I have encountered another unexpected issue in my regex parsing exercise. When parsing the one line before (it is only one string):

car seats</A>, <A href="/baby-strollers-tandem-jogger-double-triple/b/ref=by_title1_strollers?ie=UTF8&node=166842011&pf_rd_m=ATVPDKIKX0DER&pf_rd_s=center-1&pf_rd_r=08CQNMPPYCHYZW5VVZ03&pf_rd_t=101&pf_rd_p=1355317222&pf_rd_i=165796011">strollers</A>, <A href="/baby-diapering-diaper-bag-wipes-liner/b/ref=by_title1_diapering?ie=UTF8&node=166764011&pf_rd_m=ATVPDKIKX0DER&pf_rd_s=center-1&pf_rd_r=08CQNMPPYCHYZW5VVZ03&pf_rd_t=101&pf_rd_p=1355317222&pf_rd_i=165796011">diapering products</A>, <A href="/baby-feeding-highchair-bottle-bibs/b/ref=by_title1_feeding?ie=UTF8&node=166777011&pf_rd_m=ATVPDKIKX0DER&pf_rd_s=center-1&pf_rd_r=08CQNMPPYCHYZW5VVZ03&pf_rd_t=101&pf_rd_p=1355317222&pf_rd_i=165796011">feeding items

I want to obtain the ''feeding items'' at the end. But unlike in other strings, using (?<=\"\>).*  doesn't help as it starts at ''strollers''. I was thinking of using 

(?<=\"\>).*(?=\n)

or

(?<=\"\>).*(?=\^)

to evoke the end of a line, but I am not succeeding. I am using .NET, thanks for any help in advance.

There are a number of things happening here that could be going wrong.

The very first is that you are using the wrong tool: if you look back through this forum to just about any question that refers to applying a regex pattern to HTML or XML text, the response will be "use the DOM". In this case an HTML DOM library will let you parse the HTML code and then let you apply XPATH queries to it to select the (I'm guessing) inner text of a suitably selected "A" tag.

Given that you need to use a regex pattern, then you need to be careful about the '*' quantifier as it is "greedy". The actual definition of what it does is "repeat the preceding item zero or more times matching as many times as possible". The last part is what makes the quantifier greedy. Now you are applying the '*' quantifier to the '.' operator which matches any character (possibly except the newline depending on the setting of the 'singleline' option) which means in this case that your pattern will find the first place where the lookbehind matches and then match everything through to the end of the line/text. Only then will the last part of the pattern kick in to start a backtracking process to find a place where the lookahead will match. Therefore this last part of the search will be scanning backwards through the text.

This is why your match starts at "strollers" and continues through to the end.

You have several options here. The first is to use the "lazy" form of the quantifier as in '.*?'. This form has the same definition BUT ends with "...matching as few times as possible. In your case this would not actually help much as the lookbehind would still match the first instance of "> and then match all characters up to the next linefeed character: you mentioned that your text is all one line so it would still grab everything.

Therefore you need to replace the '.' with something that is more specific. Looking at your example text (and therefore this may well be quite wrong) all of the inner text of the "A" tag is immediately followed by the closing tag or the end of the text. Therefore, to let you still include any characters that may be there, I would suggest that you use the character set '[^<]' whcih will match any character EXCEPT the "<" which would start the ending tag. Therefore your pattern would look like:

(?<=\"\>)[^<]*

(You don't need a lookahead at the end in this case).

However this will match 3 items in your example text. Again you have choices: you could go back and look at the attributes of the preceding "A" tag (something that the HTML DOM and Xpath query would make very easy) to find something that narrows the selection to just what you want, or to again use the fact that what you want is always at the end of the string and use:

(?<=\"\>)[^<]*(?=\n)

Note that this may pick up any other characters at the end of the line before the "\n" - for example, in my test platform, lines are terminated by a "\r\n" pair and so the "\r" is being included.

Susan