="ply_nn17"></a>4.15 Alternative specification of lexers</H3>
|
|
828
|
|
829
|
|
830 As shown in the example, lexers are specified all within one Python module. If you want to
|
|
831 put token rules in a different module from the one in which you invoke <tt>lex()</tt>, use the
|
|
832 <tt>module</tt> keyword argument.
|
|
833
|
|
834 <p>
|
|
835 For example, you might have a dedicated module that just contains
|
|
836 the token rules:
|
|
837
|
|
838 <blockquote>
|
|
839 <pre>
|
|
840 # module: tokrules.py
|
|
841 # This module just contains the lexing rules
|
|
842
|
|
843 # List of token names. This is always required
|
|
844 tokens = (
|
|
845 'NUMBER',
|
|
846 'PLUS',
|
|
847 'MINUS',
|
|
848 'TIMES',
|
|
849 'DIVIDE',
|
|
850 'LPAREN',
|
|
851 'RPAREN',
|
|
852 )
|
|
853
|
|
854 # Regular expression rules for simple tokens
|
|
855 t_PLUS = r'\+'
|
|
856 t_MINUS = r'-'
|
|
857 t_TIMES = r'\*'
|
|
858 t_DIVIDE = r'/'
|
|
859 t_LPAREN = r'\('
|
|
860 t_RPAREN = r'\)'
|
|
861
|
|
862 # A regular expression rule with some action code
|
|
863 def t_NUMBER(t):
|
|
864 r'\d+'
|
|
865 t.value = int(t.value)
|
|
866 return t
|
|
867
|
|
868 # Define a rule so we can track line numbers
|
|
869 def t_newline(t):
|
|
870 r'\n+'
|
|
871 t.lexer.lineno += len(t.value)
|
|
872
|
|
873 # A string containing ignored characters (spaces and tabs)
|
|
874 t_ignore = ' \t'
|
|
875
|
|
876 # Error handling rule
|
|
877 def t_error(t):
|
|
878 print("Illegal character '%s'" % t.value[0])
|
|
879 t.lexer.skip(1)
|
|
880 </pre>
|
|
881 </blockquote>
|
|
882
|
|
883 Now, if you wanted to build a tokenizer from these rules from within a different module, you would do the following (shown for Python interactive mode):
|
|
884
|
|
885 <blockquote>
|
|
886 <pre>
|
|
887 >>> import tokrules
|
|
888 >>> <b>lexer = lex.lex(module=tokrules)</b>
|
|
889 >>> lexer.input("3 + 4")
|
|
890 >>> lexer.token()
|
|
891 LexToken(NUMBER,3,1,1,0)
|
|
892 >>> lexer.token()
|
|
893 LexToken(PLUS,'+',1,2)
|
|
894 >>> lexer.token()
|
|
895 LexToken(NUMBER,4,1,4)
|
|
896 >>> lexer.token()
|
|
897 None
|
|
898 >>>
|
|
899 </pre>
|
|
900 </blockquote>
|
|
901
|
|
902 The <tt>module</tt> option can also be used to define lexers from instances of a class. For example:
|
|
903
|
|
904 <blockquote>
|
|
905 <pre>
|
|
906 import ply.lex as lex
|
|
907
|
|
908 class MyLexer(object):
|
|
909 # List of token names. This is always required
|
|
910 tokens = (
|
|
911 'NUMBER',
|
|
912 'PLUS',
|
|
913 'MINUS',
|
|
914 'TIMES',
|
|
915 'DIVIDE',
|
|
916 'LPAREN',
|
|
917 'RPAREN',
|
|
918 )
|
|
919
|
|
920 # Regular expression rules for simple tokens
|
|
921 t_PLUS = r'\+'
|
|
922 t_MINUS = r'-'
|
|
923 t_TIMES = r'\*'
|
|
924 t_DIVIDE = r'/'
|
|
925 t_LPAREN = r'\('
|
|
926 t_RPAREN = r'\)'
|
|
927
|
|
928 # A regular expression rule with some action code
|
|
929 # Note addition of self parameter since we're in a class
|
|
930 def t_NUMBER(self,t):
|
|
931 r'\d+'
|
|
932 t.value = int(t.value)
|
|
933 return t
|
|
934
|
|
935 # Define a rule so we can track line numbers
|
|
936 def t_newline(self,t):
|
|
937 r'\n+'
|
|
938 t.lexer.lineno += len(t.value)
|
|
939
|
|
940 # A string containing ignored characters (spaces and tabs)
|
|
941 t_ignore = ' \t'
|
|
942
|
|
943 # Error handling rule
|
|
944 def t_error(self,t):
|
|
945 print("Illegal character '%s'" % t.value[0])
|
|
946 t.lexer.skip(1)
|
|
947
|
|
948 <b># Build the lexer
|
|
949 def build(self,**kwargs):
|
|
950 self.lexer = lex.lex(module=self, **kwargs)</b>
|
|
951
|
|
952 # Test it output
|
|
953 def test(self,data):
|
|
954 self.lexer.input(data)
|
|
955 while True:
|
|
956 tok = self.lexer.token()
|
|
957 if not tok:
|
|
958 break
|
|
959 print(tok)
|
|
960
|
|
961 # Build the lexer and try it out
|
|
962 m = MyLexer()
|
|
963 m.build() # Build the lexer
|
|
964 m.test("3 + 4") # Test it
|
|
965 </pre>
|
|
966 </blockquote>
|
|
967
|
|
968
|
|
969 When building a lexer from class, <em>you should construct the lexer from
|
|
970 an instance of the class</em>, not the class object itself. This is because
|
|
971 PLY only works properly if the lexer actions are defined by bound-methods.
|
|
972
|
|
973 <p>
|
|
974 When using the <tt>module</tt> option to <tt>lex()</tt>, PLY collects symbols
|
|
975 from the underlying object using the <tt>dir()</tt> function. There is no
|
|
976 direct access to the <tt>__dict__</tt> attribute of the object supplied as a
|
|
977 module value. </p>
|
|
978
|
|
979 <P>
|
|
980 Finally, if you want to keep things nicely encapsulated, but don't want to use a
|
|
981 full-fledged class definition, lexers can be defined using closures. For example:
|
|
982
|
|
983 <blockquote>
|
|
984 <pre>
|
|
985 import ply.lex as lex
|
|
986
|
|
987 # List of token names. This is always required
|
|
988 tokens = (
|
|
989 'NUMBER',
|
|
990 'PLUS',
|
|
991 'MINUS',
|
|
992 'TIMES',
|
|
993 'DIVIDE',
|
|
994 'LPAREN',
|
|
995 'RPAREN',
|
|
996 )
|
|
997
|
|
998 def MyLexer():
|
|
999 # Regular expression rules for simple tokens
|
|
1000 t_PLUS = r'\+'
|
|
1001 t_MINUS = r'-'
|
|
1002 t_TIMES = r'\*'
|
|
1003 t_DIVIDE = r'/'
|
|
1004 t_LPAREN = r'\('
|
|
1005 t_RPAREN = r'\)'
|
|
1006
|
|
1007 # A regular expression rule with some action code
|
|
1008 def t_NUMBER(t):
|
|
1009 r'\d+'
|
|
1010 t.value = int(t.value)
|
|
1011 return t
|
|
1012
|
|
1013 # Define a rule so we can track line numbers
|
|
1014 def t_newline(t):
|
|
1015 r'\n+'
|
|
1016 t.lexer.lineno += len(t.value)
|
|
1017
|
|
1018 # A string containing ignored characters (spaces and tabs)
|
|
1019 t_ignore = ' \t'
|
|
1020
|
|
1021 # Error handling rule
|
|
1022 def t_error(t):
|
|
1023 print("Illegal character '%s'" % t.value[0])
|
|
1024 t.lexer.skip(1)
|
|
1025
|
|
1026 # Build the lexer from my environment and return it
|
|
1027 return lex.lex()
|
|
1028 </pre>
|
|
1029 </blockquote>
|
|
1030
|
|
1031 <p>
|
|
1032 <b>Important note:</b> If you are defining a lexer using a class or closure, be aware that PLY still requires you to only
|
|
1033 define a single lexer per module (source file). There are extensive validation/error checking parts of the PLY that
|
|
1034 may falsely report error messages if you don't follow this rule.
|
|
1035 </p>
|
|
1036
|
|
1037 <H3><a name="ply_nn18"></a>4.16 Maintaining state</H3>
|
|
1038
|
|
1039
|
|
1040 In your lexer, you may want to maintain a variety of state
|
|
1041 information. This might include mode settings, symbol tables, and
|
|
1042 other details. As an example, suppose that you wanted to keep
|
|
1043 track of how many NUMBER tokens had been encountered.
|
|
1044
|
|
1045 <p>
|
|
1046 One way to do this is to keep a set of global variables in the module
|
|
1047 where you created the lexer. For example:
|
|
1048
|
|
1049 <blockquote>
|
|
1050 <pre>
|
|
1051 num_count = 0
|
|
1052 def t_NUMBER(t):
|
|
1053 r'\d+'
|
|
1054 global num_count
|
|
1055 num_count += 1
|
|
1056 t.value = int(t.value)
|
|
1057 return t
|
|
1058 </pre>
|
|
1059 </blockquote>
|
|
1060
|
|
1061 If you don't like the use of a global variable, another place to store
|
|
1062 information is inside the Lexer object created by <tt>lex()</tt>.
|
|
1063 To this, you can use the <tt>lexer</tt> attribute of tokens passed to
|
|
1064 the various rules. For example:
|
|
1065
|
|
1066 <blockquote>
|
|
1067 <pre>
|
|
1068 def t_NUMBER(t):
|
|
1069 r'\d+'
|
|
1070 t.lexer.num_count += 1 # Note use of lexer attribute
|
|
1071 t.value = int(t.value)
|
|
1072 return t
|
|
1073
|
|
1074 lexer = lex.lex()
|
|
1075 lexer.num_count = 0 # Set the initial count
|
|
1076 </pre>
|
|
1077 </blockquote>
|
|
1078
|
|
1079 This latter approach has the advantage of being simple and working
|
|
1080 correctly in applications where multiple instantiations of a given
|
|
1081 lexer exist in the same application. However, this might also feel
|
|
1082 like a gross violation of encapsulation to OO purists.
|
|
1083 Just to put your mind at some ease, all
|
|
1084 internal attributes of the lexer (with the exception of <tt>lineno</tt>) have names that are prefixed
|
|
1085 by <tt>lex</tt> (e.g., <tt>lexdata</tt>,<tt>lexpos</tt>, etc.). Thus,
|
|
1086 it is perfectly safe to store attributes in the lexer that
|
|
1087 don't have names starting with that prefix or a name that conflicts with one of the
|
|
1088 predefined methods (e.g., <tt>input()</tt>, <tt>token()</tt>, etc.).
|
|
1089
|
|
1090 <p>
|
|
1091 If you don't like assigning values on the lexer object, you can define your lexer as a class as
|
|
1092 shown in the previous section:
|
|
1093
|
|
1094 <blockquote>
|
|
1095 <pre>
|
|
1096 class MyLexer:
|
|
1097 ...
|
|
1098 def t_NUMBER(self,t):
|
|
1099 r'\d+'
|
|
1100 self.num_count += 1
|
|
1101 t.value = int(t.value)
|
|
1102 return t
|
|
1103
|
|
1104 def build(self, **kwargs):
|
|
1105 self.lexer = lex.lex(object=self,**kwargs)
|
|
1106
|
|
1107 def __init__(self):
|
|
1108 self.num_count = 0
|
|
1109 </pre>
|
|
1110 </blockquote>
|
|
1111
|
|
1112 The class approach may be the easiest to manage if your application is
|
|
1113 going to be creating multiple instances of the same lexer and you need
|
|
1114 to manage a lot of state.
|
|
1115
|
|
1116 <p>
|
|
1117 State can also be managed through closures. For example, in Python 3:
|
|
1118
|
|
1119 <blockquote>
|
|
1120 <pre>
|
|
1121 def MyLexer():
|
|
1122 num_count = 0
|
|
1123 ...
|
|
1124 def t_NUMBER(t):
|
|
1125 r'\d+'
|
|
1126 nonlocal num_count
|
|
1127 num_count += 1
|
|
1128 t.value = int(t.value)
|
|
1129 return t
|
|
1130 ...
|
|
1131 </pre>
|
|
1132 </blockquote>
|
|
1133
|
|
1134 <H3><a name="ply_nn19"></a>4.17 Lexer cloning</H3>
|
|
1135
|
|
1136
|
|
1137 <p>
|
|
1138 If necessary, a lexer object can be duplicated by invoking its <tt>clone()</tt> method. For example:
|
|
1139
|
|
1140 <blockquote>
|
|
1141 <pre>
|
|
1142 lexer = lex.lex()
|
|
1143 ...
|
|
1144 newlexer = lexer.clone()
|
|
1145 </pre>
|
|
1146 </blockquote>
|
|
1147
|
|
1148 When a lexer is cloned, the copy is exactly identical to the original lexer
|
|
1149 including any input text and internal state. However, the clone allows a
|
|
1150 different set of input text to be supplied which may be processed separately.
|
|
1151 This may be useful in situations when you are writing a parser/compiler that
|
|
1152 involves recursive or reentrant processing. For instance, if you
|
|
1153 needed to scan ahead in the input for some reason, you could create a
|
|
1154 clone and use it to look ahead. Or, if you were implementing some kind of preprocessor,
|
|
1155 cloned lexers could be used to handle different input files.
|
|
1156
|
|
1157 <p>
|
|
1158 Creating a clone is different than calling <tt>lex.lex()</tt> in that
|
|
1159 PLY doesn't regenerate any of the internal tables or regular expressions.
|
|
1160
|
|
1161 <p>
|
|
1162 Special considerations need to be made when cloning lexers that also
|
|
1163 maintain their own internal state using classes or closures. Namely,
|
|
1164 you need to be aware that the newly created lexers will share all of
|
|
1165 this state with the original lexer. For example, if you defined a
|
|
1166 lexer as a class and did this:
|
|
1167
|
|
1168 <blockquote>
|
|
1169 <pre>
|
|
1170 m = MyLexer()
|
|
1171 a = lex.lex(object=m) # Create a lexer
|
|
1172
|
|
1173 b = a.clone() # Clone the lexer
|
|
1174 </pre>
|
|
1175 </blockquote>
|
|
1176
|
|
1177 Then both <tt>a</tt> and <tt>b</tt> are going to be bound to the same
|
|
1178 object <tt>m</tt> and any changes to <tt>m</tt> will be reflected in both lexers. It's
|
|
1179 important to emphasize that <tt>clone()</tt> is only meant to create a new lexer
|
|
1180 that reuses the regular expressions and environment of another lexer. If you
|
|
1181 need to make a totally new copy of a lexer, then call <tt>lex()</tt> again.
|
|
1182
|
|
1183 <H3><a name="ply_nn20"></a>4.18 Internal lexer state</H3>
|
|
1184
|
|
1185
|
|
1186 A Lexer object <tt>lexer</tt> has a number of internal attributes that may be useful in certain
|
|
1187 situations.
|
|
1188
|
|
1189 <p>
|
|
1190 <tt>lexer.lexpos</tt>
|
|
1191 <blockquote>
|
|
1192 This attribute is an integer that contains the current position within the input text. If you modify
|
|
1193 the value, it will change the result of the next call to <tt>token()</tt>. Within token rule functions, this points
|
|
1194 to the first character <em>after</em> the matched text. If the value is modified within a rule, the next returned token will be
|
|
1195 matched at the new position.
