Documentation for Confluence 5.4.
Documentation for Confluence OnDemand and earlier versions of Confluence is available too.

Skip to end of metadata
Go to start of metadata

This page describes the special words and punctuation marks you can use to refine your search.

Matched phrase search

Use double quotes to search for content that contains the phrase 'cheese one', or a phrase where 'cheese' and 'one' are the major words:

Note: Confluence will ignore common words (stop words), including 'and', 'the', 'or', and more, even if they are included within double quotes. See the default list of stop words used by Confluence's search engine, Lucene, in the Lucene documentation.

For example:

  1. Searching for "cheese one" returns only pages in which 'one' appears as the first word (other than stop words) after 'cheese'. So it will return 'cheese for one' or 'cheese to one' or 'cheese one'. It does not return 'one cheese' or 'cheese flamingo one'.
  2. Searching for "the one" returns all pages containing 'one' because 'the' is a stop word.

If you would like to override Lucene's tokenisation and stemming, please cast your vote on this improvement request: CONF-14910.

OR search

To search for content that contains one of the terms, 'chalk' or 'cheese', use the operator OR in capital letters:


On this page:

Related pages:

AND search

To search for content that contains both the terms 'chalk' and 'cheese', use the operator AND in capital letters:

NOT search

To search for content that contains 'chalk' but NOT 'cheese', use the operator NOT in capital letters:

Excluded term search

To search for content that contains 'chalk' and 'butter' but not 'cheese':

Grouping search

To search for content that must contain 'chalk' but can contain either 'cheese' or 'butter', use brackets to group the search terms:

Title search

To search for content with 'chalk' in its title, where title is the field keyword.

Wildcard searches

Single character

To search for 'butter' or 'batter' you can use a question mark as a wildcard:


To search for 'chicken' or 'chickpea' you can use an asterisk as a wildcard:

 

Multiple characters

To search for 'chick' or 'chickpea':


You can also combine search characters to get the exact word. For example the search term below will return 'chick' but not 'chickpea':

Case sensitivity in wildcard searches

Confluence is case sensitive for wildcard searches.

Note: All the example searches given above will search across the default set of fields which are stored as lower case and therefore all searches of that style should be given lower case search terms (as shown in the examples).

However, if you were to search one of the case sensitive fields, such as 'content-name-untokenized' the case of your search term would need to match the document you are searching for.

Proximity searches

Use a tilde character followed by a number, to find two words within a certain number of words of each other.

For example, the following search will return 'Octagon blog post':

The following search is not valid:

Range search

Use the operator 'TO', in capital letters, to search for names that fall alphabetically within a specified range:

Note: You cannot use the AND keyword inside this statement.

Fuzzy search

Use a tilde character to find words spelled similarly.

To search for octagon, if unsure about spelling:

Combined search

You can also combine various search terms together:

Searching for macros

You can search Confluence content for anywhere a macro is used. To do this, just add macroName: to your search and append the macro name after the column. For example, search for all excerpt-include macros:

For more information about macroName and other search fields, see Confluence Search Fields.

Searching for labels

Use the 'labelText:' prefix to search specifically for content that has a specific label. The table below gives examples of search terms that you can enter into Confluence's search box, and the search results that you can expect.

Searching for ...

Returns content that ...

recipe labelText:chocolate

contains the word 'recipe' or has the label 'chocolate'

recipe AND labelText:chocolate

contains the word 'recipe' and has the label 'chocolate'

labelText:cake OR labelText:chocolate

has the label 'cake' or the label 'chocolate'

labelText:cake AND labelText:chocolate

has both labels 'cake' and 'chocolate'

The 'labelText:' prefix is an example of a search field. See more about Confluence Search Fields.

120 Comments

  1. Sorry about that. I meant to say...
    I've noticed that + seems to work too. AFAIK...

    chalk AND cheeze

    and

    +chalk +cheeze

    get you the same results.

    Gotta remember to hit that preview button before posting. (wink)

  2. If wildcards are invalid at the beginning of the string, how can I search for *@domain.com ?
    This would be useful to regroup emails coming from a particular source or client.

    1. Did you find a solution?

  3. I found that underscore works the same as quotations.
    So if I am searching for the exact phrase for:

    "How do I disable indexing of attachments"
    

    The following:

    How_do_I_disable_indexing_of_attachments
    

    would return the same result.

  4. Hi,

    If I have a text on pages for e.g.

    tag1 this is some temp data tag1-end
    tag2 this is some test data tag2-end
    tag3 and this is dummy data tag3-end

    and I want to searxh for "temp" which which lies between tag1 & tag1-end, what would be the query syntax ?

    Something like : contentBody:"tag1*temp*tag1-end" ???

