RegExp help: multine line var pre_replace with link doesn't work... where am I wrong?

marcnyc

Hi I have a multiline variable $var and I want to replace a specific sequence of two lines with two other lines. The problem is that these two lines contain links that change from time to time so I think I need backreferences but I am not very strong on those...

Let's say the two lines I wanna replace look like this:

<tr><td><img src='tline.gif' align='absbottom'><img src='1.gif' align='absmiddle'> <a href="http://www.chaindlk.org/whoweare/" target="top"> .: Who We Are</a> 
<tr><td><img src='2.gif' align='absbottom'> <a href="http://www.chaindlk.org/sitemap/" target="top"> :: Webzine</a>

And I want to replace them with:

<tr><td><img src='tline.gif' align='absbottom'><img src='3.gif' align='absmiddle'> <a href="http://www.chaindlk.org/whoweare/" target="top"> .: Who We Are</a> 
<tr><td><img src='4.gif' align='absbottom'> <a href="http://www.chaindlk.org/sitemap/" target="top"> :: Webzine</a>

As you can see only the images' names changed.

I was going to do this:

 $var = preg_replace('#^(<tr><td><img src=\'tline.gif\' align=\'absbottom\'><img src=\'open.gif\' align=\'absmiddle\'> ).(\n<tr><td><img src=\'open.gif\' align=\'absbottom\'> )#m','<tr><td><img src=\'tline_close.gif\' align=\'absbottom\'><img src=\'open.gif\' align=\'absmiddle\'> )\\1(\n<tr><td><img src=\'open.gif\' align=\'absbottom\'>', $var);

But I am doing something wrong I guess cause it doesn't work.

Can somebody see what I am doing wrong cause I can't see it and I am not good with backreferences and regexp?

PHPdev

One question?

Why are you using the entire replace line as the search..

Why not do preg_replace('/src\=\"1.gif\"/','src=3.gif', $var); and do a second one for the second var you want to replace?

You probably have a good reason, I just wanted to point this out..

Thanks...

PHPdev

marcnyc

Yes I need to look for that entire line because I need to replace it only when two specific lines are next to each other, like I said in the subject it's a multiline variable...

xblue

what does "doesn't work" exactly mean? parse/compilation errors? no matches? anything else?

not sure if the forum stripped them, but your pattern as it is displayed is missing quite a lot of backslashes. I'd recommend to put it in doublequotes if it contains a lot of singlequotes to save some escaping. also, remember to escape the dot where you want it to mean a dot.

PHPdev

xblue,

Where's the any key? (Homer Simpson)

PHPdev

xblue

yeah, I know that one, someone else already has got it in his signature, but I like this one as well. okay, it gets boring after some time, I probably should change it.

marcnyc

Ok, I was obviously forgetting about dots and escaping but now I thought it through and tried again like this:

$var = preg_replace("#^(<tr><td><img src='tline\.gif' align='absbottom'><img src='open_blue\.gif' align='absmiddle'>).*?(<br>\n<tr><td><img src='open_red_cont\.gif' align='absbottom'>).*?(<br>\n)#i",
		       "<tr><td><img src='tline_close.gif' align='absbottom'><img src='open_blue.gif' align='absmiddle'>\\1<br>\n<tr><td><img src='open_red_cont.gif' align='absbottom'>\\2<br>\n",$var);

PS: for some reason the forum removes the double slash from the references so even if you see \1 I actually have \1 in my code, also I notice it puts a space in before the first closing parenthesis which isn't there in my code either.

and it still doesn't work.
When I say it doesn't work I mean that it doesn't do any replacement. No error output though.

I did an echo of $var in the place right before I put the preg_replace line above to check out what the html looks like and here it is:

<tr><td><img src='tline.gif' align='absbottom'><img src='open_blue.gif' align='absmiddle'> <a href="http://www.chaindlk.org/whoweare/" target="top"> .: Who We Are</a> 
<tr><td><img src='open_red_cont.gif' align='absbottom'> <a href="http://www.chaindlk.org/sitemap/" target="top"> :: Webzine</a>

As you can see I take into account everything I have on two lines of the html but I want the links part to stay the same in the replacement so I am using the backreferences. I am also adding the \n which is there even if the html obviously doens't print an \n but I know it's there because in the html I see it on 2 lines. Also I put a ^ in the regexp because I know this is how the first line starts (should I put the ^ there or is this wrong considering that I am also taking into account the line after that?)

I'd like you to confirm that I have the right regexp syntax... my doubts are in the use of the non-greedy modifier ? (I put that there because I want \1 to be only whatever it finds until the first instance of \n only) and about the use of the backreferences (it's the first time I use them on my own so I am not sure I am using them right).

marcnyc

I think I forgot the m modifier but it still doesn't work even after adding the m.

