Revision: 8054
http://languagetool.svn.sourceforge.net/languagetool/?rev=8054&view=rev
Author: jaumeortola
Date: 2012-09-15 21:40:23 +0000 (Sat, 15 Sep 2012)
Log Message:
-----------
[ca] new rules
Modified Paths:
--------------
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/disambiguation.xml
trunk/JLanguageTool/src/main/resources/org/languagetool/rules/ca/grammar.xml
Modified:
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/disambiguation.xml
===================================================================
---
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/disambiguation.xml
2012-09-15 14:55:51 UTC (rev 8053)
+++
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/disambiguation.xml
2012-09-15 21:40:23 UTC (rev 8054)
@@ -729,7 +729,7 @@
<token postag="VM[IS]P1S0" postag_regexp="yes" />
</marker>
</pattern>
- <disambig action="filter" postag="VM[IS]P1S0" />
+ <disambig action="filter" postag="V.[IS]P1S0" />
</rule>
<rule>
<pattern>
@@ -739,7 +739,7 @@
<token postag="VM[IS]P1S0" postag_regexp="yes" />
</marker>
</pattern>
- <disambig action="filter" postag="VM[IS]P1S0" />
+ <disambig action="filter" postag="V.[IS]P1S0" />
</rule>
<rule>
<pattern>
@@ -750,7 +750,7 @@
<token postag="VM[IS]P1S0" postag_regexp="yes" />
</marker>
</pattern>
- <disambig action="filter" postag="VM[IS]P1S0" />
+ <disambig action="filter" postag="V.[IS]P1S0" />
</rule>
<rule>
<!-- un cop, aquestes tres -->
@@ -2027,9 +2027,7 @@
<token postag="V.N.*" postag_regexp="yes"></token>
</marker>
</pattern>
- <disambig>
- <match no="2" postag_regexp="yes" postag="V.N.*"></match>
- </disambig>
+ <disambig><match no="2" postag_regexp="yes"
postag="V.N.*"></match></disambig>
</rule>
<rule>
<pattern>
@@ -2038,8 +2036,7 @@
</marker>
<token postag="V.N.*" postag_regexp="yes"></token>
</pattern>
- <disambig action="add"><wd pos="_GV_"></wd>
- </disambig>
+ <disambig action="add"><wd pos="_GV_"></wd></disambig>
</rule>
<rule>
<pattern>
Modified:
trunk/JLanguageTool/src/main/resources/org/languagetool/rules/ca/grammar.xml
===================================================================
---
trunk/JLanguageTool/src/main/resources/org/languagetool/rules/ca/grammar.xml
2012-09-15 14:55:51 UTC (rev 8053)
+++
trunk/JLanguageTool/src/main/resources/org/languagetool/rules/ca/grammar.xml
2012-09-15 21:40:23 UTC (rev 8054)
@@ -1026,7 +1026,23 @@
<example type="incorrect"><marker>les principals
ganivets</marker>.</example>
<example type="correct">els principals ganivets.</example>
</rule>
- </rulegroup>
+ </rulegroup>
+ <rule id="LA_PENDENT" name="*la/el pendent">
+ <!-- Cal un tractament especial perquè pendent és nom i adjectiu.
