Announcement

Collapse
No announcement yet.

Nepomuk search documentation - search functions and operators

Collapse
This topic is closed.
X
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    Nepomuk search documentation - search functions and operators

    Some time ago, I dropped Google Desktop search, and went with Nepomuk, largely because it allowed me easily to search in a specific area of my filesystem, and thus severely constrain false hits. I am quite pleased with this tool, in comparison with Google's. I'm now wanting to be able to do more than simple single-term (or tag) searches.

    However, I have been unable to find any documentation in the installed Kubuntu Help facility (or whatever it's called). Nothing. Very odd.

    Online, I did find some good material:
    * http://userbase.kde.org/Nepomuk - this has some very good information - really tells what this is, what is it trying to do, how to deal with problems, AND it appears to be frequently updated. It give the impression of a very active project.
    * http://strigi.sourceforge.net/ [didn't work when I tried it - "the page is being remade" - I was referred to pages which either didn't exist or were useless]
    * "detailed KDE Nepomuk Manual (February 6, 2012)" - http://kdenepomukmanual.wordpress.com/ - this has a lot of information, puts things in context well, and is housed in a WordPress blog, so it will hopefully be around a while.
    * "KDE Linux Desktop Search with Strigi and Nepomuk":http://voices.yahoo.com/kde-linux-de...k-7339846.html - gives some information the resource above does not, and so is also worth consulting.

    Now we get to the crux of my problem: I want to be able to use logical operators, as one customarily does with "advanced" searches. There is, it appears, NO information about this anywhere in the universe, except in my own notes, which derive from my experiments this evening. It occurs to me that others might profit from my sharing this:

    * Nepomuk desktop search is accessed primarily in Dolphin - do this with "Ctrl+F". If you want the search results to tell you WHERE the found item is (duh) - which is really important when searching a directory tree segment containing multiple subdirectories and files, before invoking Nepomuk request display of "path" by clicking Dolphin's window column line (it typically contains "name", "size", and "date", by default), then click "other" in the modal window that appears, then "path" (it took me months to figure this out!).
    * Basics:
    ** The search target can be either the filename or its contents; there are buttons in the search interface to specifiy which.
    ** The unit of search results is the file. Location IN the file is NOT returned.
    ** Search in a directory tree goes INTO all subdirectories.
    ** What is searched for is the character string(s) you give the search engine. There is a wildcard operator, but it is often unnecessary: "up" "so*" and "*up" all return all instances of "soup".
    * Search results can be filtered -
    ** If Dolphin's "startup" settings has "show filter bar" clicked, entering a string in the text entry field will cause display of only those file names containing that string.
    ** Clicking the green "+" to the right of the search term entry field (visible after pressing Ctrl+F) allows access to additional filters - file type, date, ratings.
    * Advanced search IS possible -
    ** Text operators and, or, not are the ones I have verified - caps do not matter. Use of such operators appears to slow the search; this matters mostly in content searches, as opposed to filename searches.
    ** Examples (using a directory containing text files each containing different recipes): "bread not nuts" gives me all files (recipes) in which the word bread appears and nuts does not, which eliminates my banana bread recipe, for example. And is the default operator and need not be specified; e.g. "chocolate nuts" gets me both my chocolate pumpkin bread and my chocolate cookies (because both contain "nuts"), and "chocolate or nuts" gets my recipes in which either or both appear.
    ** Two warnings
    *** *Not* is tricky. "*not* {whatever}" appears to cause infinite looping, but {A} not {B} gives all A's which do not also contain B.
    *** Do not use parentheses - these appear to cause infinite looping, sadly.

    The best way to get a feel for all this is to test it out on a directory or directory tree containing a limited number of text files. Nepomuk can do much more than text files, but why not learn simplest materials.

    Any comments, ammendations, etcs., would be welcome.
    Last edited by tomcloyd; Sep 02, 2012, 01:15 AM.

    #2
    Try KFind. I think it offers most of what you want. Else there is nepoogle which is also very powerful. Try those tw and see if it helps. Nepoogle has really good documentation and explains nepomuk syntax nicely. Kfind should be in the repositories if not already installed.

    Nepoogle: http://kde-apps.org/content/show.php...content=145505 & https://github.com/serantes/nepoogle

    Comment


      #3
      Oh and in kde 4.9 with dolphin 2.1 the search was quite nicely updated. Just click the little green cross and a whole bunch more features available. Its still not very powerful but its not terrible.

      Comment


        #4
        Thanks -- will look at KFind, but don't think it uses a database, as does Nepomuk, so response times have to be much slower (for content searches - which surely must use grep - argh)

        As for Nepoogle: baffling. I'm a user, not a programmer (well, maybe a little on weekends, every two months...). At the Git interface I'm simply shotgunned with unfamiliar information. I actually do NOT live at Git. This is strange stuff for me, and for must users, I expect. Where's the bloody download module? Why am I not being told that since it's python (or is it all - how would I know) I don't have to compile it, which I don't know how to do anyway. This is simply NOT usable as is. That's just dumb. All that work, and it can't make it across the finish line to the actual user. (Sigh - nerdism strikes yet again...)