|
|
1196 </blockquote>
|
|
1197
|
|
1198 <p>
|
|
1199 <tt>lexer.lineno</tt>
|
|
1200 <blockquote>
|
|
1201 The current value of the line number attribute stored in the lexer. PLY only specifies that the attribute
|
|
1202 exists---it never sets, updates, or performs any processing with it. If you want to track line numbers,
|
|
1203 you will need to add code yourself (see the section on line numbers and positional information).
|
|
1204 </blockquote>
|
|
1205
|
|
1206 <p>
|
|
1207 <tt>lexer.lexdata</tt>
|
|
1208 <blockquote>
|
|
1209 The current input text stored in the lexer. This is the string passed with the <tt>input()</tt> method. It
|
|
1210 would probably be a bad idea to modify this unless you really know what you're doing.
|
|
1211 </blockquote>
|
|
1212
|
|
1213 <P>
|
|
1214 <tt>lexer.lexmatch</tt>
|
|
1215 <blockquote>
|
|
1216 This is the raw <tt>Match</tt> object returned by the Python <tt>re.match()</tt> function (used internally by PLY) for the
|
|
1217 current token. If you have written a regular expression that contains named groups, you can use this to retrieve those values.
|
|
1218 Note: This attribute is only updated when tokens are defined and processed by functions.
|
|
1219 </blockquote>
|
|
1220
|
|
1221 <H3><a name="ply_nn21"></a>4.19 Conditional lexing and start conditions</H3>
|
|
1222
|
|
1223
|
|
1224 In advanced parsing applications, it may be useful to have different
|
|
1225 lexing states. For instance, you may want the occurrence of a certain
|
|
1226 token or syntactic construct to trigger a different kind of lexing.
|
|
1227 PLY supports a feature that allows the underlying lexer to be put into
|
|
1228 a series of different states. Each state can have its own tokens,
|
|
1229 lexing rules, and so forth. The implementation is based largely on
|
|
1230 the "start condition" feature of GNU flex. Details of this can be found
|
|
1231 at <a
|
|
1232 href="http://flex.sourceforge.net/manual/Start-Conditions.html">http://flex.sourceforge.net/manual/Start-Conditions.html</a>.
|
|
1233
|
|
1234 <p>
|
|
1235 To define a new lexing state, it must first be declared. This is done by including a "states" declaration in your
|
|
1236 lex file. For example:
|
|
1237
|
|
1238 <blockquote>
|
|
1239 <pre>
|
|
1240 states = (
|
|
1241 ('foo','exclusive'),
|
|
1242 ('bar','inclusive'),
|
|
1243 )
|
|
1244 </pre>
|
|
1245 </blockquote>
|
|
1246
|
|
1247 This declaration declares two states, <tt>'foo'</tt>
|
|
1248 and <tt>'bar'</tt>. States may be of two types; <tt>'exclusive'</tt>
|
|
1249 and <tt>'inclusive'</tt>. An exclusive state completely overrides the
|
|
1250 default behavior of the lexer. That is, lex will only return tokens
|
|
1251 and apply rules defined specifically for that state. An inclusive
|
|
1252 state adds additional tokens and rules to the default set of rules.
|
|
1253 Thus, lex will return both the tokens defined by default in addition
|
|
1254 to those defined for the inclusive state.
|
|
1255
|
|
1256 <p>
|
|
1257 Once a state has been declared, tokens and rules are declared by including the
|
|
1258 state name in token/rule declaration. For example:
|
|
1259
|
|
1260 <blockquote>
|
|
1261 <pre>
|
|
1262 t_foo_NUMBER = r'\d+' # Token 'NUMBER' in state 'foo'
|
|
1263 t_bar_ID = r'[a-zA-Z_][a-zA-Z0-9_]*' # Token 'ID' in state 'bar'
|
|
1264
|
|
1265 def t_foo_newline(t):
|
|
1266 r'\n'
|
|
1267 t.lexer.lineno += 1
|
|
1268 </pre>
|
|
1269 </blockquote>
|
|
1270
|
|
1271 A token can be declared in multiple states by including multiple state names in the declaration. For example:
|
|
1272
|
|
1273 <blockquote>
|
|
1274 <pre>
|
|
1275 t_foo_bar_NUMBER = r'\d+' # Defines token 'NUMBER' in both state 'foo' and 'bar'
|
|
1276 </pre>
|
|
1277 </blockquote>
|
|
1278
|
|
1279 Alternative, a token can be declared in all states using the 'ANY' in the name.
|
|
1280
|
|
1281 <blockquote>
|
|
1282 <pre>
|
|
1283 t_ANY_NUMBER = r'\d+' # Defines a token 'NUMBER' in all states
|
|
1284 </pre>
|
|
1285 </blockquote>
|
|
1286
|
|
1287 If no state name is supplied, as is normally the case, the token is associated with a special state <tt>'INITIAL'</tt>. For example,
|
|
1288 these two declarations are identical:
|
|
1289
|
|
1290 <blockquote>
|
|
1291 <pre>
|
|
1292 t_NUMBER = r'\d+'
|
|
1293 t_INITIAL_NUMBER = r'\d+'
|
|
1294 </pre>
|
|
1295 </blockquote>
|
|
1296
|
|
1297 <p>
|
|
1298 States are also associated with the special <tt>t_ignore</tt>, <tt>t_error()</tt>, and <tt>t_eof()</tt> declarations. For example, if a state treats
|
|
1299 these differently, you can declare:</p>
|
|
1300
|
|
1301 <blockquote>
|
|
1302 <pre>
|
|
1303 t_foo_ignore = " \t\n" # Ignored characters for state 'foo'
|
|
1304
|
|
1305 def t_bar_error(t): # Special error handler for state 'bar'
|
|
1306 pass
|
|
1307 </pre>
|
|
1308 </blockquote>
|
|
1309
|
|
1310 By default, lexing operates in the <tt>'INITIAL'</tt> state. This state includes all of the normally defined tokens.
|
|
1311 For users who aren't using different states, this fact is completely transparent. If, during lexing or parsing, you want to change
|
|
1312 the lexing state, use the <tt>begin()</tt> method. For example:
|
|
1313
|
|
1314 <blockquote>
|
|
1315 <pre>
|
|
1316 def t_begin_foo(t):
|
|
1317 r'start_foo'
|
|
1318 t.lexer.begin('foo') # Starts 'foo' state
|
|
1319 </pre>
|
|
1320 </blockquote>
|
|
1321
|
|
1322 To get out of a state, you use <tt>begin()</tt> to switch back to the initial state. For example:
|
|
1323
|
|
1324 <blockquote>
|
|
1325 <pre>
|
|
1326 def t_foo_end(t):
|
|
1327 r'end_foo'
|
|
1328 t.lexer.begin('INITIAL') # Back to the initial state
|
|
1329 </pre>
|
|
1330 </blockquote>
|
|
1331
|
|
1332 The management of states can also be done with a stack. For example:
|
|
1333
|
|
1334 <blockquote>
|
|
1335 <pre>
|
|
1336 def t_begin_foo(t):
|
|
1337 r'start_foo'
|
|
1338 t.lexer.push_state('foo') # Starts 'foo' state
|
|
1339
|
|
1340 def t_foo_end(t):
|
|
1341 r'end_foo'
|
|
1342 t.lexer.pop_state() # Back to the previous state
|
|
1343 </pre>
|
|
1344 </blockquote>
|
|
1345
|
|
1346 <p>
|
|
1347 The use of a stack would be useful in situations where there are many ways of entering a new lexing state and you merely want to go back
|
|
1348 to the previous state afterwards.
|
|
1349
|
|
1350 <P>
|
|
1351 An example might help clarify. Suppose you were writing a parser and you wanted to grab sections of arbitrary C code enclosed by
|
|
1352 curly braces. That is, whenever you encounter a starting brace '{', you want to read all of the enclosed code up to the ending brace '}'
|
|
1353 and return it as a string. Doing this with a normal regular expression rule is nearly (if not actually) impossible. This is because braces can
|
|
1354 be nested and can be included in comments and strings. Thus, simply matching up to the first matching '}' character isn't good enough. Here is how
|
|
1355 you might use lexer states to do this:
|
|
1356
|
|
1357 <blockquote>
|
|
1358 <pre>
|
|
1359 # Declare the state
|
|
1360 states = (
|
|
1361 ('ccode','exclusive'),
|
|
1362 )
|
|
1363
|
|
1364 # Match the first {. Enter ccode state.
|
|
1365 def t_ccode(t):
|
|
1366 r'\{'
|
|
1367 t.lexer.code_start = t.lexer.lexpos # Record the starting position
|
|
1368 t.lexer.level = 1 # Initial brace level
|
|
1369 t.lexer.begin('ccode') # Enter 'ccode' state
|
|
1370
|
|
1371 # Rules for the ccode state
|
|
1372 def t_ccode_lbrace(t):
|
|
1373 r'\{'
|
|
1374 t.lexer.level +=1
|
|
1375
|
|
1376 def t_ccode_rbrace(t):
|
|
1377 r'\}'
|
|
1378 t.lexer.level -=1
|
|
1379
|
|
1380 # If closing brace, return the code fragment
|
|
1381 if t.lexer.level == 0:
|
|
1382 t.value = t.lexer.lexdata[t.lexer.code_start:t.lexer.lexpos+1]
|
|
1383 t.type = "CCODE"
|
|
1384 t.lexer.lineno += t.value.count('\n')
|
|
1385 t.lexer.begin('INITIAL')
|
|
1386 return t
|
|
1387
|
|
1388 # C or C++ comment (ignore)
|
|
1389 def t_ccode_comment(t):
|
|
1390 r'(/\*(.|\n)*?\*/)|(//.*)'
|
|
1391 pass
|
|
1392
|
|
1393 # C string
|
|
1394 def t_ccode_string(t):
|
|
1395 r'\"([^\\\n]|(\\.))*?\"'
|
|
1396
|
|
1397 # C character literal
|
|
1398 def t_ccode_char(t):
|
|
1399 r'\'([^\\\n]|(\\.))*?\''
|
|
1400
|
|
1401 # Any sequence of non-whitespace characters (not braces, strings)
|
|
1402 def t_ccode_nonspace(t):
|
|
1403 r'[^\s\{\}\'\"]+'
|
|
1404
|
|
1405 # Ignored characters (whitespace)
|
|
1406 t_ccode_ignore = " \t\n"
|
|
1407
|
|
1408 # For bad characters, we just skip over it
|
|
1409 def t_ccode_error(t):
|
|
1410 t.lexer.skip(1)
|
|
1411 </pre>
|
|
1412 </blockquote>
|
|
1413
|
|
1414 In this example, the occurrence of the first '{' causes the lexer to record the starting position and enter a new state <tt>'ccode'</tt>. A collection of rules then match
|
|
1415 various parts of the input that follow (comments, strings, etc.). All of these rules merely discard the token (by not returning a value).
|
|
1416 However, if the closing right brace is encountered, the rule <tt>t_ccode_rbrace</tt> collects all of the code (using the earlier recorded starting
|
|
1417 position), stores it, and returns a token 'CCODE' containing all of that text. When returning the token, the lexing state is restored back to its
|
|
1418 initial state.
|
|
1419
|
|
1420 <H3><a name="ply_nn21"></a>4.20 Miscellaneous Issues</H3>
|
|
1421
|
|
1422
|
|
1423 <P>
|
|
1424 <li>The lexer requires input to be supplied as a single input string. Since most machines have more than enough memory, this
|
|
1425 rarely presents a performance concern. However, it means that the lexer currently can't be used with streaming data
|
|
1426 such as open files or sockets. This limitation is primarily a side-effect of using the <tt>re</tt> module. You might be
|
|
1427 able to work around this by implementing an appropriate <tt>def t_eof()</tt> end-of-file handling rule. The main complication
|
|
1428 here is that you'll probably need to ensure that data is fed to the lexer in a way so that it doesn't split in in the middle
|
|
1429 of a token.</p>
|
|
1430
|
|
1431 <p>
|
|
1432 <li>The lexer should work properly with both Unicode strings given as token and pattern matching rules as
|
|
1433 well as for input text.
|
|
1434
|
|
1435 <p>
|
|
1436 <li>If you need to supply optional flags to the re.compile() function, use the reflags option to lex. For example:
|
|
1437
|
|
1438 <blockquote>
|
|
1439 <pre>
|
|
1440 lex.lex(reflags=re.UNICODE)
|
|
1441 </pre>
|
|
1442 </blockquote>
|
|
1443
|
|
1444 <p>
|
|
1445 <li>Since the lexer is written entirely in Python, its performance is
|
|
1446 largely determined by that of the Python <tt>re</tt> module. Although
|
|
1447 the lexer has been written to be as efficient as possible, it's not
|
|
1448 blazingly fast when used on very large input files. If
|
|
1449 performance is concern, you might consider upgrading to the most
|
|
1450 recent version of Python, creating a hand-written lexer, or offloading
|
|
1451 the lexer into a C extension module.
|
|
1452
|
|
1453 <p>
|
|
1454 If you are going to create a hand-written lexer and you plan to use it with <tt>yacc.py</tt>,
|
|
1455 it only needs to conform to the following requirements:
|
|
1456
|
|
1457 <ul>
|
|
1458 <li>It must provide a <tt>token()</tt> method that returns the next token or <tt>None</tt> if no more
|
|
1459 tokens are available.
|
|
1460 <li>The <tt>token()</tt> method must return an object <tt>tok</tt> that has <tt>type</tt> and <tt>value</tt> attributes. If
|
|
1461 line number tracking is being used, then the token should also define a <tt>lineno</tt> attribute.
|
|
1462 </ul>
|
|
1463
|
|
1464 <H2><a name="ply_nn22"></a>5. Parsing basics</H2>
|
|
1465
|
|
1466
|
|
1467 <tt>yacc.py</tt> is used to parse language syntax. Before showing an
|
|
1468 example, there are a few important bits of background that must be
|
|
1469 mentioned. First, <em>syntax</em> is usually specified in terms of a BNF grammar.
|
|
1470 For example, if you wanted to parse
|
|
1471 simple arithmetic expressions, you might first write an unambiguous
|
|
1472 grammar specification like this:
|
|
1473
|
|
1474 <blockquote>
|
|
1475 <pre>
|
|
1476 expression : expression + term
|
|
1477 | expression - term
|
|
1478 | term
|
|
1479
|
|
1480 term : term * factor
|
|
1481 | term / factor
|
|
1482 | factor
|
|
1483
|
|
1484 factor : NUMBER
|
|
1485 | ( expression )
|
|
1486 </pre>
|
|
1487 </blockquote>
|
|
1488
|
|
1489 In the grammar, symbols such as <tt>NUMBER</tt>, <tt>+</tt>, <tt>-</tt>, <tt>*</tt>, and <tt>/</tt> are known
|
|
1490 as <em>terminals</em> and correspond to raw input tokens. Identifiers such as <tt>term</tt> and <tt>factor</tt> refer to
|
|
1491 grammar rules comprised of a collection of terminals and other rules. These identifiers are known as <em>non-terminals</em>.
