Revision: 7665
http://languagetool.svn.sourceforge.net/languagetool/?rev=7665&view=rev
Author: jaumeortola
Date: 2012-07-07 18:36:10 +0000 (Sat, 07 Jul 2012)
Log Message:
-----------
[ca] Improved rule: TANT_TAN.
Modified Paths:
--------------
trunk/JLanguageTool/src/resource/ca/disambiguation.xml
trunk/JLanguageTool/src/rules/ca/grammar.xml
Modified: trunk/JLanguageTool/src/resource/ca/disambiguation.xml
===================================================================
--- trunk/JLanguageTool/src/resource/ca/disambiguation.xml 2012-07-07
11:22:32 UTC (rev 7664)
+++ trunk/JLanguageTool/src/resource/ca/disambiguation.xml 2012-07-07
18:36:10 UTC (rev 7665)
@@ -1901,9 +1901,9 @@
<disambig action="filter" postag="A.*"></disambig>
</rule>
<rule id="adv_adjverb" name="adverbi + adjectiu/verb">
- <!-- tan complexa -->
+ <!-- tan complexa, tant planes de sabana com regions de jungla; tan
baixa com -->
<pattern>
- <token postag="RG"></token>
+ <token postag="RG" regexp="yes">tant?</token>
<marker>
<and>
<token postag="V.[MIS].*" postag_regexp="yes"></token>
@@ -1912,7 +1912,7 @@
</and>
</marker>
</pattern>
- <disambig action="filter" postag="A.*"></disambig>
+ <disambig action="filter" postag="[^V].*"></disambig>
</rule>
<rule id="DA_PX" name="article + possessiu(nostre)">
<pattern>
@@ -3100,7 +3100,7 @@
<pattern>
<unify>
<feature id="nombre"/>
- <token postag="D[^R].*" postag_regexp="yes"></token>
+ <token postag="D[AD].*" postag_regexp="yes"></token>
<and>
<token postag="N.*" postag_regexp="yes"><exception
postag="V.[MSI].*" postag_regexp="yes"/></token>
<token postag="AQ.CS.*" postag_regexp="yes"></token>
@@ -3518,7 +3518,7 @@
<disambig action="filter" postag="[^V].*"></disambig>
</rule>
<!-- gerundi/adj + nom "Ocupant llocs" Canviat de lloc per: "D'eixa
important època" -->
- <rule id="adv_gerundi_adjectiu" name="adverbi + gerundi/adjectiu + nom">
+ <rule id="adv_geradj_nom" name="adverbi + gerundi/adjectiu + nom">
<pattern>
<token postag="RG"></token>
<marker>
@@ -4614,6 +4614,18 @@
</pattern>
<disambig action="filter" postag="A.*|V.P.*"></disambig>
</rule>
+ <rule id="tant_nom_com_nom" name="desambigua tant+nom+com+nom">
+ <!-- tant grecs com romans. És excep. de la regla següent -->
+ <pattern>
+ <token>tant</token>
+ <token postag="NC.*" postag_regexp="yes"/>
+ <token>com</token>
+ <marker>
+ <token postag="NC.*" postag_regexp="yes"/>
+ </marker>
+ </pattern>
+ <disambig action="filter" postag="[^V].*"></disambig>
+ </rule>
<rule id="VerbNomAdj" name="verb i adjectiu i nom: verb">
<pattern>
<marker>
@@ -4694,8 +4706,9 @@
</pattern>
<disambig action="filter" postag="[^V].*"></disambig>
</rule>
- <rule id="VerbAdjectiu" name="verb i adjectiu que no ha concordat"> <!--
fer exàmens -->
+ <rule id="VerbAdjectiu" name="verb i adjectiu que no ha concordat"> <!--
fer exàmens. excep: tan regular (adj) -->
<pattern>
+ <token postag="RG" negate_pos="yes"/>
<marker>
<and>
<token postag="A.*" postag_regexp="yes"><exception
postag="_GN_.*" postag_regexp="yes" /><exception
regexp="yes">\p{Lu}+[\p{Ll}·]*</exception><exception
regexp="yes">[\p{Ll}·]+(e|[aei]sses|[ae]res|era)</exception></token>
@@ -4981,18 +4994,22 @@
-->
<rule id="InfGer_Nom" name="infinitiu o gerundi / nom que no ha concordat">
- <!-- ser, deixant, etc.-->
+ <!-- ser, deixant, etc. excep: tan regular (adj)-->
<pattern>
- <and>
- <token postag="V.[NG].*" postag_regexp="yes"></token>
- <token postag="N.*" postag_regexp="yes"><exception
