X-Git-Url: http://iramuteq.org/git?p=iramuteq;a=blobdiff_plain;f=textdist.py;h=33dd2ced2c678bae50760b8972606ac364d3d3d6;hp=c7fc1337b3843db8e1c114ec912d89cf70ced446;hb=9b78e6210e7fc88a7e77d178c4090aabb23580d9;hpb=8fa853a25a9d62b1446e1bc543e5a3a4d0e03dcf

diff --git a/textdist.py b/textdist.py
index c7fc133..33dd2ce 100644
--- a/textdist.py
+++ b/textdist.py
@@ -2,54 +2,66 @@
 # -*- coding: utf-8 -*-
 #Author: Pierre Ratinaud
 #Copyright (c) 2008-2009 Pierre Ratinaud
-#Lisense: GNU/GPL
+#License: GNU/GPL
 
 from chemins import ConstructPathOut, ConstructAfcUciPath, ChdTxtPathOut
 from corpus import Corpus
 from OptionAlceste import OptionPam
+from analysetxt import AnalyseText
 import wx
 import os
 from ConfigParser import *
 import sys
 from functions import print_liste, exec_rcode, CreateIraFile, progressbar, check_Rresult, BugDialog
 from layout import PrintRapport
-from PrintRScript import AlcesteTxtProf, RPamTxt
+from PrintRScript import ReinertTxtProf, RPamTxt
 from openanalyse import OpenAnalyse
 from time import time, sleep
 
 
-class AnalysePam:
-    def __init__(self, parent, corpus, cmd = False):
-        t1 = time()
-        self.parent = parent
-        self.corpus = corpus
-        self.cmd = cmd
+class AnalysePam(AnalyseText) :
+#    def __init__(self, parent, corpus, cmd = False):
+    def doanalyse(self) :
+        self.parametres['type'] = 'pamtxt'
+        self.pathout.basefiles(ChdTxtPathOut)
+        self.actives, lim = self.corpus.make_actives_nb(self.parametres['max_actives'], 1)
+        self.parametres['eff_min_forme'] = lim
+        self.parametres['nbactives'] = len(self.actives)
+        if self.parametres['classif_mode'] == 0 :
+            self.corpus.make_and_write_sparse_matrix_from_uces(self.actives, self.pathout['TableUc1'], self.pathout['listeuce1'])
+        elif self.parametres['classif_mode'] == 1 :
+            self.corpus.make_and_write_sparse_matrix_from_uci(self.actives, self.pathout['TableUc1'], self.pathout['listeuce1'])
+        RPamTxt(self.corpus, self.parent.RscriptsPath) 
+        #t1 = time()
+        #self.parent = parent
+        #self.corpus = corpus
+        #self.cmd = cmd
         if not self.cmd :
             self.dlg = progressbar(self, 9)
-        else :
-            self.dlg = None
-        ucis_txt, ucis_paras_txt = self.corpus.start_analyse(self.parent, dlg = self.dlg, cmd = self.cmd)
-        self.corpus.make_len_uce(self.corpus.get_tot_occ_from_ucis_txt(ucis_txt))
-        del ucis_txt
-        if not self.cmd :
-            self.dlg.Update(5, '%i ucis - Construction des uces' % len(self.corpus.ucis))        
-        if self.corpus.parametre['type'] == 0 :
-            self.corpus.make_ucis_paras_uces(ucis_paras_txt, make_uce = True)
-        else :
-            self.corpus.make_ucis_paras_uces(ucis_paras_txt, make_uce = False)
-        del ucis_paras_txt
+        #else :
+        #    self.dlg = None
+        #ucis_txt, ucis_paras_txt = self.corpus.start_analyse(self.parent, dlg = self.dlg, cmd = self.cmd)
+        #self.corpus.make_len_uce(self.corpus.get_tot_occ_from_ucis_txt(ucis_txt))
+        #del ucis_txt
+        #if not self.cmd :
+        #    self.dlg.Update(5, '%i ucis - Construction des uces' % len(self.corpus.ucis))        
+        #if self.corpus.parametre['type'] == 0 :
+        #    self.corpus.make_ucis_paras_uces(ucis_paras_txt, make_uce = True)
+        #else :
+        #    self.corpus.make_ucis_paras_uces(ucis_paras_txt, make_uce = False)
+        #del ucis_paras_txt
 