    Thnx

    1. Hi Ankur,

      Lucene will basically consider this, tag1*temp*tag1-end, as a full text and does not consider on other random characters. On the other hand, if you were to remove the quote, the hyphen from tag1-end, will be tokenized, hence, displaying results from pages consisting of tag1 and end as well.

      Regards,
      Mei

  5. Hi,

    Noticed a couple of typos on this (otherwise excellent) wiki page, plus I had a question about usage.

    First, notice the typo serch just ahead of the other typo, Mutliple, in Mutliple characters on this wiki page.

    Also, regarding the following:

    You can also combine search characters to get the exact word. For example the search term below will return "chic" yet not "chickpea":

    c*c?

    So does ? stand in for a single character but include no characters? I'm having mixed results. For example, I tried:

    version?

    and

    JIRA?

    and am getting far fewer results than I expect (based on previous searches).

    Thanks, and don't forget about those typos! (smile)  

    Rosie

    1. Hallo Rosie

      Thank you for pointing out the typos. I've fixed them. I hope someone else will pick up on your questions (smile)

      Cheers — Sarah

    2. Hi Rosie,

      I believe the "?" character represents a single character, hence the line that reads

      should be read as

      I have made the changes to this page. Thank you for pointing that out.

      Cheers,
      Azwandi

  6. Is it possible to run a pure date search.

     I want to list all modified documents within the last week.

    Andreas

    1. Hi Andreas,

      Currently, Confluence doesn't supported this feature. However, there are few similar feature requests being raised at:

      Please cast your vote and add yourself as a watcher for future updates to the reported request. Thanks.

      Regards,
      MG

  7. Anonymous

    Search by "user" or "author"

    How do I search by "user" or "author"

    Can't find it. I would like to know who authored what quickly.

    Please let me know by emailing to gchutrau@gmail.com

    Thank you

    1. Hi there,

      There is an improvement request at CONF-2328 to cater this requirement. Please note that this has been implemented in Confluence version 2.9, which is scheduled to be released in the near future. Hope this helps.

      Cheers,
      Azwandi

  8. Hi

    How do I search against a date?

    Specifically I'm trying to do this in an advanced search field, but can't work it out on standard search either

    What I want to do is created date > something

    ie created: >"29 Jul 2008"

    Regards

    H

    1. Hi Hamish,

      This feature is not supported in Confluence currently. However, there is a similar feature request being raised at:

      Please cast your vote and add yourself as a watcher for future updates to the reported request. Thanks.

      Regards,
      MG

  9. Anonymous

    Hi ,

    Is it possible to search pages for lastest modified

    1. Try this:

      (recently-updated:spaces=YOUR SPACE NAME|types=page,comment|width=100%) Substitute the brackets with flower brackets.

      Its not necessarily a search, but it elegently displays recently updates pages at one convenient location.

  10. Anonymous

    Hi,

    working in software, and many of the docs relating directly to code, we often want to search exact strings with underscores (funct names, macros, defines, enums etc).

    But it appears Confluence uses the underscore to split words up, so the search "ERR_CONDITION_REGISTER" is searched as "ERR or CONDITION or REGISTER".

    Is there a way to make it search without fragmenting on the underscore?

    Nick...

    1. Hi,

      This behavior is actually controlled by lucene analyzer that is used by Confluence. You need to customize Confluence core code to do this as this is a default behavior of the search engine.

      Cheers,

    2. Anonymous

      I found searching "ERR_CONDITION_REGISTER"~1 (a proximity search) is a decent work around.

      JRA-5567 may also be helpful to vote on

  11. Anonymous

    Hello,

    how do I search in the wiki markup source. I deleted a flash player and now I want to search for all wikis/pages which contain a link to the player flvplayer.swf to find all broken links.

    Kind regards, Marc

    1. Hi Marc,

      Thanks for your query. Confluence's search functions actually search the wiki markup source itself. Hence, if you conducted a search for flvplayer.swf, all pages containing that string should appear in your search results.

      Bear in mind that search results only present text content, so your search term flvplayer.swf would not be shown in these results. Hope this information helps.

      Kind regards,

      Giles Gaskell
      Technical Writer
      ggaskell@atlassian.com
      ATLASSIAN - http://www.atlassian.com

  12. In a Feb 04, 2008 comment, Roy Hartono noted that for search, underscore between words in a group works the same as quotations around the group of words (searching for a_b works same as "a b").  Is there any way to search for underscore_separated without returning pages that just have underscore and/or separated standing on their own?

    1. Hi Ken Ho,

      Unfortunately Confluence does not have that type of searching because Confluence Indexing service(Lucene) does not support this feature. If you are keen to have this feature, feel free to raise an improvement request at http://jira.atlassian.com. Vote for it and add yourself as a watcher.

      Cheers,
      Komathi

      1. Thanks for your response.  I created the issue http://jira.atlassian.com/browse/CONF-14554 but fyi, one is apparently not allowed to vote for an issue that one reported.