|
|
1492 <P>
|
|
1493
|
|
1494 The semantic behavior of a language is often specified using a
|
|
1495 technique known as syntax directed translation. In syntax directed
|
|
1496 translation, attributes are attached to each symbol in a given grammar
|
|
1497 rule along with an action. Whenever a particular grammar rule is
|
|
1498 recognized, the action describes what to do. For example, given the
|
|
1499 expression grammar above, you might write the specification for a
|
|
1500 simple calculator like this:
|
|
1501
|
|
1502 <blockquote>
|
|
1503 <pre>
|
|
1504 Grammar Action
|
|
1505 -------------------------------- --------------------------------------------
|
|
1506 expression0 : expression1 + term expression0.val = expression1.val + term.val
|
|
1507 | expression1 - term expression0.val = expression1.val - term.val
|
|
1508 | term expression0.val = term.val
|
|
1509
|
|
1510 term0 : term1 * factor term0.val = term1.val * factor.val
|
|
1511 | term1 / factor term0.val = term1.val / factor.val
|
|
1512 | factor term0.val = factor.val
|
|
1513
|
|
1514 factor : NUMBER factor.val = int(NUMBER.lexval)
|
|
1515 | ( expression ) factor.val = expression.val
|
|
1516 </pre>
|
|
1517 </blockquote>
|
|
1518
|
|
1519 A good way to think about syntax directed translation is to
|
|
1520 view each symbol in the grammar as a kind of object. Associated
|
|
1521 with each symbol is a value representing its "state" (for example, the
|
|
1522 <tt>val</tt> attribute above). Semantic
|
|
1523 actions are then expressed as a collection of functions or methods
|
|
1524 that operate on the symbols and associated values.
|
|
1525
|
|
1526 <p>
|
|
1527 Yacc uses a parsing technique known as LR-parsing or shift-reduce parsing. LR parsing is a
|
|
1528 bottom up technique that tries to recognize the right-hand-side of various grammar rules.
|
|
1529 Whenever a valid right-hand-side is found in the input, the appropriate action code is triggered and the
|
|
1530 grammar symbols are replaced by the grammar symbol on the left-hand-side.
|
|
1531
|
|
1532 <p>
|
|
1533 LR parsing is commonly implemented by shifting grammar symbols onto a
|
|
1534 stack and looking at the stack and the next input token for patterns that
|
|
1535 match one of the grammar rules.
|
|
1536 The details of the algorithm can be found in a compiler textbook, but the
|
|
1537 following example illustrates the steps that are performed if you
|
|
1538 wanted to parse the expression
|
|
1539 <tt>3 + 5 * (10 - 20)</tt> using the grammar defined above. In the example,
|
|
1540 the special symbol <tt>$</tt> represents the end of input.
|
|
1541
|
|
1542
|
|
1543 <blockquote>
|
|
1544 <pre>
|
|
1545 Step Symbol Stack Input Tokens Action
|
|
1546 ---- --------------------- --------------------- -------------------------------
|
|
1547 1 3 + 5 * ( 10 - 20 )$ Shift 3
|
|
1548 2 3 + 5 * ( 10 - 20 )$ Reduce factor : NUMBER
|
|
1549 3 factor + 5 * ( 10 - 20 )$ Reduce term : factor
|
|
1550 4 term + 5 * ( 10 - 20 )$ Reduce expr : term
|
|
1551 5 expr + 5 * ( 10 - 20 )$ Shift +
|
|
1552 6 expr + 5 * ( 10 - 20 )$ Shift 5
|
|
1553 7 expr + 5 * ( 10 - 20 )$ Reduce factor : NUMBER
|
|
1554 8 expr + factor * ( 10 - 20 )$ Reduce term : factor
|
|
1555 9 expr + term * ( 10 - 20 )$ Shift *
|
|
1556 10 expr + term * ( 10 - 20 )$ Shift (
|
|
1557 11 expr + term * ( 10 - 20 )$ Shift 10
|
|
1558 12 expr + term * ( 10 - 20 )$ Reduce factor : NUMBER
|
|
1559 13 expr + term * ( factor - 20 )$ Reduce term : factor
|
|
1560 14 expr + term * ( term - 20 )$ Reduce expr : term
|
|
1561 15 expr + term * ( expr - 20 )$ Shift -
|
|
1562 16 expr + term * ( expr - 20 )$ Shift 20
|
|
1563 17 expr + term * ( expr - 20 )$ Reduce factor : NUMBER
|
|
1564 18 expr + term * ( expr - factor )$ Reduce term : factor
|
|
1565 19 expr + term * ( expr - term )$ Reduce expr : expr - term
|
|
1566 20 expr + term * ( expr )$ Shift )
|
|
1567 21 expr + term * ( expr ) $ Reduce factor : (expr)
|
|
1568 22 expr + term * factor $ Reduce term : term * factor
|
|
1569 23 expr + term $ Reduce expr : expr + term
|
|
1570 24 expr $ Reduce expr
|
|
1571 25 $ Success!
|
|
1572 </pre>
|
|
1573 </blockquote>
|
|
1574
|
|
1575 When parsing the expression, an underlying state machine and the
|
|
1576 current input token determine what happens next. If the next token
|
|
1577 looks like part of a valid grammar rule (based on other items on the
|
|
1578 stack), it is generally shifted onto the stack. If the top of the
|
|
1579 stack contains a valid right-hand-side of a grammar rule, it is
|
|
1580 usually "reduced" and the symbols replaced with the symbol on the
|
|
1581 left-hand-side. When this reduction occurs, the appropriate action is
|
|
1582 triggered (if defined). If the input token can't be shifted and the
|
|
1583 top of stack doesn't match any grammar rules, a syntax error has
|
|
1584 occurred and the parser must take some kind of recovery step (or bail
|
|
1585 out). A parse is only successful if the parser reaches a state where
|
|
1586 the symbol stack is empty and there are no more input tokens.
|
|
1587
|
|
1588 <p>
|
|
1589 It is important to note that the underlying implementation is built
|
|
1590 around a large finite-state machine that is encoded in a collection of
|
|
1591 tables. The construction of these tables is non-trivial and
|
|
1592 beyond the scope of this discussion. However, subtle details of this
|
|
1593 process explain why, in the example above, the parser chooses to shift
|
|
1594 a token onto the stack in step 9 rather than reducing the
|
|
1595 rule <tt>expr : expr + term</tt>.
|
|
1596
|
|
1597 <H2><a name="ply_nn23"></a>6. Yacc</H2>
|
|
1598
|
|
1599
|
|
1600 The <tt>ply.yacc</tt> module implements the parsing component of PLY.
|
|
1601 The name "yacc" stands for "Yet Another Compiler Compiler" and is
|
|
1602 borrowed from the Unix tool of the same name.
|
|
1603
|
|
1604 <H3><a name="ply_nn24"></a>6.1 An example</H3>
|
|
1605
|
|
1606
|
|
1607 Suppose you wanted to make a grammar for simple arithmetic expressions as previously described. Here is
|
|
1608 how you would do it with <tt>yacc.py</tt>:
|
|
1609
|
|
1610 <blockquote>
|
|
1611 <pre>
|
|
1612 # Yacc example
|
|
1613
|
|
1614 import ply.yacc as yacc
|
|
1615
|
|
1616 # Get the token map from the lexer. This is required.
|
|
1617 from calclex import tokens
|
|
1618
|
|
1619 def p_expression_plus(p):
|
|
1620 'expression : expression PLUS term'
|
|
1621 p[0] = p[1] + p[3]
|
|
1622
|
|
1623 def p_expression_minus(p):
|
|
1624 'expression : expression MINUS term'
|
|
1625 p[0] = p[1] - p[3]
|
|
1626
|
|
1627 def p_expression_term(p):
|
|
1628 'expression : term'
|
|
1629 p[0] = p[1]
|
|
1630
|
|
1631 def p_term_times(p):
|
|
1632 'term : term TIMES factor'
|
|
1633 p[0] = p[1] * p[3]
|
|
1634
|
|
1635 def p_term_div(p):
|
|
1636 'term : term DIVIDE factor'
|
|
1637 p[0] = p[1] / p[3]
|
|
1638
|
|
1639 def p_term_factor(p):
|
|
1640 'term : factor'
|
|
1641 p[0] = p[1]
|
|
1642
|
|
1643 def p_factor_num(p):
|
|
1644 'factor : NUMBER'
|
|
1645 p[0] = p[1]
|
|
1646
|
|
1647 def p_factor_expr(p):
|
|
1648 'factor : LPAREN expression RPAREN'
|
|
1649 p[0] = p[2]
|
|
1650
|
|
1651 # Error rule for syntax errors
|
|
1652 def p_error(p):
|
|
1653 print("Syntax error in input!")
|
|
1654
|
|
1655 # Build the parser
|
|
1656 parser = yacc.yacc()
|
|
1657
|
|
1658 while True:
|
|
1659 try:
|
|
1660 s = raw_input('calc > ')
|
|
1661 except EOFError:
|
|
1662 break
|
|
1663 if not s: continue
|
|
1664 result = parser.parse(s)
|
|
1665 print(result)
|
|
1666 </pre>
|
|
1667 </blockquote>
|
|
1668
|
|
1669 In this example, each grammar rule is defined by a Python function
|
|
1670 where the docstring to that function contains the appropriate
|
|
1671 context-free grammar specification. The statements that make up the
|
|
1672 function body implement the semantic actions of the rule. Each function
|
|
1673 accepts a single argument <tt>p</tt> that is a sequence containing the
|
|
1674 values of each grammar symbol in the corresponding rule. The values
|
|
1675 of <tt>p[i]</tt> are mapped to grammar symbols as shown here:
|
|
1676
|
|
1677 <blockquote>
|
|
1678 <pre>
|
|
1679 def p_expression_plus(p):
|
|
1680 'expression : expression PLUS term'
|
|
1681 # ^ ^ ^ ^
|
|
1682 # p[0] p[1] p[2] p[3]
|
|
1683
|
|
1684 p[0] = p[1] + p[3]
|
|
1685 </pre>
|
|
1686 </blockquote>
|
|
1687
|
|
1688 <p>
|
|
1689 For tokens, the "value" of the corresponding <tt>p[i]</tt> is the
|
|
1690 <em>same</em> as the <tt>p.value</tt> attribute assigned in the lexer
|
|
1691 module. For non-terminals, the value is determined by whatever is
|
|
1692 placed in <tt>p[0]</tt> when rules are reduced. This value can be
|
|
1693 anything at all. However, it probably most common for the value to be
|
|
1694 a simple Python type, a tuple, or an instance. In this example, we
|
|
1695 are relying on the fact that the <tt>NUMBER</tt> token stores an
|
|
1696 integer value in its value field. All of the other rules simply
|
|
1697 perform various types of integer operations and propagate the result.
|
|
1698 </p>
|
|
1699
|
|
1700 <p>
|
|
1701 Note: The use of negative indices have a special meaning in
|
|
1702 yacc---specially <tt>p[-1]</tt> does not have the same value
|
|
1703 as <tt>p[3]</tt> in this example. Please see the section on "Embedded
|
|
1704 Actions" for further details.
|
|
1705 </p>
|
|
1706
|
|
1707 <p>
|
|
1708 The first rule defined in the yacc specification determines the
|
|
1709 starting grammar symbol (in this case, a rule for <tt>expression</tt>
|
|
1710 appears first). Whenever the starting rule is reduced by the parser
|
|
1711 and no more input is available, parsing stops and the final value is
|
|
1712 returned (this value will be whatever the top-most rule placed
|
|
1713 in <tt>p[0]</tt>). Note: an alternative starting symbol can be
|
|
1714 specified using the <tt>start</tt> keyword argument to
|
|
1715 <tt>yacc()</tt>.
|
|
1716
|
|
1717 <p>The <tt>p_error(p)</tt> rule is defined to catch syntax errors.
|
|
1718 See the error handling section below for more detail.
|
|
1719
|
|
1720 <p>
|
|
1721 To build the parser, call the <tt>yacc.yacc()</tt> function. This
|
|
1722 function looks at the module and attempts to construct all of the LR
|
|
1723 parsing tables for the grammar you have specified. The first
|
|
1724 time <tt>yacc.yacc()</tt> is invoked, you will get a message such as
|
|
1725 this:
|
|
1726
|
|
1727 <blockquote>
|
|
1728 <pre>
|
|
1729 $ python calcparse.py
|
|
1730 Generating LALR tables
|
|
1731 calc >
|
|
1732 </pre>
|
|
1733 </blockquote>
|
|
1734
|
|
1735 <p>
|
|
1736 Since table construction is relatively expensive (especially for large
|
|
1737 grammars), the resulting parsing table is written to
|
|
1738 a file called <tt>parsetab.py</tt>. In addition, a
|
|
1739 debugging file called <tt>parser.out</tt> is created. On subsequent
|
|
1740 executions, <tt>yacc</tt> will reload the table from
|
|
1741 <tt>parsetab.py</tt> unless it has detected a change in the underlying
|
|
1742 grammar (in which case the tables and <tt>parsetab.py</tt> file are
|
|
1743 regenerated). Both of these files are written to the same directory
|
|
1744 as the module in which the parser is specified.
|
|
1745 The name of the <tt>parsetab</tt> module can be changed using the
|
|
1746 <tt>tabmodule</tt> keyword argument to <tt>yacc()</tt>. For example:
|
|
1747 </p>
|
|
1748
|
|
1749 <blockquote>
|
|
1750 <pre>
|
|
1751 parser = yacc.yacc(tabmodule='fooparsetab')
|
|
1752 </pre>
|
|
1753 </blockquote>
|
|
1754
|
|
1755 <p>
|
|
1756 If any errors are detected in your grammar specification, <tt>yacc.py</tt> will produce
|
|
1757 diagnostic messages and possibly raise an exception. Some of the errors that can be detected include:
|
|
1758
|
|
1759 <ul>
|
|
1760 <li>Duplicated function names (if more than one rule function have the same name in the grammar file).
|
|
1761 <li>Shift/reduce and reduce/reduce conflicts generated by ambiguous grammars.
|
|
1762 <li>Badly specified grammar rules.
|
|
1763 <li>Infinite recursion (rules that can never terminate).
|
|
1764 <li>Unused rules and tokens
|
|
1765 <li>Undefined rules and tokens
|
|
1766 </ul>
|
|
1767
|
|
1768 The next few sections discuss grammar specification in more detail.
|
|
1769
|
|
1770 <p>
|
|
1771 The final part of the example shows how to actually run the parser
|
|
1772 created by
|
|
1773 <tt>yacc()</tt>. To run the parser, you simply have to call
|
|
1774 the <tt>parse()</tt> with a string of input text. This will run all
|
|
1775 of the grammar rules and return the result of the entire parse. This
|
|
1776 result return is the value assigned to <tt>p[0]</tt> in the starting
|
|
1777 grammar rule.
