X-Git-Url: http://iramuteq.org/git?p=iramuteq;a=blobdiff_plain;f=tools.py;h=669337b548f5296f800066bfea5f34c14df6736d;hp=e35ea4b790f1fe9146aece68f2e4c5d5b1f6c6b1;hb=eb6087e90df00aee22d69cada9ef473331e85344;hpb=eb614c725930bc65a7ad43eda1b769b504433f88

diff --git a/tools.py b/tools.py
index e35ea4b..669337b 100644
--- a/tools.py
+++ b/tools.py
@@ -22,6 +22,12 @@ def istext(line) :
     else :
         return False
 
+def isthem(line):
+    if line.startswith(u'-*') :
+        return True
+    else :
+        return False
+
 def testvar(line, variable) :
     line = line.split()
     varmod = [val.split('_') for val in line[1:]]
@@ -48,12 +54,16 @@ class Extract :
             parametres = dial.make_param()
             if option == 'splitvar' :
                 SplitFromVar(parametres)
-            else :
+            elif option == 'mods' :
                 ExtractMods(parametres)
-        dial.Destroy()
-        dial = wx.MessageDialog(parent, 'Done !', style = wx.OK)
-        dial.ShowModal()
-        dial.Destroy()
+            elif option == 'them' :
+                SplitFromThem(parametres)
+            dial.Destroy()
+            dial = wx.MessageDialog(parent, 'Done !', style = wx.OK)
+            dial.ShowModal()
+            dial.Destroy()
+        else :
+            dial.Destroy()
 
 class SplitFromVar :
     def __init__(self, parametres) :
@@ -84,6 +94,43 @@ class SplitFromVar :
         for f in filedict :
             filedict[f].close()
 
+class SplitFromThem :
+    def __init__(self, parametres) :
+        self.filein = parametres['filein']
+        self.them = parametres['them']
+        self.encodein = parametres['encodein']
+        self.encodeout = parametres['encodeout']
+        self.basepath = os.path.dirname(self.filein)
+        self.pathout = os.path.join(self.basepath, '_'.join([them.replace(u'-*','') for them in self.them]))
+        self.fileout = open(self.pathout, 'w')
+        self.doparse()
+        self.fileout.close()
+    
+    def doparse(self):
+        text = ''
+        keepline = False
+        lastet = ''
+        with codecs.open(self.filein, 'r', self.encodein) as fin :
+            for line in fin :
+                if istext(line) :
+                    self.writetext(self.fileout, lastet, text)
+                    text = ''
+                    lastet = line
+                if isthem(line) :
+                    l = line.strip().rstrip('\n\r')
+                    if l in self.them :
+                        keepline = True
+                    else :
+                        keepline = False
+                if keepline :
+                    text += line
+            self.writetext(self.fileout, lastet, text)
+    
+    def writetext(self, fileout, lastet, text):
+        if text != '' :
+            self.fileout.write(lastet.encode(self.encodeout) + text.encode(self.encodeout))
+            
+
 class ExtractMods :
     def __init__(self, parametres) :
         self.onefile = parametres.get('onefile', False)
@@ -167,3 +214,4 @@ def converttabletocorpus(table, fileout, enc='UTF8') :
 if __name__ == '__main__' :
     #SplitFromVar(parametres)
     ExtractMods(parametres, True)
+    #end