        1. I've added my vote to this ticket as well. (And to http://jira.atlassian.com/browse/CONF-3036, which also includes the same problem, and has been rated Major for quite a long time now).

          Anybody in any software-related industry needs to be able to search for tokens with underscores.

  13. Anonymous

    Hello.

    I was wondering whether there is any way to exclude the searching of markup? For example if a user were interesting in searching for "panel" for the purposes of finding mentions of topics like "panel discussion," "window panel" etc, currently their search result would include every page that uses {panel} in the markup.

    I'd like to suggest that default searches should only search on visible content. If the average user searches a wiki for the word "color" it looks like the search function is broken when almost every page in the wiki is returned as a result, yet none of them have the word "color" on the page. It isn't generally an expected behavior for a search engine to return results where the matches are invisible to the searcher. (smile)

    An option to perform markup searching for a given search would still be desirable, of course.

    Thanks,

    Kellie

  14. We are having a hard time with search. Trying to search for "GUID" always returns results containing "guide" and no results containing GUID. I tried: GUID not guide, and "guid " but it always returns Guide results. It also has the recommendation "Did you mean Guide" which if you click on returns no results, even though it returned 277 results for guid all of which have Guide in them ...

    It seems very broken. If you search for CDS you will get all the CD results because it thinks you just pluralized CD. Is there some quoting method that will give you EXACT results?

    1. We've upgraded to the latest version and the search is still broken.

      "GUID"  -> Returns only "Guide" results. But still says "Did you mean Guide?" But there's no "No, I did not mean guide" option
      "GUID " -> Trying to game the system now. No luck
      %GUID%  -> This actually says "Did you mean %Guide%?" and still returns only Guide results
      "GUID" not guide -> Same results :(
      

      Here's the best one. I create a page that has the term GUID on it and not guide and filter the results based only on pages I created.

      GUI?   -> No results
      gui?   -> Finds the page. Even though on the page it's GUID!
      
  15. Anonymous

    In 3.0.1, searching for whatever, and with Who: adam to ben produces a stack trace.

    1. Hi

      I found a bug report in JIRA:

      Please add your comments to the discussion, vote on it and add yourself as a watcher for future updates. Also, please bear in mind the following document on how we schedule features for inclusion in our products: Atlassian Bug Fixing Policy.

      Hope that helps.
      Best rgds,
      Zed

  16. Need some help here.

    I got a user report reported that when he search by input exact word, the return result does not show the exact word that he input.

    Example: Create a page contenting "pathed". Then the search the page by key in "pathed" at the search box (with the double quote, as I want the exact word), the return result show path and paths. The search result does not return the exact "pathed".

    Please help.

    Thanks.

  17. The wildcard search is not working for me. I added three pages to the Sandbox: Lot27, Lot27a, Lot27b. When I search for Lot27, I only get the first page. When I search for Lot27*, I expect to see all three pages, but instead I see 0 results. See http://sandbox.onconfluence.com/dashboard.action as an example.

    What am I doing wrong?

    Thanks in advance for any help! Carrie

    1. Anonymous

      If you search for lot27*, does it work?

      Gerald

    2. I'm having a very similar issue with search box as well ... Seems that if I type in the ENTIRE string I'm looking for, I get a hit ... if I type in partial, no result is returned?  ie, String is jeff.caron ... search for jeff.caron, and I get a hit ... search for jeff, no results found .... search for jeff*, no result set found ...

      I do beleive we're using the search correctly, any other thoughts?

      1. Anonymous

        We have the same problem with wildcard searches using Confluence 3.1.1. For example, on a page, we have the term serveServletsByClassnameEnabled. If we search for the entire term serveServletsByClassnameEnabled we get 1 hit. However, if we search for serveServlets* then we do not get any hits.. Do we need to configure anything someplace?

        1. Anonymous
          That seems to be a bug in Lucene. I just tried it and I stopped the execution in a debugger and fired queries straight at the index and wild-cards work, however it seems that string does not return any documents. So unfortunately there is nothing that can be enabled to make this work.

          For Jeffrey, I notice you have a full-stop in your query. If you remove the full stop your query should work as expected. Punctuation is usually removed when things are indexed, together with stop-words (words with little meaning, or very frequently occurring, like 'a' or 'the'). I do not know why it fails when the full stop is present but removing it fixes the problem.

  18. Anonymous

    Is there any way you can search FILENAMES only. I have tried to put in the following string

    It displays a list of attachments living up to the criteria, but it also searches within the files (PDF as it is).

    I have also tried the "title" parameter, but it returns no results on attachments.

    1. You are only limiting the search in the filename field to AAA or BBB (and not ccc), however "Examplename" can exist anywhere in the document, which is why it also looks at the text inside pdf documents say. I believe if you are after a search on just filenames you want to do filename: (Examplename1 OR Examplename2) note that the filename field is tokenised so you can do searched like filename: pdf to return all pdf documents.