postag="_GN.*" postag_regexp="yes"/></token>
- </and>
+ <token postag="RG" negate_pos="yes"/>
+ <marker>
+ <and>
+ <token postag="V.[NG].*" postag_regexp="yes"></token>
+ <token postag="N.*" postag_regexp="yes"><exception
postag="_GN.*" postag_regexp="yes"/></token>
+ </and>
+ </marker>
</pattern>
<disambig action="filter" postag="V.*"></disambig>
</rule>
<rule id="Adv_Nom" name="adverbi i nom: adverbi">
- <!-- llavors company-->
+ <!-- llavors company. Excepció: tant com, tant (de) mal com -->
<pattern>
+ <token negate="yes">tant</token>
<marker>
<and>
<token postag="RG" postag_regexp="yes"></token>
@@ -5000,10 +5017,11 @@
</and>
</marker>
</pattern>
- <disambig action="filter" postag="RG"></disambig>
+ <disambig action="filter" postag="[^N].*"></disambig>
</rule>
- <rule id="AdjAdv" name="adjectiu i adverbi"> <!-- correctament primer -->
+ <rule id="AdjAdv" name="adjectiu i adverbi"> <!-- correctament primer.
Excep: tant (de) mal -->
<pattern>
+ <token negate="yes">tant</token>
<marker>
<and>
<token postag="RG" postag_regexp="yes"></token>
@@ -5143,7 +5161,7 @@
</pattern>
<disambig action="filter" postag="SPS00"></disambig>
</rule>
- <rule id="tant_x_com" name="tant + __ + com">
+<!-- <rule id="tant_x_com" name="tant + __ + com">
<pattern>
<marker>
<token>tant</token>
@@ -5154,7 +5172,7 @@
<disambig>
<match no="1" postag_regexp="yes" postag="RG"></match>
</disambig>
- </rule>
+ </rule> -->
<rule id="elsuns_x_elsaltres" name=" (els) uns + prep + els altres">
<pattern>
<token skip="1">uns</token>
@@ -5312,4 +5330,90 @@
<disambig action="add"><wd pos="allow_masculine"></wd></disambig>
</rule>
</rulegroup>
+ <rulegroup id="tant_x_com" name="tant + __ + com + __">
+ <rule>
+ <!-- Dificultats: és tan(t) important com bonic -->
+ <pattern>
+ <marker>
+ <token>tant</token>
+ </marker>
+ <token postag="A.*|V.P.*" postag_regexp="yes" skip="10"/>
+ <token>com</token>
+ <token postag="A.*|V.P.*" postag_regexp="yes"/>
+ </pattern>
+ <disambig action="add"><wd pos="allow_tant"></wd></disambig>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>tant</token>
+ </marker>
+ <token postag="A.*|V.P.*" postag_regexp="yes" skip="10"/>
+ <token>com</token>
+ <token>no</token>
+ <token postag="A.*|V.P.*" postag_regexp="yes"/>
+ </pattern>
+ <disambig action="add"><wd pos="allow_tant"></wd></disambig>
+ </rule>
+ <rule>
+ <!-- tant partidaris com opositors -->
+ <pattern>
+ <marker>
+ <token>tant</token>
+ </marker>
+ <token postag="NC.*" postag_regexp="yes" skip="10"/>
+ <token>com</token>
+ <token postag="NC.*" postag_regexp="yes"/>
+ </pattern>
+ <disambig action="add"><wd pos="allow_tant"></wd></disambig>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>tant</token>
+ </marker>
+ <token postag="NC.*" postag_regexp="yes" skip="10"/>
+ <token>com</token>
+ <token>no</token>
+ <token postag="NC.*" postag_regexp="yes"/>
+ </pattern>
+ <disambig action="add"><wd pos="allow_tant"></wd></disambig>
+ </rule>
+ <rule>
+ <!-- Tant fora com dins del Japó -->
+ <pattern>
+ <marker>
+ <token>tant</token>
+ </marker>
+ <token postag="RG" skip="10"/>
+ <token>com</token>
+ <token postag="RG"/>
+ </pattern>
+ <disambig action="add"><wd pos="allow_tant"></wd></disambig>
+ </rule>
+ <rule>
+ <!-- Tant jurídicament com en la realitat -->
+ <pattern>
+ <marker>
+ <token>tant</token>
+ </marker>
+ <token postag="RG" skip="10"/>
+ <token>com</token>
+ <token postag="SPS00"/>