|
|
1778
|
|
1779 <H3><a name="ply_nn25"></a>6.2 Combining Grammar Rule Functions</H3>
|
|
1780
|
|
1781
|
|
1782 When grammar rules are similar, they can be combined into a single function.
|
|
1783 For example, consider the two rules in our earlier example:
|
|
1784
|
|
1785 <blockquote>
|
|
1786 <pre>
|
|
1787 def p_expression_plus(p):
|
|
1788 'expression : expression PLUS term'
|
|
1789 p[0] = p[1] + p[3]
|
|
1790
|
|
1791 def p_expression_minus(t):
|
|
1792 'expression : expression MINUS term'
|
|
1793 p[0] = p[1] - p[3]
|
|
1794 </pre>
|
|
1795 </blockquote>
|
|
1796
|
|
1797 Instead of writing two functions, you might write a single function like this:
|
|
1798
|
|
1799 <blockquote>
|
|
1800 <pre>
|
|
1801 def p_expression(p):
|
|
1802 '''expression : expression PLUS term
|
|
1803 | expression MINUS term'''
|
|
1804 if p[2] == '+':
|
|
1805 p[0] = p[1] + p[3]
|
|
1806 elif p[2] == '-':
|
|
1807 p[0] = p[1] - p[3]
|
|
1808 </pre>
|
|
1809 </blockquote>
|
|
1810
|
|
1811 In general, the doc string for any given function can contain multiple grammar rules. So, it would
|
|
1812 have also been legal (although possibly confusing) to write this:
|
|
1813
|
|
1814 <blockquote>
|
|
1815 <pre>
|
|
1816 def p_binary_operators(p):
|
|
1817 '''expression : expression PLUS term
|
|
1818 | expression MINUS term
|
|
1819 term : term TIMES factor
|
|
1820 | term DIVIDE factor'''
|
|
1821 if p[2] == '+':
|
|
1822 p[0] = p[1] + p[3]
|
|
1823 elif p[2] == '-':
|
|
1824 p[0] = p[1] - p[3]
|
|
1825 elif p[2] == '*':
|
|
1826 p[0] = p[1] * p[3]
|
|
1827 elif p[2] == '/':
|
|
1828 p[0] = p[1] / p[3]
|
|
1829 </pre>
|
|
1830 </blockquote>
|
|
1831
|
|
1832 When combining grammar rules into a single function, it is usually a good idea for all of the rules to have
|
|
1833 a similar structure (e.g., the same number of terms). Otherwise, the corresponding action code may be more
|
|
1834 complicated than necessary. However, it is possible to handle simple cases using len(). For example:
|
|
1835
|
|
1836 <blockquote>
|
|
1837 <pre>
|
|
1838 def p_expressions(p):
|
|
1839 '''expression : expression MINUS expression
|
|
1840 | MINUS expression'''
|
|
1841 if (len(p) == 4):
|
|
1842 p[0] = p[1] - p[3]
|
|
1843 elif (len(p) == 3):
|
|
1844 p[0] = -p[2]
|
|
1845 </pre>
|
|
1846 </blockquote>
|
|
1847
|
|
1848 If parsing performance is a concern, you should resist the urge to put
|
|
1849 too much conditional processing into a single grammar rule as shown in
|
|
1850 these examples. When you add checks to see which grammar rule is
|
|
1851 being handled, you are actually duplicating the work that the parser
|
|
1852 has already performed (i.e., the parser already knows exactly what rule it
|
|
1853 matched). You can eliminate this overhead by using a
|
|
1854 separate <tt>p_rule()</tt> function for each grammar rule.
|
|
1855
|
|
1856 <H3><a name="ply_nn26"></a>6.3 Character Literals</H3>
|
|
1857
|
|
1858
|
|
1859 If desired, a grammar may contain tokens defined as single character literals. For example:
|
|
1860
|
|
1861 <blockquote>
|
|
1862 <pre>
|
|
1863 def p_binary_operators(p):
|
|
1864 '''expression : expression '+' term
|
|
1865 | expression '-' term
|
|
1866 term : term '*' factor
|
|
1867 | term '/' factor'''
|
|
1868 if p[2] == '+':
|
|
1869 p[0] = p[1] + p[3]
|
|
1870 elif p[2] == '-':
|
|
1871 p[0] = p[1] - p[3]
|
|
1872 elif p[2] == '*':
|
|
1873 p[0] = p[1] * p[3]
|
|
1874 elif p[2] == '/':
|
|
1875 p[0] = p[1] / p[3]
|
|
1876 </pre>
|
|
1877 </blockquote>
|
|
1878
|
|
1879 A character literal must be enclosed in quotes such as <tt>'+'</tt>. In addition, if literals are used, they must be declared in the
|
|
1880 corresponding <tt>lex</tt> file through the use of a special <tt>literals</tt> declaration.
|
|
1881
|
|
1882 <blockquote>
|
|
1883 <pre>
|
|
1884 # Literals. Should be placed in module given to lex()
|
|
1885 literals = ['+','-','*','/' ]
|
|
1886 </pre>
|
|
1887 </blockquote>
|
|
1888
|
|
1889 <b>Character literals are limited to a single character</b>. Thus, it is not legal to specify literals such as <tt>'<='</tt> or <tt>'=='</tt>. For this, use
|
|
1890 the normal lexing rules (e.g., define a rule such as <tt>t_EQ = r'=='</tt>).
|
|
1891
|
|
1892 <H3><a name="ply_nn26"></a>6.4 Empty Productions</H3>
|
|
1893
|
|
1894
|
|
1895 <tt>yacc.py</tt> can handle empty productions by defining a rule like this:
|
|
1896
|
|
1897 <blockquote>
|
|
1898 <pre>
|
|
1899 def p_empty(p):
|
|
1900 'empty :'
|
|
1901 pass
|
|
1902 </pre>
|
|
1903 </blockquote>
|
|
1904
|
|
1905 Now to use the empty production, simply use 'empty' as a symbol. For example:
|
|
1906
|
|
1907 <blockquote>
|
|
1908 <pre>
|
|
1909 def p_optitem(p):
|
|
1910 'optitem : item'
|
|
1911 ' | empty'
|
|
1912 ...
|
|
1913 </pre>
|
|
1914 </blockquote>
|
|
1915
|
|
1916 Note: You can write empty rules anywhere by simply specifying an empty
|
|
1917 right hand side. However, I personally find that writing an "empty"
|
|
1918 rule and using "empty" to denote an empty production is easier to read
|
|
1919 and more clearly states your intentions.
|
|
1920
|
|
1921 <H3><a name="ply_nn28"></a>6.5 Changing the starting symbol</H3>
|
|
1922
|
|
1923
|
|
1924 Normally, the first rule found in a yacc specification defines the starting grammar rule (top level rule). To change this, simply
|
|
1925 supply a <tt>start</tt> specifier in your file. For example:
|
|
1926
|
|
1927 <blockquote>
|
|
1928 <pre>
|
|
1929 start = 'foo'
|
|
1930
|
|
1931 def p_bar(p):
|
|
1932 'bar : A B'
|
|
1933
|
|
1934 # This is the starting rule due to the start specifier above
|
|
1935 def p_foo(p):
|
|
1936 'foo : bar X'
|
|
1937 ...
|
|
1938 </pre>
|
|
1939 </blockquote>
|
|
1940
|
|
1941 The use of a <tt>start</tt> specifier may be useful during debugging
|
|
1942 since you can use it to have yacc build a subset of a larger grammar.
|
|
1943 For this purpose, it is also possible to specify a starting symbol as
|
|
1944 an argument to <tt>yacc()</tt>. For example:
|
|
1945
|
|
1946 <blockquote>
|
|
1947 <pre>
|
|
1948 parser = yacc.yacc(start='foo')
|
|
1949 </pre>
|
|
1950 </blockquote>
|
|
1951
|
|
1952 <H3><a name="ply_nn27"></a>6.6 Dealing With Ambiguous Grammars</H3>
|
|
1953
|
|
1954
|
|
1955 The expression grammar given in the earlier example has been written
|
|
1956 in a special format to eliminate ambiguity. However, in many
|
|
1957 situations, it is extremely difficult or awkward to write grammars in
|
|
1958 this format. A much more natural way to express the grammar is in a
|
|
1959 more compact form like this:
|
|
1960
|
|
1961 <blockquote>
|
|
1962 <pre>
|
|
1963 expression : expression PLUS expression
|
|
1964 | expression MINUS expression
|
|
1965 | expression TIMES expression
|
|
1966 | expression DIVIDE expression
|
|
1967 | LPAREN expression RPAREN
|
|
1968 | NUMBER
|
|
1969 </pre>
|
|
1970 </blockquote>
|
|
1971
|
|
1972 Unfortunately, this grammar specification is ambiguous. For example,
|
|
1973 if you are parsing the string "3 * 4 + 5", there is no way to tell how
|
|
1974 the operators are supposed to be grouped. For example, does the
|
|
1975 expression mean "(3 * 4) + 5" or is it "3 * (4+5)"?
|
|
1976
|
|
1977 <p>
|
|
1978 When an ambiguous grammar is given to <tt>yacc.py</tt> it will print
|
|
1979 messages about "shift/reduce conflicts" or "reduce/reduce conflicts".
|
|
1980 A shift/reduce conflict is caused when the parser generator can't
|
|
1981 decide whether or not to reduce a rule or shift a symbol on the
|
|
1982 parsing stack. For example, consider the string "3 * 4 + 5" and the
|
|
1983 internal parsing stack:
|
|
1984
|
|
1985 <blockquote>
|
|
1986 <pre>
|
|
1987 Step Symbol Stack Input Tokens Action
|
|
1988 ---- --------------------- --------------------- -------------------------------
|
|
1989 1 $ 3 * 4 + 5$ Shift 3
|
|
1990 2 $ 3 * 4 + 5$ Reduce : expression : NUMBER
|
|
1991 3 $ expr * 4 + 5$ Shift *
|
|
1992 4 $ expr * 4 + 5$ Shift 4
|
|
1993 5 $ expr * 4 + 5$ Reduce: expression : NUMBER
|
|
1994 6 $ expr * expr + 5$ SHIFT/REDUCE CONFLICT ????
|
|
1995 </pre>
|
|
1996 </blockquote>
|
|
1997
|
|
1998 In this case, when the parser reaches step 6, it has two options. One
|
|
1999 is to reduce the rule <tt>expr : expr * expr</tt> on the stack. The
|
|
2000 other option is to shift the token <tt>+</tt> on the stack. Both
|
|
2001 options are perfectly legal from the rules of the
|
|
2002 context-free-grammar.
|
|
2003
|
|
2004 <p>
|
|
2005 By default, all shift/reduce conflicts are resolved in favor of
|
|
2006 shifting. Therefore, in the above example, the parser will always
|
|
2007 shift the <tt>+</tt> instead of reducing. Although this strategy
|
|
2008 works in many cases (for example, the case of
|
|
2009 "if-then" versus "if-then-else"), it is not enough for arithmetic expressions. In fact,
|
|
2010 in the above example, the decision to shift <tt>+</tt> is completely
|
|
2011 wrong---we should have reduced <tt>expr * expr</tt> since
|
|
2012 multiplication has higher mathematical precedence than addition.
|
|
2013
|
|
2014 <p>To resolve ambiguity, especially in expression
|
|
2015 grammars, <tt>yacc.py</tt> allows individual tokens to be assigned a
|
|
2016 precedence level and associativity. This is done by adding a variable
|
|
2017 <tt>precedence</tt> to the grammar file like this:
|
|
2018
|
|
2019 <blockquote>
|
|
2020 <pre>
|
|
2021 precedence = (
|
|
2022 ('left', 'PLUS', 'MINUS'),
|
|
2023 ('left', 'TIMES', 'DIVIDE'),
|
|
2024 )
|
|
2025 </pre>
|
|
2026 </blockquote>
|
|
2027
|
|
2028 This declaration specifies that <tt>PLUS</tt>/<tt>MINUS</tt> have the
|
|
2029 same precedence level and are left-associative and that
|
|
2030 <tt>TIMES</tt>/<tt>DIVIDE</tt> have the same precedence and are
|
|
2031 left-associative. Within the <tt>precedence</tt> declaration, tokens
|
|
2032 are ordered from lowest to highest precedence. Thus, this declaration
|
|
2033 specifies that <tt>TIMES</tt>/<tt>DIVIDE</tt> have higher precedence
|
|
2034 than <tt>PLUS</tt>/<tt>MINUS</tt> (since they appear later in the
|
|
2035 precedence specification).
|
|
2036
|
|
2037 <p>
|
|
2038 The precedence specification works by associating a numerical
|
|
2039 precedence level value and associativity direction to the listed
|
|
2040 tokens. For example, in the above example you get:
|
|
2041
|
|
2042 <blockquote>
|
|
2043 <pre>
|
|
2044 PLUS : level = 1, assoc = 'left'
|
|
2045 MINUS : level = 1, assoc = 'left'
|
|
2046 TIMES : level = 2, assoc = 'left'
|
|
2047 DIVIDE : level = 2, assoc = 'left'
|
|
2048 </pre>
|
|
2049 </blockquote>
|
|
2050
|
|
2051 These values are then used to attach a numerical precedence value and
|
|
2052 associativity direction to each grammar rule. <em>This is always
|
|
2053 determined by looking at the precedence of the right-most terminal
|
|
2054 symbol.</em> For example:
|
|
2055
|
|
2056 <blockquote>
|
|
2057 <pre>
|
|
2058 expression : expression PLUS expression # level = 1, left
|
|
2059 | expression MINUS expression # level = 1, left
|
|
2060 | expression TIMES expression # level = 2, left
|
|
2061 | expression DIVIDE expression # level = 2, left
|
|
2062 | LPAREN expression RPAREN # level = None (not specified)
|
|
2063 | NUMBER # level = None (not specified)
|
|
2064 </pre>
|
|
2065 </blockquote>
|
|
2066
|
|
2067 When shift/reduce conflicts are encountered, the parser generator resolves the conflict by
|
|
2068 looking at the precedence rules and associativity specifiers.
|
|
2069
|
|
2070 <p>
|
|
2071 <ol>
|
|
2072 <li>If the current token has higher precedence than the rule on the stack, it is shifted.
|
|
2073 <li>If the grammar rule on the stack has higher precedence, the rule is reduced.
|
|
2074 <li>If the current token and the grammar rule have the same precedence, the
|
|
2075 rule is reduced for left associativity, whereas the token is shifted for right associativity.
|
|
2076 <li>If nothing is known about the precedence, shift/reduce conflicts are resolved in
|
|
2077 favor of shifting (the default).
|
|
2078 </ol>
|
|
2079
|
|
2080 For example, if "expression PLUS expression" has been parsed and the
|
|
2081 next token is "TIMES", the action is going to be a shift because
|
|
2082 "TIMES" has a higher precedence level than "PLUS". On the other hand,
|
|
2083 if "expression TIMES expression" has been parsed and the next token is
|
|
2084 "PLUS", the action is going to be reduce because "PLUS" has a lower
|
|
2085 precedence than "TIMES."
|
|
2086
|
|
2087 <p>
|
|
2088 When shift/reduce conflicts are resolved using the first three
|
|
2089 techniques (with the help of precedence rules), <tt>yacc.py</tt> will
|
|
2090 report no errors or conflicts in the grammar (although it will print
|
|
2091 some information in the <tt>parser.out</tt> debugging file).
|
|
2092
|
|
2093 <p>
|
|
2094 One problem with the precedence specifier technique is that it is
|
|
2095 sometimes necessary to change the precedence of an operator in certain
|
|
2096 contexts. For example, consider a unary-minus operator in "3 + 4 *
|
|
2097 -5". Mathematically, the unary minus is normally given a very high
|
|
2098 precedence--being evaluated before the multiply. However, in our
|
|
2099 precedence specifier, MINUS has a lower precedence than TIMES. To
|
|
2100 deal with this, precedence rules can be given for so-called "fictitious tokens"
|
|
2101 like this:
|
|
2102
|
|
2103 <blockquote>
|
|
2104 <pre>
|
|
2105 precedence = (
|
|
2106 ('left', 'PLUS', 'MINUS'),
|
|
2107 ('left', 'TIMES', 'DIVIDE'),
|
|
2108 ('right', 'UMINUS'), # Unary minus operator
|
|
2109 )
|
|
2110 </pre>
|
|
2111 </blockquote>
|
|
2112
|
|
2113 Now, in the grammar file, we can write our unary minus rule like this:
|
|
2114
|
|
2115 <blockquote>
|
|
2116 <pre>
|
|
2117 def p_expr_uminus(p):
|
|
2118 'expression : MINUS expression %prec UMINUS'
|
|
2119 p[0] = -p[2]
|
|
2120 </pre>
|
|
2121 </blockquote>
|
|
2122
|
|
2123 In this case, <tt>%prec UMINUS</tt> overrides the default rule precedence--setting it to that
|
|
2124 of UMINUS in the precedence specifier.