  19. I'd like to allow users to use natural language queries, such as, "How do I create a new page?" and have the search return the most relevant results. Is there a plugin for that?

    1. I am not aware of any such plugin. I did have a very brief look on http://plugins.atlassian.com but could not find any. Maybe an opportunity for you to write one?!

  20. I want to create a search macro and use the lucene api to search confluence.is there any such macro available.if not then can you Please tell us to use the lucene API in the mcaro to do the search.

    Thanks

    Amarendra Thakur

    1. Hello Amarebdra,
      There are a number of macros available that makes use of searching. Depending on what you want to do you have
      {livesearch} which gives you a searchbox similar to the quicknav at the top right hand corner. It looks like this:

      {pagetreesearch}
      Which lets you search a page hierarchy, similar to the search box to the left on this page.

      You can also include just the search results on a page like this
      {search:query=Confluence 3.4|maxLimit=2}
      Which gives you this:

      Found 2 search result(s) for Confluence 3.4.

      File Confluence 3.4 Complete Documentation (HTML) DOC-20101013.zip (Atlassian Documentation > Confluence Documentation Directory)
      ... Confluence 3.4 complete documentation (HTML
      Zip Archive - 95.84 MB - Oct 13, 2010 - Download - Attachments

      If neither of these suits your needs you can look on plugins.atlassian.com for search. Of course if you still can not find something that suits your needs you can write your own macro. You will want a number of classes injected com.atlassian.confluence.search.v2.SearchManager and com.atlassian.confluence.search.service.PredefinedSearchBuilder Please see the JAVA doc for more information on the specifics of these classes. With these two classes the simplest you can do is
      (I've shortened this code a bit for readability, but you should be able to get the gist)

      Then of course it is up to you what you do with the search results that are returned.

  21. Anonymous

    I am trying to search pages, but want to exclude 'Home' page from the search results...how can I do that ?

    What query should I write for this. Currently I am using the following query but I am not sure of the placement of 'NOT'

    {search-input:type=hidden|match=query|value=Home}

    {search-input:type=hidden|match=query|value=NOT (Home)} - this doesnt seem to work.

    Please advise.

    1. The query to search for the string "examples and tutorials" but excluding the home page from the results would be
      examples and tutorials NOT(title:Home)

  22. Anonymous

    Hello:

    There are a couple of questions unanswered about wildcards here and I would like to reactivate them.

    Why is not allowed to use wildcard at the beginning of the search term?

    In my situation, we store list of names separated by commas (George,Josh,Michael,Paul,etc) and I need to find Josh! (search: *Josh*).

    1. Hello,
      The reason you can not use a wildcard at the beginning of the search term is performance. If we allowed that the search would have to iterate over every single term in the index. That is a very very costly operation and for the great majority it is better to not allow it, as a single search search could then severely impact the entire site.

      I don't know where you store your list of names, but if they are part of the body say like this
      "Lorem ipsum dolor sit amet, consectetuer adipiscing elit. Aliquam fermentum vestibulum est. Cras rhoncus. Pellentesque habitant morbi tristique senectus et netus et malesuada fames ac turpis egestas. Sed quis tortor. Donec non ipsum. Mauris condimentum, odio nec porta tristique, ante neque malesuada massa, in dignissim eros velit at tellus. Donec et risus in ligula eleifend consectetuer. Donec volutpat eleifend augue. Integer gravida sodales leo. Nunc vehicula neque ac erat. Vivamus non nisl. Fusce ac magna. Suspendisse euismod libero eget mauris.

      George,Josh,Michael,Paul,etc"

      You can simply search for "Josh" no need for wildcards. Chances are that you can do that even if you have a third party plugin that puts this information in the index.

  23. Anonymous

    This might be a really silly question, but how do you search for part of a word?

    For example, if I have a serial number on a wiki page (QEBRQ-123JB-23432A), I have to search for the entire serial number. I can't just put in 'QEBRQ' and have the page come up.

    What am I missing?

    1. You need to put a wildcard in. So in this case the query would be 'QEBRQ*'

      1. Anonymous

        Adding a wildcard does not work. At least, it does not work for me.

        I tried with and without quotes, just in case that made a difference.

        1. Ah, yes sorry I missed that you also need to convert it to lower case. so 'qebrq*' (without quotes) should find you the page.

  24. Anonymous

    Is there a wildcard that restricts the search to numbers and excludes alpha-characters?  I'm hoping for something a little less clumsy than 0 OR 1 OR 2 OR 3 ... OR 9.  Thanks.

    1. Unfortunately there is no such capabilities available to the end user. If you are a plugin developer you can look at the RegexQuery which is available in the Lucene contrib packages (and thus not bundled with Confluence) and build your own search macro that allows regular expression syntax.