+ </pattern>
+ <disambig action="add"><wd pos="allow_tant"></wd></disambig>
+ </rule>
+ <rule>
+ <!-- coneixements tant científics com de les més variades
disciplines-->
+ <pattern>
+ <marker>
+ <token>tant</token>
+ </marker>
+ <token postag="A.*|V.P.*" postag_regexp="yes" skip="10"/>
+ <token>com</token>
+ <token inflected="yes">de</token>
+ </pattern>
+ <disambig action="add"><wd pos="allow_tant"></wd></disambig>
+ </rule>
+ </rulegroup>
</rules>
\ No newline at end of file
Modified: trunk/JLanguageTool/src/rules/ca/grammar.xml
===================================================================
--- trunk/JLanguageTool/src/rules/ca/grammar.xml 2012-07-07 11:22:32 UTC
(rev 7664)
+++ trunk/JLanguageTool/src/rules/ca/grammar.xml 2012-07-07 18:36:10 UTC
(rev 7665)
@@ -1705,22 +1705,62 @@
</rule>
</rulegroup>
<rulegroup id="TANT_TAN" name="Comprova: «tan/tant»">
- <!-- cal controlar les exp. tant ... com-->
<rule>
<pattern>
<marker>
+ <token>un</token>
<token>tant</token>
</marker>
- <token postag="[AR].*"
postag_regexp="yes"><exception>com</exception><exception postag="N.[MC][SN].*"
postag_regexp="yes"></exception></token>
+ <token postag="AQ.*|V.P.*" postag_regexp="yes"/>
</pattern>
+ <message>Hauria de ser: <suggestion>una mica</suggestion>,
<suggestion>més o menys</suggestion>
+ , <suggestion>més aviat</suggestion>, <suggestion>poc o
molt</suggestion>
+ , <suggestion>un pèl</suggestion>.</message>
+ <short>Expressió errònia.</short>
+ <example type="incorrect">Una qüestió <marker>un tant</marker>
particular.</example>
+ <example type="correct">Una qüestió una mica
particular.</example>
+ </rule>
+ <rule>
+ <pattern>
+ <token>tan</token>
+ <token>punt</token>
+ </pattern>
+ <message>Hauria de ser: <suggestion>tan bon punt</suggestion>
o <suggestion>bon punt</suggestion>.</message>
+ <short>Expressió errònia.</short>
+ <example type="incorrect"><marker>tan punt</marker> va
arribar</example>
+ <example type="correct">tan bon punt va arribar</example>
+ </rule>
+ <rule>
+ <pattern>
+ <token negate="yes">per</token>
+ <marker>
+ <token>tant<exception postag="allow_tant"/></token>
+ </marker>
+ <token postag="[AR].*" postag_regexp="yes"><exception
regexp="yes">abans|ja|on|quan|que|com|menys|doncs|però|ara|emperò|ans</exception><exception
postag="N.[MC][SN].*" postag_regexp="yes"/></token>
+ </pattern>
<message>'Tan' davant d'adjectiu o adverbi. Podria ser:
<suggestion>tan</suggestion>.</message>
<short>'Tan' i no 'tant' davant d'adjectiu o adverbi.</short>
<example type="incorrect">És <marker>tant</marker>
bonic.</example>
<example type="correct">És tan bonic.</example>
+ <example type="correct">Tant quan sigui així com quan no ho
sigui.</example>
</rule>
<rule>
<pattern>
+ <token>per</token>
<marker>
+ <token>tant</token>
+ </marker>
+ <token postag="AQ.*|V.P.*" postag_regexp="yes"/>
+ <token postag="NC.*" postag_regexp="yes"/>
+ </pattern>
+ <message>'Tan' davant d'adjectiu o adverbi. Podria ser:
<suggestion>tan</suggestion>.</message>
+ <short>'Tan' i no 'tant' davant d'adjectiu o adverbi.</short>
+ <example type="incorrect">Felicitats per <marker>tant</marker>
noble tasca.</example>
+ <example type="correct">Felicitats per tan noble
tasca.</example>
+ </rule>
+ <rule>
+ <pattern>
+ <marker>
<token inflected="yes">tant</token>
</marker>
<token inflected="yes">poc</token>
@@ -1748,7 +1788,7 @@
<marker>
<token>tan</token>
</marker>
- <token postag="N.*" postag_regexp="yes"><exception
