|
|
<html><head><title>pcre2perform specification</title></head><body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB"><h1>pcre2perform man page</h1><p>Return to the <a href="index.html">PCRE2 index page</a>.</p><p>This page is part of the PCRE2 HTML documentation. It was generatedautomatically from the original man page. If there is any nonsense in it,please consult the man page, in case the conversion went wrong.<br><ul><li><a name="TOC1" href="#SEC1">PCRE2 PERFORMANCE</a><li><a name="TOC2" href="#SEC2">COMPILED PATTERN MEMORY USAGE</a><li><a name="TOC3" href="#SEC3">STACK AND HEAP USAGE AT RUN TIME</a><li><a name="TOC4" href="#SEC4">PROCESSING TIME</a><li><a name="TOC5" href="#SEC5">AUTHOR</a><li><a name="TOC6" href="#SEC6">REVISION</a></ul><br><a name="SEC1" href="#TOC1">PCRE2 PERFORMANCE</a><br><P>Two aspects of performance are discussed below: memory usage and processingtime. The way you express your pattern as a regular expression can affect bothof them.</P><br><a name="SEC2" href="#TOC1">COMPILED PATTERN MEMORY USAGE</a><br><P>Patterns are compiled by PCRE2 into a reasonably efficient interpretive code,so that most simple patterns do not use much memory for storing the compiledversion. However, there is one case where the memory usage of a compiledpattern can be unexpectedly large. If a parenthesized group has a quantifierwith a minimum greater than 1 and/or a limited maximum, the whole group isrepeated in the compiled code. For example, the pattern<pre> (abc|def){2,4}</pre>is compiled as if it were<pre> (abc|def)(abc|def)((abc|def)(abc|def)?)?</pre>(Technical aside: It is done this way so that backtrack points within each ofthe repetitions can be independently maintained.)</P><P>For regular expressions whose quantifiers use only small numbers, this is notusually a problem. However, if the numbers are large, and particularly if suchrepetitions are nested, the memory usage can become an embarrassment. Forexample, the very simple pattern<pre> ((ab){1,1000}c){1,3}</pre>uses over 50KiB when compiled using the 8-bit library. When PCRE2 iscompiled with its default internal pointer size of two bytes, the size limit ona compiled pattern is 65535 code units in the 8-bit and 16-bit libraries, andthis is reached with the above pattern if the outer repetition is increasedfrom 3 to 4. PCRE2 can be compiled to use larger internal pointers and thushandle larger compiled patterns, but it is better to try to rewrite yourpattern to use less memory if you can.</P><P>One way of reducing the memory usage for such patterns is to make use ofPCRE2's<a href="pcre2pattern.html#subpatternsassubroutines">"subroutine"</a>facility. Re-writing the above pattern as<pre> ((ab)(?2){0,999}c)(?1){0,2}</pre>reduces the memory requirements to around 16KiB, and indeed it remains under20KiB even with the outer repetition increased to 100. However, this kind ofpattern is not always exactly equivalent, because any captures withinsubroutine calls are lost when the subroutine completes. If this is not aproblem, this kind of rewriting will allow you to process patterns that PCRE2cannot otherwise handle. The matching performance of the two different versionsof the pattern are roughly the same. (This applies from release 10.30 - thingswere different in earlier releases.)</P><br><a name="SEC3" href="#TOC1">STACK AND HEAP USAGE AT RUN TIME</a><br><P>From release 10.30, the interpretive (non-JIT) version of <b>pcre2_match()</b>uses very little system stack at run time. In earlier releases recursivefunction calls could use a great deal of stack, and this could cause problems,but this usage has been eliminated. Backtracking positions are now explicitlyremembered in memory frames controlled by the code.</P><P>The size of each frame depends on the size of pointer variables and the numberof capturing parenthesized groups in the pattern being matched. On a 64-bitsystem the frame size for a pattern with no captures is 128 bytes. For eachcapturing group the size increases by 16 bytes.</P><P>Until release 10.41, an initial 20KiB frames vector was allocated on the systemstack, but this still caused some issues for multi-thread applications whereeach thread has a very small stack. From release 10.41 backtracking memoryframes are always held in heap memory. An initial heap allocation is obtainedthe first time any match data block is passed to <b>pcre2_match()</b>. This isremembered with the match data block and re-used if that block is used foranother match. It is freed when the match data block itself is freed.</P><P>The size of the initial block is the larger of 20KiB or ten times the pattern'sframe size, unless the heap limit is less than this, in which case the heaplimit is used. If the initial block proves to be too small during matching, itis replaced by a larger block, subject to the heap limit. The heap limit ischecked only when a new block is to be allocated. Reducing the heap limitbetween calls to <b>pcre2_match()</b> with the same match data block does notaffect the saved block.</P><P>In contrast to <b>pcre2_match()</b>, <b>pcre2_dfa_match()</b> does use recursivefunction calls, but only for processing atomic groups, lookaround assertions,and recursion within the pattern. The original version of the code used toallocate quite large internal workspace vectors on the stack, which caused someproblems for some patterns in environments with small stacks. From release10.32 the code for <b>pcre2_dfa_match()</b> has been re-factored to use heapmemory when necessary for internal workspace when recursing, though recursivefunction calls are still used.</P><P>The "match depth" parameter can be used to limit the depth of functionrecursion, and the "match heap" parameter to limit heap memory in<b>pcre2_dfa_match()</b>.</P><br><a name="SEC4" href="#TOC1">PROCESSING TIME</a><br><P>Certain items in regular expression patterns are processed more efficientlythan others. It is more efficient to use a character class like [aeiou] than aset of single-character alternatives such as (a|e|i|o|u). In general, thesimplest construction that provides the required behaviour is usually the mostefficient. Jeffrey Friedl's book contains a lot of useful general discussionabout optimizing regular expressions for efficient performance. This documentcontains a few observations about PCRE2.