|
|
2125
|
|
2126 <p>
|
|
2127 At first, the use of UMINUS in this example may appear very confusing.
|
|
2128 UMINUS is not an input token or a grammar rule. Instead, you should
|
|
2129 think of it as the name of a special marker in the precedence table. When you use the <tt>%prec</tt> qualifier, you're simply
|
|
2130 telling yacc that you want the precedence of the expression to be the same as for this special marker instead of the usual precedence.
|
|
2131
|
|
2132 <p>
|
|
2133 It is also possible to specify non-associativity in the <tt>precedence</tt> table. This would
|
|
2134 be used when you <em>don't</em> want operations to chain together. For example, suppose
|
|
2135 you wanted to support comparison operators like <tt><</tt> and <tt>></tt> but you didn't want to allow
|
|
2136 combinations like <tt>a < b < c</tt>. To do this, simply specify a rule like this:
|
|
2137
|
|
2138 <blockquote>
|
|
2139 <pre>
|
|
2140 precedence = (
|
|
2141 ('nonassoc', 'LESSTHAN', 'GREATERTHAN'), # Nonassociative operators
|
|
2142 ('left', 'PLUS', 'MINUS'),
|
|
2143 ('left', 'TIMES', 'DIVIDE'),
|
|
2144 ('right', 'UMINUS'), # Unary minus operator
|
|
2145 )
|
|
2146 </pre>
|
|
2147 </blockquote>
|
|
2148
|
|
2149 <p>
|
|
2150 If you do this, the occurrence of input text such as <tt> a < b < c</tt> will result in a syntax error. However, simple
|
|
2151 expressions such as <tt>a < b</tt> will still be fine.
|
|
2152
|
|
2153 <p>
|
|
2154 Reduce/reduce conflicts are caused when there are multiple grammar
|
|
2155 rules that can be applied to a given set of symbols. This kind of
|
|
2156 conflict is almost always bad and is always resolved by picking the
|
|
2157 rule that appears first in the grammar file. Reduce/reduce conflicts
|
|
2158 are almost always caused when different sets of grammar rules somehow
|
|
2159 generate the same set of symbols. For example:
|
|
2160
|
|
2161 <blockquote>
|
|
2162 <pre>
|
|
2163 assignment : ID EQUALS NUMBER
|
|
2164 | ID EQUALS expression
|
|
2165
|
|
2166 expression : expression PLUS expression
|
|
2167 | expression MINUS expression
|
|
2168 | expression TIMES expression
|
|
2169 | expression DIVIDE expression
|
|
2170 | LPAREN expression RPAREN
|
|
2171 | NUMBER
|
|
2172 </pre>
|
|
2173 </blockquote>
|
|
2174
|
|
2175 In this case, a reduce/reduce conflict exists between these two rules:
|
|
2176
|
|
2177 <blockquote>
|
|
2178 <pre>
|
|
2179 assignment : ID EQUALS NUMBER
|
|
2180 expression : NUMBER
|
|
2181 </pre>
|
|
2182 </blockquote>
|
|
2183
|
|
2184 For example, if you wrote "a = 5", the parser can't figure out if this
|
|
2185 is supposed to be reduced as <tt>assignment : ID EQUALS NUMBER</tt> or
|
|
2186 whether it's supposed to reduce the 5 as an expression and then reduce
|
|
2187 the rule <tt>assignment : ID EQUALS expression</tt>.
|
|
2188
|
|
2189 <p>
|
|
2190 It should be noted that reduce/reduce conflicts are notoriously
|
|
2191 difficult to spot simply looking at the input grammar. When a
|
|
2192 reduce/reduce conflict occurs, <tt>yacc()</tt> will try to help by
|
|
2193 printing a warning message such as this:
|
|
2194
|
|
2195 <blockquote>
|
|
2196 <pre>
|
|
2197 WARNING: 1 reduce/reduce conflict
|
|
2198 WARNING: reduce/reduce conflict in state 15 resolved using rule (assignment -> ID EQUALS NUMBER)
|
|
2199 WARNING: rejected rule (expression -> NUMBER)
|
|
2200 </pre>
|
|
2201 </blockquote>
|
|
2202
|
|
2203 This message identifies the two rules that are in conflict. However,
|
|
2204 it may not tell you how the parser arrived at such a state. To try
|
|
2205 and figure it out, you'll probably have to look at your grammar and
|
|
2206 the contents of the
|
|
2207 <tt>parser.out</tt> debugging file with an appropriately high level of
|
|
2208 caffeination.
|
|
2209
|
|
2210 <H3><a name="ply_nn28"></a>6.7 The parser.out file</H3>
|
|
2211
|
|
2212
|
|
2213 Tracking down shift/reduce and reduce/reduce conflicts is one of the finer pleasures of using an LR
|
|
2214 parsing algorithm. To assist in debugging, <tt>yacc.py</tt> creates a debugging file called
|
|
2215 'parser.out' when it generates the parsing table. The contents of this file look like the following:
|
|
2216
|
|
2217 <blockquote>
|
|
2218 <pre>
|
|
2219 Unused terminals:
|
|
2220
|
|
2221
|
|
2222 Grammar
|
|
2223
|
|
2224 Rule 1 expression -> expression PLUS expression
|
|
2225 Rule 2 expression -> expression MINUS expression
|
|
2226 Rule 3 expression -> expression TIMES expression
|
|
2227 Rule 4 expression -> expression DIVIDE expression
|
|
2228 Rule 5 expression -> NUMBER
|
|
2229 Rule 6 expression -> LPAREN expression RPAREN
|
|
2230
|
|
2231 Terminals, with rules where they appear
|
|
2232
|
|
2233 TIMES : 3
|
|
2234 error :
|
|
2235 MINUS : 2
|
|
2236 RPAREN : 6
|
|
2237 LPAREN : 6
|
|
2238 DIVIDE : 4
|
|
2239 PLUS : 1
|
|
2240 NUMBER : 5
|
|
2241
|
|
2242 Nonterminals, with rules where they appear
|
|
2243
|
|
2244 expression : 1 1 2 2 3 3 4 4 6 0
|
|
2245
|
|
2246
|
|
2247 Parsing method: LALR
|
|
2248
|
|
2249
|
|
2250 state 0
|
|
2251
|
|
2252 S' -> . expression
|
|
2253 expression -> . expression PLUS expression
|
|
2254 expression -> . expression MINUS expression
|
|
2255 expression -> . expression TIMES expression
|
|
2256 expression -> . expression DIVIDE expression
|
|
2257 expression -> . NUMBER
|
|
2258 expression -> . LPAREN expression RPAREN
|
|
2259
|
|
2260 NUMBER shift and go to state 3
|
|
2261 LPAREN shift and go to state 2
|
|
2262
|
|
2263
|
|
2264 state 1
|
|
2265
|
|
2266 S' -> expression .
|
|
2267 expression -> expression . PLUS expression
|
|
2268 expression -> expression . MINUS expression
|
|
2269 expression -> expression . TIMES expression
|
|
2270 expression -> expression . DIVIDE expression
|
|
2271
|
|
2272 PLUS shift and go to state 6
|
|
2273 MINUS shift and go to state 5
|
|
2274 TIMES shift and go to state 4
|
|
2275 DIVIDE shift and go to state 7
|
|
2276
|
|
2277
|
|
2278 state 2
|
|
2279
|
|
2280 expression -> LPAREN . expression RPAREN
|
|
2281 expression -> . expression PLUS expression
|
|
2282 expression -> . expression MINUS expression
|
|
2283 expression -> . expression TIMES expression
|
|
2284 expression -> . expression DIVIDE expression
|
|
2285 expression -> . NUMBER
|
|
2286 expression -> . LPAREN expression RPAREN
|
|
2287
|
|
2288 NUMBER shift and go to state 3
|
|
2289 LPAREN shift and go to state 2
|
|
2290
|
|
2291
|
|
2292 state 3
|
|
2293
|
|
2294 expression -> NUMBER .
|
|
2295
|
|
2296 $ reduce using rule 5
|
|
2297 PLUS reduce using rule 5
|
|
2298 MINUS reduce using rule 5
|
|
2299 TIMES reduce using rule 5
|
|
2300 DIVIDE reduce using rule 5
|
|
2301 RPAREN reduce using rule 5
|
|
2302
|
|
2303
|
|
2304 state 4
|
|
2305
|
|
2306 expression -> expression TIMES . expression
|
|
2307 expression -> . expression PLUS expression
|
|
2308 expression -> . expression MINUS expression
|
|
2309 expression -> . expression TIMES expression
|
|
2310 expression -> . expression DIVIDE expression
|
|
2311 expression -> . NUMBER
|
|
2312 expression -> . LPAREN expression RPAREN
|
|
2313
|
|
2314 NUMBER shift and go to state 3
|
|
2315 LPAREN shift and go to state 2
|
|
2316
|
|
2317
|
|
2318 state 5
|
|
2319
|
|
2320 expression -> expression MINUS . expression
|
|
2321 expression -> . expression PLUS expression
|
|
2322 expression -> . expression MINUS expression
|
|
2323 expression -> . expression TIMES expression
|
|
2324 expression -> . expression DIVIDE expression
|
|
2325 expression -> . NUMBER
|
|
2326 expression -> . LPAREN expression RPAREN
|
|
2327
|
|
2328 NUMBER shift and go to state 3
|
|
2329 LPAREN shift and go to state 2
|
|
2330
|
|
2331
|
|
2332 state 6
|
|
2333
|
|
2334 expression -> expression PLUS . expression
|
|
2335 expression -> . expression PLUS expression
|
|
2336 expression -> . expression MINUS expression
|
|
2337 expression -> . expression TIMES expression
|
|
2338 expression -> . expression DIVIDE expression
|
|
2339 expression -> . NUMBER
|
|
2340 expression -> . LPAREN expression RPAREN
|
|
2341
|
|
2342 NUMBER shift and go to state 3
|
|
2343 LPAREN shift and go to state 2
|
|
2344
|
|
2345
|
|
2346 state 7
|
|
2347
|
|
2348 expression -> expression DIVIDE . expression
|
|
2349 expression -> . expression PLUS expression
|
|
2350 expression -> . expression MINUS expression
|
|
2351 expression -> . expression TIMES expression
|
|
2352 expression -> . expression DIVIDE expression
|
|
2353 expression -> . NUMBER
|
|
2354 expression -> . LPAREN expression RPAREN
|
|
2355
|
|
2356 NUMBER shift and go to state 3
|
|
2357 LPAREN shift and go to state 2
|
|
2358
|
|
2359
|
|
2360 state 8
|
|
2361
|
|
2362 expression -> LPAREN expression . RPAREN
|
|
2363 expression -> expression . PLUS expression
|
|
2364 expression -> expression . MINUS expression
|
|
2365 expression -> expression . TIMES expression
|
|
2366 expression -> expression . DIVIDE expression
|
|
2367
|
|
2368 RPAREN shift and go to state 13
|
|
2369 PLUS shift and go to state 6
|
|
2370 MINUS shift and go to state 5
|
|
2371 TIMES shift and go to state 4
|
|
2372 DIVIDE shift and go to state 7
|
|
2373
|
|
2374
|
|
2375 state 9
|
|
2376
|
|
2377 expression -> expression TIMES expression .
|
|
2378 expression -> expression . PLUS expression
|
|
2379 expression -> expression . MINUS expression
|
|
2380 expression -> expression . TIMES expression
|
|
2381 expression -> expression . DIVIDE expression
|
|
2382
|
|
2383 $ reduce using rule 3
|
|
2384 PLUS reduce using rule 3
|
|
2385 MINUS reduce using rule 3
|
|
2386 TIMES reduce using rule 3
|
|
2387 DIVIDE reduce using rule 3
|
|
2388 RPAREN reduce using rule 3
|
|
2389
|
|
2390 ! PLUS [ shift and go to state 6 ]
|
|
2391 ! MINUS [ shift and go to state 5 ]
|
|
2392 ! TIMES [ shift and go to state 4 ]
|
|
2393 ! DIVIDE [ shift and go to state 7 ]
|
|
2394
|
|
2395 state 10
|
|
2396
|
|
2397 expression -> expression MINUS expression .
|
|
2398 expression -> expression . PLUS expression
|
|
2399 expression -> expression . MINUS expression
|
|
2400 expression -> expression . TIMES expression
|
|
2401 expression -> expression . DIVIDE expression
|
|
2402
|
|
2403 $ reduce using rule 2
|
|
2404 PLUS reduce using rule 2
|
|
2405 MINUS reduce using rule 2
|
|
2406 RPAREN reduce using rule 2
|
|
2407 TIMES shift and go to state 4
|
|
2408 DIVIDE shift and go to state 7
|
|
2409
|
|
2410 ! TIMES [ reduce using rule 2 ]
|
|
2411 ! DIVIDE [ reduce using rule 2 ]
|
|
2412 ! PLUS [ shift and go to state 6 ]
|
|
2413 ! MINUS [ shift and go to state 5 ]
|
|
2414
|
|
2415 state 11
|
|
2416
|
|
2417 expression -> expression PLUS expression .
|
|
2418 expression -> expression . PLUS expression
|
|
2419 expression -> expression . MINUS expression
|
|
2420 expression -> expression . TIMES expression
|
|
2421 expression -> expression . DIVIDE expression
|
|
2422
|
|
2423 $ reduce using rule 1
|
|
2424 PLUS reduce using rule 1
|
|
2425 MINUS reduce using rule 1
|
|
2426 RPAREN reduce using rule 1
|
|
2427 TIMES shift and go to state 4
|
|
2428 DIVIDE shift and go to state 7
|
|
2429
|
|
2430 ! TIMES [ reduce using rule 1 ]
|
|
2431 ! DIVIDE [ reduce using rule 1 ]
|
|
2432 ! PLUS [ shift and go to state 6 ]
|
|
2433 ! MINUS [ shift and go to state 5 ]
|
|
2434
|
|
2435 state 12
|
|
2436
|
|
2437 expression -> expression DIVIDE expression .
|
|
2438 expression -> expression . PLUS expression
|
|
2439 expression -> expression . MINUS expression
|
|
2440 expression -> expression . TIMES expression
|
|
2441 expression -> expression . DIVIDE expression
|
|
2442
|
|
2443 $ reduce using rule 4
|
|
2444 PLUS reduce using rule 4
|
|
2445 MINUS reduce using rule 4
|
|
2446 TIMES reduce using rule 4
|
|
2447 DIVIDE reduce using rule 4
|
|
2448 RPAREN reduce using rule 4
|
|
2449
|
|
2450 ! PLUS [ shift and go to state 6 ]
|
|
2451 ! MINUS [ shift and go to state 5 ]
|
|
2452 ! TIMES [ shift and go to state 4 ]
|
|
2453 ! DIVIDE [ shift and go to state 7 ]
|
|
2454
|
|
2455 state 13
|
|
2456
|
|
2457 expression -> LPAREN expression RPAREN .
|
|
2458
|
|
2459 $ reduce using rule 6
|
|
2460 PLUS reduce using rule 6
|
|
2461 MINUS reduce using rule 6
|
|
2462 TIMES reduce using rule 6
|
|
2463 DIVIDE reduce using rule 6
|
|
2464 RPAREN reduce using rule 6
|
|
2465 </pre>
|
|
2466 </blockquote>
|
|
2467
|
|
2468 The different states that appear in this file are a representation of
|
|
2469 every possible sequence of valid input tokens allowed by the grammar.
