PHP Regex match between two patterns

Question

I'm attempting to parse through a log file containing numerous traces, some of which have multiple lines to them.

Example:

[trace-123] <request>This is a log line</request>
[trace-124] <reply>This is another log line

this is part of "[trace-124]" still.</reply>
[trace-125] <request>final log line.</request>

I'm attempting to use preg_match_all to get an array of all the traces.

$file = file_get_contents("traces.txt");
$tracePattern = "/(\[trace-[0-9]*+\]+[\s\S]*)(?<=\<\/reply>|\<\/request>)/";

preg_match_all($tracePattern,$file,$lines);

echo "<pre>";print_r($lines);echo "</pre>";

Ideally, I'd like my results to look like this:

Array
(
    [0] => [trace-123] <request>This is a log line</request>
    [1] => [trace-124] <reply>This is another log line

this is part of "[trace-124]" still.</reply>
    [2] => [trace-125] <request>final log line.</request>
)

but when I run it, I get an array with everything in 1 element of the array. When I wrote the expression, my goal was to basically look for:

[trace-\[0-9]*\]

and find everything from that match to the next match of it.

I found that

\[trace-[0-9]*+\].*

works pretty well, but breaks down when there are line breaks.

I fail to see why what you are doing would require regex at all. You can simply loop through each line of the file looking for [trace- at the beginning of the string. Each time you encounter this value, start adding the lines to a string at the next array position in teh array you are building. You stop adding to this string the next time you encounter either another line beginning with [trace- or the next time you encounter a line beginning with some other non-trace signature (like for instance if you have lines like [error- or whatever). — Mike Brant
– Mike Brant, Commented Nov 14, 2013 at 20:35
To contiune my comment... requiring multi-line regex is likely going to require you put put your ENTIRE log file into memory (either that or at least store portions of it into memory until next [trace- is seen, which basically requires the implementation I suggested above to be to execute). This may not be feasible for larger log files. You likely should focus on a solution that allows you to parse and work with a single line at a time. — Mike Brant
– Mike Brant, Commented Nov 14, 2013 at 20:37
@Mike - With multi-gig ram capacity, if a file is too big. reading line by line might take a protracted period of time and building an array would have to be offloaded as well. Another approach would be read in 10,000 lines at a time, process records with a multi-line regex, capture the last record start, put at front of buffer, read another 10,000 lines (or, like 10 megs), repeat. — user557597
– user557597, Commented Nov 14, 2013 at 22:12
@sin True, but oftentimes you want a very process that is very light on RAM, as the servers may be taking production load. In these cases, you cannot afford to dedicate large chunks of RAM to a single process. You are right that building the array might be too expensive on RAM as well depending on number of overall lines where this trace data is present. — Mike Brant
– Mike Brant, Commented Nov 14, 2013 at 23:43
@Mike - Yeah thats true, but in a server environment each virtual machine might be given a gig or two virtual ram. But thats a lot of disk seeks doing it line by line on partitioned drives. Reading in 10 meg chunks and using a multi-line regex is a good balance when it comes to resources. Flat out, its at least 10 or more times faster. — user557597
– user557597, Commented Nov 15, 2013 at 0:43

hwnd · Accepted Answer · 2013-11-14 20:38:19Z

3

The following would probably be a better approach here.

$results = preg_split('/\R(?=\[trace[^\]]*\])/', $text);
print_r($results);

See working demo

Output

Array
(
    [0] => [trace-123] <request>This is a log line</request>
    [1] => [trace-124] <reply>This is another log line

this is part of "[trace-124]" still.</reply>
    [2] => [trace-125] <request>final log line.</request>
)

answered Nov 14, 2013 at 20:38

hwnd

70.9k4 gold badges100 silver badges135 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

elixenide · Accepted Answer · 2013-11-14 20:30:05Z

2

Use this:

$file = '[trace-123] <request>This is a log line</request>
[trace-124] <reply>This is another log line

this is part of "[trace-124]" still.</reply>
[trace-125] <request>final log line.</request>';

$tracePattern = "/\[trace-[0-9]*+\]+\s*<(?:reply|request)>.*?<\/(?:reply|request)>/s";

preg_match_all($tracePattern,$file,$lines);

$lines = $lines[0]; // by defaults, $lines[0] will be an array of the matches, so get that

echo "<pre>";print_r($lines);echo "</pre>";

Working demo: http://ideone.com/n8n5r3

answered Nov 14, 2013 at 20:30

elixenide

44.9k16 gold badges79 silver badges103 bronze badges

Comments

score 2 · Accepted Answer · 2013-11-14 21:09:10Z

2

This works in MULTI_LINE mode. Trims leading spaces and trailing newline's.

Edit: This assumes an anchor that is [trace- ] and is either at the beginning of
the line or beginning plus non-newline whitespace until 'trace'. This is the
only discernable record separator.

 #  ^[^\S\n]*(\[trace-[^]]*\][^\n]*(?:(?!\s+\[trace-[^]]*\])\n[^\n]*)*)

 ^ [^\S\n]* 
 (
      \[trace- [^]]* \] [^\n]* 

      (?:
           (?! \s+ \[trace- [^]]* \] )
           \n [^\n]* 
      )*
 )

Output (in single quotes)

 '[trace-123] <request>This is a log line</request>'
 '[trace-124] <reply>This is another log line

 this is part of "[trace-124]" still.</reply>'
 '[trace-125] <request>final log line.</request>'

edited Nov 14, 2013 at 21:09

answered Nov 14, 2013 at 20:50

user557597

Comments

bukart · Accepted Answer · 2013-11-15 07:54:18Z

2

I'd recommend a solution via preg_split

preg_split('/\R+(?=\[trace-\d+])/', $str)

this results in the following

Array
(
    [0] => [trace-123] <request>This is a log line</request>
    [1] => [trace-124] <reply>This is another log line

this is part of "[trace-124]" still.</reply>
    [2] => [trace-125] <request>final log line.</request>
)

edited Nov 15, 2013 at 7:54

answered Nov 14, 2013 at 20:32

bukart

4,9162 gold badges23 silver badges40 bronze badges

Comments

56ka · Accepted Answer · 2013-11-14 20:25:46Z

0

The symbol . means every char except line breaks \n, you can try to change it with (.|\s) this way :

#\[trace-[0-9]*+\](.|\s)*#

Note : you can use non-capturant parenthesis (?: )

Easyer, add the flag "s"

#\[trace-[0-9]*+\].*#s

answered Nov 14, 2013 at 20:25

56ka

1,5851 gold badge22 silver badges37 bronze badges

1 Comment

Glavić Over a year ago

No need, he can add pattern modifier s (PCRE_DOTALL).

paulotorrens · Accepted Answer · 2013-11-14 20:28:48Z

0

You should use a reluctant quantifier (??, +? or *?).

I believe this regex /(\[trace-[0-9]*\]\s*(?m:.*?)<\/(?:reply|request)>)/ should do it... the (?m:.*?) part is the secret. :)

answered Nov 14, 2013 at 20:28

paulotorrens

2,33921 silver badges32 bronze badges

Comments

Enissay · Accepted Answer · 2013-11-14 20:33:24Z

0

This should do with the flag s on:

(\[trace-[0-9]+\].*?<\/(?:reply|request)>)

Live DEMO

answered Nov 14, 2013 at 20:33

Enissay

4,9633 gold badges33 silver badges59 bronze badges

Collectives™ on Stack Overflow

PHP Regex match between two patterns

7 Answers 7

Comments

Comments

Comments

Comments

1 Comment

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

7 Answers 7

Comments

Comments

Comments

Comments

1 Comment

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related