  25. Can Confluence search for mixt word (uppercase and lowercase).

    I have lots of case were these kind of words appears: "EOFException".

    When i put EOFException in the search box, it returns no hit.

    I'm not sure how to search for this ?

    Thanks for help.

    Michael

    1. That should work. I just tried putting "EOFException" in a page and I can find it with a mixture of cases in the search query. I also tried putting "com.company.EOFException" and then it is not found when searching for just "EOFException" however, if i instead use the query "com.company.eof*" I find it. When using wildcards everything is lowercased this has been reported already. I know there are currently a few problems around using fully qualified classnames like "com.company.Exception" and then searching for a just "Exception", at the moment the workaround is to use wildcards and lowercase queries (due to CONF-13846)

      1. Hi Daniel,

        Yes you are right. It works with wildcard and lowercase or the whole string for the classname.

        thanks for the tip.

      2. Hi Daniel,

        Yes you are right. It works with wildcard and lowercase or the whole string for the classname.

        thanks for the tip.

      3. As mentionned, your tips is working.

        However, lots of people doesnt know the whole classname  (xxxxx.yyyyy.zzzzzz) and thus, hard to search...

        As it is not possible to use wildcard in front of word, can I search for only a subset of a classnamen and how ?

        1. Hello Michael,
          This is a known problem and at present I am not aware of any workarounds. I am however working on improving the search in Confluence and I know this is a problem. At present I have no clear path to improve this but I am aware and I am hoping it will be improved in a future release of Confluence. We have discussed allowing "n-gram" or "substring" searching but have not yet made a decision on it due to performance and index size reasons.

          1. ok.

            Thanks Daniel for the input.

  26. Anonymous

    Hello

     How to search for texts like c:\ ... cause i want to search for all  lokal mappings in wiki and replace them with unc path names

    1. That string contains quite a few reserved words so you need to escape them. In the case of finding c:\ the query should be (without quotes) "c\:\\"

  27. Hello,

    Not sure if this is possible.

    I would like to search for pages that what created or edited by a specific user containing a specific word.

     

    I was looking to something like

    wordtofind AND username:"John Doe"
    
    

    Thanks for help.

     

    Sincerly,

    Michael

    1. Hello,

      You are very close, but for pages the field that contains the creator name is "creatorName" for a list of supported fields look at Confluence Search Fields.

      You query should be(assuming username jdoe)

      wordToFind AND (creatorName:jdoe OR lastModifiers: jdoe)
      1. Yeah, close (wink)

        thanks for helping.

         

        Mike

  28. Help with the following would be appreciated.

    I am trying to search for the text string 'PDF/A' (without the quotes).

    I have tried of the following permutations:

    PDF*A

    PDF?A

    PDF\/\A

    None of which work.

    Anyone able to point me in the right direction please?

    thanks.

    1. I think you are facing two problems here. The first is that '/' is a token separator, so PDF/A will be considered as two words in the index, 'PDF' and 'A', I do not think you can prevent that from happening. Second is that 'A' is a stopword in English, since it is very common it is stripped from the index and thus all that remains is 'PDF' so you should be able to search for PDF and then maybe filter it to find your page.

      1. Hi Daniel, 

        thanks very much for your prompt response.

        Would be good if we could wrap this string text in to quotations so as to avoid circumvent this Stopword issue (i.e. "PDF/A" or "PDF\/\A").

        Possibility of an enhancement?

        cheers

        Alan.

        1. Yes it would be good. Of course you could do such a thing in an extractor and stick it in your own field that is not analysed, and then do a fielded search. Very cumbersome but possible. I am hoping that we can revamp our search infrastructure a bit in the near future but as always many things are in need of attention so I do not know when such an improvement could happen.

          1. Yes, sounds good.

            We already use a component field, but of course this information is only relevant if actually populated in the first place.

            We look forward to future builds.

            cheers

            Alan.

      2. Our organization is new to Confluence and we're still getting up to speed with the search ability.  2 Q's:

        1--Where can I find the stopword list that is being used now?  I found the Lucene list but it's not clear whether that is being used or is only under consideration.  example: Lucene list shows single numerals being 'stopped', but I can search and get hits on "5" (no quotes).

        2–Is there a definitive list explaining how Confluence treats nonalphanumeric characters in filenames?  We have many legacy docs we uploaded, with widely varying naming styles: hyphens, spaces, underscores...  Users are frustrated because they expect a filename to be searchable and the results are all over the map.  I'd like to explain the mystery and recommend a naming convention that can be searched reliably. 

        1. Hello Claire.

          1. I could not find the list of stop words listed here, so I dug it up out of the source and here is the list  "a", "and", "are", "as", "at", "be", "but", "by", "for", "if", "in", "into", "is", "it", "no", "not", "of", "on", "or", "s", "such", "t", "that", "the", "their", "then", "there", "these", "they", "this", "to", "was", "will", "with"
          2. No, I am not aware of such a list. Perhaps you can create a support ticket and get support to give you recommendations and help?
  29. Anonymous

    Hi Daniel,

    I was hoping you could help me out. We are experiencing some problems with searching for a combination of letters and numbers.