|
|
2470 When receiving input tokens, the parser is building up a stack and
|
|
2471 looking for matching rules. Each state keeps track of the grammar
|
|
2472 rules that might be in the process of being matched at that point. Within each
|
|
2473 rule, the "." character indicates the current location of the parse
|
|
2474 within that rule. In addition, the actions for each valid input token
|
|
2475 are listed. When a shift/reduce or reduce/reduce conflict arises,
|
|
2476 rules <em>not</em> selected are prefixed with an !. For example:
|
|
2477
|
|
2478 <blockquote>
|
|
2479 <pre>
|
|
2480 ! TIMES [ reduce using rule 2 ]
|
|
2481 ! DIVIDE [ reduce using rule 2 ]
|
|
2482 ! PLUS [ shift and go to state 6 ]
|
|
2483 ! MINUS [ shift and go to state 5 ]
|
|
2484 </pre>
|
|
2485 </blockquote>
|
|
2486
|
|
2487 By looking at these rules (and with a little practice), you can usually track down the source
|
|
2488 of most parsing conflicts. It should also be stressed that not all shift-reduce conflicts are
|
|
2489 bad. However, the only way to be sure that they are resolved correctly is to look at <tt>parser.out</tt>.
|
|
2490
|
|
2491 <H3><a name="ply_nn29"></a>6.8 Syntax Error Handling</H3>
|
|
2492
|
|
2493
|
|
2494 If you are creating a parser for production use, the handling of
|
|
2495 syntax errors is important. As a general rule, you don't want a
|
|
2496 parser to simply throw up its hands and stop at the first sign of
|
|
2497 trouble. Instead, you want it to report the error, recover if possible, and
|
|
2498 continue parsing so that all of the errors in the input get reported
|
|
2499 to the user at once. This is the standard behavior found in compilers
|
|
2500 for languages such as C, C++, and Java.
|
|
2501
|
|
2502 In PLY, when a syntax error occurs during parsing, the error is immediately
|
|
2503 detected (i.e., the parser does not read any more tokens beyond the
|
|
2504 source of the error). However, at this point, the parser enters a
|
|
2505 recovery mode that can be used to try and continue further parsing.
|
|
2506 As a general rule, error recovery in LR parsers is a delicate
|
|
2507 topic that involves ancient rituals and black-magic. The recovery mechanism
|
|
2508 provided by <tt>yacc.py</tt> is comparable to Unix yacc so you may want
|
|
2509 consult a book like O'Reilly's "Lex and Yacc" for some of the finer details.
|
|
2510
|
|
2511 <p>
|
|
2512 When a syntax error occurs, <tt>yacc.py</tt> performs the following steps:
|
|
2513
|
|
2514 <ol>
|
|
2515 <li>On the first occurrence of an error, the user-defined <tt>p_error()</tt> function
|
|
2516 is called with the offending token as an argument. However, if the syntax error is due to
|
|
2517 reaching the end-of-file, <tt>p_error()</tt> is called with an
|
|
2518 argument of <tt>None</tt>.
|
|
2519 Afterwards, the parser enters
|
|
2520 an "error-recovery" mode in which it will not make future calls to <tt>p_error()</tt> until it
|
|
2521 has successfully shifted at least 3 tokens onto the parsing stack.
|
|
2522
|
|
2523 <p>
|
|
2524 <li>If no recovery action is taken in <tt>p_error()</tt>, the offending lookahead token is replaced
|
|
2525 with a special <tt>error</tt> token.
|
|
2526
|
|
2527 <p>
|
|
2528 <li>If the offending lookahead token is already set to <tt>error</tt>, the top item of the parsing stack is
|
|
2529 deleted.
|
|
2530
|
|
2531 <p>
|
|
2532 <li>If the entire parsing stack is unwound, the parser enters a restart state and attempts to start
|
|
2533 parsing from its initial state.
|
|
2534
|
|
2535 <p>
|
|
2536 <li>If a grammar rule accepts <tt>error</tt> as a token, it will be
|
|
2537 shifted onto the parsing stack.
|
|
2538
|
|
2539 <p>
|
|
2540 <li>If the top item of the parsing stack is <tt>error</tt>, lookahead tokens will be discarded until the
|
|
2541 parser can successfully shift a new symbol or reduce a rule involving <tt>error</tt>.
|
|
2542 </ol>
|
|
2543
|
|
2544 <H4><a name="ply_nn30"></a>6.8.1 Recovery and resynchronization with error rules</H4>
|
|
2545
|
|
2546
|
|
2547 The most well-behaved approach for handling syntax errors is to write grammar rules that include the <tt>error</tt>
|
|
2548 token. For example, suppose your language had a grammar rule for a print statement like this:
|
|
2549
|
|
2550 <blockquote>
|
|
2551 <pre>
|
|
2552 def p_statement_print(p):
|
|
2553 'statement : PRINT expr SEMI'
|
|
2554 ...
|
|
2555 </pre>
|
|
2556 </blockquote>
|
|
2557
|
|
2558 To account for the possibility of a bad expression, you might write an additional grammar rule like this:
|
|
2559
|
|
2560 <blockquote>
|
|
2561 <pre>
|
|
2562 def p_statement_print_error(p):
|
|
2563 'statement : PRINT error SEMI'
|
|
2564 print("Syntax error in print statement. Bad expression")
|
|
2565
|
|
2566 </pre>
|
|
2567 </blockquote>
|
|
2568
|
|
2569 In this case, the <tt>error</tt> token will match any sequence of
|
|
2570 tokens that might appear up to the first semicolon that is
|
|
2571 encountered. Once the semicolon is reached, the rule will be
|
|
2572 invoked and the <tt>error</tt> token will go away.
|
|
2573
|
|
2574 <p>
|
|
2575 This type of recovery is sometimes known as parser resynchronization.
|
|
2576 The <tt>error</tt> token acts as a wildcard for any bad input text and
|
|
2577 the token immediately following <tt>error</tt> acts as a
|
|
2578 synchronization token.
|
|
2579
|
|
2580 <p>
|
|
2581 It is important to note that the <tt>error</tt> token usually does not appear as the last token
|
|
2582 on the right in an error rule. For example:
|
|
2583
|
|
2584 <blockquote>
|
|
2585 <pre>
|
|
2586 def p_statement_print_error(p):
|
|
2587 'statement : PRINT error'
|
|
2588 print("Syntax error in print statement. Bad expression")
|
|
2589 </pre>
|
|
2590 </blockquote>
|
|
2591
|
|
2592 This is because the first bad token encountered will cause the rule to
|
|
2593 be reduced--which may make it difficult to recover if more bad tokens
|
|
2594 immediately follow.
|
|
2595
|
|
2596 <H4><a name="ply_nn31"></a>6.8.2 Panic mode recovery</H4>
|
|
2597
|
|
2598
|
|
2599 An alternative error recovery scheme is to enter a panic mode recovery in which tokens are
|
|
2600 discarded to a point where the parser might be able to recover in some sensible manner.
|
|
2601
|
|
2602 <p>
|
|
2603 Panic mode recovery is implemented entirely in the <tt>p_error()</tt> function. For example, this
|
|
2604 function starts discarding tokens until it reaches a closing '}'. Then, it restarts the
|
|
2605 parser in its initial state.
|
|
2606
|
|
2607 <blockquote>
|
|
2608 <pre>
|
|
2609 def p_error(p):
|
|
2610 print("Whoa. You are seriously hosed.")
|
|
2611 if not p:
|
|
2612 print("End of File!")
|
|
2613 return
|
|
2614
|
|
2615 # Read ahead looking for a closing '}'
|
|
2616 while True:
|
|
2617 tok = parser.token() # Get the next token
|
|
2618 if not tok or tok.type == 'RBRACE':
|
|
2619 break
|
|
2620 parser.restart()
|
|
2621 </pre>
|
|
2622 </blockquote>
|
|
2623
|
|
2624 <p>
|
|
2625 This function simply discards the bad token and tells the parser that the error was ok.
|
|
2626
|
|
2627 <blockquote>
|
|
2628 <pre>
|
|
2629 def p_error(p):
|
|
2630 if p:
|
|
2631 print("Syntax error at token", p.type)
|
|
2632 # Just discard the token and tell the parser it's okay.
|
|
2633 parser.errok()
|
|
2634 else:
|
|
2635 print("Syntax error at EOF")
|
|
2636 </pre>
|
|
2637 </blockquote>
|
|
2638
|
|
2639 <P>
|
|
2640 More information on these methods is as follows:
|
|
2641 </p>
|
|
2642
|
|
2643 <p>
|
|
2644 <ul>
|
|
2645 <li><tt>parser.errok()</tt>. This resets the parser state so it doesn't think it's in error-recovery
|
|
2646 mode. This will prevent an <tt>error</tt> token from being generated and will reset the internal
|
|
2647 error counters so that the next syntax error will call <tt>p_error()</tt> again.
|
|
2648
|
|
2649 <p>
|
|
2650 <li><tt>parser.token()</tt>. This returns the next token on the input stream.
|
|
2651
|
|
2652 <p>
|
|
2653 <li><tt>parser.restart()</tt>. This discards the entire parsing stack and resets the parser
|
|
2654 to its initial state.
|
|
2655 </ul>
|
|
2656
|
|
2657 <p>
|
|
2658 To supply the next lookahead token to the parser, <tt>p_error()</tt> can return a token. This might be
|
|
2659 useful if trying to synchronize on special characters. For example:
|
|
2660
|
|
2661 <blockquote>
|
|
2662 <pre>
|
|
2663 def p_error(p):
|
|
2664 # Read ahead looking for a terminating ";"
|
|
2665 while True:
|
|
2666 tok = parser.token() # Get the next token
|
|
2667 if not tok or tok.type == 'SEMI': break
|
|
2668 parser.errok()
|
|
2669
|
|
2670 # Return SEMI to the parser as the next lookahead token
|
|
2671 return tok
|
|
2672 </pre>
|
|
2673 </blockquote>
|
|
2674
|
|
2675 <p>
|
|
2676 Keep in mind in that the above error handling functions,
|
|
2677 <tt>parser</tt> is an instance of the parser created by
|
|
2678 <tt>yacc()</tt>. You'll need to save this instance someplace in your
|
|
2679 code so that you can refer to it during error handling.
|
|
2680 </p>
|
|
2681
|
|
2682 <H4><a name="ply_nn35"></a>6.8.3 Signalling an error from a production</H4>
|
|
2683
|
|
2684
|
|
2685 If necessary, a production rule can manually force the parser to enter error recovery. This
|
|
2686 is done by raising the <tt>SyntaxError</tt> exception like this:
|
|
2687
|
|
2688 <blockquote>
|
|
2689 <pre>
|
|
2690 def p_production(p):
|
|
2691 'production : some production ...'
|
|
2692 raise SyntaxError
|
|
2693 </pre>
|
|
2694 </blockquote>
|
|
2695
|
|
2696 The effect of raising <tt>SyntaxError</tt> is the same as if the last symbol shifted onto the
|
|
2697 parsing stack was actually a syntax error. Thus, when you do this, the last symbol shifted is popped off
|
|
2698 of the parsing stack and the current lookahead token is set to an <tt>error</tt> token. The parser
|
|
2699 then enters error-recovery mode where it tries to reduce rules that can accept <tt>error</tt> tokens.
|
|
2700 The steps that follow from this point are exactly the same as if a syntax error were detected and
|
|
2701 <tt>p_error()</tt> were called.
|
|
2702
|
|
2703 <P>
|
|
2704 One important aspect of manually setting an error is that the <tt>p_error()</tt> function will <b>NOT</b> be
|
|
2705 called in this case. If you need to issue an error message, make sure you do it in the production that
|
|
2706 raises <tt>SyntaxError</tt>.
|
|
2707
|
|
2708 <P>
|
|
2709 Note: This feature of PLY is meant to mimic the behavior of the YYERROR macro in yacc.
|
|
2710
|
|
2711 <H4><a name="ply_nn38"></a>6.8.4 When Do Syntax Errors Get Reported</H4>
|
|
2712
|
|
2713
|
|
2714 <p>
|
|
2715 In most cases, yacc will handle errors as soon as a bad input token is
|
|
2716 detected on the input. However, be aware that yacc may choose to
|
|
2717 delay error handling until after it has reduced one or more grammar
|
|
2718 rules first. This behavior might be unexpected, but it's related to
|
|
2719 special states in the underlying parsing table known as "defaulted
|
|
2720 states." A defaulted state is parsing condition where the same
|
|
2721 grammar rule will be reduced regardless of what <em>valid</em> token
|
|
2722 comes next on the input. For such states, yacc chooses to go ahead
|
|
2723 and reduce the grammar rule <em>without reading the next input
|
|
2724 token</em>. If the next token is bad, yacc will eventually get around to reading it and
|
|
2725 report a syntax error. It's just a little unusual in that you might
|
|
2726 see some of your grammar rules firing immediately prior to the syntax
|
|
2727 error.
|
|
2728 </p>
|
|
2729
|
|
2730 <p>
|
|
2731 Usually, the delayed error reporting with defaulted states is harmless
|
|
2732 (and there are other reasons for wanting PLY to behave in this way).
|
|
2733 However, if you need to turn this behavior off for some reason. You
|
|
2734 can clear the defaulted states table like this:
|
|
2735 </p>
|
|
2736
|
|
2737 <blockquote>
|
|
2738 <pre>
|
|
2739 parser = yacc.yacc()
|
|
2740 parser.defaulted_states = {}
|
|
2741 </pre>
|
|
2742 </blockquote>
|
|
2743
|
|
2744 <p>
|
|
2745 Disabling defaulted states is not recommended if your grammar makes use
|
|
2746 of embedded actions as described in Section 6.11.</p>
|
|
2747
|
|
2748 <H4><a name="ply_nn32"></a>6.8.5 General comments on error handling</H4>
|
|
2749
|
|
2750
|
|
2751 For normal types of languages, error recovery with error rules and resynchronization characters is probably the most reliable
|
|
2752 technique. This is because you can instrument the grammar to catch errors at selected places where it is relatively easy
|
|
2753 to recover and continue parsing. Panic mode recovery is really only useful in certain specialized applications where you might want
|
|
2754 to discard huge portions of the input text to find a valid restart point.
|
|
2755
|
|
2756 <H3><a name="ply_nn33"></a>6.9 Line Number and Position Tracking</H3>
|
|
2757
|
|
2758
|
|
2759 Position tracking is often a tricky problem when writing compilers.
|
|
2760 By default, PLY tracks the line number and position of all tokens.
|
|
2761 This information is available using the following functions:
|
|
2762
|
|
2763 <ul>
|
|
2764 <li><tt>p.lineno(num)</tt>. Return the line number for symbol <em>num</em>
|
|
2765 <li><tt>p.lexpos(num)</tt>. Return the lexing position for symbol <em>num</em>
|
|
2766 </ul>
|
|
2767
|
|
2768 For example:
|
|
2769
|
|
2770 <blockquote>
|
|
2771 <pre>
|
|
2772 def p_expression(p):
|
|
2773 'expression : expression PLUS expression'
|
|
2774 line = p.lineno(2) # line number of the PLUS token
|
|
2775 index = p.lexpos(2) # Position of the PLUS token
|
|
2776 </pre>
|
|
2777 </blockquote>
|
|
2778
|
|
2779 As an optional feature, <tt>yacc.py</tt> can automatically track line
|
|
2780 numbers and positions for all of the grammar symbols as well.
