Re: [EasyHack] #44681 port to CLucene from java/Lucene
Great to hear that it works & that it will make it into master :-). Any thoughts on this one: 3) When creating the CLucene FileReader (HelpIndexer.cxx), the path is converted to plain ASCII, that's probably dangerous. There is probably a way to work around this, but I haven't gotten around to it yet. Is that a problem? On 20-02-12, Caolán McNamara wrote: > On Sun, 2012-02-19 at 18:49 +0100, Gert van Valkenhoef wrote: > > Thanks again for the help. Attached a new series of patches (cumulative > > with the previously sent ones and Caolan's), in which (I think) all the > > Java invocations have been removed in favor of using the C++ components: > > Attached is an additional patch to stick together the code to date with > internal clucene and the "missing link" to use the OUString ctor that > takes UCS-4 strings. So with this applied additionally you should now be > able to type stuff help's search and see a list of results, so vital bit > apparently works :-) > > I'll integrate all of this to master in the next day or two. > > C. > > ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
License for my contributions
Hi all, I hereby declare that my past and future contributions to LibreOffice are licensed under the LGPL v3+ and the MPL v1.1. Best wishes, Gert van Valkenhoef ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [EasyHack] #44681 port to CLucene from java/Lucene
Great, thanks. That looks like it fills in most of the missing pieces from my patch. On 15-02-12, Caolán McNamara wrote: > On Tue, 2012-02-14 at 12:01 +, Caolán McNamara wrote: > > On Tue, 2012-02-14 at 09:45 +0100, G.H.M.Valkenhoef, van wrote: > > > > > > > Yes, I found that java code (the HelpIndexer I refer to). I'll work on > > > a patch to replace the XInvocations of the Java code with calls to my > > > code. > > > > I can try and knock together a skeleton of a conversion of that Java > > component to a C++ component for you to integration the clucene stuff > > into. > > FWIW, here's a suggested skeleton for replacing the java help component, > with this applied going to help and typing something into "search" > should print out on stderr "implement me" and same for the other case of > indexing help of an extension on-demand if it doesn't come with an index > already. > > C. > > ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [EasyHack] #44681 port to CLucene from java/Lucene
On 14-02-12, Andras Timar wrote: > 2012/2/14 Norbert Thiebaud : > > Just a word of encouragement: thanks for working on that, I'm looking > > forward to see the impact on the build with LANG=all :-) > > Me too, and also I wonder, if it fixes > https://bugs.freedesktop.org/show_bug.cgi?id=40665 > > I noticed that CJK-based indexing is only enabled for the Japanese language. Maybe this can be fixed by adding more languages to be CJK-indexed. > > Cheers, > Andras > > ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [EasyHack] #44681 port to CLucene from java/Lucene
Thanks for the answers, responses below. On 13-02-12, Caolán McNamara wrote: > The xmlhelp/source/com/sun/star/help/*.java route is the one that sets > the "bExtensionMode". I think this one is which third party extensions > can insert their help into our help system. The cxxhelp is the > straightforward backend of the "search in help" dialogs. > Yes, I found that java code (the HelpIndexer I refer to). I'll work on a patch to replace the XInvocations of the Java code with calls to my code. > > couldn't the ZIP creation just always be replaced by this alternative > > code path? > > hmph, indeed, seems that way on the face of it. Lets try that. > Great! I'll work on that too. > > * This implementation is using the master branch of CLucene's git, > > with clucene-contribs-lib enabled (for CJK support). The released > > version of CLucene is compatible with Lucene 1.9.x, whereas LibreOffice > > uses Lucene 2.3. > > I don't *think* compatibility between java and c++ file formats matters > to us, if that's what you're getting at here. > Ok, that's good to know. I'm not sure if the 1.9.x-compatible version has all the required functionality (but I just sent another message to the list about this). > > * I'm not sure exactly how to make my code build as part of the LO > > build, but could probably figure it out as long as the previous point is > > addressed. > > Presumably just editing l10ntools/source/help/makefile.mk and adding > another target or so in there will do the trick. I can hook this up and > see if how it goes. > Great, send me a patch if you get it going, then I can work on some of the other stuff. > > * CLucene (like Java) uses wide characters throughout, and defines > > it's own TCHAR type for that. Can we make this play nice with how LO > > handles strings? > > > > * I'm using some Unix headers, are these available on windows or > > should I use some kind of LO equivalent of them? > > Should be basically cross-platform stuff in sal/inc/ to handle any of > that stuff. > Ok, will check it out, thanks. > > * I tried replacing the HelpIndexerTool in > > helpcontent2/util/target.pmk, which seems to work fine, except that I'm > > returning an error code when the content/caption directory doesn't exist > > (unlike HelpIndexerTool), which breaks on "shared". > > I'll see if I can hook up what you got to our build system, ignoring the > lack of clucene in our tree and assuming availablility of system > clucene, and see how that goes. > I've got an update on this: I managed to create all the indexes and doing a few searches on both the Java-generated an the C++-generated indexes seems to give identical results (at least if I pipe the results through sort). Gert ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [EasyHack] #44681 port to CLucene from java/Lucene
On 13-02-12, Rene Engelhard wrote: > Hi, > > first of all: thanks for this effort, which should have been done when this > lucene dependency was introduced in the first place in OOo times. > > But... > > On Mon, Feb 13, 2012 at 04:17:49PM +0100, Radek Doulik wrote: > > > * This implementation is using the master branch of CLucene's git, > > > with clucene-contribs-lib enabled (for CJK support). The released > > > version of CLucene is compatible with Lucene 1.9.x, whereas LibreOffice > > > uses Lucene 2.3. > > This is bad. Will that get somewhen released? And is the clucene-contribs-lib > included? (And if separate, how hard is to enable it? Patching it into > "proper" > clucene is a no-go.) > > As Radek says, we should (if it was only me: must) support building > against "standard" libclucene. And relying on a git snapshot is bad... > I can't guess at if or when there will be a clucene release. However the released version is ancient, and the developers themselves say the git version is stable and should be used. On the other hand, I don't think distributions carry it. I could look into using the released 1.9.x-compatible version instead, but I expect there was a reason for using Lucene 2.3.x. There is a tag for clucene-src-2.3.3.4 for which there are also tarballs on the sourceforge page. We could depend on that version -- it is a small dependency and takes under a minute to build. Regarding the contribs: it is part of their git repository, and can be enabled as part of the standard build. > > > * Can someone help to figure out how to make CLucene part of the LO > > > build process? CLucene is using CMake and there seems to be no way to > > > 'make install' the clucene-contribs-lib, so this might be tricky. > > > > This usually done like this, you either use system libraries if > > available or build the package (CLucene in this case) inside LO build > > tree. Look into configure.in, search for cairo for example. Cairo is > > graphic library where we link against system one or build one inside LO. > > Giving Cc to _rene_ and pmladek who know a lot about build process. > > Exactly. > > But he didn't post *any* makefile, so trying to write a configure check > is moot right now anyways ;-) > I'm planning to work on that later today. (But also see the post by Caolan, which you may have missed.) > Regards, > > Rene > > ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice