X-Git-Url: http://iramuteq.org/git?a=blobdiff_plain;f=autres%2Fcable.py;fp=autres%2Fcable.py;h=bbfa37f6a55c9f514e4899da87e1c5703a33b1ab;hb=eaa044d1147e26b82942ce56d5965c83fdddf069;hp=0000000000000000000000000000000000000000;hpb=10d67a5cd48583c060b6a0e77e87c41f80671027;p=iramuteq

diff --git a/autres/cable.py b/autres/cable.py
new file mode 100644
index 0000000..bbfa37f
--- /dev/null
+++ b/autres/cable.py
@@ -0,0 +1,68 @@
+# -*- coding: utf-8 -*-
+#Author: Pierre Ratinaud
+#Copyright (c) 2008-2020 Pierre Ratinaud
+#Lisense: GNU/GPL
+# usage ?
+# encodage est utilisé - mais utf-8 est par défaut dans PY3
+
+#------------------------------------
+# import des modules python
+#------------------------------------
+import codecs
+
+
+filein = '/home/pierre/fac/cablegate/allcables-all.txt'
+enc = 'utf-8'
+
+infile = codecs.open(filein, 'r', enc)
+content = []
+
+class BigCorpus :
+    def __init__(self, parent) :
+        self.parent = parent
+        self.parametre = {'syscoding': sys.getdefaultencoding()}
+        self.content = None
+        self.ucis = None
+        self.formes = {}
+        self.lems = {}
+        self.ucenb = None
+        self.etoiles = None
+        self.etintxt = {}
+        self.ucis_paras_uces = None
+        self.lc = None
+        self.lc0 = None
+        self.actives = None
+        self.supp = None
+        #self.supplementaires = []
+        self.lenuc1 = None
+        self.lenuc2 = None
+        self.lexique = None
+    
+    def open_corpus(self) :
+        return codecs.open(self.parametre['filename'], "r", self.parametre['encodage'])
+    
+    def buildcorpus(self) :
+        i = 0
+        ucifile = os.path.join(os.path.basedir(self.parametre['filename']), 'ucis.txt')
+        uci = open(ucifile, 'w')
+        ucinb = 0
+        for line in self.open_corpus() :
+            if line.startswith(u'****') and i==0 :
+                uci.write(line)
+                i += 1
+            elif line.startswith(u'****') and i=!0 :
+                uci.write(line)
+                parse_uci()
+
+                write_uci()
+                uci[ucinb] = i
+                ucinb += 1
+                i += 1
+            else :
+                addlinetouci(uci, prepare(line))
+                line = line.lower().replace(u'\'','\' ').replace(u'’','\' ').replace('...',u' £ ').replace('?',' ? ').replace('.',' . ').replace('!', ' ! ').replace(',',' , ').replace(';', ' ; ').replace(':', ' : ').strip()
+                line = line.replace('\n', ' ').replace('\r', ' ')
+                line = line.split()
+                content[-1].append(line)
+            i += 1
+print len(content)