|
|
2781 However, this extra tracking requires extra processing and can
|
|
2782 significantly slow down parsing. Therefore, it must be enabled by
|
|
2783 passing the
|
|
2784 <tt>tracking=True</tt> option to <tt>yacc.parse()</tt>. For example:
|
|
2785
|
|
2786 <blockquote>
|
|
2787 <pre>
|
|
2788 yacc.parse(data,tracking=True)
|
|
2789 </pre>
|
|
2790 </blockquote>
|
|
2791
|
|
2792 Once enabled, the <tt>lineno()</tt> and <tt>lexpos()</tt> methods work
|
|
2793 for all grammar symbols. In addition, two additional methods can be
|
|
2794 used:
|
|
2795
|
|
2796 <ul>
|
|
2797 <li><tt>p.linespan(num)</tt>. Return a tuple (startline,endline) with the starting and ending line number for symbol <em>num</em>.
|
|
2798 <li><tt>p.lexspan(num)</tt>. Return a tuple (start,end) with the starting and ending positions for symbol <em>num</em>.
|
|
2799 </ul>
|
|
2800
|
|
2801 For example:
|
|
2802
|
|
2803 <blockquote>
|
|
2804 <pre>
|
|
2805 def p_expression(p):
|
|
2806 'expression : expression PLUS expression'
|
|
2807 p.lineno(1) # Line number of the left expression
|
|
2808 p.lineno(2) # line number of the PLUS operator
|
|
2809 p.lineno(3) # line number of the right expression
|
|
2810 ...
|
|
2811 start,end = p.linespan(3) # Start,end lines of the right expression
|
|
2812 starti,endi = p.lexspan(3) # Start,end positions of right expression
|
|
2813
|
|
2814 </pre>
|
|
2815 </blockquote>
|
|
2816
|
|
2817 Note: The <tt>lexspan()</tt> function only returns the range of values up to the start of the last grammar symbol.
|
|
2818
|
|
2819 <p>
|
|
2820 Although it may be convenient for PLY to track position information on
|
|
2821 all grammar symbols, this is often unnecessary. For example, if you
|
|
2822 are merely using line number information in an error message, you can
|
|
2823 often just key off of a specific token in the grammar rule. For
|
|
2824 example:
|
|
2825
|
|
2826 <blockquote>
|
|
2827 <pre>
|
|
2828 def p_bad_func(p):
|
|
2829 'funccall : fname LPAREN error RPAREN'
|
|
2830 # Line number reported from LPAREN token
|
|
2831 print("Bad function call at line", p.lineno(2))
|
|
2832 </pre>
|
|
2833 </blockquote>
|
|
2834
|
|
2835 <p>
|
|
2836 Similarly, you may get better parsing performance if you only
|
|
2837 selectively propagate line number information where it's needed using
|
|
2838 the <tt>p.set_lineno()</tt> method. For example:
|
|
2839
|
|
2840 <blockquote>
|
|
2841 <pre>
|
|
2842 def p_fname(p):
|
|
2843 'fname : ID'
|
|
2844 p[0] = p[1]
|
|
2845 p.set_lineno(0,p.lineno(1))
|
|
2846 </pre>
|
|
2847 </blockquote>
|
|
2848
|
|
2849 PLY doesn't retain line number information from rules that have already been
|
|
2850 parsed. If you are building an abstract syntax tree and need to have line numbers,
|
|
2851 you should make sure that the line numbers appear in the tree itself.
|
|
2852
|
|
2853 <H3><a name="ply_nn34"></a>6.10 AST Construction</H3>
|
|
2854
|
|
2855
|
|
2856 <tt>yacc.py</tt> provides no special functions for constructing an
|
|
2857 abstract syntax tree. However, such construction is easy enough to do
|
|
2858 on your own.
|
|
2859
|
|
2860 <p>A minimal way to construct a tree is to simply create and
|
|
2861 propagate a tuple or list in each grammar rule function. There
|
|
2862 are many possible ways to do this, but one example would be something
|
|
2863 like this:
|
|
2864
|
|
2865 <blockquote>
|
|
2866 <pre>
|
|
2867 def p_expression_binop(p):
|
|
2868 '''expression : expression PLUS expression
|
|
2869 | expression MINUS expression
|
|
2870 | expression TIMES expression
|
|
2871 | expression DIVIDE expression'''
|
|
2872
|
|
2873 p[0] = ('binary-expression',p[2],p[1],p[3])
|
|
2874
|
|
2875 def p_expression_group(p):
|
|
2876 'expression : LPAREN expression RPAREN'
|
|
2877 p[0] = ('group-expression',p[2])
|
|
2878
|
|
2879 def p_expression_number(p):
|
|
2880 'expression : NUMBER'
|
|
2881 p[0] = ('number-expression',p[1])
|
|
2882 </pre>
|
|
2883 </blockquote>
|
|
2884
|
|
2885 <p>
|
|
2886 Another approach is to create a set of data structure for different
|
|
2887 kinds of abstract syntax tree nodes and assign nodes to <tt>p[0]</tt>
|
|
2888 in each rule. For example:
|
|
2889
|
|
2890 <blockquote>
|
|
2891 <pre>
|
|
2892 class Expr: pass
|
|
2893
|
|
2894 class BinOp(Expr):
|
|
2895 def __init__(self,left,op,right):
|
|
2896 self.type = "binop"
|
|
2897 self.left = left
|
|
2898 self.right = right
|
|
2899 self.op = op
|
|
2900
|
|
2901 class Number(Expr):
|
|
2902 def __init__(self,value):
|
|
2903 self.type = "number"
|
|
2904 self.value = value
|
|
2905
|
|
2906 def p_expression_binop(p):
|
|
2907 '''expression : expression PLUS expression
|
|
2908 | expression MINUS expression
|
|
2909 | expression TIMES expression
|
|
2910 | expression DIVIDE expression'''
|
|
2911
|
|
2912 p[0] = BinOp(p[1],p[2],p[3])
|
|
2913
|
|
2914 def p_expression_group(p):
|
|
2915 'expression : LPAREN expression RPAREN'
|
|
2916 p[0] = p[2]
|
|
2917
|
|
2918 def p_expression_number(p):
|
|
2919 'expression : NUMBER'
|
|
2920 p[0] = Number(p[1])
|
|
2921 </pre>
|
|
2922 </blockquote>
|
|
2923
|
|
2924 The advantage to this approach is that it may make it easier to attach more complicated
|
|
2925 semantics, type checking, code generation, and other features to the node classes.
|
|
2926
|
|
2927 <p>
|
|
2928 To simplify tree traversal, it may make sense to pick a very generic
|
|
2929 tree structure for your parse tree nodes. For example:
|
|
2930
|
|
2931 <blockquote>
|
|
2932 <pre>
|
|
2933 class Node:
|
|
2934 def __init__(self,type,children=None,leaf=None):
|
|
2935 self.type = type
|
|
2936 if children:
|
|
2937 self.children = children
|
|
2938 else:
|
|
2939 self.children = [ ]
|
|
2940 self.leaf = leaf
|
|
2941
|
|
2942 def p_expression_binop(p):
|
|
2943 '''expression : expression PLUS expression
|
|
2944 | expression MINUS expression
|
|
2945 | expression TIMES expression
|
|
2946 | expression DIVIDE expression'''
|
|
2947
|
|
2948 p[0] = Node("binop", [p[1],p[3]], p[2])
|
|
2949 </pre>
|
|
2950 </blockquote>
|
|
2951
|
|
2952 <H3><a name="ply_nn35"></a>6.11 Embedded Actions</H3>
|
|
2953
|
|
2954
|
|
2955 The parsing technique used by yacc only allows actions to be executed at the end of a rule. For example,
|
|
2956 suppose you have a rule like this:
|
|
2957
|
|
2958 <blockquote>
|
|
2959 <pre>
|
|
2960 def p_foo(p):
|
|
2961 "foo : A B C D"
|
|
2962 print("Parsed a foo", p[1],p[2],p[3],p[4])
|
|
2963 </pre>
|
|
2964 </blockquote>
|
|
2965
|
|
2966 <p>
|
|
2967 In this case, the supplied action code only executes after all of the
|
|
2968 symbols <tt>A</tt>, <tt>B</tt>, <tt>C</tt>, and <tt>D</tt> have been
|
|
2969 parsed. Sometimes, however, it is useful to execute small code
|
|
2970 fragments during intermediate stages of parsing. For example, suppose
|
|
2971 you wanted to perform some action immediately after <tt>A</tt> has
|
|
2972 been parsed. To do this, write an empty rule like this:
|
|
2973
|
|
2974 <blockquote>
|
|
2975 <pre>
|
|
2976 def p_foo(p):
|
|
2977 "foo : A seen_A B C D"
|
|
2978 print("Parsed a foo", p[1],p[3],p[4],p[5])
|
|
2979 print("seen_A returned", p[2])
|
|
2980
|
|
2981 def p_seen_A(p):
|
|
2982 "seen_A :"
|
|
2983 print("Saw an A = ", p[-1]) # Access grammar symbol to left
|
|
2984 p[0] = some_value # Assign value to seen_A
|
|
2985
|
|
2986 </pre>
|
|
2987 </blockquote>
|
|
2988
|
|
2989 <p>
|
|
2990 In this example, the empty <tt>seen_A</tt> rule executes immediately
|
|
2991 after <tt>A</tt> is shifted onto the parsing stack. Within this
|
|
2992 rule, <tt>p[-1]</tt> refers to the symbol on the stack that appears
|
|
2993 immediately to the left of the <tt>seen_A</tt> symbol. In this case,
|
|
2994 it would be the value of <tt>A</tt> in the <tt>foo</tt> rule
|
|
2995 immediately above. Like other rules, a value can be returned from an
|
|
2996 embedded action by simply assigning it to <tt>p[0]</tt>
|
|
2997
|
|
2998 <p>
|
|
2999 The use of embedded actions can sometimes introduce extra shift/reduce conflicts. For example,
|
|
3000 this grammar has no conflicts:
|
|
3001
|
|
3002 <blockquote>
|
|
3003 <pre>
|
|
3004 def p_foo(p):
|
|
3005 """foo : abcd
|
|
3006 | abcx"""
|
|
3007
|
|
3008 def p_abcd(p):
|
|
3009 "abcd : A B C D"
|
|
3010
|
|
3011 def p_abcx(p):
|
|
3012 "abcx : A B C X"
|
|
3013 </pre>
|
|
3014 </blockquote>
|
|
3015
|
|
3016 However, if you insert an embedded action into one of the rules like this,
|
|
3017
|
|
3018 <blockquote>
|
|
3019 <pre>
|
|
3020 def p_foo(p):
|
|
3021 """foo : abcd
|
|
3022 | abcx"""
|
|
3023
|
|
3024 def p_abcd(p):
|
|
3025 "abcd : A B C D"
|
|
3026
|
|
3027 def p_abcx(p):
|
|
3028 "abcx : A B seen_AB C X"
|
|
3029
|
|
3030 def p_seen_AB(p):
|
|
3031 "seen_AB :"
|
|
3032 </pre>
|
|
3033 </blockquote>
|
|
3034
|
|
3035 an extra shift-reduce conflict will be introduced. This conflict is
|
|
3036 caused by the fact that the same symbol <tt>C</tt> appears next in
|
|
3037 both the <tt>abcd</tt> and <tt>abcx</tt> rules. The parser can either
|
|
3038 shift the symbol (<tt>abcd</tt> rule) or reduce the empty
|
|
3039 rule <tt>seen_AB</tt> (<tt>abcx</tt> rule).
|
|
3040
|
|
3041 <p>
|
|
3042 A common use of embedded rules is to control other aspects of parsing
|
|
3043 such as scoping of local variables. For example, if you were parsing C code, you might
|
|
3044 write code like this:
|
|
3045
|
|
3046 <blockquote>
|
|
3047 <pre>
|
|
3048 def p_statements_block(p):
|
|
3049 "statements: LBRACE new_scope statements RBRACE"""
|
|
3050 # Action code
|
|
3051 ...
|
|
3052 pop_scope() # Return to previous scope
|
|
3053
|
|
3054 def p_new_scope(p):
|
|
3055 "new_scope :"
|
|
3056 # Create a new scope for local variables
|
|
3057 s = new_scope()
|
|
3058 push_scope(s)
|
|
3059 ...
|
|
3060 </pre>
|
|
3061 </blockquote>
|
|
3062
|
|
3063 In this case, the embedded action <tt>new_scope</tt> executes
|
|
3064 immediately after a <tt>LBRACE</tt> (<tt>{</tt>) symbol is parsed.
|
|
3065 This might adjust internal symbol tables and other aspects of the
|
|
3066 parser. Upon completion of the rule <tt>statements_block</tt>, code
|
|
3067 might undo the operations performed in the embedded action
|
|
3068 (e.g., <tt>pop_scope()</tt>).
|
|
3069
|
|
3070 <H3><a name="ply_nn36"></a>6.12 Miscellaneous Yacc Notes</H3>
|
|
3071
|
|
3072
|
|
3073 <ul>
|
|
3074
|
|
3075 <li>By default, <tt>yacc.py</tt> relies on <tt>lex.py</tt> for tokenizing. However, an alternative tokenizer
|
|
3076 can be supplied as follows:
|
|
3077
|
|
3078 <blockquote>
|
|
3079 <pre>
|
|
3080 parser = yacc.parse(lexer=x)
|
|
3081 </pre>
|
|
3082 </blockquote>
|
|
3083 in this case, <tt>x</tt> must be a Lexer object that minimally has a <tt>x.token()</tt> method for retrieving the next
|
|
3084 token. If an input string is given to <tt>yacc.parse()</tt>, the lexer must also have an <tt>x.input()</tt> method.
|
|
3085
|
|
3086 <p>
|
|
3087 <li>By default, the yacc generates tables in debugging mode (which produces the parser.out file and other output).
|
|
3088 To disable this, use
|
|
3089
|
|
3090 <blockquote>
|
|
3091 <pre>
|
|
3092 parser = yacc.yacc(debug=False)
|
|
3093 </pre>
|
|
3094 </blockquote>
|
|
3095
|
|
3096 <p>
|
|
3097 <li>To change the name of the <tt>parsetab.py</tt> file, use:
|
|
3098
|
|
3099 <blockquote>
|
|
3100 <pre>
|
|
3101 parser = yacc.yacc(tabmodule="foo")
|
|
3102 </pre>
|
|
3103 </blockquote>
|
|
3104
|
|
3105 <P>
|
|
3106 Normally, the <tt>parsetab.py</tt> file is placed into the same directory as
|
|
3107 the module where the parser is defined. If you want it to go somewhere else, you can
|
|
3108 given an absolute package name for <tt>tabmodule</tt> instead. In that case, the
|
|
3109 tables will be written there.
|
|
3110 </p>
|
|
3111
|
|
3112 <p>
|
|
3113 <li>To change the directory in which the <tt>parsetab.py</tt> file (and other output files) are written, use:
|
|
3114 <blockquote>
|
|
3115 <pre>
|
|
3116 parser = yacc.yacc(tabmodule="foo",outputdir="somedirectory")
|
|
3117 </pre>
|
|
3118 </blockquote>
|
|
3119
|
|
3120 <p>
|
|
3121 Note: Be aware that unless the directory specified is also on Python's path (<tt>sys.path</tt>), subsequent
|
|
3122 imports of the table file will fail. As a general rule, it's better to specify a destination using the
|
|
3123 <tt>tabmodule</tt> argument instead of directly specifying a directory using the <tt>outputdir</tt> argument.
