capitalization can cause porterstemmer_sbp_excerpt_match() to miss matches

This is correct. It only currently affects people using Porter Stemmer with Search by Page, because the core search module didn't get this excerpt fix into D7 (much less D6).

The problem is that when the SBP excerpt function lets Porter Stemmer highlight words, Porter Stemmer could miss some matches because it doesn't lower case the words before stemming them. There's a suggested fix over on that other issue.

Probably this can be made more efficient by lower casing once instead every time inside the if(). And it needs a test. And a port to Drupal 7. But it's along the right idea.

You're probably right about that, and I don't think the same key would be likely to come in more than once. I was thinking it was in a loop, but I was working from memory rather than looking at the code.

Anyway, I do very much appreciate the patch, as well as of course finding the bug in the first place. :)

And just for the record, my work on search is not tireless. I am actually very very very tired of the way things have and haven't been accepted into Drupal 7. Sigh. Anyway, not an issue with Porter Stemmer or Search by Page. :)

This prompted me to write a test for the excerpt matching function, which uncovered some more issues with it. They're now fixed in the dev branch for Drupal 6, and here's the patch I ended up with.

Care to give it a try gpk? Then I can put out a new release...

I also just made this change in the 7.x-dev branch.

Hmmm. I'm not too concerned about waLk, since I think it's unlikely to be entered, but I will add a test for a key of Walk and verify that it matches too...

Well. I am going to first see if the search module automatically lower cases search strings before you even get to this point. If it does that, then we don't need the test, but if it doesn't, then we need the test.

I get it. I just wanted to check to see whether search.module lowercases all search query keywords before we even get to the point of finding the excerpt. If it does, then we don't actually have to worry about $key being anything but lowercase. If it doesn't, then we need the drupal_strtolower($key) and a new test.

Let's see... In core search, the answer is no -- the keywords are not lowercased before the extract. So you are correct, additional tests are needed.

OK. I've added tests for the keyword capitalization as well, to both D6 and D7 versions. New test:

    // Test with upper-case in the keyword.
    $result1b = porterstemmer_sbp_excerpt_match('waLk', $text, $offset, $boundary);
    $this->assertEqual($result1b['where'], 2, 'Found match 1b in right place', 'Excerpt');
    $this->assertEqual($result1b['keyword'], 'walked', 'Found right keyword for match 1b', 'Excerpt');