Highlighter returns unexpected results if one term is found within another

[witten]

from haystack.utils import Highlighter
Highlighter(query='portland or').highlight('portland')

Actual result:

u'<span class="highlighted">portland</span><span class="highlighted">or</span>tland'

Note that the text to highlight is just "portland", but the highlighter is tacking "or" + "tland" onto the end as well.

Expected result:

u'<span class="highlighted">portland</span>'

Or perhaps:

u'<span class="highlighted">p<span class="highlighted">or</span>tland</span>'

I'm using haystack 1.2.4.

[witten]

I added one potential solution which makes the output look like the expected result above. Basically, if there would be two highlight areas that overlap or coincide, it just highlights the first one.

[witten]

I added a test for this fix. Without the fix, the test fails. After the fix is added, the test passes.

[seddonym]

I ran into this issue too. IMHO this fix is not addressing the correct problem, which is that Highlighter.find_highlightable_words is returning the wrong results. Here's a different version that uses regular expression word boundaries to return the correct results:

import re

class Highlighter(object):
    ...        
    def find_highlightable_words(self):
        word_positions = {}

        lower_text_block = self.text_block.lower()

        for word in self.query_words:
            # Use a regular expression to search by whole words
            # \b corresponds to a word boundary
            matches = re.finditer(r'\b%s\b' % word, lower_text_block)

            if matches:
                word_positions[word] = [match.start() for match in matches]

        return word_positions

Let me know if you'd like me to make a pull request.

Incidentally in the meantime if anyone wants to correct this behaviour themselves, they can use a custom highlighter - see this gist.

[jw]

I agree with @seddonym. Maybe just close this issue?