-        if not self.cmd :
-            self.dlg.Update(6, u'Dictionnaires')        
-        uces, orderuces = self.corpus.make_forms_and_uces()
-        self.corpus.ucenb = len(uces)
-        self.corpus.make_lems(self.parent.lexique)
-        self.corpus.min_eff_formes()
-        self.corpus.make_var_actives() 
-        self.corpus.make_var_supp()
+        #if not self.cmd :
+        #    self.dlg.Update(6, u'Dictionnaires')        
+        #uces, orderuces = self.corpus.make_forms_and_uces()
+        #self.corpus.ucenb = len(uces)
+        #self.corpus.make_lems(self.parent.lexique)
+        #self.corpus.min_eff_formes()
+        #self.corpus.make_var_actives() 
+        #self.corpus.make_var_supp()
 
-        if not self.cmd :
-            self.dlg.Update(7, u'Creation des tableaux')
+        #if not self.cmd :
+        #    self.dlg.Update(7, u'Creation des tableaux')
         if self.corpus.parametre['type'] == 0:
             tabuc1 = self.corpus.make_table_with_uce(orderuces)
             uc1 = None
@@ -63,58 +75,61 @@ class AnalysePam:
         self.corpus.write_tab(tabuc1,self.corpus.dictpathout['TableUc1'])
         self.corpus.lenuc1 = len(tabuc1)
         del tabuc1, uc1
-        RPamTxt(self.corpus, self.parent.RscriptsPath)
-        pid = exec_rcode(self.parent.RPath,self.corpus.dictpathout['Rchdtxt'], wait = False)
-        while pid.poll() == None :
-            if not self.cmd :
-                self.dlg.Pulse(u'CHD...')
-                sleep(0.2)
-            else :
-                pass
-        check_Rresult(self.parent, pid)
-        ucecl = self.corpus.read_uce_from_R(self.corpus.dictpathout['uce'])
-        ucecl0 = [cl for uce,cl in ucecl if cl != 0]
-        clnb = len(list(set(ucecl0)))
-        classes = [cl for uce, cl in ucecl]
-        uces1 = [val for val, i in uces1] 
-        self.corpus.make_lc(uces1, classes, clnb)
-        self.corpus.build_profile(clnb, classes, self.corpus.actives, self.corpus.dictpathout['Contout']) 
+        RPamTxt(self, self.parent.RscriptsPath)
+        self.DoR(self.pathout['Rchdtxt'], dlg = self.dlg, message = 'R...')
+        #pid = exec_rcode(self.parent.RPath,self.pathout['Rchdtxt'], wait = False)
+        #while pid.poll() == None :
+        #    if not self.cmd :
+        #        self.dlg.Pulse(u'CHD...')
+        #        sleep(0.2)
+        #    else :
+        #        pass
+        #check_Rresult(self.parent, pid)
+        self.corpus.make_ucecl_from_R(self.pathout['uce'])
+        #ucecl0 = [cl for uce,cl in ucecl if cl != 0]
+        #clnb = len(list(set(ucecl0)))
+        #classes = [cl for uce, cl in ucecl]
+        #uces1 = [val for val, i in uces1] 
+        #self.corpus.make_lc(uces1, classes, clnb)
+        #self.corpus.build_profile(clnb, classes, self.corpus.actives, self.corpus.dictpathout['Contout']) 
+        self.corpus.make_and_write_profile(self.actives, self.corpus.lc, self.pathout['Contout'])
+        self.sup, lim = self.corpus.make_actives_nb(self.parametres['max_actives'], 2)
+        self.corpus.make_and_write_profile(self.sup, self.corpus.lc, self.pathout['ContSupOut'])
+        self.corpus.make_and_write_profile_et(self.corpus.lc, self.pathout['ContEtOut'])
+        self.clnb = len(self.corpus.lc)
+        self.parametres['clnb'] = self.clnb
         