¿Passa igual amb altres paraules? -->
+ <pattern>
+ <token><exception postag="NCF.000"
postag_regexp="yes"/></token>
+ <token postag="_GN_F." postag_regexp="yes"
regexp="yes">pendents?</token>
+ </pattern>
+ <message>'Pendent' és un nom masculí</message>
+ <short>És nom masculí</short>
+ <example type="incorrect">en <marker>la pendent</marker>
descendent dels últims anys.</example>
+ <example type="correct">en el pendent descendent dels últims
anys.</example>
+ <example type="correct">La reforma pendent.</example>
+ <example type="incorrect">Hi ha una <marker>forta
pendent</marker>.</example>
+ <example type="correct">Hi ha un fort pendent.</example>
+ <example type="incorrect">la manca d'aigua en <marker>les
pendents</marker> superiors</example>
+ <example type="correct">Són reptes pendents.</example>
+ </rule>
</category>
<category name="Possessius">
<rulegroup id="POSSESSIU_SENSE_ARTICLE" name="possessiu sense article:
de meua casa">
@@ -2211,31 +2227,123 @@
<marker>
<token>tan</token>
</marker>
- <token postag="AQ.*|V.P.*" postag_regexp="yes" skip="10"/>
+ <token postag="AQ.[MC][SN].*|V.P.*SM" postag_regexp="yes"
skip="10"/>
<token>com</token>
- <token postag="AQ.*|V.P.*" postag_regexp="yes"/>
+ <token postag="AQ.[MC][SN].*|V.P.*SM" postag_regexp="yes"/>
</pattern>
<message>Hauria de ser:
<suggestion>tant</suggestion>.</message>
<short>Confusió.</short>
+ <example type="incorrect"><marker>tan</marker> físic com
estètic.</example>
+ <example type="correct">Tant físic com estètic.</example>
+ <example type="correct">tan coneixedor com servidora</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>tan</token>
+ </marker>
+ <token postag="AQ.[FC][SN].*|V.P.*SF" postag_regexp="yes"
skip="10"/>
+ <token>com</token>
+ <token postag="AQ.[FC][SN].*|V.P.*SF" postag_regexp="yes"/>
+ </pattern>
+ <message>Hauria de ser:
<suggestion>tant</suggestion>.</message>
+ <short>Confusió.</short>
+ <example type="incorrect"><marker>tan</marker> física com
estètica.</example>
+ <example type="correct">Tant física com estètica.</example>
+ <example type="correct">tan coneixedor com servidora</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>tan</token>
+ </marker>
+ <token postag="AQ.[MC][PN].*|V.P.*PM" postag_regexp="yes"
skip="10"/>
+ <token>com</token>
+ <token postag="AQ.[MC][PN].*|V.P.*PM" postag_regexp="yes"/>
+ </pattern>
+ <message>Hauria de ser:
<suggestion>tant</suggestion>.</message>
+ <short>Confusió.</short>
+ <example type="incorrect"><marker>tan</marker> físics com
estètics.</example>
+ <example type="correct">Tant físics com estètics.</example>
+ <example type="correct">tan coneixedor com servidora</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>tan</token>
+ </marker>
+ <token postag="AQ.[FC][PN].*|V.P.*PF" postag_regexp="yes"
skip="10"/>
+ <token>com</token>
+ <token postag="AQ.[FC][PN].*|V.P.*PF" postag_regexp="yes"/>
+ </pattern>
+ <message>Hauria de ser:
<suggestion>tant</suggestion>.</message>
+ <short>Confusió.</short>
<example type="incorrect"><marker>tan</marker> físiques com
estètiques.</example>
<example type="correct">Tant físiques com estètiques.</example>
+ <example type="correct">tan coneixedor com servidora</example>
</rule>
<rule>
<pattern>
<marker>
<token>tan</token>
</marker>
- <token postag="AQ.*|V.P.*" postag_regexp="yes" skip="10"/>
+ <token postag="AQ.[MC][SN].*|V.P.*SM" postag_regexp="yes"
skip="10"/>
<token>com</token>
<token>no</token>
- <token postag="AQ.*|V.P.*" postag_regexp="yes"/>
