You are viewing limited content. For full access, please sign in.

Question

Question

Exact search with hyphen

asked on May 22, 2014

We are trying to search an exact text string that contains a hyphen eg. 2013-1. We are trying to find the results in the full text search, not in a template field.

 

Quotes don't make a difference. We have a user that claims that it used to work fine prior to ver 9 but we have no proof of that.

 

its just skipping over the hyphen and returning results like "2013. 1", "2013 1", "2013, (1"

So basically its treating the hyphen as a space. Is there a way to get around this?

 

0 0

Answer

APPROVED ANSWER
replied on May 25, 2014 Show version history

The feature of configurable delimiter set is supported since 8.3. It works on 9. You can add/remove delimiters in delimiters.txt. For example, if you want the section sign (§, Unicode U+00A7) searchable, you could add it to the delimiter set in delimiters.txt, restart LFFTS and re-index, then you could search §. But some characters must be delimiters, and unfortunately '-' is one of them. When the delimiters.txt removes a character which is required to be a delimiter, default delimiters will be used, and there will be an error message "Fail to parse the delimiter file at line 5098. Default delimiter set will be used." in the application log in event viewer.

 

I suppose it means 8.0 in this post on the old forums.

0 0

Replies

replied on May 22, 2014 Show version history

For full text searches, punctuation in the text you are searching is indeed treated the same as a space. This is because the full-text search engine does not index punctuation. This is behavior is not new to version 9.

 

Since there is no way to search document text other than a full-text search, there is not a workaround. If you wish to be able to search on exact punctuation, you would need to store that information in a (non-indexed) field.

0 0
replied on May 22, 2014
So with version 9 the number of non indexed locations is pretty low I would think. Thanks so much.
0 0
replied on May 22, 2014

Some additional points about the changes made to searching in version 9:

  • The number of non-indexed fields is entirely dependent on your setup, as indexing of fields is controlled by administrators of the database
  • Except for document text, all other indexed content (such as annotation text and indexed fields) can always be searched using a non-indexed search
1 0
replied on May 22, 2014

I had a similar problem and just had it resolved. we have to start and end our search with starts. see attached pic.

replied on May 23, 2014

I have found under the LFFTS folder on the server delimiters.txt this looks like a file that lets me set skipped words etc in the full text search. However it doesn't seem to do anything when I change it, I also restart and re-index after making the change but nothing. This used to be a feature in the GUI back in version 6 at least. So just wondering if I am doing it wrong, or if this file is still here but doesn't do anything anymore.

0 0
replied on May 23, 2014 Show version history

I believe that the ability to customize delimiters was removed starting in version 8: a similar question was asked on the old forums and Michael Allen responded that word delimiters could not be modified in Laserfiche 8.

 

The delimiters.txt file that you mention was not present on the clean installs of the Laserfiche I checked, so it may be that the file is simply leftover after upgrading.

0 0
APPROVED ANSWER
replied on May 25, 2014 Show version history

The feature of configurable delimiter set is supported since 8.3. It works on 9. You can add/remove delimiters in delimiters.txt. For example, if you want the section sign (§, Unicode U+00A7) searchable, you could add it to the delimiter set in delimiters.txt, restart LFFTS and re-index, then you could search §. But some characters must be delimiters, and unfortunately '-' is one of them. When the delimiters.txt removes a character which is required to be a delimiter, default delimiters will be used, and there will be an error message "Fail to parse the delimiter file at line 5098. Default delimiter set will be used." in the application log in event viewer.

 

I suppose it means 8.0 in this post on the old forums.

0 0
You are not allowed to follow up in this post.

Sign in to reply to this post.