I have also tried adding and removing the ^ and the &. I have tried this:

$var = preg_replace("# (<tr><td><img src='tline.gif' align='absbottom'><img src='open_blue.gif' align='absmiddle'>).?( \n$ <tr><td><img src='open_red_cont.gif' align='absbottom'>).?( \n)$ #im",
"<tr><td><img src='tline_close.gif' align='absbottom'><img src='open_blue.gif' align='absmiddle'>\1 \n<tr><td><img src='open_red_cont.gif' align='absbottom'>\2 \n",$var);

and also this:

$var = preg_replace("# (<tr><td><img src='tline.gif' align='absbottom'><img src='open_blue.gif' align='absmiddle'>).?( \n<tr><td><img src='open_red_cont.gif' align='absbottom'>).?( \n)$ #im",
"<tr><td><img src='tline_close.gif' align='absbottom'><img src='open_blue.gif' align='absmiddle'>\1 \n<tr><td><img src='open_red_cont.gif' align='absbottom'>\2 \n",$var);

and even without any at all:

$var = preg_replace("#(<tr><td><img src='tline.gif' align='absbottom'><img src='open_blue.gif' align='absmiddle'>).?( \n<tr><td><img src='open_red_cont.gif' align='absbottom'>).?( \n)#im",
"<tr><td><img src='tline_close.gif' align='absbottom'><img src='open_blue.gif' align='absmiddle'>\1 \n<tr><td><img src='open_red_cont.gif' align='absbottom'>\2 \n",$var);

Without any at all or with just the initial one ^ and the m modified that I had forgetten in the previous post there is some replacement going on but the wrong result is achieved and the links get lost (they don't show at all, which makes me think I am doing something wrong with the backreferences).

Please help me, after this my script is done!

Again it's this vBulletin forum that adds the space after my ^ and & symbols, they aren't there in my code

xblue

yes, the forum does have its nasty habits, anyway for me the community is so nice it does not have to be perfect.

ok, one thing: $^ will never match anything even with the m-modifier set, these sign just mark end and the beginning of a string or line, and still there has to be a newline character in between them (this is from the tutorial I linked to in your last thread on regex).
And in your second and third approach the \n may fail on windows' style newline that consists of 2 characters \r and \n, but that's just a guess.

yet another: you still don't seem to escape the dots (that's not contributing to you current problem, though, keep it for later if you want, but keep it in mind).

then: if your pattern doesn't find anything, cut it down to its pieces.

test parts of it, best with preg_match_all($pattern, $string, $matches); and print out print_r($matches); to see if it matches what you want it to.

no joke, this is the best way I know to debug patterns that return less results then one expects.

marcnyc

Hey thanks for the suggestion about the preg_match_all and print_r... that helped me figure out that my search string was indeed right and the the real problem must lay in the backreferences or at least in the replacement string. I also made sure that there were no \r after the \n.

Now that I know that my code in question actually finds what I want it to find HOW do I get it to return what I want it to return, which is what I already posted but keeping the links in there unaffected???

Please suggest something!

xblue

you can reuse anything that gets called as a separate array element in your array $matches in the replacement argument.

I usually use the $0, $1 notation instead of \0, \1.

btw, what do you need the backreference for?
from your example, it seems that you want to replace something static (the numbers in the filenames) with something static (some different number) and the rest is supposed to be unchanged, so I don't really see a need for backreferences here.

marcnyc

Ok, maybe I haven't been clear enoguh so let me explain everything correctly breaking down the code:

I have a multiline varibale where every line is made like this:

I have links of the form of:
<a href="........." target=">something</a>

Then I have stuff before these links and after these links:
BEFORE: <tr><td><image src="...."> etc...
AFTER: </td></tr>

Now, I need to MATCH two lines according to WHAT TYPE OF IMAGE THEY CONTAIN and every time there is a match I need to REPLACE the match with two different lines which contain DIFFERENT IMAGES but the SAME LINKS.

Because these links change at EVERY LINE I thought I was supposed to use backreferences to put the links back into the new line (since I know what images I am gonna replace the matched line with, but I don't know what links are gonna be in the replacement string because those link change everytime on every line).

Therefore, again, I need to MATCH two CONSEQUENT lines based on WHAT TYPE of IMAGE they contain and REPLACE the matched line with a SIMILAR LINE where ONLY THE IMAGES are DIFFERENT but the LINKS are the SAME as in the matched line.

This has to happen on the entire multiline variable, whose number of lines changes too.

I hope I have been able to explain myself corectly. Sorry if I wasn't clear enough before and sorry if I wasn't supposed to use backreferences and if that did mislead you... I really hope that with this new clarification you will be able to explain to me how to do this thing.

xblue

nope, sorry. what you say now does not match the example you gave earlier:

two different lines which contain DIFFERENT IMAGES but the SAME LINKS

<tr><td><img src='tline.gif' align='absbottom'><img src='1.gif' align='absmiddle'> <a href="http://www.chaindlk.org/whoweare/" target="_top"> .: Who We Are</a><br> 
<tr><td><img src='2.gif' align='absbottom'> <a href="http://www.chaindlk.org/sitemap/" target="_top"> :: Webzine</a><br>

the links are different, aren't they?

anyway, the regex you posted earlier:

#^(<tr><td><img src='tline.gif' align='absbottom'><img src='open_blue.gif' align='absmiddle'> ).*?(<br>\n<tr><td><img src='open_red_cont.gif' align='absbottom'> ).*?(<br>\n)#i

you say this (or a similar version, probably with m modifier) works fine when used in preg_match_all? that's a good starting point then.

as far as I can see, the links would be supposed to go into .*?, right?

so, to reuse this part, first thing is to put it in parentheses. Actually you don't need all the parentheses you put around every other part.

then count opening parenthesis from left to right until you reach (.*?), and this number you use in the replacement argument.

marcnyc

For some reason I haven't been receiving any notifications from PHPBuilder (any changes/problems with the site?) so today I found the solution and I came back to post it in case somebody, besides the very kind and available xblue, was following this thread... then I have seen his(/her?) post so an answer to that is due:

Dear Xblue, thanks for your continuous help!
What I meant is that the links are different on every line but I need them to stay the same after the replacement, you know what I mean?

Anyway, you have pointed out one of the two errors I was making: I didn't know I had to use parenthesis... if I had read your post before spending a couple of course trying things blindfolded I would have saved time! ;-)

Anyway, the parenthesis were one thing, the other thing was that I was using the wrong backreferences... I thought I was supposed to use \1 and \2 but now I have found out it works with \2 and \4, which I guess (now that you mention it) is due to the fact that I am using the other parenthesis that you say are useless (which makes me also think that if I removed those then \1 and \2 might have worked, right?)...

Anyway, here is the code that worked for me (remember that vBulleting might do some strange things to it):

$var = preg_replace("#(<tr><td><img src='tline2\.gif' align='absbottom'><img src='open_blue\.gif' align='absmiddle'>)(.*?)(</td></tr>\n<tr><td><img src='open_red_cont\.gif' align='absbottom'>)(.*?)(</td></tr>\n)#im",
		      "<tr><td><img src='tline2_close.gif' align='absbottom'><img src='open_blue.gif' align='absmiddle'>\\2<br>\n<tr><td><img src='open_red_cont.gif' align='absbottom'>\\4</td></tr>\n",$var);

Now that I have it all worked out let me offer you the results of my second ever PHP script (I am so proud of myself!!) ;-)

I dedicate this script to xblue who helped me so much ;-)

Here it is:

http://www.chaindlk.org/sitemap

Here is what you are looking at:
basically the sitemap you are looking at is created dynamically and on the fly (real time) from the dropdown menu that you see at the top of the page.
The drop down menu is one of the best and most cross-platform I have found. It is called UDM (Ultimate Dropdown Menu) and is done by the fine folks at www.brothercake.com
It is javascripting and DHTML.
In case somebody uses UDM or will be using UDM and wants to use my script I will allow for use.

Basically I process it and create a 2-column site map...
I also added the following features to my script, here is a list, so you too can be proud of me ;-)

MAIN folders now are in red
SUB and child folder are in blue
Non-Linked Descriptive MAIN folders printed out in bold to avoid an entry in the sitemap that is linked to nothing (no dead links such as <a href="">)!!!
all link's targets ("_top", "blank", "whatever") are retained
ability to display the sitemap as one column or as two
real time dynamic creation of the sitemap from the current custom.js file (you never have to manually edit anything in this script, it always mirrors custom.js)
the entire sitemap is split in half and divided into two columns and the cool thing is that it divides in half respecting the main menus so that you want have part of a main menu on the bottom of the left column and the other half on the top of the right column (this was really hard for me! ;-)
backward compatible with both UDM 3.7 and 3.8
optional new color-emphasis feature (just added!!!) - check out my sitemap page to understand... If you click on any of the MAIN menus (either from UDM or from the SiteMap) the SiteMap page will color-emphasize the menu you actually clicked!!! (tricky to achieve, but can be kind of cool for cases like my site where the MAIN menus of UDM all go to the same SiteMap page and you want to actually change that SiteMap page to emphasize where you clicked...
if somebody has an extra MAIN menu like mine (see where it says "<<-- MAIN MENU (click here)!!!"???) that can be removed also to avoid creating an useless entry in the SiteMap (see it is now mirrored!)

...now to understand the following features take a look at the images I have used (there are 7 of them, and if you download them all to your disk you will see what I am talking about!!!) ha ha

the first red MAIN folder at the top of every column uses a different image to avoid that it will seem like it is branching out from nowhere!
the last folder at the bottom of both columns (regardless if it happens to be a red MAIN, a blue SUB or a blue CHILD folder) has a different image to avoid that it will seem like it keeps branching out when nothing actually comes after it!
every time a SUB menu's instances of CHILD menus are finished the last of the blue CHILD folders has a different image to avoid making it look like it keeps branching out when the SUB menu is really finished - this will be in effect regardless of the fact that the last CHILD folder may precede a new SUB menu or a new MAIN menu (this was another really hard thing to implement... always trying to look ahead of where the script's pointer is and finding out what comes after etc... wow, really hard for me to get that ;-))

Well that is it, I pulled it off, thanks to you at PHPBuilder and to xblue!!
THANK YOU VERY MUCH AGAIN!!!