Revision: 8300
http://languagetool.svn.sourceforge.net/languagetool/?rev=8300&view=rev
Author: jaumeortola
Date: 2012-11-04 15:51:51 +0000 (Sun, 04 Nov 2012)
Log Message:
-----------
[ca] Fixed some false alarms.
Modified Paths:
--------------
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/catalan.dict
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/catalan_synth.dict
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/disambiguation.xml
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/multiwords.txt
trunk/JLanguageTool/src/main/resources/org/languagetool/rules/ca/grammar.xml
Modified:
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/catalan.dict
===================================================================
(Binary files differ)
Modified:
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/catalan_synth.dict
===================================================================
(Binary files differ)
Modified:
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/disambiguation.xml
===================================================================
---
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/disambiguation.xml
2012-11-04 15:41:10 UTC (rev 8299)
+++
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/disambiguation.xml
2012-11-04 15:51:51 UTC (rev 8300)
@@ -22,18 +22,40 @@
<token postag="(PX.|N.|[ADP].{2}|V.{5})[FC].*|RG.*"
postag_regexp="yes"></token>
</equivalence>
</unification>
- <rule id="PUNTUACIO" name="signes de puntuació">
- <pattern>
- <token regexp="yes">[.;:!?¡¿—–…()\[\]]</token>
- </pattern>
- <disambig action="add"><wd pos="PUNCT"></wd></disambig>
- </rule>
- <rule id="PUNTUACIO2" name="signes de puntuació que no trenquen els
sintagmes">
- <pattern>
- <token regexp="yes">[,"‘’“”«»]</token>
- </pattern>
- <disambig action="add"><wd pos="PUNCT_CONT"></wd></disambig>
- </rule>
+ <rulegroup id="PUNTUACIO" name="signes de puntuació">
+ <rule>
+ <pattern>
+ <token regexp="yes">[.;:!?¡¿—–…()\[\]]</token>
+ </pattern>
+ <disambig action="add"><wd pos="PUNCT"></wd></disambig>
+ </rule>
+ <rule id="PUNTUACIO2" name="signes de puntuació que no trenquen els
sintagmes">
+ <pattern>
+ <token regexp="yes">[,"‘’“”«»]</token>
+ </pattern>
+ <disambig action="add"><wd pos="PUNCT_CONT"></wd></disambig>
+ </rule>
+ <rule id="coma_decimal" name="coma decimal">
+ <pattern>
+ <token regexp="yes">\d+</token>
+ <marker>
+ <token spacebefore="no">,</token>
+ </marker>
+ <token spacebefore="no" regexp="yes">\d+</token>
+ </pattern>
+ <disambig action="remove"><wd pos="PUNCT_CONT"></wd></disambig>
+ </rule>
+ <rule id="coma_decimal2" name="coma decimal">
+ <pattern>
+ <token regexp="yes">\d+</token>
+ <marker>
+ <token spacebefore="no">,</token>
+ </marker>
+ <token spacebefore="no" regexp="yes">\d+</token>
+ </pattern>
+ <disambig action="add"><wd pos="decimal_comma"></wd></disambig>
+ </rule>
+ </rulegroup>
<rule id="no_noms_error_diccionari" name="no haurien de ser noms, error
del diccionari">
<pattern>
<token regexp="yes">empaquetat|emmotll?at</token>
@@ -529,7 +551,7 @@
<token>entre</token>
</marker>
</pattern>
- <disambig action="filter" postag="SPS000"/>
+ <disambig action="filter" postag="SPS00"/>
</rule>
<!-- <rule>
<pattern>
@@ -555,7 +577,7 @@
<pattern>
<token><exception
postag="NP..SP.*|P0.{6}|PP3CN000|PP3NN000|PP3..A00|PP3CP000|PP3CSD00"
postag_regexp="yes" regexp="yes">[^'-].+</exception></token>
<marker>
- <token postag="V.I[PI]3S0*" postag_regexp="yes"
inflected="yes" skip="12">fer<exception scope="next"
regexp="yes">,|que|en</exception><exception postag="V.*" postag_regexp="yes"
scope="next"/></token>
+ <token postag="V.I[PI]3S0*" postag_regexp="yes"
inflected="yes" skip="12">fer<exception scope="next"
regexp="yes">que|en</exception><exception scope="next" postag="PUNCT.*"
postag_regexp="yes">,</exception><exception postag="V.*" postag_regexp="yes"
scope="next"/></token>
</marker>
<token
regexp="yes">segons?|minuts?|hor(a|es)|di(a|es)|setman(a|es)|mes|mesos|(tri|quadri|se)mestres?|anys?|lustres?|dècad(a|es)|segles?|mil·lennis?|moments?|instants?|temps</token>
</pattern>
@@ -699,7 +721,19 @@
</marker>
</pattern>
<disambig action="add"><wd pos="NPCN000"></wd></disambig>
- </rule>
+ </rule>
+ <rule id="nom_persona_o_geogràfic" name="nom de persona o geogràfic: a
l'Àfrica?">
+ <pattern>
+ <token skip="1">a</token>
+ <marker>
+ <and>
+ <token postag="NP..G00" postag_regexp="yes"/>
+ <token postag="NP..SP0" postag_regexp="yes"/>
+ </and>
+ </marker>
+ </pattern>
+ <disambig action="filter" postag="NP..G00" />
+ </rule>
</rulegroup>
<!--crea etiquetes per als adverbis acabats en -ment-->
<!-- <rulegroup id="adverbis_ment" name="marca adverbis acabats en ment">
@@ -5635,15 +5669,13 @@
</marker>
</pattern>
<disambig action="filter"
postag="_GN_.*|A.*|N.*|D.*|V.P.*|PX.*"></disambig>
- </rule>
-
-
+ </rule>
</rulegroup>
<rule id="VERB_PREPOSICIO" name="verb + preposició">
<!-- mogut per "haver acabat"; excepció: per tal d'aconseguir els seus
fins; per a aconseguir fins dignes d'aconseguir-se-->
<!-- mogut per "aconseguir fins dignes";-->
<pattern>
- <token postag="V.[^P].*" postag_regexp="yes"><exception
postag="N.*|_GN_.*" postag_regexp="yes"/></token>
+ <token postag="V.[SING].*|_GV_" postag_regexp="yes"><exception
postag="N.*|_GN_.*|RG.*" postag_regexp="yes"/></token>
<marker>
<token postag="SPS00" postag_regexp="yes"><exception
postag="_GN_.*|PI.*|DI.*" postag_regexp="yes"/></token>
</marker>
@@ -6389,14 +6421,14 @@
<pattern>
<token postag="_GV_"/>
<marker>
- <and>
- <token postag="_GV_"/>
- <token postag="_GN_.*" postag_regexp="yes"/>
- </and>
+ <and>
+ <token postag="_GV_"/>
+ <token postag="_GN_.*" postag_regexp="yes"/>
+ </and>
</marker>
<token postag="_GN_.*" postag_regexp="yes"/>
<and>
- <token postag="NC.*" postag_regexp="yes"/>
+ <token postag="NC.*" postag_regexp="yes"><exception
regexp="yes">cop|vegada|volta</exception></token>
<token postag="_GN_.*" postag_regexp="yes"/>
</and>
</pattern>
@@ -6847,7 +6879,7 @@
<pattern>
<token><exception
postag="NP..SP.*|P0.{6}|PP3CN000|PP3NN000|PP3..A00|PP3CP000|PP3CSD00"
postag_regexp="yes" regexp="yes">[^'-].+</exception></token>
<marker>
- <token postag="V.I[PI]3S0*" postag_regexp="yes"
inflected="yes" skip="12">fer<exception scope="next"
regexp="yes">,|que|en</exception><exception postag="V.*" postag_regexp="yes"
scope="next"/></token>
+ <token postag="V.I[PI]3S0*" postag_regexp="yes"
inflected="yes" skip="12">fer<exception scope="next"
regexp="yes">que|en</exception><exception scope="next" postag="PUNCT.*"
postag_regexp="yes">,</exception><exception postag="V.*" postag_regexp="yes"
scope="next"/></token>
</marker>
<token
regexp="yes">segons?|minuts?|hor(a|es)|di(a|es)|setman(a|es)|mes|mesos|(tri|quadri|se)mestres?|anys?|lustres?|dècad(a|es)|segles?|mil·lennis?|moments?|instants?|temps</token>
</pattern>
@@ -7192,6 +7224,28 @@
<disambig action="add"><wd pos="allow_masculine"></wd></disambig>
</rule>
</rulegroup>
+ <rulegroup id="una_vegada" name="locució una vegada/volta/cop">
+ <rule>
+ <pattern>
+ <token>una</token>
+ <marker>
+ <token regexp="yes">vegada|volta</token>
+ </marker>
+ <token postag="_GN_.*|V.[SI].*" postag_regexp="yes"/>
+ </pattern>
+ <disambig action="add"><wd pos="locucio_unavegada"></wd></disambig>
+ </rule>
+ <rule>
+ <pattern>
+ <token>un</token>
+ <marker>
+ <token>cop</token>
+ </marker>
+ <token postag="_GN_.*|V.[SI].*" postag_regexp="yes"/>
+ </pattern>
+ <disambig action="add"><wd pos="locucio_unavegada"></wd></disambig>
+ </rule>
+ </rulegroup>
<rulegroup id="tant_x_com" name="tant + __ + com + __">
<rule>
<!-- Dificultats: és tan(t) important com bonic -->
@@ -7470,7 +7524,7 @@
<rulegroup id="HAC_ASPIRADA" name="hac aspirada inicial">
<rule>
<pattern>
- <token
regexp="yes">heav.*|Hitch.*|Hong|hongk.+|Haifa|herbert.*|hinden.*|hein.*|humbold.*|hardware|Hampton|Husserl|Hill|high|Heiden.*|Hans|Hereford|His|Heitz|h[òo]bbits?|hawk.*|Heisen.*|Hohen.*|Havilland|Higgs|Howard|hilbert.*|hoover.*|harry|hopkins|hearst|hamilton.*|harrods?|handicaps?|harolds?|Helen|hollywood.*|Houston|Heard|hawai.*|Hudson|Harvard|Hannover|Hayworth|Hepburn|Henry|Hubble|Harrison|Hamas|Hamàs|hegel.*|henna|hennes|Hezboll.+|Hirado|hitler.*|h[oò]ldings?|husseinit.+|Haruki|H[ou]f.*|Hartmann?|hall.*</token>
+ <token
regexp="yes">Honolu.+|heav.*|Hitch.*|Hong|hongk.+|Haifa|herbert.*|hinden.*|hein.*|humbold.*|hardware|Hampton|Husserl|Hill|high|Heiden.*|Hans|Hereford|His|Heitz|h[òo]bbits?|hawk.*|Heisen.*|Hohen.*|Havilland|Higgs|Howard|hilbert.*|hoover.*|harry|hopkins|hearst|hamilton.*|harrods?|handicaps?|harolds?|Helen|hollywood.*|Houston|Heard|hawai.*|Hudson|Harvard|Hannover|Hayworth|Hepburn|Henry|Hubble|Harrison|Hamas|Hamàs|hegel.*|henna|hennes|Hezboll.+|Hirado|hitler.*|h[oò]ldings?|husseinit.+|Haruki|H[ou]f.*|Hartmann?|hall.*</token>
</pattern>
<disambig action="add"><wd pos="hac_aspirada"></wd></disambig>
</rule>
Modified:
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/multiwords.txt
===================================================================
---
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/multiwords.txt
2012-11-04 15:41:10 UTC (rev 8299)
+++
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/multiwords.txt
2012-11-04 15:51:51 UTC (rev 8300)
@@ -76,7 +76,10 @@
D' un cap a l' altre LOC_ADV
D' un costat a l' altre LOC_ADV
D' un dia a l' altre LOC_ADV
+D' un en un LOC_ADV
D' un plegat LOC_ADV
+D' una a una LOC_ADV
+D' una en una LOC_ADV
D' una vegada LOC_ADV
D' upa LOC_ADV
D' última hora LOC_ADV
@@ -223,9 +226,11 @@
Tots sols LOC_ADV
Tour de force NCMS000
Ultra mesura LOC_ADV
+Un per un LOC_ADV
Un poc LOC_ADV
Un xic LOC_ADV
Una mica LOC_ADV
+Una per una LOC_ADV
Vist i no vist LOC_ADV
a dies LOC_ADV
a l capdavall LOC_ADV
@@ -304,7 +309,10 @@
d' un cap a l' altre LOC_ADV
d' un costat a l' altre LOC_ADV
d' un dia a l' altre LOC_ADV
+d' un en un LOC_ADV
d' un plegat LOC_ADV
+d' una a una LOC_ADV
+d' una en una LOC_ADV
d' una vegada LOC_ADV
d' upa LOC_ADV
d' última hora LOC_ADV
@@ -450,7 +458,9 @@
tots sols LOC_ADV
tour de force NCMS000
ultra mesura LOC_ADV
+un per un LOC_ADV
un poc LOC_ADV
un xic LOC_ADV
una mica LOC_ADV
+una per una LOC_ADV
vist i no vist LOC_ADV
Modified:
trunk/JLanguageTool/src/main/resources/org/languagetool/rules/ca/grammar.xml
===================================================================
---
trunk/JLanguageTool/src/main/resources/org/languagetool/rules/ca/grammar.xml
2012-11-04 15:41:10 UTC (rev 8299)
+++
trunk/JLanguageTool/src/main/resources/org/languagetool/rules/ca/grammar.xml
2012-11-04 15:51:51 UTC (rev 8300)
@@ -290,12 +290,13 @@
<token>dos</token>
</marker>
<token regexp="yes">i|o|ni|,</token>
- <token postag="_GN_FP"/>
+ <token postag="_GN_FP"><exception
postag="_GN_MP"/></token>
</pattern>
<message>Error de concordança:
<suggestion>dues</suggestion>.</message>
<short>Error de concordança</short>
<example type="incorrect"><marker>dos</marker> o tres
coses</example>
<example type="correct">dos o tres homes</example>
+ <example type="correct">dos o tres mascles</example>
</rule>
</rulegroup>
<rule id="CONCORDANCES_PRONOM_NEUTRE" name="concordances amb pronom
neutre">
@@ -4181,7 +4182,7 @@
<token regexp="yes">presa|preses</token>
</marker>
</pattern>
- <message>¿Volíeu dir <suggestion><match no="1"
regexp_match="pres(.+)" regexp_replace="press$1"/></suggestion>?</message>
+ <message>¿Volíeu dir <suggestion><match no="2"
regexp_match="pres(.+)" regexp_replace="press$1"/>
(=ànsia)</suggestion>?</message>
<short>Possible confusió</short>
<example type="incorrect">sense
<marker>presa</marker></example>
<example type="correct">sense pressa</example>
@@ -5298,7 +5299,7 @@
<rule>
<pattern>
<token>el</token>
- <token regexp="yes">h?[aeiouàèéíòóú].*<exception
regexp="yes">h?[ui][aeioàèéóòu].+</exception><exception
inflected="yes">ordinal</exception><exception
postag="hac_aspirada|mot_foraster|UNKNOWN" postag_regexp="yes"/></token>
+ <token regexp="yes">h?[aeiouàèéíòóú].*<exception
regexp="yes">h?[ui][aeioàèéóòu].+|o|i</exception><exception
inflected="yes">ordinal</exception><exception
postag="hac_aspirada|mot_foraster|UNKNOWN" postag_regexp="yes"/></token>
</pattern>
<message>Cal escriure:<suggestion>l'<match
no="2"></match></suggestion>.</message>
<short>Error ortogràfic</short>
@@ -5307,6 +5308,7 @@
<example type="correct">el Hubble</example>
<example type="correct">el atletismo</example>
<example type="correct">el iugoslau</example>
+ <example type="correct">el o els conqueridors</example>
</rule>
<rule>
<pattern>
@@ -5327,7 +5329,7 @@
<pattern>
<token>de</token>
<token spacebefore="no">l</token>
- <token regexp="yes">h?[aeiouàèéíòóú].*<exception
regexp="yes">h?[ui][aeioàèéóòu].+</exception><exception
inflected="yes">ordinal</exception><exception
postag="hac_aspirada|mot_foraster|UNKNOWN" postag_regexp="yes"/></token>
+ <token regexp="yes">h?[aeiouàèéíòóú].*<exception
regexp="yes">h?[ui][aeioàèéóòu].+|o|i</exception><exception
inflected="yes">ordinal</exception><exception
postag="hac_aspirada|mot_foraster|UNKNOWN" postag_regexp="yes"/></token>
</pattern>
<message>Cal escriure:<suggestion>de l'<match
no="3"></match></suggestion>.</message>
<short>Error ortogràfic</short>
@@ -5341,7 +5343,7 @@
<pattern>
<token>a</token>
<token spacebefore="no">l</token>
- <token regexp="yes">h?[aeiouàèéíòóú].*<exception
regexp="yes">h?[ui][aeioàèéóòu].+</exception><exception
inflected="yes">ordinal</exception><exception
postag="hac_aspirada|mot_foraster|UNKNOWN" postag_regexp="yes"/></token>
+ <token regexp="yes">h?[aeiouàèéíòóú].*<exception
regexp="yes">h?[ui][aeioàèéóòu].+|o|i</exception><exception
inflected="yes">ordinal</exception><exception
postag="hac_aspirada|mot_foraster|UNKNOWN" postag_regexp="yes"/></token>
</pattern>
<message>Cal escriure:<suggestion>a l'<match
no="3"></match></suggestion>.</message>
<short>Error ortogràfic</short>
@@ -5355,7 +5357,7 @@
<pattern>
<token>pe</token>
<token spacebefore="no">l</token>
- <token regexp="yes">h?[aeiouàèéíòóú].*<exception
regexp="yes">h?[ui][aeioàèéóòu].+</exception><exception
inflected="yes">ordinal</exception><exception
postag="hac_aspirada|mot_foraster|UNKNOWN" postag_regexp="yes"/></token>
+ <token regexp="yes">h?[aeiouàèéíòóú].*<exception
regexp="yes">h?[ui][aeioàèéóòu].+|o|i</exception><exception
inflected="yes">ordinal</exception><exception
postag="hac_aspirada|mot_foraster|UNKNOWN" postag_regexp="yes"/></token>
</pattern>
<message>Cal escriure:<suggestion>per l'<match
no="3"></match></suggestion>.</message>
<short>Error ortogràfic</short>
@@ -13769,7 +13771,7 @@
<pattern>
<token postag="SENT_START|CC" postag_regexp="yes"
skip="-1"><exception postag="PR.*|PT.*|DT.*|.*LOC_CONJ.*|CS"
postag_regexp="yes" scope="next"/><exception scope="next"
regexp="yes">perquè|com</exception></token>
<marker>
- <token postag="V.[^NGP].*" postag_regexp="yes"
regexp="yes" skip="-1">[\p{Ll}·]+<exception postag="_GN_.*|N.*|A.*|V.[NGP].*"
postag_regexp="yes"/><exception>di</exception><exception scope="next"
regexp="yes">[:;,.!?\(\)–—\[\]]|com|perquè|que</exception><exception
postag="C.*|PR.*|PT.*|DT.*|.*LOC_CONJ.*" postag_regexp="yes"
scope="next"/></token>
+ <token postag="V.[^NGP].*" postag_regexp="yes"
regexp="yes" skip="-1">[\p{Ll}·]+<exception postag="_GN_.*|N.*|A.*|V.[NGP].*"
postag_regexp="yes"/><exception>di</exception><exception scope="next"
regexp="yes">[:;,.!?\(\)–—\[\]]|com|perquè|que</exception><exception
postag="C.*|PR.*|PT.*|DT.*|.*LOC_CONJ.*|locucio_unavegada" postag_regexp="yes"
scope="next"/></token>
<token postag="V.[^NGP].*" postag_regexp="yes"
regexp="yes">[\p{Ll}·]+<exception postag="_GN_.*|N.*|A.*|V.[NGP].*|.*LOC.*"
postag_regexp="yes"/><exception regexp="yes">di|cap</exception></token>
</marker>
</pattern>
@@ -13793,6 +13795,7 @@
<example type="correct">L'ésser ja no es troba allí. </example>
<example type="correct">Era la feina més dura.</example>
<example type="correct">Ara fa quatre anys em vaig atrevir a
publicar.</example>
+ <example type="correct">Els mascles tenen una tendència a
dominar la caça un cop han atrapat la presa.</example>
</rule>
<rule> <!-- verb forma simple a l'inici de frase (en majúscula) -->
<pattern>
This was sent by the SourceForge.net collaborative development platform, the
world's largest Open Source development site.
------------------------------------------------------------------------------
LogMeIn Central: Instant, anywhere, Remote PC access and management.
Stay in control, update software, and manage PCs from one command center
Diagnose problems and improve visibility into emerging IT issues
Automate, monitor and manage. Do more in less time with Central
http://p.sf.net/sfu/logmein12331_d2d
_______________________________________________
Languagetool-commits mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/languagetool-commits
