1PPI::Tokenizer(3) User Contributed Perl Documentation PPI::Tokenizer(3)
2
6 PPI::Tokenizer - The Perl Document Tokenizer
7
9 # Create a tokenizer for a file, array or string
10 $Tokenizer = PPI::Tokenizer->new( 'filename.pl' );
11 $Tokenizer = PPI::Tokenizer->new( \@lines );
12 $Tokenizer = PPI::Tokenizer->new( \$source );
13 # Return all the tokens for the document
15 my $tokens = $Tokenizer->all_tokens;
16 # Or we can use it as an iterator
18 while ( my $Token = $Tokenizer->get_token ) {
19 print "Found token '$Token'\n";
20 }
21 # If we REALLY need to manually nudge the cursor, you
23 # can do that to (The lexer needs this ability to do rollbacks)
24 $is_incremented = $Tokenizer->increment_cursor;
25 $is_decremented = $Tokenizer->decrement_cursor;
26
28 PPI::Tokenizer is the class that provides Tokenizer objects for use in
29 breaking strings of Perl source code into Tokens.
30 By the time you are reading this, you probably need to know a little
32 about the difference between how perl parses Perl "code" and how PPI
33 parsers Perl "documents".
34 "perl" itself (the interpreter) uses a heavily modified lex
36 specification to specify its parsing logic, maintains several types of
37 state as it goes, and incrementally tokenizes, lexes AND EXECUTES at
38 the same time.
39 In fact, it is provably impossible to use perl's parsing method without
41 simultaneously executing code. A formal mathematical proof has been
42 published demonstrating the method.
43 This is where the truism "Only perl can parse Perl" comes from.
45 PPI uses a completely different approach by abandoning the (impossible)
47 ability to parse Perl the same way that the interpreter does, and
48 instead parsing the source as a document, using a document structure
49 independantly derived from the Perl documentation and approximating the
50 perl interpreter interpretation as closely as possible.
51 It was touch and go for a long time whether we could get it close
53 enough, but in the end it turned out that it could be done.
54 In this approach, the tokenizer "PPI::Tokenizer" is implemented
56 separately from the lexer PPI::Lexer.
57 The job of "PPI::Tokenizer" is to take pure source as a string and
59 break it up into a stream/set of tokens, and contains most of the
60 "black magic" used in PPI. By comparison, the lexer implements a
61 relatively straight forward tree structure, and has an implementation
62 that is uncomplicated (compared to the insanity in the tokenizer at
63 least).
64 The Tokenizer uses an immense amount of heuristics, guessing and cruft,
66 supported by a very VERY flexible internal API, but fortunately it was
67 possible to largely encapsulate the black magic, so there is not a lot
68 that gets exposed to people using the "PPI::Tokenizer" itself.
69
71 Despite the incredible complexity, the Tokenizer itself only exposes a
72 relatively small number of methods, with most of the complexity
73 implemented in private methods.
74 new $file | \@lines | \$source
76 The main "new" constructor creates a new Tokenizer object. These
77 objects have no configuration parameters, and can only be used once, to
78 tokenize a single perl source file.
79 It takes as argument either a normal scalar containing source code, a
81 reference to a scalar containing source code, or a reference to an
82 ARRAY containing newline-terminated lines of source code.
83 Returns a new "PPI::Tokenizer" object on success, or throws a
85 PPI::Exception exception on error.
86 get_token
88 When using the PPI::Tokenizer object as an iterator, the "get_token"
89 method is the primary method that is used. It increments the cursor and
90 returns the next Token in the output array.
91 The actual parsing of the file is done only as-needed, and a line at a
93 time. When "get_token" hits the end of the token array, it will cause
94 the parser to pull in the next line and parse it, continuing as needed
95 until there are more tokens on the output array that get_token can then
96 return.
97 This means that a number of Tokenizer objects can be created, and won't
99 consume significant CPU until you actually begin to pull tokens from
100 it.
101 Return a PPI::Token object on success, 0 if the Tokenizer had reached
103 the end of the file, or "undef" on error.
104 all_tokens
106 When not being used as an iterator, the "all_tokens" method tells the
107 Tokenizer to parse the entire file and return all of the tokens in a
108 single ARRAY reference.
109 It should be noted that "all_tokens" does NOT interfere with the use of
111 the Tokenizer object as an iterator (does not modify the token cursor)
112 and use of the two different mechanisms can be mixed safely.
113 Returns a reference to an ARRAY of PPI::Token objects on success or
115 throws an exception on error.
116 increment_cursor
118 Although exposed as a public method, "increment_method" is implemented
119 for expert use only, when writing lexers or other components that work
120 directly on token streams.
121 It manually increments the token cursor forward through the file, in
123 effect "skipping" the next token.