    For example: when searching for 'd560', all pages that contain items (terms) starting with d followed by a number are listed in the search results (e.g., d577, d701, d60). Searching for 'd AND 560': same results. Solely using 'd': same results. Solely using numbers gives no results at all.

    Since we use confluence for factory manuals, these queries (they are numbers of specific machines) are quite important to us. Would you know what we can do / how this can be fixed?

    Thanks in advance,

    Joanne

    1. Hello Joanne,

      I don't know what version of Confluence you are running, but when I tried it on the latest (development) build of Confluence 4.2 the search worked as expected. Granted I only edited 5 pages and added in the machine numbers to those pages, so my dataset is small and not really representative to your use case. However, Confluence found and highlighted the correct page and term. Could you create a question on https://answers.atlassian.com/ with the relevant details as you are far more likely to get a response there. If you e-mail me the URL with the question I will look at it as well.

      1. Hi Daniel,

        Thanks a lot for your prompt response and testing of the issue. We are running confluence 3.5, that may work differently...?

        I have added my question to the website you listed, and will email your the url.

  30. Special characters searches! We need to search for the plus sign symbol (+). We have documentation on, for example, Push+ and we want to be able to search for the string "Push+" and get results that contain instances of the string "Push+" listed at the top of the results. The way the search results are now, the first document listed with an instance of Push+ is on page 5 of our search results, behind loads of documents that contain the word "Push," all of which are irrelevant. I have a couple of questions:

    1. How can we search for "+" special character as a string? (JIRA allows for this using '\\+')
    2. How do we influence the organization of the search results so that when searching for documents with "Push+", all exact matches of "Push+" are at the top of the results list, and not listed behind matches with documents containing the word "push?"

    Thanks in advance for any information on the matter. If I have something wrong, please let me know. I'm a Confluence newbie.

    1. Hello Susan,

      I think you are out of luck with those searches and the current search functionality we have. The current tokenising algorithms will treat '+' as a delimiter and remove it. So in the index 'Push+', 'Push' and 'push' are all seen as the same token.

      1. Your query is almost correct to search for special characters you escape it with a single back slash, so \+ but I believe that + is a token delimiter so it may not give you any improve results :/
      2. Labelling is usually the way to go. Labels influence the ranks quite a bit. However I have not tested your specific use-case, it could be something worth trying.
  31. I'm trying to search for "guid". The search engine apparently interprets "guid" as a misspelling of "guide" because, in addition to all the pages that contain "guid" I get all the pages that contain "guide". How do I turn this off?

    But when I search for guid NOT guide (or guid -guide), I get no results. Why is this?

    Can anyone help with either question?

    1. Hello Charley,

      I'll answer both your questions. What you have run into is a problem with stemming. Confluence uses the porter stemmer algorithm to stem words at index time. Stemming essentially means stripping the word down to its root and index that word. By doing this we allow you to find pages that contains "guide", "guides" when you search for "guide". Of course in your particular case it gets in the way. It is currently not possible to turn this feature off.

      But when I search for guid NOT guide (or guid -guide), I get no results. Why is this?

      With the above explanation we can do the transformation that Confluence makes to your query and I think it becomes clear why no results show up. You query stemmed would be "guid NOT guid" which of course can not return any results!

  32. At the top of the page under "Exact phrase search" it says:
    1. Searching for "The One" returns all pages containing "One" because "The" is a stop word.

    Is this accurate? Why would the search engine still continue to ignore the word "the" even when you put it in quotes to specifically say "I want to search for this phrase containing the word 'the'"? Isn't the whole point of using quotation marks to tell the engine to ignore its normal operation and search only for exactly what you typed?
    1. That is accurate. The reason is that the stop words are removed at index time, and at query time. I am well aware for the need of this and there are a number of issues over on jira.atlassian.com touching on this subject. Right now though there is no solution to this problem. I would like to point out that the problem lies in Lucene and not in Confluence itself. Having said that I think that we should be the ones addressing this through the API that Lucene provides. Sorry I can't give a better response.

  33. Anonymous

    Is it possible to search for a JIRA issue(by summary of an issue) from JIRA issue macro in a confluence page?

    1. You can use the "Insert JIRA Issue" dialog to search the summary JIRA issues. Go to Insert->JIRA issue. Select the search tab, then input your search terms and press enter. You can also use JQL as well.

      Docs here:

      Inserting JIRA Issues

  34. Anonymous

    Hi All,

    I have the following text (with a link to google) in confluence:
    visit http://www.google.ie/ for details

    However, it is not found when I do a search.