postag="A.*" postag_regexp="yes"></exception></token>
+ <token postag="N.*" postag_regexp="yes"><exception
postag="A.*|V.P.*" postag_regexp="yes"/></token>
</pattern>
<message>'Tant' davant de nom. Podria ser:
<suggestion>tant</suggestion>.</message>
<short>'Tant' i no 'tan' davant de nom.</short>
@@ -1773,16 +1813,34 @@
<marker>
<token>tan</token>
</marker>
- <token>com</token>
+ <token
regexp="yes">abans|ja|on|quan|que|com|menys|doncs|però|ara|emperò|ans</token>
</pattern>
- <message>'Tant com'. Podria ser:
<suggestion>tant</suggestion>.</message>
- <short>'Tant com'.</short>
+ <message>Podria ser: <suggestion>tant</suggestion>.</message>
+ <short>Confusió.</short>
<example type="incorrect">Menja <marker>tan</marker> com
vulguis.</example>
<example type="correct">Menja tant com vulguis.</example>
+ <example type="incorrect">M'agrada <marker>tan</marker> que me
l'enduc.</example>
+ <example type="correct">M'agrada tant que me l'enduc.</example>
+ <example type="incorrect"><marker>Tan</marker> quan sigui així
com quan no ho sigui.</example>
+ <example type="correct">Tant quan sigui així com quan no ho
sigui.</example>
</rule>
<rule>
<pattern>
<marker>
+ <token>tan</token>
+ </marker>
+ <token>,</token>
+ <token>que</token>
+ </pattern>
+ <message>Hauria de ser:
<suggestion>tant</suggestion>.</message>
+ <short>Confusió.</short>
+ <example type="incorrect">M'agrada <marker>tan</marker>, que
me l'enduc.</example>
+ <example type="correct">M'agrada tant, que me
l'enduc.</example>
+ </rule>
+ <rule>
+ <pattern>
+ <token negate="yes">per</token>
+ <marker>
<token>tant</token>
</marker>
<token postag="<LOC_ADV>"></token>
@@ -1803,32 +1861,96 @@
<short>'Tant' quan indica quantitat.</short>
<example type="incorrect"><marker>Tan</marker> de bo!</example>
<example type="correct">Tant de bo!</example>
+ <example type="correct">Anava tan de pressa!</example>
</rule>
<rule>
<pattern>
+ <token>per</token>
<marker>
<token>tan</token>
</marker>
- <token>que</token>
+ <token>,</token>
</pattern>
<message>Hauria de ser:
<suggestion>tant</suggestion>.</message>
<short>Confusió.</short>
- <example type="incorrect">M'agrada <marker>tan</marker> que me
l'enduc.</example>
- <example type="correct">M'agrada <marker>tant</marker> que me
l'enduc.</example>
+ <example type="incorrect">Per <marker>tan</marker>, és
correcte.</example>
+ <example type="correct">Per tant, és correcte.</example>
</rule>
+ <!-- tan(t) en frases distributives tant + __ +com +__ -->
<rule>
<pattern>
<marker>
<token>tan</token>
</marker>
- <token>,</token>
- <token>que</token>
+ <token postag="AQ.*|V.P.*" postag_regexp="yes" skip="10"/>
+ <token>com</token>
+ <token postag="AQ.*|V.P.*" postag_regexp="yes"/>
</pattern>
<message>Hauria de ser:
<suggestion>tant</suggestion>.</message>
<short>Confusió.</short>
- <example type="incorrect">M'agrada <marker>tan</marker>, que
me l'enduc.</example>
- <example type="correct">M'agrada <marker>tant</marker>, que me
l'enduc.</example>
+ <example type="incorrect"><marker>tan</marker> físiques com
estètiques.</example>
+ <example type="correct">Tant físiques com estètiques.</example>
</rule>
+ <rule>
+ <pattern>
+ <marker>
+ <token>tan</token>
+ </marker>
+ <token postag="AQ.*|V.P.*" postag_regexp="yes" skip="10"/>
+ <token>com</token>
+ <token>no</token>
+ <token postag="AQ.*|V.P.*" 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>
+ </rule>
+ <rule>