+ <token postag="AQ.[MC][SN].*|V.P.*SM" postag_regexp="yes"/>
</pattern>
<message>Hauria de ser:
<suggestion>tant</suggestion>.</message>
<short>Confusió.</short>
- <example type="incorrect"><marker>tan</marker> físiques com no
físiques.</example>
- <example type="correct">Tant físiques com no
físiques.</example>
+ <example type="incorrect"><marker>tan</marker> físic com no
estètic.</example>
+ <example type="correct">Tant físic com no estètic.</example>
</rule>
<rule>
+ <pattern>
+ <marker>
+ <token>tan</token>
+ </marker>
+ <token postag="AQ.[FC][SN].*|V.P.*SF" postag_regexp="yes"
skip="10"/>
+ <token>com</token>
+ <token>no</token>
+ <token postag="AQ.[FC][SN].*|V.P.*SF" postag_regexp="yes"/>
+ </pattern>
+ <message>Hauria de ser:
<suggestion>tant</suggestion>.</message>
+ <short>Confusió.</short>
+ <example type="incorrect"><marker>tan</marker> física com no
estètica.</example>
+ <example type="correct">Tant física com no estètica.</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>tan</token>
+ </marker>
+ <token postag="AQ.[MC][PN].*|V.P.*PM" postag_regexp="yes"
skip="10"/>
+ <token>com</token>
+ <token>no</token>
+ <token postag="AQ.[MC][PN].*|V.P.*PM" postag_regexp="yes"/>
+ </pattern>
+ <message>Hauria de ser:
<suggestion>tant</suggestion>.</message>
+ <short>Confusió.</short>
+ <example type="incorrect"><marker>tan</marker> físics com no
estètics.</example>
+ <example type="correct">Tant físics com no estètics.</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>tan</token>
+ </marker>
+ <token postag="AQ.[FC][PN].*|V.P.*PF" postag_regexp="yes"
skip="10"/>
+ <token>com</token>
+ <token>no</token>
+ <token postag="AQ.[FC][PN].*|V.P.*PF" postag_regexp="yes"/>
+ </pattern>
+ <message>Hauria de ser:
<suggestion>tant</suggestion>.</message>
+ <short>Confusió.</short>
+ <example type="incorrect"><marker>tan</marker> físiques com no
estètiques.</example>
+ <example type="correct">Tant físiques com no
estètiques.</example>
+ </rule>
+
+ <rule>
<!-- Tant fora com dins del Japó -->
<pattern>
<marker>
@@ -2509,7 +2617,7 @@
<message>¿Volíeu dir <suggestion>sense cap</suggestion>?</message>
<example type="incorrect"><marker>sense cap ni un</marker>
treballador</example>
<example type="correct">sense cap treballador</example>
- <example type="correct">O sense cap mi un.</example>
+ <example type="correct">O sense cap ni un.</example>
</rule>
</category>
<category name="Numerals">
@@ -2597,7 +2705,7 @@
<example type="incorrect">¿<marker>Cada quan</marker> et
dutxes?</example>
<example type="correct">¿Cada quant et dutxes?</example>
</rule>
- <rule id="APROP" name="*aprop/">
+ <rule id="APROP" name="*aprop/a prop">
<pattern>
<marker>
<token>aprop</token>
@@ -4864,6 +4972,105 @@
<example type="correct">Les hi volem donar.</example>
</rule>
</rulegroup>
+ <rulegroup id="PRONOM_FEBLE_EN" name="Comprovacions del pronom feble
'en'">
+ <rule>
+ <pattern>
+ <marker>
+ <token postag="SENT_START" skip="-1"><exception
postag="PP3CN000" scope="next" inflected="yes">en</exception></token>
+ <token postag="V[^S][SI].*" postag_regexp="yes"/>
+ <token regexp="yes"
inflected="yes">numeral|un|algun</token>
+ </marker>
+ <token regexp="yes">[.;?!]</token>
+ <!-- <token regexp="yes" inflected="yes"
negate="yes">numeral|un|algun</token> -->
+ </pattern>
+ <message>Probablement falta el pronom 'en'</message>
+ <example type="incorrect"><marker>Jo em tinc
dos</marker>.</example>
+ <example type="correct">Jo en tinc dos.</example>