124 Return true if the cursor is incremented, 0 if already at the end of
126 the file, or "undef" on error.
127 decrement_cursor
129 Although exposed as a public method, "decrement_method" is implemented
130 for expert use only, when writing lexers or other components that work
131 directly on token streams.
132 It manually decrements the token cursor backwards through the file, in
134 effect "rolling back" the token stream. And indeed that is what it is
135 primarily intended for, when the component that is consuming the token
136 stream needs to implement some sort of "roll back" feature in its use
137 of the token stream.
138 Return true if the cursor is decremented, 0 if already at the beginning
140 of the file, or "undef" on error.
141
143 How the Tokenizer Works
144 Understanding the Tokenizer is not for the feint-hearted. It is by far
145 the most complex and twisty piece of perl I've ever written that is
146 actually still built properly and isn't a terrible spaghetti-like mess.
147 In fact, you probably want to skip this section.
148 But if you really want to understand, well then here goes.
150 Source Input and Clean Up
152 The Tokenizer starts by taking source in a variety of forms, sucking it
153 all in and merging into one big string, and doing our own internal line
154 split, using a "universal line separator" which allows the Tokenizer to
155 take source for any platform (and even supports a few known types of
156 broken newlines caused by mixed mac/pc/*nix editor screw ups).
157 The resulting array of lines is used to feed the tokenizer, and is also
159 accessed directly by the heredoc-logic to do the line-oriented part of
160 here-doc support.
161 Doing Things the Old Fashioned Way
163 Due to the complexity of perl, and after 2 previously aborted parser
164 attempts, in the end the tokenizer was fashioned around a line-buffered
165 character-by-character method.
166 That is, the Tokenizer pulls and holds a line at a time into a line
168 buffer, and then iterates a cursor along it. At each cursor position, a
169 method is called in whatever token class we are currently in, which
170 will examine the character at the current position, and handle it.
171 As the handler methods in the various token classes are called, they
173 build up a output token array for the source code.
174 Various parts of the Tokenizer use look-ahead, arbitrary-distance look-
176 behind (although currently the maximum is three significant tokens), or
177 both, and various other heuristic guesses.
178 I've been told it is officially termed a "backtracking parser with
180 infinite lookaheads".
181 State Variables
183 Aside from the current line and the character cursor, the Tokenizer
184 maintains a number of different state variables.
185 Current Class
187 The Tokenizer maintains the current token class at all times. Much
188 of the time is just going to be the "Whitespace" class, which is
189 what the base of a document is. As the tokenizer executes the
190 various character handlers, the class changes a lot as it moves a
191 long. In fact, in some instances, the character handler may not
192 handle the character directly itself, but rather change the
193 "current class" and then hand off to the character handler for the
194 new class.
195 Because of this, and some other things I'll deal with later, the
197 number of times the character handlers are called does not in fact
198 have a direct relationship to the number of actual characters in
199 the document.
200 Current Zone
202 Rather than create a class stack to allow for infinitely nested
203 layers of classes, the Tokenizer recognises just a single layer.
204 To put it a different way, in various parts of the file, the
206 Tokenizer will recognise different "base" or "substrate" classes.
207 When a Token such as a comment or a number is finalised by the
208 tokenizer, it "falls back" to the base state.
209 This allows proper tokenization of special areas such as __DATA__
211 and __END__ blocks, which also contain things like comments and
212 POD, without allowing the creation of any significant Tokens inside
213 these areas.
214 For the main part of a document we use PPI::Token::Whitespace for
216 this, with the idea being that code is "floating in a sea of
217 whitespace".
218 Current Token
220 The final main state variable is the "current token". This is the
221 Token that is currently being built by the Tokenizer. For certain
222 types, it can be manipulated and morphed and change class quite a
223 bit while being assembled, as the Tokenizer's understanding of the
224 token content changes.
225 When the Tokenizer is confident that it has seen the end of the
227 Token, it will be "finalized", which adds it to the output token
228 array and resets the current class to that of the zone that we are
229 currently in.
230 I should also note at this point that the "current token" variable
232 is optional. The Tokenizer is capable of knowing what class it is
233 currently set to, without actually having accumulated any
234 characters in the Token.
235 Making It Faster
237 As I'm sure you can imagine, calling several different methods for each
238 character and running regexes and other complex heuristics made the
239 first fully working version of the tokenizer extremely slow.
240 During testing, I created a metric to measure parsing speed called
242 LPGC, or "lines per gigacycle" . A gigacycle is simple a billion CPU
243 cycles on a typical single-core CPU, and so a Tokenizer running at
244 "1000 lines per gigacycle" should generate around 1200 lines of
245 tokenized code when running on a 1200 MHz processor.