    Is there any way around this?
    Many thanks,
    Dermot

    1. What search terms are you using? As always I think you are more likely to get accurate and prompt help on https://answers.atlassian.com/

  35. Ist there a way to enable wildcards in front of the search term and look if the performance impact is to high or tolerable? Thanks!

    1. No, that is unfortunately not possible. Unless you have a very small index the performance impact is quite significant.

      1. Hi, this is not realy clear to me. I talked to our erp and database administrators and they told me, that they have database with a size of hundreds of gb and they perform very well, even with the ability to search with wildcards in front. Why don´t you allow the customer to decide if the impact is too high or not?

        1. The problem lies in Lucene, Lucence is not a database, a database have very different performance characteristics compared to a search index.

          From the Lucene documentation on leading wild cards:

          "Note that this can be an expensive operation: it requires scanning the list of tokens in the index in its entirety to look for those that match the pattern. "

          Perhaps it should be configurable, but realistically you need a lot of memory, and a small index for it to perform at an acceptable level. As far as I know there are no plans to make this configurable as enabling it will cause both performance degradation and possibly even bring your instance down.

          1. The ticket in question to watch is

            CONF-10412 - Support substring (partial) searches for terms used in Confluence: page title, attachment names, etc. Open

            From my understanding this is currently available as an experimental feature for our OnDemand offering. It may still be removed as the performance impact we've observed is quite significant.

          2. Hi Daniel,

            memory shouldn´t be the problem. Why don´t but the whole index into the memory? I can´t belive, that the size is too big...is there a possibility to see the size of the index? From my point of view lucence isn´t state of the art regarding this topic. I mean if we have to consume 2 or 4 GB memory (500 confluence users) to put the index into memory this would be absolutly no problem! Thanks for your efforts.

  36. Anonymous

    I have pages with content IR/R/2389.4 . I have tried various searches and got mixed results. At first I thought this was because the words were being split on the backslash, but this does not seem to be the case - the results seem to treat IR/R/2389.4 as a single token.

    IR/R/2389.4          returns pages with content IR/R/2389.4

    ir/r/2389                does not return pages with content IR/R/2389.4

    ir*                          returns pages including those with content IR/R/2389.4

    ir/r/2389.?             does not return pages with content IR/R/2389.4

    ir/r/*                       does not return pages with content IR/R/2389.4

    ir/r/2*                     does not return pages with content IR/R/2389.4

    ir/r                         does not return pages with content IR/R/2389.4

    Could you please explain why the wildcard searches do not work? Is there any way of doing this sort of search?

  37. Anonymous

    I have pages with content IR/R/2389.4 . I have tried various searches and got mixed results. At first I thought this was because the words were being split on the backslash, but this does not seem to be the case - the results seem to treat IR/R/2389.4 as a single token.

    IR/R/2389.4          returns pages with content IR/R/2389.4

    ir/r/2389                does not return pages with content IR/R/2389.4

    ir*                          returns pages including those with content IR/R/2389.4

    ir/r/2389.?             does not return pages with content IR/R/2389.4

    ir/r/*                       does not return pages with content IR/R/2389.4

    ir/r/2*                     does not return pages with content IR/R/2389.4

    ir/r                         does not return pages with content IR/R/2389.4

    Could you please explain why the wildcard searches do not work? Is there any way of doing this sort of search?

  38. You say:

    You can use wildcards anywhere within a word, even at the very beginning:

     

    But my Confluence doesn't let me start search queries with an asterisk..

    It shows me an Error box stating:

    Search term(s) beginning with an asterisk character (*) are not supported.

    1. Hello,

      Leading wildcards are only available in our OnDemand offering. I'm guessing you are running a downloaded version, if that is the case it is not yet available for you. I have updated the docs to make this clearer.

      1. That's right, we're running the downloaded version. Is there a specific reason to this? I mean, isn't OnDemand a kind of lightweight version of Confluence? I thought that it has a reduced feature set in comparison to the donwnloaded version.... Seems like the opposite is the case..

  39. Hi, is there any way to get the current logged in user as a search option? For example, if I wanted to use the Livesearch macro to generate a list of pages where the logged in user is mentioned. In JIRA you can use currentUser()

    Thanks!
    -Sam 

  40. Hi Sam,

    Unfortunately, no, the Livesearch Macro was written for a different purpose. However, it sounds like the Search Results Macro may be of help.

  41. Anonymous

    I want to search only for guid but am getting back results for guide and guid.  I tried searching guid -guide but getting no getting any results.  How can I narrow this search?

    1. I do not think you can. What you are experiencing is related to stemming of the word guid. I can not think of  a way at the top of my head of how to address that.

  42. Hi,

    Special characters in "search" result in problems. 