        OK...found it. Right after firing off the first draft of this semi-rant.

        There's a button labeled "ZIP". On a mouseover, the tooltip reads "download this repository as a zip file". BAD INTERFACE, people. The button needs to read: DOWNLOAD ZIP FILE. If I have to think about it, it's a failure, remember?

        So many ways to fail, when you're smart (I should know...it happens to me, too, in my profession!)

        Thanks.
        Last edited by tomcloyd; Sep 02, 2012, 10:13 AM. Reason: clarification & update

        Comment


          #5
          Uh...I believe I noted this in my original post. Thanks anyway.

          Comment


            #6
            Go to: http://kde-apps.org/content/show.php...content=145505

            Scroll to the bottom of the changelog and you'll see a download button. Just grab the file. Extract it somewhere. Open the folder and just click/double-click on the file called nepoogle. If it doesn't run you might need to make it executable (right click > permissions > executable). Enjoy. Nothing more to it.

            The Git link wasn't there to download from, it was really just a place where you can see all the bits and pieces, see its development etc. I have a habit of always checking out gits to check if huge changes have come which haven't been pushed downstream yet. So yeah. That should do it for you.

            Comment


              #7
              Fantastic! Thanks...

              Comment


                #8
                No problem. Nepoogle is incredibly powerful, and it will take some getting used to but just take the time and try learn the syntax. You'll find it quite easy and productive.

                I've always found that KDE has always really lacked an all powerful search program that ties all of nepomuk and the KDE ecosystem as a whole together which is both quick and simple to use while allowing us to really use the potential of nepomuk to its fullest.

                Comment


                  #9
                  tomcloyd@

                  You might want to take a look at Recoll. It's in the repositories, so can be installed using your favorite package manager or from the CLI using apt-get. It's powerful and very fast once the initial indexing has been done.
                  Windows no longer obstructs my view.
                  Using Kubuntu Linux since March 23, 2007.
                  "It is a capital mistake to theorize before one has data." - Sherlock Holmes

                  Comment


                    #10
                    Recoll looks as if its basically does what nepomuk+strigi does but adds a nice gui. I actually quite like the concept but I think it might be a bit redundant with KDE.

                    As a disclaimer though, I've never really tried to search inside documents so much as I search for their context or metadata or tags (basically I apply the idea of semantic searching) but I feel Recoll might be better at traditional searched i.e. no context but powerful depth of search abilities. It seems to be a competitor as much as I see it to be a complementary program.

                    Comment


                      #11
                      OK, OK, I get it! We don't need documentation. We have this forum. It's just amazing the wealth of response I have gotten to my post, which was more information than question. Thanks to you all. I'm especially interested in checking out recoll, as it looks like it's precisely what I'm wanting. Nepoogle also looks quite interesting. Thanks again!!!

                      Comment


                        #12
                        Thanks to all for the great response to my post. I'm currently checking out recoll and nepoogle. The former looks very good, but I'm not yet able to tell it what to index and what not to. Hoping it's me and not the program. I also need to be able to search just in a directory and not get the whole index db as a search target - if I wanted that, Google Desktop is fine. It appears that Nepoogle DOES off these functions. Am pouring over the recoll documentation to try to get it to behave.

                        Ooops. Sorry for the double posting. I couldn't see my immediately previous post. It appeared right after posting this one. Strange...

                        Comment


                          #13
                          Yeah, there's a problem with recoll, so far as I can see. I have specified that it index one specific directory tree, so I can quickly run some test queries. It ignores my specifications, and is indexing (right now) my email database. That's exactly what it is NOT supposed to do. Nepomuk desktop give you a clear filesystem tree, with checkboxes for each branch, so you can be very specific about what to index and what not to index. I don't see this capability anywhere in recoll. It has to be there for me to use it. Am I missing something

                          In addition, it is not at all clear how I might search just one directory, and I also have to be able to do that. Again, am I being dense somehow?

                          Comment


                            #14
                            Recoll user Manual

                            Yeah, you have to read the thing.
                            Windows no longer obstructs my view.
                            Using Kubuntu Linux since March 23, 2007.
                            "It is a capital mistake to theorize before one has data." - Sherlock Holmes

                            Comment


                              #15
                              OK, I can read. Used to be able to, anyway. And...doing so, I find that configuration files can be manually edited to quite specifically direct the indexing. What was confusing was that the GUI appeared to be offering that functionality, but clearly had no effect. Made no sense to me.

                              On the plus side, this thing has magnificent documentation (lack of which was my original complaint about Nepomuk), and, from initial appears, all the power one might ask for. Definitely the power user's indexing/search tool. Hard to go elsewhere, seeing this, so I'll stay and fiddle with it. I am persuaded, as they say. Thanks for the direction to this tool.

                              Comment

                              Working...
                              X