|
|
3124 </p>
|
|
3125
|
|
3126 <p>
|
|
3127 <li>To prevent yacc from generating any kind of parser table file, use:
|
|
3128 <blockquote>
|
|
3129 <pre>
|
|
3130 parser = yacc.yacc(write_tables=False)
|
|
3131 </pre>
|
|
3132 </blockquote>
|
|
3133
|
|
3134 Note: If you disable table generation, yacc() will regenerate the parsing tables
|
|
3135 each time it runs (which may take awhile depending on how large your grammar is).
|
|
3136
|
|
3137 <P>
|
|
3138 <li>To print copious amounts of debugging during parsing, use:
|
|
3139
|
|
3140 <blockquote>
|
|
3141 <pre>
|
|
3142 parser = yacc.parse(debug=True)
|
|
3143 </pre>
|
|
3144 </blockquote>
|
|
3145
|
|
3146 <p>
|
|
3147 <li>Since the generation of the LALR tables is relatively expensive, previously generated tables are
|
|
3148 cached and reused if possible. The decision to regenerate the tables is determined by taking an MD5
|
|
3149 checksum of all grammar rules and precedence rules. Only in the event of a mismatch are the tables regenerated.
|
|
3150
|
|
3151 <p>
|
|
3152 It should be noted that table generation is reasonably efficient, even for grammars that involve around a 100 rules
|
|
3153 and several hundred states. </li>
|
|
3154
|
|
3155
|
|
3156 <p>
|
|
3157 <li>Since LR parsing is driven by tables, the performance of the parser is largely independent of the
|
|
3158 size of the grammar. The biggest bottlenecks will be the lexer and the complexity of the code in your grammar rules.
|
|
3159 </li>
|
|
3160 </p>
|
|
3161
|
|
3162 <p>
|
|
3163 <li><tt>yacc()</tt> also allows parsers to be defined as classes and as closures (see the section on alternative specification of
|
|
3164 lexers). However, be aware that only one parser may be defined in a single module (source file). There are various
|
|
3165 error checks and validation steps that may issue confusing error messages if you try to define multiple parsers
|
|
3166 in the same source file.
|
|
3167 </li>
|
|
3168 </p>
|
|
3169
|
|
3170 </ul>
|
|
3171 </p>
|
|
3172
|
|
3173
|
|
3174 <H2><a name="ply_nn37"></a>7. Multiple Parsers and Lexers</H2>
|
|
3175
|
|
3176
|
|
3177 In advanced parsing applications, you may want to have multiple
|
|
3178 parsers and lexers.
|
|
3179
|
|
3180 <p>
|
|
3181 As a general rules this isn't a problem. However, to make it work,
|
|
3182 you need to carefully make sure everything gets hooked up correctly.
|
|
3183 First, make sure you save the objects returned by <tt>lex()</tt> and
|
|
3184 <tt>yacc()</tt>. For example:
|
|
3185
|
|
3186 <blockquote>
|
|
3187 <pre>
|
|
3188 lexer = lex.lex() # Return lexer object
|
|
3189 parser = yacc.yacc() # Return parser object
|
|
3190 </pre>
|
|
3191 </blockquote>
|
|
3192
|
|
3193 Next, when parsing, make sure you give the <tt>parse()</tt> function a reference to the lexer it
|
|
3194 should be using. For example:
|
|
3195
|
|
3196 <blockquote>
|
|
3197 <pre>
|
|
3198 parser.parse(text,lexer=lexer)
|
|
3199 </pre>
|
|
3200 </blockquote>
|
|
3201
|
|
3202 If you forget to do this, the parser will use the last lexer
|
|
3203 created--which is not always what you want.
|
|
3204
|
|
3205 <p>
|
|
3206 Within lexer and parser rule functions, these objects are also
|
|
3207 available. In the lexer, the "lexer" attribute of a token refers to
|
|
3208 the lexer object that triggered the rule. For example:
|
|
3209
|
|
3210 <blockquote>
|
|
3211 <pre>
|
|
3212 def t_NUMBER(t):
|
|
3213 r'\d+'
|
|
3214 ...
|
|
3215 print(t.lexer) # Show lexer object
|
|
3216 </pre>
|
|
3217 </blockquote>
|
|
3218
|
|
3219 In the parser, the "lexer" and "parser" attributes refer to the lexer
|
|
3220 and parser objects respectively.
|
|
3221
|
|
3222 <blockquote>
|
|
3223 <pre>
|
|
3224 def p_expr_plus(p):
|
|
3225 'expr : expr PLUS expr'
|
|
3226 ...
|
|
3227 print(p.parser) # Show parser object
|
|
3228 print(p.lexer) # Show lexer object
|
|
3229 </pre>
|
|
3230 </blockquote>
|
|
3231
|
|
3232 If necessary, arbitrary attributes can be attached to the lexer or parser object.
|
|
3233 For example, if you wanted to have different parsing modes, you could attach a mode
|
|
3234 attribute to the parser object and look at it later.
|
|
3235
|
|
3236 <H2><a name="ply_nn38"></a>8. Using Python's Optimized Mode</H2>
|
|
3237
|
|
3238
|
|
3239 Because PLY uses information from doc-strings, parsing and lexing
|
|
3240 information must be gathered while running the Python interpreter in
|
|
3241 normal mode (i.e., not with the -O or -OO options). However, if you
|
|
3242 specify optimized mode like this:
|
|
3243
|
|
3244 <blockquote>
|
|
3245 <pre>
|
|
3246 lex.lex(optimize=1)
|
|
3247 yacc.yacc(optimize=1)
|
|
3248 </pre>
|
|
3249 </blockquote>
|
|
3250
|
|
3251 then PLY can later be used when Python runs in optimized mode. To make this work,
|
|
3252 make sure you first run Python in normal mode. Once the lexing and parsing tables
|
|
3253 have been generated the first time, run Python in optimized mode. PLY will use
|
|
3254 the tables without the need for doc strings.
|
|
3255
|
|
3256 <p>
|
|
3257 Beware: running PLY in optimized mode disables a lot of error
|
|
3258 checking. You should only do this when your project has stabilized
|
|
3259 and you don't need to do any debugging. One of the purposes of
|
|
3260 optimized mode is to substantially decrease the startup time of
|
|
3261 your compiler (by assuming that everything is already properly
|
|
3262 specified and works).
|
|
3263
|
|
3264 <H2><a name="ply_nn44"></a>9. Advanced Debugging</H2>
|
|
3265
|
|
3266
|
|
3267 <p>
|
|
3268 Debugging a compiler is typically not an easy task. PLY provides some
|
|
3269 advanced diagostic capabilities through the use of Python's
|
|
3270 <tt>logging</tt> module. The next two sections describe this:
|
|
3271
|
|
3272 <H3><a name="ply_nn45"></a>9.1 Debugging the lex() and yacc() commands</H3>
|
|
3273
|
|
3274
|
|
3275 <p>
|
|
3276 Both the <tt>lex()</tt> and <tt>yacc()</tt> commands have a debugging
|
|
3277 mode that can be enabled using the <tt>debug</tt> flag. For example:
|
|
3278
|
|
3279 <blockquote>
|
|
3280 <pre>
|
|
3281 lex.lex(debug=True)
|
|
3282 yacc.yacc(debug=True)
|
|
3283 </pre>
|
|
3284 </blockquote>
|
|
3285
|
|
3286 Normally, the output produced by debugging is routed to either
|
|
3287 standard error or, in the case of <tt>yacc()</tt>, to a file
|
|
3288 <tt>parser.out</tt>. This output can be more carefully controlled
|
|
3289 by supplying a logging object. Here is an example that adds
|
|
3290 information about where different debugging messages are coming from:
|
|
3291
|
|
3292 <blockquote>
|
|
3293 <pre>
|
|
3294 # Set up a logging object
|
|
3295 import logging
|
|
3296 logging.basicConfig(
|
|
3297 level = logging.DEBUG,
|
|
3298 filename = "parselog.txt",
|
|
3299 filemode = "w",
|
|
3300 format = "%(filename)10s:%(lineno)4d:%(message)s"
|
|
3301 )
|
|
3302 log = logging.getLogger()
|
|
3303
|
|
3304 lex.lex(debug=True,debuglog=log)
|
|
3305 yacc.yacc(debug=True,debuglog=log)
|
|
3306 </pre>
|
|
3307 </blockquote>
|
|
3308
|
|
3309 If you supply a custom logger, the amount of debugging
|
|
3310 information produced can be controlled by setting the logging level.
|
|
3311 Typically, debugging messages are either issued at the <tt>DEBUG</tt>,
|
|
3312 <tt>INFO</tt>, or <tt>WARNING</tt> levels.
|
|
3313
|
|
3314 <p>
|
|
3315 PLY's error messages and warnings are also produced using the logging
|
|
3316 interface. This can be controlled by passing a logging object
|
|
3317 using the <tt>errorlog</tt> parameter.
|
|
3318
|
|
3319 <blockquote>
|
|
3320 <pre>
|
|
3321 lex.lex(errorlog=log)
|
|
3322 yacc.yacc(errorlog=log)
|
|
3323 </pre>
|
|
3324 </blockquote>
|
|
3325
|
|
3326 If you want to completely silence warnings, you can either pass in a
|
|
3327 logging object with an appropriate filter level or use the <tt>NullLogger</tt>
|
|
3328 object defined in either <tt>lex</tt> or <tt>yacc</tt>. For example:
|
|
3329
|
|
3330 <blockquote>
|
|
3331 <pre>
|
|
3332 yacc.yacc(errorlog=yacc.NullLogger())
|
|
3333 </pre>
|
|
3334 </blockquote>
|
|
3335
|
|
3336 <H3><a name="ply_nn46"></a>9.2 Run-time Debugging</H3>
|
|
3337
|
|
3338
|
|
3339 <p>
|
|
3340 To enable run-time debugging of a parser, use the <tt>debug</tt> option to parse. This
|
|
3341 option can either be an integer (which simply turns debugging on or off) or an instance
|
|
3342 of a logger object. For example:
|
|
3343
|
|
3344 <blockquote>
|
|
3345 <pre>
|
|
3346 log = logging.getLogger()
|
|
3347 parser.parse(input,debug=log)
|
|
3348 </pre>
|
|
3349 </blockquote>
|
|
3350
|
|
3351 If a logging object is passed, you can use its filtering level to control how much
|
|
3352 output gets generated. The <tt>INFO</tt> level is used to produce information
|
|
3353 about rule reductions. The <tt>DEBUG</tt> level will show information about the
|
|
3354 parsing stack, token shifts, and other details. The <tt>ERROR</tt> level shows information
|
|
3355 related to parsing errors.
|
|
3356
|
|
3357 <p>
|
|
3358 For very complicated problems, you should pass in a logging object that
|
|
3359 redirects to a file where you can more easily inspect the output after
|
|
3360 execution.
|
|
3361
|
|
3362 <H2><a name="ply_nn49"></a>10. Packaging Advice</H2>
|
|
3363
|
|
3364
|
|
3365 <p>
|
|
3366 If you are distributing a package that makes use of PLY, you should
|
|
3367 spend a few moments thinking about how you want to handle the files
|
|
3368 that are automatically generated. For example, the <tt>parsetab.py</tt>
|
|
3369 file generated by the <tt>yacc()</tt> function.</p>
|
|
3370
|
|
3371 <p>
|
|
3372 Starting in PLY-3.6, the table files are created in the same directory
|
|
3373 as the file where a parser is defined. This means that the
|
|
3374 <tt>parsetab.py</tt> file will live side-by-side with your parser
|
|
3375 specification. In terms of packaging, this is probably the easiest and
|
|
3376 most sane approach to manage. You don't need to give <tt>yacc()</tt>
|
|
3377 any extra arguments and it should just "work."</p>
|
|
3378
|
|
3379 <p>
|
|
3380 One concern is the management of the <tt>parsetab.py</tt> file itself.
|
|
3381 For example, should you have this file checked into version control (e.g., GitHub),
|
|
3382 should it be included in a package distribution as a normal file, or should you
|
|
3383 just let PLY generate it automatically for the user when they install your package?
|
|
3384 </p>
|
|
3385
|
|
3386 <p>
|
|
3387 As of PLY-3.6, the <tt>parsetab.py</tt> file should be compatible across all versions
|
|
3388 of Python including Python 2 and 3. Thus, a table file generated in Python 2 should
|
|
3389 work fine if it's used on Python 3. Because of this, it should be relatively harmless
|
|
3390 to distribute the <tt>parsetab.py</tt> file yourself if you need to. However, be aware
|
|
3391 that older/newer versions of PLY may try to regenerate the file if there are future
|
|
3392 enhancements or changes to its format.
|
|
3393 </p>
|
|
3394
|
|
3395 <p>
|
|
3396 To make the generation of table files easier for the purposes of installation, you might
|
|
3397 way to make your parser files executable using the <tt>-m</tt> option or similar. For
|
|
3398 example:
|
|
3399 </p>
|
|
3400
|
|
3401 <blockquote>
|
|
3402 <pre>
|
|
3403 # calc.py
|
|
3404 ...
|
|
3405 ...
|
|
3406 def make_parser():
|
|
3407 parser = yacc.yacc()
|
|
3408 return parser
|
|
3409
|
|
3410 if __name__ == '__main__':
|
|
3411 make_parser()
|
|
3412 </pre>
|
|
3413 </blockquote>
|
|
3414
|
|
3415 <p>
|
|
3416 You can then use a command such as <tt>python -m calc.py</tt> to generate the tables. Alternatively,
|
|
3417 a <tt>setup.py</tt> script, can import the module and use <tt>make_parser()</tt> to create the
|
|
3418 parsing tables.
|
|
3419 </p>
|
|
3420
|
|
3421 <p>
|
|
3422 If you're willing to sacrifice a little startup time, you can also instruct PLY to never write the
|
|
3423 tables using <tt>yacc.yacc(write_tables=False, debug=False)</tt>. In this mode, PLY will regenerate
|
|
3424 the parsing tables from scratch each time. For a small grammar, you probably won't notice. For a
|
|
3425 large grammar, you should probably reconsider--the parsing tables are meant to dramatically speed up this process.
|
|
3426 </p>
|
|
3427
|
|
3428 <p>
|
|
3429 During operation, is is normal for PLY to produce diagnostic error
|
|
3430 messages (usually printed to standard error). These are generated
|
|
3431 entirely using the <tt>logging</tt> module. If you want to redirect
|
|
3432 these messages or silence them, you can provide your own logging
|
|
3433 object to <tt>yacc()</tt>. For example:
|
|
3434 </p>
|
|
3435
|
|
3436 <blockquote>
|
|
3437 <pre>
|
|
3438 import logging
|
|
3439 log = logging.getLogger('ply')
|
|
3440 ...
|
|
3441 parser = yacc.yacc(errorlog=log)
|
|
3442 </pre>
|
|
3443 </blockquote>
|
|
3444
|
|
3445 <H2><a name="ply_nn39"></a>11. Where to go from here?</H2>
|
|
3446
|
|
3447
|
|
3448 The <tt>examples</tt> directory of the PLY distribution contains several simple examples. Please consult a
|
|
3449 compilers textbook for the theory and underlying implementation details or LR parsing.
|
|
3450
|
|
3451 </body>
|
|
3452 </html>
|
|
3453
|
|
3454
|
|
3455
|
|
3456
|
|
3457
|
|
3458
|
|
3459
|