is there a conditional/selection regex?

jipe

hi,

i'm looking for a regex that adheres to the following specification, but i'm not sure if it's possible to do such a selection. any help would be greatly appreciated.

i want to replace all occurences of a comma with the phrase "COMMA_DELIMITER", if and only if the comma is located within quotes.

examples:

tom said "hello, world"

becomes

tom said "helloCOMMA_DELIMITER world"

but 

tom said, "hello world"

would stay the same.

is this possible?

the following regex would replace an entire quoted string with COMMA_DELIMITER:

$sentence = preg_replace("/(\"[^\"]*\")|('[^']*')/", "COMMA_DELIMITER", $sentence);

is there a way to tweak the above (or just a better way in general!) to replace a comma, if and only if it is located within quotes?

i guess an english equivalent to the problem would be

if ((there's an opening quote and an aribitrary number of characters before a comma) 
&& (there's an arbitrary number of characters and a closing quote after the comma))
{
       replace the comma with COMMA_DELIMITER
}

thank you!

xblue

looks a little like csv file parsing, is it? if so, some examples should be available somewhere, i think - may they be good or poor, i personally found that sometimes examples may give you ideas about the overall approach you simply had not thought of.

anyway, back to your problem: if opening and closing quotes are the same symbols, it's hard to detect which one's which.

2 possibilities I can think of right now:
- "masking" them with something else as you seem to be doing with the commas already. however, trouble is ahead if you cannot garantee that your placeholder will never appear in the original data.
- match all quoted strings no matter whether the contain commas or not in order to keep track of opening and closing quotes, then use another function to actually replace the commas. i'd probably go for preg_replace_callback(), though preg_replace with e-modifier might work out as well.

jipe

thanks for your reply xblue 🙂

i'm actually not doing csv file parsing, as such. rather it's going to form part of a css/xhtml accessibility diagnostic tool i hope to write. but i will definitely look into csv file parsing for possible ideas.

however, the fact that i gave an example of a comma is not really too important. the main thing is that i want to perform an operation on a certain character only if that character is within quotes. i just happened to give an example of a comma.

i was hoping that someone with some backreferencing regex experience may have some ideas for me. i was trying something like:

$sentence = preg_replace("/(\"[^\"]*),([^\"]*\")/", "$1COMMA_DELIMITER$2", $sentence);

where $1 is a backreference to everything before the comma - i.e. (\"[^\"]) - and $2 is a backreference to everything after the comma - i.e. ([^\"]\")

but that does not seem to work. am i using those back reference variables correctly? i'll look into the csv parsing you spoke of in the meantime.

thanks!