</P><P>Using Unicode character properties (the \p, \P, and \X escapes) is slow,because PCRE2 has to use a multi-stage table lookup whenever it needs acharacter's property. If you can find an alternative pattern that does not usecharacter properties, it will probably be faster.</P><P>By default, the escape sequences \b, \d, \s, and \w, and the POSIXcharacter classes such as [:alpha:] do not use Unicode properties, partly forbackwards compatibility, and partly for performance reasons. However, you canset the PCRE2_UCP option or start the pattern with (*UCP) if you want Unicodecharacter properties to be used. This can double the matching time for itemssuch as \d, when matched with <b>pcre2_match()</b>; the performance loss isless with a DFA matching function, and in both cases there is not muchdifference for \b.</P><P>When a pattern begins with .* not in atomic parentheses, nor in parenthesesthat are the subject of a backreference, and the PCRE2_DOTALL option is set,the pattern is implicitly anchored by PCRE2, since it can match only at thestart of a subject string. If the pattern has multiple top-level branches, theymust all be anchorable. The optimization can be disabled by thePCRE2_NO_DOTSTAR_ANCHOR option, and is automatically disabled if the patterncontains (*PRUNE) or (*SKIP).</P><P>If PCRE2_DOTALL is not set, PCRE2 cannot make this optimization, because thedot metacharacter does not then match a newline, and if the subject stringcontains newlines, the pattern may match from the character immediatelyfollowing one of them instead of from the very start. For example, the pattern<pre> .*second</pre>matches the subject "first\nand second" (where \n stands for a newlinecharacter), with the match starting at the seventh character. In order to dothis, PCRE2 has to retry the match starting after every newline in the subject.</P><P>If you are using such a pattern with subject strings that do not containnewlines, the best performance is obtained by setting PCRE2_DOTALL, or startingthe pattern with ^.* or ^.*? to indicate explicit anchoring. That saves PCRE2from having to scan along the subject looking for a newline to restart at.</P><P>Beware of patterns that contain nested indefinite repeats. These can take along time to run when applied to a string that does not match. Consider thepattern fragment<pre> ^(a+)*</pre>This can match "aaaa" in 16 different ways, and this number increases veryrapidly as the string gets longer. (The * repeat can match 0, 1, 2, 3, or 4times, and for each of those cases other than 0 or 4, the + repeats can matchdifferent numbers of times.) When the remainder of the pattern is such that theentire match is going to fail, PCRE2 has in principle to try every possiblevariation, and this can take an extremely long time, even for relatively shortstrings.</P><P>An optimization catches some of the more simple cases such as<pre> (a+)*b</pre>where a literal character follows. Before embarking on the standard matchingprocedure, PCRE2 checks that there is a "b" later in the subject string, and ifthere is not, it fails the match immediately. However, when there is nofollowing literal this optimization cannot be used. You can see the differenceby comparing the behaviour of<pre> (a+)*\d</pre>with the pattern above. The former gives a failure almost instantly whenapplied to a whole line of "a" characters, whereas the latter takes anappreciable time with strings longer than about 20 characters.</P><P>In many cases, the solution to this kind of performance issue is to use anatomic group or a possessive quantifier. This can often reduce memoryrequirements as well. As another example, consider this pattern:<pre> ([^<]|<(?!inet))+</pre>It matches from wherever it starts until it encounters "<inet" or the end ofthe data, and is the kind of pattern that might be used when processing an XMLfile. Each iteration of the outer parentheses matches either one character thatis not "<" or a "<" that is not followed by "inet". However, each time aparenthesis is processed, a backtracking position is passed, so thisformulation uses a memory frame for each matched character. For a long string,a lot of memory is required. Consider now this rewritten pattern, which matchesexactly the same strings:<pre> ([^<]++|<(?!inet))+</pre>This runs much faster, because sequences of characters that do not contain "<"are "swallowed" in one item inside the parentheses, and a possessive quantifieris used to stop any backtracking into the runs of non-"<" characters. Thisversion also uses a lot less memory because entry to a new set of parentheseshappens only when a "<" character that is not followed by "inet" is encountered(and we assume this is relatively rare).</P><P>This example shows that one way of optimizing performance when matching longsubject strings is to write repeated parenthesized subpatterns to match morethan one character whenever possible.</P><br><b>SETTING RESOURCE LIMITS</b><br><P>You can set limits on the amount of processing that takes place when matching,and on the amount of heap memory that is used. The default values of the limitsare very large, and unlikely ever to operate. They can be changed when PCRE2 isbuilt, and they can also be set when <b>pcre2_match()</b> or<b>pcre2_dfa_match()</b> is called. For details of these interfaces, see the<a href="pcre2build.html"><b>pcre2build</b></a>documentation and the section entitled<a href="pcre2api.html#matchcontext">"The match context"</a>in the<a href="pcre2api.html"><b>pcre2api</b></a>documentation.</P><P>The <b>pcre2test</b> test program has a modifier called "find_limits" which, ifapplied to a subject line, causes it to find the smallest limits that allow apattern to match. This is done by repeatedly matching with different limits.</P><br><a name="SEC5" href="#TOC1">AUTHOR</a><br><P>Philip Hazel<br>Retired from University Computing Service<br>Cambridge, England.<br></P><br><a name="SEC6" href="#TOC1">REVISION</a><br><P>Last updated: 27 July 2022<br>Copyright © 1997-2022 University of Cambridge.<br><p>Return to the <a href="index.html">PCRE2 index page</a>.</p>
|
