Hi David,
the tokenization process triggered by the "contains text" keywords should never accept substrings as results. Because of that, the following query will return an "empty sequence" (i.e., no hits):
'c:\songs1\songbird review'[. contains text 'song']
Do you have a small example that allows us to reproduce the issue? What's exactly inside your database (the file paths itself, or meta data on the quoted files?)
If I got you wrong, feel free to give me an update. Christian
On Tue, Feb 8, 2011 at 9:02 PM, David leigh david@leighweb.com wrote:
When I search against lyrics in my song database using "contains text", text is found where ever the input string is found, even if the string is simply part of a word. So "wor" would find "word". If I search with base-uri($song)[ . contains text "wor"] the results are interesting.
Say I have the following files in my database:
c:\song\music in the air c:\songs1\songbird review c:\songs2\songbird cafe c:\lyrics\songbirds singing c:\Songs3\I like to sing c:\Songs4\Songs worth singing
searching with base-uri($song)[ . contains text "song"] will return: c:\song\music in the air c:\songs1\songbird review c:\songs2\songbird cafe c:\lyrics\songbirds singing c:\songs4\Songs worth singing
searching with base-uri($song)[ . contains text "songs"] will return: c:\songs4\Songs worth singing
My conclusion is that after the last slash (and in the database, of course, they are forward slashes) the string can be found as a substring but before the last slash, the string must be found as that whole section of the path (I've tried this with multi-word path sections and the entire path section must be specified or that node is not selected.)
This is not particularly troubling, but I'd like to confirm that the behavior is expected. It would appear that the indexing is different for these file system path sections than for the actual file name at the end of the path (which appears to be indexed like all the other text). Can someone confirm this?
Thanks!
BaseX-Talk mailing list BaseX-Talk@mailman.uni-konstanz.de https://mailman.uni-konstanz.de/mailman/listinfo/basex-talk