+ <!-- Tant fora com dins del Japó -->
+ <pattern>
+ <marker>
+ <token>tan</token>
+ </marker>
+ <token postag="RG" skip="10"><exception
regexp="yes">abans|ja|on|quan|que|com|menys|doncs|però|ara|emperò|ans</exception></token>
+ <token>com</token>
+ <token postag="RG"><exception
regexp="yes">abans|ja|on|quan|que|com|menys|doncs|però|ara|emperò|ans</exception></token>
+ </pattern>
+ <message>Hauria de ser:
<suggestion>tant</suggestion>.</message>
+ <short>Confusió.</short>
+ <example type="incorrect"><marker>tan</marker> dins com fora
del Japó.</example>
+ <example type="correct">Tant dins com fora del Japó.</example>
+ </rule>
+ <rule>
+ <!-- coneixements tant científics com de les més variades
disciplines-->
+ <pattern>
+ <marker>
+ <token>tan</token>
+ </marker>
+ <token postag="A.*|V.P.*" postag_regexp="yes" skip="10"/>
+ <token>com</token>
+ <token inflected="yes">de</token>
+ <token postag="NC.*" postag_regexp="yes"/>
+ </pattern>
+ <message>Hauria de ser:
<suggestion>tant</suggestion>.</message>
+ <short>Confusió.</short>
+ <example type="incorrect">coneixements <marker>tan</marker>
científics com de sentit comú</example>
+ <example type="correct">coneixements tant científics com de
sentit comú</example>
+ </rule>
+ <!-- 'tan' davant d'article -->
+ <rule>
+ <pattern>
+ <marker>
+ <token>tan</token>
+ </marker>
+ <token postag="D.*" postag_regexp="yes"><exception
inflected="yes">poc</exception></token>
+ </pattern>
+ <message>Hauria de ser:
<suggestion>tant</suggestion>.</message>
+ <short>Confusió.</short>
+ <example type="incorrect"><marker>tan</marker> l'un com
l'altre</example>
+ <example type="incorrect"><marker>tan</marker> aquest com
aquell</example>
+ <example type="correct">tant aquest com aquell</example>
+ </rule>
</rulegroup>
<rulegroup id="ADV_MENT" name="dos adverbis acabats en -ment">
<rule>
@@ -6281,7 +6403,7 @@
<example type="incorrect"><marker>quant més</marker> en portin,
millor</example>
<example type="correct"><marker>com més</marker> en portin,
millor</example>
</rule>
- <rule id="SEMBLA_SER" name="sembla ser">
+ <rule id="SEMBLA_SER" name="*sembla/pareix ser">
<pattern>
<marker>
<token inflected="yes"
regexp="yes">semblar|parèixer</token>
@@ -6305,7 +6427,17 @@
<example type="incorrect"><marker>tal i com</marker> vam quedar
ahir</example>
<example type="correct"><marker>tal com</marker> vam quedar
ahir</example>
</rule>
- <rulegroup id="UN_A_UN" name="un a un">
+ <rule id="TANMATEIX" name="*tan(t) mateix/tanmateix">
+ <pattern>
+ <token regexp="yes">tant?</token>
+ <token>mateix</token>
+ </pattern>
+ <message>Cal dir <suggestion>tanmateix</suggestion>.</message>
+ <short>Expressió incorrecta</short>
+ <example type="incorrect"><marker>Tant mateix</marker> cal
fer-ho.</example>
+ <example type="correct">Tanmateix cal fer-ho.</example>
+ </rule>
+ <rulegroup id="UN_A_UN" name="*un a un">
<rule>
<pattern>
<token negate="yes">d'</token>
@@ -6335,7 +6467,7 @@
<example type="correct">passeu <marker>d'una a
una</marker></example>
</rule>
</rulegroup>
- <rulegroup id="ALS_TEMPS" name="als + expr. de temps">
+ <rulegroup id="ALS_TEMPS" name="*als + expr. de temps">
<rule>
<pattern>
<token>a</token>
This was sent by the SourceForge.net collaborative development platform, the
world's largest Open Source development site.
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Languagetool-cvs mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/languagetool-cvs