246 The first working version of the tokenizer ran at only 350 LPGC, so to
248 tokenize a typical large module such as ExtUtils::MakeMaker took 10-15
249 seconds. This sluggishness made it unpractical for many uses.
250 So in the current parser, there are multiple layers of optimisation
252 very carefully built in to the basic. This has brought the tokenizer up
253 to a more reasonable 1000 LPGC, at the expense of making the code quite
254 a bit twistier.
255 Making It Faster - Whole Line Classification
257 The first step in the optimisation process was to add a hew handler to
258 enable several of the more basic classes (whitespace, comments) to be
259 able to be parsed a line at a time. At the start of each line, a
260 special optional handler (only supported by a few classes) is called to
261 check and see if the entire line can be parsed in one go.
262 This is used mainly to handle things like POD, comments, empty lines,
264 and a few other minor special cases.
265 Making It Faster - Inlining
267 The second stage of the optimisation involved inlining a small number
268 of critical methods that were repeated an extremely high number of
269 times. Profiling suggested that there were about 1,000,000 individual
270 method calls per gigacycle, and by cutting these by two thirds a
271 significant speed improvement was gained, in the order of about 50%.
272 You may notice that many methods in the "PPI::Tokenizer" code look very
274 nested and long hand. This is primarily due to this inlining.
275 At around this time, some statistics code that existed in the early
277 versions of the parser was also removed, as it was determined that it
278 was consuming around 15% of the CPU for the entire parser, while making
279 the core more complicated.
280 A judgment call was made that with the difficulties likely to be
282 encountered with future planned enhancements, and given the relatively
283 high cost involved, the statistics features would be removed from the
284 Tokenizer.
285 Making It Faster - Quote Engine
287 Once inlining had reached diminishing returns, it became obvious from
288 the profiling results that a huge amount of time was being spent
289 stepping a char at a time though long, simple and "syntactically
290 boring" code such as comments and strings.
291 The existing regex engine was expanded to also encompass quotes and
293 other quote-like things, and a special abstract base class was added
294 that provided a number of specialised parsing methods that would "scan
295 ahead", looking out ahead to find the end of a string, and updating the
296 cursor to leave it in a valid position for the next call.
297 This is also the point at which the number of character handler calls
299 began to greatly differ from the number of characters. But it has been
300 done in a way that allows the parser to retain the power of the
301 original version at the critical points, while skipping through the
302 "boring bits" as needed for additional speed.
303 The addition of this feature allowed the tokenizer to exceed 1000 LPGC
305 for the first time.
306 Making It Faster - The "Complete" Mechanism
308 As it became evident that great speed increases were available by using
309 this "skipping ahead" mechanism, a new handler method was added that
310 explicitly handles the parsing of an entire token, where the structure
311 of the token is relatively simple. Tokens such as symbols fit this
312 case, as once we are passed the initial sigil and word char, we know
313 that we can skip ahead and "complete" the rest of the token much more
314 easily.
315 A number of these have been added for most or possibly all of the
317 common cases, with most of these "complete" handlers implemented using
318 regular expressions.
319 In fact, so many have been added that at this point, you could arguably
321 reclassify the tokenizer as a "hybrid regex, char-by=char heuristic
322 tokenizer". More tokens are now consumed in "complete" methods in a
323 typical program than are handled by the normal char-by-char methods.
324 Many of the these complete-handlers were implemented during the writing
326 of the Lexer, and this has allowed the full parser to maintain around
327 1000 LPGC despite the increasing weight of the Lexer.
328 Making It Faster - Porting To C (In Progress)
330 While it would be extraordinarily difficult to port all of the
331 Tokenizer to C, work has started on a PPI::XS "accelerator" package
332 which acts as a separate and automatically-detected add-on to the main
333 PPI package.
334 PPI::XS implements faster versions of a variety of functions scattered
336 over the entire PPI codebase, from the Tokenizer Core, Quote Engine,
337 and various other places, and implements them identically in XS/C.
338 In particular, the skip-ahead methods from the Quote Engine would
340 appear to be extremely amenable to being done in C, and a number of
341 other functions could be cherry-picked one at a time and implemented in
342 C.
343 Each method is heavily tested to ensure that the functionality is
345 identical, and a versioning mechanism is included to ensure that if a
346 function gets out of sync, PPI::XS will degrade gracefully and just not
347 replace that single method.
348
350 - Add an option to reset or seek the token stream...
351 - Implement more Tokenizer functions in PPI::XS
353
355 See the support section in the main module.
356
358 Adam Kennedy <adamk@cpan.org>
359
361 Copyright 2001 - 2009 Adam Kennedy.
362 This program is free software; you can redistribute it and/or modify it
364 under the same terms as Perl itself.
365 The full text of the license can be found in the LICENSE file included
367 with this module.
368perl v5.10.1 2009-08-08 PPI::Tokenizer(3)