+ <example type="incorrect"><marker>Em porten
dos</marker>.</example>
+ <example type="correct">Em porten dos plats.</example>
+ <example type="correct">Ells són set.</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token postag="SENT_START" skip="-1"><exception
postag="PP3CN000" scope="next" inflected="yes">en</exception></token>
+ <token postag="V[^S][SI].*" postag_regexp="yes"/>
+ <token regexp="yes"
inflected="yes">numeral|un|algun</token>
+ </marker>
+ <token>,</token>
+ <token inflected="yes" negate="yes">numeral</token>
+ </pattern>
+ <message>Probablement falta el pronom 'en'</message>
+ <example type="incorrect"><marker>Jo em tinc dos</marker>, i
què</example>
+ <example type="correct">la molècula conté dos, tres o més
grups funcionals</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token postag="SENT_START" skip="-1"><exception
postag="PP3CN000" scope="next" inflected="yes">en</exception></token>
+ <token postag="VA.*" postag_regexp="yes" regexp="yes"
inflected="yes">anar|haver</token>
+ <token postag="V.[PGN].*"
postag_regexp="yes"><exception postag="VS.*" postag_regexp="yes"/></token>
+ <token regexp="yes"
inflected="yes">numeral|un|algun</token>
+ </marker>
+ <token regexp="yes">[.,;?!]</token>
+ </pattern>
+ <message>Probablement falta el pronom 'en'</message>
+ <example type="incorrect"><marker>Jo em vaig tenir
dos</marker>.</example>
+ <example type="correct">Jo en vaig tenir dos.</example>
+ <example type="incorrect"><marker>M'han portat
dos</marker>.</example>
+ <example type="correct">M'han portat dos plats.</example>
+ <example type="correct">Me n'han portat dos plats.</example>
+ <example type="correct">Ells han estat set.</example>
+ </rule>
+ </rulegroup>
+ <rulegroup id="PRONOM_FEBLE_HI" name="Comprovacions del pronom feble
'hi'">
+ <!-- Per a fer la norma més robusta, caldria comptar des del verb cap
enrere (hauria de ser una regla en Java) -->
+ <rule>
+ <pattern>
+ <token postag="SENT_START" skip="-1"><exception
postag="PP3CN000" scope="next" inflected="yes">hi</exception><exception
postag="PP3..A00" scope="next" postag_regexp="yes" /></token>
+ <marker>
+ <token regexp="yes">haver|havent<exception
postag="_GN_MS"/></token>
+ </marker>
+ <token><exception
postag="V.P.*|P0.{6}|PP3CN000|PP3NN000|PP3..A00|PP3CP000|PP3CSD00|RG"
postag_regexp="yes"/><exception regexp="yes">de|d'</exception></token>
+ </pattern>
+ <message>Falta el pronom 'hi'. <suggestion><match
no="2"/>-hi</suggestion></message>
+ <example type="incorrect" correction="haver-hi">Va
<marker>haver</marker> una reunió.</example>
+ <example type="incorrect"
correction="Havent-hi"><marker>Havent</marker> aigua, no cal res més.</example>
+ <example type="correct">Haver cantat bé és magnífic.</example>
+ <example type="correct">Haver-lo dut bé és magnífic.</example>
+ <example type="correct">va haver de retirar-se</example>
+ <example type="correct">després d'haver finalment abandonat la
teoria</example>
+ <example type="correct">la guineu, quan no les pot haver, diu
que són verdes</example>
+ </rule>
+ <rule>
+ <pattern>
+ <token postag="SENT_START" skip="-1"><exception
postag="PP3CN000" scope="next" inflected="yes">hi</exception></token>
+ <marker>
+ <token regexp="yes">haver|havent<exception
postag="_GN_MS"/></token>
+ </marker>
+ <token postag="RG"></token>
+ <token><exception postag="V.P.*"
postag_regexp="yes"/><exception regexp="yes">de|d'</exception></token>
+ </pattern>
+ <message>Falta el pronom 'hi'. <suggestion><match
no="2"/>-hi</suggestion></message>