    If I enter an umlaut character in Search, like "ä" I get "ä" in the resulting search result page. I also get zero hits.
    Eg. in Confluence we go a user called "väättänen". I don't get a hit when I search for "väättänen". It results in: No results found for väättänen. Same result when searching for 'Gábor' which results in 'No results found for Gábor.'

    So search of non-ascii characters seems to be messed up.

    How can this be fixed?

    BR, Kjell


    1. Hell Kjell,

      I just tried your scenario on a recent Confluence (in OnDemand), and I had no problem. Looking at what your results it looks like tomcat has not been configured properly (it is not using utf-8 for urls). Could you please have a read of Configuring URL Encoding on Tomcat Application Server and verify that your tomcat configuration matches the recommended one?

      1. Hi Daniel,

        This is our tomcat config:

        <Connector port="10060" protocol="HTTP/1.1"
        address="*"
        connectionTimeout="20000"
        proxyName="xx.yy.zz"
        scheme="https"
        URIencoding="UTF-8"
        proxyPort="443

        To my understanding it should be correct? In what other configuration files could the utf-8 be found? The database and Confluence are configured to use utf-8.

        ///Kjell

        1. hello,

          Dont forget to look also for database character encoding: Configuring Database Character Encoding

          Sincerly,

           

          Michael

          1. Hi Michael,

            I double checked the database configuration and it is below:

            CHARACTER_SET_CLIENTutf8
            CHARACTER_SET_CONNECTIONutf8
            CHARACTER_SET_DATABASEutf8
            CHARACTER_SET_RESULTSutf8
            CHARACTER_SET_SERVERutf8
            CHARACTER_SET_SYSTEMutf8

            ///Kjell

        2. Hi Kjell,

          That does indeed look correct. I think you have a misconfiguration perhaps you could contact support to get them to help you sort this out. I don't think I can give you efficient help here.

          1. Hi Daniel,

            Thanks for the help. I made a support case about this.

            ///Kjell

  43. ok.

    Are you doing a quick search or a site search (do both have the same results).

    And are you using another add-on to do the search live live-search or something else ?

    I remember having trouble with an old add-on that what not compliant and was disturbing the indexing and search...

    Michael

    1. Both quick search and site search give the same results.

      No add-on used for doing the search.

      ///Kjell

  44. I can't seem to get the NOT operator to function.

    The search I'm attempting to do is, entering into the search box,

    thisdomain.some.tld NOT otherdomain.foreign.tld

     

    Top of the search results it asks "Did you mean: thisdomain.some.tld Now otherdomain.foreign.tld" as if it simply doesn't recognise NOT at all.

    It seems to find all pages containing thisdomain.some.tld but also returns those that contain otherdomain.foreign.tld which is what I don't want.

    We're using a stand-alone version 4.2.11-1

    Matthew

     

  45.  I am creating a repository for services we have. When a user comes to the main page he should see a search button like Google on the page. The child page of the main page will have list of service links. When user will search a service from main page,  it will run search through its child pages and should show result on the main page under Search box .

      

    Any suggestions ?

  46. I still can't get the exact word match to work.  I'm using version 4.  I only want results for "localization", but I'm getting results for "local".

    So what should I enter according to the example below?

    You can also combine search characters to get the exact word. For example the search term below will return 'chick' but not 'chickpea':

     

    c*c?

    Thanks!

    Gina

  47. Hi there,

    I have the same problem as Gina W: searching for the word "pricer" also returns matches for "price". Is there a way to force the system to give only back exact matches?

     

    Best regards,

     

    Peter

    1. Hello Gina and Peter,

      What you are hitting is a problem with stemming. At present it is not possible to turn off stemming, or return only exact matches.

  48. Anonymous

    The information about stemming is very useful. Pity I didn't know about it sooner.

    There is another problem. When a word is entered the search insists on including close results and there does not appear to be a way to disable it. Search for the word "mai" and it will include "may". It doesn't matter if you put it in single or double quotes, add -may or -"may" or any other combination and you still can't find pages with the word "mai". 

  49. We are running Confluence 5.0 and have run into an issue with searching for a number:

    Content contains "FIX ID:  1000525"

    When I search for 1000525, no results are returned.  Also, if I search for "FIX ID" nothing comes back.  However, if I search for "ID:" I get the document back.

    Why won't it return the results for the number search?

    Thanks,

    Rhys

    1. We discovered through trial and error that the source content lines were seperated by a soft return and that when it was indexed the next line after "FIX ID: 1000525" was appended to the line above it changing the tokenized value to 1000525NEXTLINE

      Lucene must not identify soft returns as whitespace??

  50. Anonymous

    Hi,

    I'm trying to search for specific IP address in my documentation but no results are returned. I'm using OnDemand service.
    Do you have any idea or workaround?

    Thanks,

    Karel V.

    1. Anonymous

      It seems, that the problem is only with IP addresses within links (like http://10.20.30.40/).

      Thanks,

      Karel V.