         #passives = [lem for lem in self.corpus.lems if lem not in self.corpus.actives]
-        self.corpus.build_profile(clnb, classes, self.corpus.supp, self.corpus.dictpathout['ContSupOut'])
-        self.corpus.make_etoiles(self.corpus.para_coords)
-        self.corpus.build_profile_et(clnb, classes, uces1, self.corpus.dictpathout['ContEtOut'])
-        AlcesteTxtProf(self.corpus.dictpathout, self.parent.RscriptsPath, clnb, '0.9')
-        pid = exec_rcode(self.parent.RPath, self.corpus.dictpathout['RTxtProfGraph'], wait = False)
-        while pid.poll() == None :
-            if not self.cmd :
-                self.dlg.Pulse(u'AFC...')
-                sleep(0.2)
-            else :
-                pass
-        check_Rresult(self.parent, pid)
-        temps = time() - t1
-        self.corpus.minutes, self.corpus.seconds = divmod(temps, 60)
-        self.corpus.hours, self.corpus.minutes = divmod(self.corpus.minutes, 60)
-        PrintRapport(self.corpus, 'txt')
-        CreateIraFile(self.corpus.dictpathout, clnb, os.path.basename(self.corpus.parametre['filename']))
+        #self.corpus.build_profile(clnb, classes, self.corpus.supp, self.corpus.dictpathout['ContSupOut'])
+        #self.corpus.make_etoiles(self.corpus.para_coords)
+        #self.corpus.build_profile_et(clnb, classes, uces1, self.corpus.dictpathout['ContEtOut'])
+        AlcesteTxtProf(self.pathout, self.parent.RscriptsPath, clnb, '0.9')
+        self.doR(self.pathout['RTxtProfGraph'], dlg = self.dlg, message = 'profils et A.F.C. ...')
+        #pid = exec_rcode(self.parent.RPath, self.corpus.dictpathout['RTxtProfGraph'], wait = False)
+        #while pid.poll() == None :
+        #    if not self.cmd :
+        #        self.dlg.Pulse(u'AFC...')
+        #        sleep(0.2)
+        #    else :
+        #        pass
+        #check_Rresult(self.parent, pid)
+        #temps = time() - t1
+        #self.corpus.minutes, self.corpus.seconds = divmod(temps, 60)
+        #self.corpus.hours, self.corpus.minutes = divmod(self.corpus.minutes, 60)
+        PrintRapport(self, self.corpus, self.parametres)
+        #CreateIraFile(self.corpus.dictpathout, clnb, os.path.basename(self.corpus.parametre['filename']))
         self.corpus.save_corpus(self.corpus.dictpathout['db'])
         afc_graph_list = [[os.path.basename(self.corpus.dictpathout['AFC2DL_OUT']), u'Variables actives - coordonnées - facteurs 1 / 2'],
           [os.path.basename(self.corpus.dictpathout['AFC2DSL_OUT']), u'variables supplémentaires - coordonnées - facteurs 1 / 2'],
           [os.path.basename(self.corpus.dictpathout['AFC2DEL_OUT']), u'Variables illustratives - Coordonnées - facteur 1 / 2'],
-          [os.path.basename(self.corpus.dictpathout['AFC2DCL_OUT']), u'Classes - Coordonnées - facteur 1 / 2'],
-          [os.path.basename(self.corpus.dictpathout['AFC2DCoul']), u'Variables actives - Corrélation - facteur 1/2'],
-          [os.path.basename(self.corpus.dictpathout['AFC2DCoulSup']), u'Variables supplémentaires - Corrélation - facteur 1 / 2'],
-          [os.path.basename(self.corpus.dictpathout['AFC2DCoulEt']), u'Variables illustratives - Corrélations - facteur 1 / 2'],
-          [os.path.basename(self.corpus.dictpathout['AFC2DCoulCl']), u'Classes - Corrélations - facteurs 1 / 2'],]
+          [os.path.basename(self.corpus.dictpathout['AFC2DCL_OUT']), u'Classes - Coordonnées - facteur 1 / 2']]
         chd_graph_list = [[os.path.basename(self.corpus.dictpathout['arbre1']), u'résultats de la classification']]
-        print_liste(self.corpus.dictpathout['liste_graph_afc'],afc_graph_list)
-        print_liste(self.corpus.dictpathout['liste_graph_chd'],chd_graph_list)
-        if not self.cmd :
-            self.dlg.Update(9, u'fin')
-            self.dlg.Destroy()
-            OpenAnalyse(self.parent, self.corpus.dictpathout['ira'])
-        print 'fini'
+        print_liste(self.pathout['liste_graph_afc'],afc_graph_list)
+        print_liste(self.pathout['liste_graph_chd'],chd_graph_list)
+        #if not self.cmd :
+        #    self.dlg.Update(9, u'fin')
+        #    self.dlg.Destroy()
+        #print 'fini'