Revision: 6126
http://languagetool.svn.sourceforge.net/languagetool/?rev=6126&view=rev
Author: dominikoeo
Date: 2012-01-01 09:40:24 +0000 (Sun, 01 Jan 2012)
Log Message:
-----------
[br] - detect *daou gementi?\195?\177* ?\226?\134?\146 daougementi?\195?\177
- fix suggestion when detecting missing soft mutation
of words that start in goui.* or goue.*:
*Dirolla?\195?\177 da gouela?\195?\177* ?\226?\134?\146
"Dirolla?\195?\177 da ouela?\195?\177"
(c?\226?\128?\153houela?\195?\177 was suggested instead)
- detect more words with missing dash such as:
*dreist holl* ?\226?\134?\146 "dreist-holl" (etc.)
Modified Paths:
--------------
trunk/JLanguageTool/src/rules/br/grammar.xml
Modified: trunk/JLanguageTool/src/rules/br/grammar.xml
===================================================================
--- trunk/JLanguageTool/src/rules/br/grammar.xml 2011-12-31 15:47:50 UTC
(rev 6125)
+++ trunk/JLanguageTool/src/rules/br/grammar.xml 2012-01-01 09:40:24 UTC
(rev 6126)
@@ -86,8 +86,7 @@
<rule>
<pattern mark_from="1">
<token regexp="yes">[au]r</token>
- <token regexp="yes" postag="N f s.*|N m p t.*"
postag_regexp="yes">g[^w].*
- <exception regexp="yes" postag="N f s.*|N m p t.*"
postag_regexp="yes">gou[ei].*</exception>
+ <token regexp="yes" postag="N f s.*|N m p t.*"
postag_regexp="yes">g(?!ou[ei])[^w].*
<exception regexp="yes" inflected="yes">k.*</exception>
<exception
regexp="yes">(gar|gavotenn|greg)(ig)?(-(mañ|se|hont))?</exception>
<exception postag="N m s.*" postag_regexp="yes"/>
@@ -280,7 +279,7 @@
<exception postag="P|L e" postag_regexp="yes">e</exception>
</token>
<and>
- <token regexp="yes" inflected="yes">g[^w].*
+ <token regexp="yes" inflected="yes">g(?!ou[ie])[^w].*
<exception inflected="yes" regexp="yes">k.*</exception>
<exception postag="UNKNOWN"/>
</token>
@@ -300,6 +299,23 @@
<exception postag="P|L e" postag_regexp="yes">e</exception>
</token>
<and>
+ <token regexp="yes" inflected="yes">gou[ie].*
+ <exception inflected="yes" regexp="yes">k.*</exception>
+ <exception postag="UNKNOWN"/>
+ </token>
+ <token regexp="yes">[gk].*</token>
+ </and>
+ </pattern>
+ <message>Ur c’hemmadur dre vlotaat a zlefe bezañ goude ar ger «\1». Ha
fellout a rae deoc’h skrivañ <suggestion><match no="2" regexp_match=".(.*)"
regexp_replace="$1"/></suggestion>?</message>
+ <example type="incorrect">Dirollañ da
<marker>gouelañ</marker></example>
+ <example type="correct">Dirollañ da <marker>ouelañ</marker></example>
+ </rule>
+ <rule>
+ <pattern mark_from="1">
+ <token
regexp="yes">2|[ae]|aba|da|daou|div|dre|eme|endra|gwall|ne|pa|re|seul
+ <exception postag="P|L e" postag_regexp="yes">e</exception>
+ </token>
+ <and>
<token regexp="yes" inflected="yes" postag="UNKNOWN"
negate_pos="yes">d.*
<exception inflected="yes">da</exception>
<exception regexp="yes" inflected="yes">t.*</exception>
@@ -492,7 +508,7 @@
<token>en</token>
<token regexp="yes">em|ur<exception postag="D e
sp">ur</exception></token>
<and>
- <token postag="V.*" postag_regexp="yes" regexp="yes"
inflected="yes">g[^w].*</token>
+ <token postag="V.*" postag_regexp="yes" regexp="yes"
inflected="yes">g(?!ou[ei])[^w].*</token>
<token postag="V.*" postag_regexp="yes" regexp="yes">[gk].*</token>
</and>
</pattern>
@@ -506,6 +522,19 @@
<token>en</token>
<token regexp="yes">em|ur<exception postag="D e
sp">ur</exception></token>
<and>
+ <token postag="V.*" postag_regexp="yes" regexp="yes"
inflected="yes">gou[ei].*</token>
+ <token postag="V.*" postag_regexp="yes" regexp="yes">[gk].*</token>
+ </and>
+ </pattern>
+ <message>Ur c’hemmadur dre vlotaat a zlefe bezañ goude «\1 \2». Ha
fellout a rae deoc’h skrivañ <suggestion><match no="3" regexp_match=".(.*)"
regexp_replace="$1"/></suggestion>?</message>
+ <example type="incorrect">En ur <marker>gouestlañ</marker></example>
+ <example type="correct">En ur ouestlañ</example>
+ </rule>
+ <rule>
+ <pattern mark_from="2">
+ <token>en</token>
+ <token regexp="yes">em|ur<exception postag="D e
sp">ur</exception></token>
+ <and>
<token postag="V.*" postag_regexp="yes" regexp="yes"
inflected="yes">d.*</token>
<token postag="V.*" postag_regexp="yes" regexp="yes">[dt].*</token>
</and>
@@ -2157,7 +2186,7 @@
<example type="correct">Echu gant se.</example>
<example type="correct">Ne c’hallan ket ober se.</example>
</rule>
- <rule id="SETU_ME" name="setu me → setu-me">
+ <rule id="SETU_ME" name="setu me → setu-me (hag all)">
<pattern>
<token>setu</token>
<token regexp="yes">me|te|eñ|hi|ni|c’hwi|int</token>
@@ -2166,7 +2195,7 @@
<example type="incorrect"><marker>Setu me</marker> o komz
brezhoneg.</example>
<example type="correct">Setu-me o komz brezhoneg.</example>
</rule>
- <rulegroup id="GANIN_ME" name="ganin me → ganin-me">
+ <rulegroup id="GANIN_ME" name="ganin me → ganin-me (hag all)">
<rule>
<pattern>
<token regexp="yes">ganin|evidon</token>
@@ -2204,6 +2233,51 @@
<example type="correct"><marker>Ganeoc’h-c’hwi</marker></example>
</rule>
</rulegroup>
+ <rule id="BARRENNIG_STAGAN" name="barrennig stagañ → barrennig-stagañ">
+ <pattern>
+ <token regexp="yes">[bvp]arrennig</token>
+ <token>stagañ</token>
+ </pattern>
+ <message>Implijit <suggestion>\1-\2</suggestion>.</message>
+ <example type="incorrect">Ar <marker>varrennig stagañ</marker></example>
+ <example type="correct">Ar <marker>varrennig-stagañ</marker></example>
+ </rule>
+ <rule id="DREIST_HOLL" name="dreist holl → dreist-holl">
+ <pattern>
+ <token>dreist</token>
+ <token regexp="yes">holl|kemm|penn</token>
+ </pattern>
+ <message>Implijit <suggestion>\1-\2</suggestion>.</message>
+ <example type="incorrect">Kalz tud a oa, ha <marker>dreist holl</marker>
kalz a douristed.</example>
+ <example type="correct">Kalz tud a oa, ha dreist-holl kalz a
douristed.</example>
+ </rule>
+ <rule id="ASKEL_GROCHEN" name="askell groc’hen → askell-groc’hen">
+ <pattern>
+ <token regexp="yes">[ae]skell</token>
+ <token regexp="yes">[gk]roc’hen</token>
+ </pattern>
+ <message>Implijit <suggestion>\1-\2</suggestion>.</message>
+ <example type="incorrect">Un <marker>askell groc’hen</marker></example>
+ <example type="correct">Un askell-groc’hen</example>
+ </rule>
+ <rule id="HENT_HOUARN" name="hent houarn → hent-houarn (hag all)">
+ <pattern>
+ <token regexp="yes">hent(où)?</token>
+ <token
regexp="yes">bleiz|bras|dall|don|gwenn|houarn|karr|kirri|kroaz|meur|sant-(Jakez|Jalm)|treuz|tro|troad</token>
+ </pattern>
+ <message>Implijit <suggestion>\1-\2</suggestion>.</message>
+ <example type="incorrect">Labourat a ra en <marker>hent
houarn</marker></example>
+ <example type="correct">Labourat a ra en
<marker>hent-houarn</marker></example>
+ </rule>
+ <rule id="TOUR_TAN" name="tour tan → tour-tan">
+ <pattern>
+ <token regexp="yes">tour(ioù)?</token>
+ <token regexp="yes">dour|iliz|meur|tan</token>
+ </pattern>
+ <message>Implijit <suggestion>\1-\2</suggestion>.</message>
+ <example type="incorrect">Uhel eo an <marker>tour tan</marker>.</example>
+ <example type="correct">Uhel eo an <marker>tour-tan</marker>.</example>
+ </rule>
<rule id="BLEIZ_MOR" name="bleiz mor → bleiz-mor">
<pattern>
<token regexp="yes">bleizi?</token>
@@ -2222,6 +2296,33 @@
<example type="incorrect">Ur <marker>grilh mor</marker> am eus
tapet.</example>
<example type="correct">Ur grilh-mor am eus tapet.</example>
</rule>
+ <rule id="ANV_BIHAN" name="anv bihan → anv-bihan (hag all)">
+ <pattern>
+ <token regexp="yes">anv(ioù)?</token>
+ <token
regexp="yes">badez|bihan|familh|gwan|kadarn|pluenn|stroll|stiegezh|verb</token>
+ </pattern>
+ <message>Implijit <suggestion>\1-\2</suggestion>.</message>
+ <example type="incorrect">Petra zo da <marker>anv
bihan</marker>?</example>
+ <example type="correct">Petra zo da <marker>anv-bihan</marker>?</example>
+ </rule>
+ <rule id="HOCH_GOUEZ" name="hoc’h gouez → hoc’h-gouez (hag all)">
+ <pattern>
+ <token regexp="yes">hoc’h(ed)?|moc’h|pemoc’h|porc’hell|tourc’h</token>
+ <token>gouez</token>
+ </pattern>
+ <message>Implijit <suggestion>\1-\2</suggestion>.</message>
+ <example type="incorrect">An <marker>hoc’h gouez</marker></example>
+ <example type="correct">An <marker>hoc’h-gouez</marker></example>
+ </rule>
+ <rule id="GER_MELL" name="ger mell → ger-mell (hag all)">
+ <pattern>
+ <token regexp="yes">ger(ioù)?</token>
+ <token regexp="yes">mell|rener|stur|tremen</token>
+ </pattern>
+ <message>Implijit <suggestion>\1-\2</suggestion>.</message>
+ <example type="incorrect">Ar <marker>ger mell</marker>.</example>
+ <example type="correct">Ar <marker>ger-mell</marker>.</example>
+ </rule>
<rule id="EVEL_JUST" name="evel just → evel-just">
<pattern>
<token>evel</token>
@@ -2276,6 +2377,15 @@
<example type="incorrect">Ho pilhed <marker>karr nij</marker> mar
plij.</example>
<example type="correct">Ho pilhed karr-nij mar plij.</example>
</rule>
+ <rule id="PLANKENN_RUILH" name="plankenn ruiklh → plankenn-ruilh">
+ <pattern>
+ <token regexp="yes">[pbf]lankenn(où)?</token>
+ <token regexp="yes">ruilh|skoaz|dre-lien</token>
+ </pattern>
+ <message>Implijit <suggestion>\1-\2</suggestion>.</message>
+ <example type="incorrect">Ur <marker>plankenn ruilh</marker></example>
+ <example type="correct">Ur <marker>plankenn-ruilh</marker></example>
+ </rule>
<rule id="MENEZ_TAN" name="menez tan → menez-tan">
<pattern>
<token regexp="yes">menez(ioù)?</token>
@@ -2334,10 +2444,19 @@
<example type="correct">kemer perzh enno e meur a du.</example>
<example type="correct">Meur a zoare a zo da gemer perzh</example>
</rule>
+ <rule id="MA_UNAN" name="ma unan → ma-unan (hag all)">
+ <pattern>
+ <token regexp="yes">ma|da|e|hec’h|hon|hoc’h|o</token>
+ <token regexp="yes">unan(ig)?</token>
+ </pattern>
+ <message>Implijit <suggestion>\1-\2</suggestion>.</message>
+ <example type="incorrect">Desket em eus <marker>ma
unan</marker>.</example>
+ <example type="correct">Desket em eus <marker>ma-unan</marker>.</example>
+ </rule>
<rule id="E_KREIZ" name="e kreiz → e-kreiz (hag all)">
<pattern>
<token>e</token>
- <token
regexp="yes">barzh|douare|giz(-(all|mañ|se|hon))?|harz|hunan|keit(-se)?|kerz|kichen|krec’h|kreiz|leizh|maez|pad|sell|ser|tailh|tal|touez|tro|unan</token>
+ <token
regexp="yes">barzh|douare|giz(-(all|mañ|se|hon))?|harz|hunan|keit(-se)?|kerz|kichen|krec’h|kreiz|leizh|maez|pad|sell|ser|tailh|tal|touez|tro</token>
</pattern>
<message>Implijit <suggestion>\1-\2</suggestion>.</message>
<example type="incorrect"><marker>E kichen</marker> Gwitalmeze emañ o
chom.</example>
@@ -2494,6 +2613,34 @@
<example type="incorrect"><marker>Serr prim</marker> ma chupenn</example>
<example type="correct"><marker>Serr-prim</marker> ma chupenn</example>
</rule>
+ <rulegroup id="MARCHAD_MAD" name="marc’had mat → marc’had-mat">
+ <rule>
+ <pattern>
+ <token regexp="yes">[mv]arc’had</token>
+ <token>mat</token>
+ </pattern>
+ <message>Implijit <suggestion>\1-\2</suggestion>.</message>
+ <example type="incorrect"><marker>Marc’had mat</marker> eo.</example>
+ <example type="correct"><marker>Marc’had-mat</marker> eo.</example>
+ </rule>
+ <rule>
+ <pattern>
+ <token regexp="yes">[mv]arc’had</token>
+ <token regexp="yes">mat(oc’h|añ)</token>
+ </pattern>
+ <message>Implijit <suggestion>\1\2</suggestion>.</message>
+ <example type="incorrect"><marker>Marc’had matoc’h</marker>
eo.</example>
+ <example type="correct"><marker>Marc’hadmatoc’h</marker> eo.</example>
+ </rule>
+ <rule>
+ <pattern>
+ <token regexp="yes">[mv]arc’had-mat(oc’h|añ)</token>
+ </pattern>
+ <message>Implijit <suggestion><match no="1" regexp_match="-"
regexp_replace=""/></suggestion>.</message>
+ <example type="incorrect"><marker>Marc’had-matoc’h</marker>
eo.</example>
+ <example type="correct"><marker>Marc’hadmatoc’h</marker> eo.</example>
+ </rule>
+ </rulegroup>
<rule id="MARCH_HOUARN" name="marc’h houarn → marc’h-houarn">
<pattern>
<token regexp="yes">[vm]arc’h</token>
@@ -3028,7 +3175,7 @@
<rule id="BRO" name="Broioù">
<pattern>
<token regexp="yes">[BVP]ro</token>
- <token
regexp="yes">Alamagn|Amerika|C’hall|Dreger|Gembre|Gerne|Iwerzhon|Japan|Leon|Naoned|Raez|Roazhon|Saoz|Sant-Brieg|Sant-Maloù|Skos|Suis|Ulad|Wened</token>
+ <token
regexp="yes">Alamagn|Amerika|C’hall|Dreger|Friz|Gembre|Gerne|Iwerzhon|Japan|Leon|Naoned|Raez|Roazhon|Saoz|Sant-Brieg|Sant-Maloù|Skos|Suis|Ulad|Wened</token>
</pattern>
<message>Implijit <suggestion><match no="1" case_conversion
="startupper"/>-<match no="2" case_conversion
="startupper"/></suggestion>.</message>
<example type="incorrect">Labour a ra e <marker>Bro
Amerika</marker>.</example>
@@ -3291,7 +3438,7 @@
<rule>
<pattern mark_from="1">
<token inflected="yes"
regexp="yes">(abadenn|c’hoariva|film|kantik|levr|embannadur|kanaouenn|keloù|krennlavar|musik|pladenn|romant?|sonerezh|skrid)(ig)?</token>
- <token
regexp="yes">alamaneg|albaneg|arabeg|belaruseg|brezhoneg|daneg|esperanteg|euskareg|finneg|frizeg|galizeg|gallaoueg|galleg|islandeg|italianeg|iwerzhoneg|katalaneg|kembraeg|kerneveureg|korseg|lojbaneg|manaveg|nederlandeg|norveveg|okitaneg|poloneg|portugaleg|roumaneg|ruseg|rusianeg|saozneg|sardeg|skoseg|skoteg|skovakeg|sorabeg|spagnoleg|svedeg|tchekeg|turkeg|ukraineg</token>
+ <token
regexp="yes">alamaneg|albaneg|arabeg|belaruseg|brezhoneg|daneg|esperanteg|euskareg|finneg|frizeg|galizeg|gallaoueg|galleg|gouezeleg|islandeg|italianeg|iwerzhoneg|katalaneg|kembraeg|kerneveureg|korseg|lojbaneg|manaveg|nederlandeg|norveveg|okitaneg|poloneg|portugaleg|roumaneg|ruseg|rusianeg|saozneg|sardeg|skoseg|skoteg|skovakeg|sorabeg|spagnoleg|svedeg|tchekeg|turkeg|ukraineg</token>
</pattern>
<message>Ha fellout a rae deoc’h skrivañ <suggestion><match no="2"
regexp_match="(.*)g" regexp_replace="$1k"/></suggestion>?</message>
<example type="incorrect">Ul levr <marker>brezhoneg</marker>
eo.</example>
@@ -3300,7 +3447,7 @@
<rule>
<pattern mark_from="1">
<token regexp="yes">a[nr]</token>
- <token
regexp="yes">alamanek|albanek|arabek|belarusek|brezhonek|danek|esperantek|euskarek|finnek|frizek|galizek|gallaouek|gallek|islandek|italianek|iwerzhonek|c’hatalanek|c’hembraek|c’herneveurek|c’horsek|lojbanek|manavek|nederlandek|norvevek|okitanek|polonek|portugalek|roumanek|rusek|rusianek|saoznek|sardek|skosek|skotek|slovakek|sorabek|spagnolek|svedek|tchekek|turkek|ukrainek</token>
+ <token
regexp="yes">alamanek|albanek|arabek|belarusek|brezhonek|danek|esperantek|euskarek|finnek|frizek|galizek|gallaouek|gallek|gouezelek|islandek|italianek|iwerzhonek|c’hatalanek|c’hembraek|c’herneveurek|c’horsek|lojbanek|manavek|nederlandek|norvevek|okitanek|polonek|portugalek|roumanek|rusek|rusianek|saoznek|sardek|skosek|skotek|slovakek|sorabek|spagnolek|svedek|tchekek|turkek|ukrainek</token>
</pattern>
<message>Ha fellout a rae deoc’h skrivañ <suggestion><match no="2"
regexp_match="(.*)k" regexp_replace="$1g"/></suggestion>?</message>
<example type="incorrect">Ar <marker>brezhonek</marker> a zo ur yezh
keltiek.</example>
@@ -4092,11 +4239,11 @@
<rule id="DOUAGEMENTIN" name="daou gementiñ → daougementiñ">
<pattern>
<token regexp="yes">daou|tri|pevar</token>
- <token regexp="yes">gement(iñ|et)</token>
+ <token
regexp="yes">(c’hement|gement)(iñ|et)?|(ilin|gromm)(añ|et)?|roadek|bik|lamm(et)?|dortet</token>
</pattern>
<message>Ha fellout a rae deoc’h skrivañ
<suggestion>\1\2</suggestion>?</message>
- <example type="incorrect">Ar pal zo <marker>daou gementiñ</marker> an
niver a dud o teskiñ brezhoneg.</example>
- <example type="correct">Ar pal zo <marker>daougementiñ</marker> an niver
a dud o teskiñ brezhoneg.</example>
+ <example type="incorrect">Ar pal zo <marker>daou c’hementiñ</marker> an
niver a dud o teskiñ brezhoneg.</example>
+ <example type="correct">Ar pal zo <marker>daouc’hementiñ</marker> an
niver a dud o teskiñ brezhoneg.</example>
</rule>
</category>
This was sent by the SourceForge.net collaborative development platform, the
world's largest Open Source development site.
------------------------------------------------------------------------------
Ridiculously easy VDI. With Citrix VDI-in-a-Box, you don't need a complex
infrastructure or vast IT resources to deliver seamless, secure access to
virtual desktops. With this all-in-one solution, easily deploy virtual
desktops for less than the cost of PCs and save 60% on VDI infrastructure
costs. Try it free! http://p.sf.net/sfu/Citrix-VDIinabox
_______________________________________________
Languagetool-cvs mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/languagetool-cvs
