LibreOffice is hard to search for on Chinese Desktops

[Kent Baxley]

Summary:

If a Chinese end user wants to find the office application (i.e. Libre Office), it should be easy for them to type in "office" in both English and Chinese and see the resultant applications. On an English desktop, this works. On a Chinese-translated desktop, this doesn't work unless the user explicitly types "libre" or "libreoffice" in the search bar.

Steps to reproduce:
1) Install 11.10.
2) Make Simplified Chinese the default language.
3) Boot into either unity 2d or unity.
4) Click the lens button and type "office" in either English or Chinese.

Actual Results: Nothing is returned if "office" is typed into the search bar.
Expected Results: Users should see a list of office applications in the search results like they would for an English desktop.

I will also attach screenshots of what a Chinese user sees when attempting to type 'office' into the search bar.

I'm not sure if this is a Unity issue or a translation problem with LibreOffice (or both). Please forward to the appropriate teams if this is not a Unity problem. Thanks.

ProblemType: Bug
DistroRelease: Ubuntu 11.10
Package: unity 4.24.0-0ubuntu2b1
ProcVersionSignature: Ubuntu 3.0.0-13.22somerville1-generic 3.0.6
Uname: Linux 3.0.0-13-generic x86_64
NonfreeKernelModules: nvidia
ApportVersion: 1.23-0ubuntu4
Architecture: amd64
CompizPlugins: [core,bailer,detection,composite,opengl,decor,regex,vpswitch,gnomecompat,imgpng,compiztoolbox,move,place,mousepoll,grid,unitymtgrabhandles,resize,snap,animation,session,expo,workarounds,wall,fade,ezoom,scale,unityshell]
Date: Tue Nov 22 23:25:08 2011
DistributionChannelDescriptor:
 # This is a distribution channel descriptor
 # For more information see http://wiki.ubuntu.com/DistributionChannelDescriptor
 canonical-oem-somerville-oneiric-amd64-20111116-1
DistroCodename: oneiric
DistroVariant: ubuntu
InstallationMedia: Ubuntu 11.10 "Oneiric" - Build amd64 LIVE Binary 20111116-18:24
ProcEnviron:
 LANGUAGE=zh_CN:
 PATH=(custom, no user)
 LANG=zh_CN.UTF-8
 SHELL=/bin/bash
SourcePackage: unity
UpgradeStatus: No upgrade log present (probably fresh install)

[Kent Baxley (kentb)]

Searching for 'office' in Chinese

[Kent Baxley (kentb)]

Searching for office in English on a Chinese desktop

[Sebastien Bacher (seb128)]

Thank you for your bug report, I didn't try with a chinese locale so it could be a cjk issue but the dash doesn't do substring matching, i.e it will not match "office" in "libreoffice", it does search in the name and description keys though (but only in the locale, not in english):

The libreoffice-startcenter has that comment string:
Comment[zh_CN]=兼容开放标准 ODF 的办公软件套装。由 The Document Foundation 支持。

I don't speak chinese but the glyphs seems different from the one you used

The libreoffice-writer.desktop uses:

Comment[zh_CN]=创建编辑文档、报告、信件和网页中的文本和图形。

Does any of those has the word you looked for?

[Sebastien Bacher (seb128)]

bug #752259 is the request for substring matching in the dash

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Kent Baxley (kentb)]

Would this be something that can be SRU'ed for Unity in 11.10? Thanks.

[Sebastien Bacher (seb128)]

Kent, could you reply to my question? It doesn't seem the bug report has currently enough informations to be qualified as a bug in unity, it could be simply that the word you are looking for is not used in the libreoffice description in which case the description could need to be updated

[Kent Baxley (kentb)]

Sebastien,

oops...Sorry for the delay. I'm waiting to hear back from my user in China that reported the problem. I'll update this issue as soon as he replies.

[Kent Baxley (kentb)]

Sebastien,

Here is what I got from my user in China regarding your question:

For what I type in Chinese is "办公软件" is same as the substring description in "兼容开放标准 ODF 的''''''办公软件''''''套装。由 The Document Foundation 支持" .

I tried several English descriptions, such like "Document" can get the result. Also, I tried several other words in Chinese included in comment string, some of them can show the result, but some can't. It's hard for me to identify which kind of string should be input.

[Sebastien Bacher (seb128)]

Kent, I don't speak chinese but since it doesn't do substring matching your example would work:

Comment[zh_CN]=兼容开放标准 ODF 的办公软件套装。由 The Document Foundation 支持。

"的办公软件套装 doesn't start with "办公软件" so that's probably why typing that doesn't work

Does typing "兼容开" or "的办公" or "ODF" work?

[Kent Baxley (kentb)]

Here's what my Chinese user told me:

Typing "兼容开" or "的办公" or "ODF" individually works, but, even so these strings are no meaning for Chinese.

..I believe what he's saying is that those do work, but, it's not something that would mean anything to a Chinese user because they would most likely not use those when trying to search for LibreOffice.

[Sebastien Bacher (seb128)]

Does it mean that the libreoffice description doesn't make sense and should be updated then?

[Kent Baxley (kentb)]

Hi Sebastien,

I think updating the libreoffice description might be worth a look. If some substrings are working, yet, the ones that would make most sense don't, then maybe the description needs to be looked at.

[Kent Baxley (kentb)]

Hi Sebastien,

Did you want this bug moved to libreoffice or do you think there are other issues with unity that also need to be looked at? Thanks.

[Sebastien Bacher (seb128)]

Not sure, I will let some of the unity hackers comment on it but it seems the issue is rather in the description, that would need confirmation on what is wrong in the description from a chinese user though

[Anthony Wong (anthonywong)]

Hi Sebastien,

Modifying libreoffice description can only fix the problem of searching libreoffice, but not all the other applications. The root cause is in how unity lens searching works. The libreoffice description in Chinese is actually fine, so modifying it such that unity can search it is just a hack. Asking users to search "兼容开" or "的办公" is like asking English users to search by "suite compatible", which is not a complete phrase and rather meaningless.

All in all, I think CJK support needs substring/tokenization, and likely it's covered by bug 822743. Anyone knows the progress?

[Anthony Wong (anthonywong)]

Having checked Precise and confirmed that this bug is no longer there.