+ <example type="incorrect" correction="haver-hi">després
d'<marker>haver</marker> finalment sentència</example>
+ <example type="correct">després d'haver finalment abandonat la
teoria</example>
+ </rule>
+ <rule>
+ <pattern>
+ <token postag="SENT_START" skip="-1"><exception
postag="PP3CN000" scope="next" inflected="yes">hi</exception></token>
+ <marker>
+ <token regexp="yes">haver|havent</token>
+ <token>-ne</token>
+ </marker>
+ <token><exception
postag="V.P.*|P0.{6}|PP3CN000|PP3NN000|PP3..A00|PP3CP000|PP3CSD00"
postag_regexp="yes"/><exception regexp="yes">de|d'</exception></token>
+ </pattern>
+ <message>Falta el pronom 'hi'. <suggestion><match
no="2"/>-n'hi</suggestion></message>
+ <example type="incorrect" correction="haver-n'hi">Va
<marker>haver-ne</marker> una.</example>
+ <example type="incorrect"
correction="Havent-n'hi"><marker>Havent-ne</marker> prou, no cal res
més.</example>
+ <example type="correct">Haver-ne dut tres és
magnífic.</example>
+ <!-- False alarm Qui pogués haver-ne una peça -->
+ </rule>
+ </rulegroup>
</category>
<category name="«lo» neutre">
<!-- es podria fer una llista molt més completa de solucions-->
@@ -7865,6 +8072,35 @@
<example type="incorrect">Ens fa <marker>molt de
por</marker></example>
<example type="correct">Ens fa molta por</example>
</rule>
+ <rulegroup id="CREUATS_CROATS" name="*creuats/croats">
+ <!-- Podria anar en WrongWordInContext ? -->
+ <rule>
+ <pattern>
+ <token postag="D[DAIN].*|AO.*" postag_regexp="yes"></token>
+ <marker>
+ <token regexp="yes">creua(t|ts|da|des)</token>
+ </marker>
+ </pattern>
+ <message>¿Volíeu dir <suggestion><match no="2"
regexp_match="creua(.+)" regexp_replace="croa$1"/></suggestion>?</message>
+ <example type="incorrect" correction="croats">els
<marker>creuats</marker> assetjaren la ciutat</example>
+ <example type="correct">els croats assetjaren la
ciutat</example>
+ <example type="correct">té els cables creuats</example>
+ <example type="incorrect" correction="croada" >la
<marker>creuada</marker> albigesa</example>
+ <example type="incorrect" correction="croada" >la primera
<marker>creuada</marker></example>
+ </rule>
+ <rule>
+ <pattern>
+ <token
regexp="yes">soldats?|exèrcits?|hosts?|cavallers?|guerrers?</token>
+ <marker>
+ <token regexp="yes">creua(t|ts|da|des)</token>
+ </marker>
+ </pattern>
+ <message>¿Volíeu dir <suggestion><match no="2"
regexp_match="creua(.+)" regexp_replace="croa$1"/></suggestion>?</message>
+ <example type="incorrect" correction="croats">els exèrcits
<marker>creuats</marker> assetjaren la ciutat</example>
+ <example type="correct">els exèrcits croats assetjaren la
ciutat</example>
+ <example type="correct">té els cables creuats</example>
+ </rule>
+ </rulegroup>
</category>
<category name="Sintaxi">
<rulegroup id="FALTA_ELEMENT" name="Falta un element entre verbs: hi
ha una cosa *(que) és clara">
This was sent by the SourceForge.net collaborative development platform, the
world's largest Open Source development site.
------------------------------------------------------------------------------
How fast is your code?
3 out of 4 devs don\\\'t know how their code performs in production.
Find out how slow your code is with AppDynamics Lite.
http://ad.doubleclick.net/clk;262219672;13503038;z?
http://info.appdynamics.com/FreeJavaPerformanceDownload.html
_______________________________________________
Languagetool-cvs mailing list
Languagetool-cvs@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/languagetool-cvs
