teisendused

tools/visdic.py

@@ -20,6 +20,7 @@ import codecs
 from lxml import etree
 
 from settings import PWN_PATH
+from settings import FIN_PATH
 
 POINTER_SYMBOLS = { # (symbol, pos):
     ('!','n'):{'gloss':'Antonym','vis':'antonym'},
@@ -114,6 +115,15 @@ def deadj(iStr):
     else:
         return iStr
 
+def translate(orig,table=None):
+    """Translation of orig to replacement in table
+    """
+    if table:
+        return map(lambda x: x[-1], 
+                   filter(lambda y: y[0]==orig,table))[0]
+    else:
+        return orig
+
 def parse_wn_index(path): 
     """parses index.sense file of Princeton WN
 
@@ -456,18 +466,20 @@ def process_file(f):
     return out
 
 
-def make_visdic_xml(dataDict,xml):
+def make_visdic_xml(dataDict,xml,translation=None):
     ID_PREFIX = 'eng-30'
     # do not care about what xml has already
     for i in dataDict:
         try:
             idText = '{}-{}-{}'.format(ID_PREFIX,
-                                       i['synset_offset'],
+                                       translate(i['synset_offset'],
+                                                 translation),
                                        i['ss_type']
                                        )
         except ValueError:
             idText = '%s-%s-%s' % (ID_PREFIX,
-                                   i['synset_offset'],
+                                   translate(i['synset_offset'],
+                                             translation),
                                    i['ss_type']
                                    )
         posText = i['ss_type']
@@ -496,12 +508,14 @@ def make_visdic_xml(dataDict,xml):
             ilr = etree.SubElement(synset, 'ILR')
             try:
                 ilrText = '{}-{}-{}'.format(ID_PREFIX,
-                                            j['synset_offset'],
+                                            translate(j['synset_offset'],
+                                                      translation),
                                             j['pos']
                                             )
             except ValueError:
                 ilrText = '%s-%s-%s' % (ID_PREFIX,
-                                        j['synset_offset'],
+                                        translate(j['synset_offset'],
+                                                  translation),
                                         j['pos']
                                         )
 
@@ -551,10 +565,27 @@ def oldermain():
     parser.add_option('-r','--prefix',
                       help='outfile prefix'
                       )
+    parser.add_option('-l','--lang',
+                      help='wordnet language'
+                      )
+    parser.add_option('-t','--trans',
+                      help='Translation tables for ILI')
+
     (options, args) = parser.parse_args()
 
     b = WN()
 
+    if options.lang and options.lang == 'fin':
+        PWN_PATH = FIN_PATH
+
+    transD = None
+
+    if options.trans:
+        f = open(options.trans, 'r')
+        transD = map(lambda x: x.strip().split(':'),f.readlines())
+        f.close()
+        print transD
+
     oList = wnparser(PWN_PATH,options.pos,
                      int(options.start),int(options.end)
                      )
@@ -595,6 +626,8 @@ def main():
     parser.add_argument('-p','--pos', dest='pos',
                         required=True,
                         help='part of speech')
+    parser.add_argument('-t','--trans',
+                        help='Translation tables for ILI')
 
     args = parser.parse_args()
 

tools/vistnoun.sh

@@ -9,9 +9,11 @@ i=$1
 if [ "$3" == fin ]; then
     wnpath=$HOME/Wordnet/FinWN/fiwn-2.0/dict
     prefix="fin-20"
+    lang="$3"
 else
     wnpath=$HOME/Wordnet/WordNet-3.0/dict/
     prefix="pwn-30"
+    lang="eng"
 fi
 
 ulimit=$[$(cat $wnpath/data.$2|egrep -v '^  '|wc -l)-1]
@@ -24,6 +26,6 @@ do
     then j=$[$ulimit+1]
     fi 
     echo $i $j
-    python visdic.py -p $2 -r $prefix -s $i -e $j
+    python visdic.py -p $2 -r $prefix -s $i -e $j -l $lang -t fin2pwn-$2.txt
     i=$[$i+200]
 done