X-Git-Url: http://iramuteq.org/git?p=iramuteq;a=blobdiff_plain;f=analysetxt.py;h=95fcfa2df03c2a250d235c828f7fc1f03422e404;hp=f5d927350dfea9d661f575c3a2738138ecec643c;hb=e84160b7f61eb5b05cc12339e44a61d67b499e15;hpb=8fa853a25a9d62b1446e1bc543e5a3a4d0e03dcf

diff --git a/analysetxt.py b/analysetxt.py
index f5d9273..95fcfa2 100644
--- a/analysetxt.py
+++ b/analysetxt.py
@@ -1,91 +1,80 @@
 # -*- coding: utf-8 -*-
-#Author: Pierre Ratinaud
-#from corpusNG import Corpus
+# Author: Pierre Ratinaud
+# lisence : GNU GPL
+# copyright : 2012-2013 (c) Pierre Ratinaud
+
 import logging
-from chemins import PathOut, ChdTxtPathOut
-from functions import exec_rcode, check_Rresult, DoConf, print_liste
+from chemins import PathOut
+from functions import exec_rcode, check_Rresult, DoConf, ReadDicoAsDico
+from shutil import copy
 from time import time, sleep
 from uuid import uuid4
 import os
-#ALCESTE
-from PrintRScript import RchdTxt, AlcesteTxtProf
-from OptionAlceste import OptionAlc 
-from layout import PrintRapport
 from openanalyse import OpenAnalyse
-from time import time
-######################################
-print '#######LOGGING TEST###########'
-log = logging.getLogger('iramuteq.analyse')
-#formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
-#ch = logging.StreamHandler()
-#ch.setFormatter(formatter)
-#log.addHandler(ch)
-#log.setLevel(logging.INFO)
-#######################################
-
-#def make_ucecl_from_R(filein) :
-#    with open(filein, 'rU') as f :
-#        c = f.readlines()
-#    c.pop(0)
-#    ucecl = []
-#    for line in c :
-#        line = line.replace('\n', '').replace('"', '').split(';')
-#        ucecl.append([int(line[0]) - 1, int(line[1])])
-#    classesl = [val[1] for val in ucecl]
-#    clnb = max(classesl)
-#    ucecl = sorted(ucecl, key=itemgetter(1))
-#    ucecl = [[uce[0] for uce in ucecl if uce[1] == i] for i in range(clnb+1)]
-#    return ucecl
-#
-#def make_lc(self, uces, classes, clnb) :    
-#    self.lc = [[] for classe in range(0,clnb)]
-#    for i in range(0,clnb):
-#        self.lc[i] = [uce for j, uce in enumerate(uces) if i+1 == classes[j]]
-#    self.lc0 = [uce for j, uce in enumerate(uces) if 0 == classes[j]]
+from dialog import StatDialog
 
+log = logging.getLogger('iramuteq.analyse')
 
 class AnalyseText :
-    def __init__(self, ira, corpus, parametres = None, dlg = False) :
+    def __init__(self, ira, corpus, parametres=None, dlg=False, lemdial=True) :
         self.corpus = corpus
         self.ira = ira
         self.parent = ira
         self.dlg = dlg
         self.dialok = True
         self.parametres = parametres
-        self.pathout = PathOut(corpus.parametres['originalpath'], analyse_type = parametres['type'], dirout = corpus.parametres['pathout'])
-        self.parametres = self.make_config(parametres)
+        self.lemdial = lemdial
+        self.val = False
+        self.keys = DoConf(self.ira.ConfigPath['key']).getoptions()
+        if not 'pathout' in self.parametres :
+            self.pathout = PathOut(corpus.parametres['originalpath'], analyse_type=parametres['type'], dirout=corpus.parametres['pathout'])
+        else :
+            self.pathout = PathOut(filename=corpus.parametres['originalpath'], dirout=self.parametres['pathout'], analyse_type=self.parametres['type'])
+        self.parametres = self.lemparam()
+        if self.parametres is not None :
+            self.parametres = self.make_config(parametres)
         log.info(self.pathout.dirout)
         if self.parametres is not None :
             self.keys = DoConf(self.ira.ConfigPath['key']).getoptions()
-            gramact = [k for k in keys if keys[k] == 1]
-            gramsup = [k for k in keys if keys[k] == 2]
-            #FIXME
-            if not 'lem' in self.parametres :
-                self.parametres['lem'] = 1
+            gramact = [k for k in self.keys if self.keys[k] == 1]
+            gramsup = [k for k in self.keys if self.keys[k] == 2]
             self.parametres['pathout'] = self.pathout.mkdirout()
-            self.pathout = PathOut(dirout = self.parametres['pathout'])
+            self.pathout = PathOut(dirout=self.parametres['pathout'])
             self.pathout.createdir(self.parametres['pathout'])
             self.parametres['corpus'] = self.corpus.parametres['uuid']
             self.parametres['uuid'] = str(uuid4())
             self.parametres['name'] = os.path.split(self.parametres['pathout'])[1]
             self.parametres['type'] = parametres['type']
+            self.parametres['encoding'] = self.ira.syscoding
             self.t1 = time()
-            #if self.corpus.lems is None :
-            self.corpus.make_lems(lem = self.parametres['lem'])
-            corpus.parse_active(gramact, gramsup)
-            self.doanalyse()
-            self.time = time() - self.t1
-            minutes, seconds = divmod(self.time, 60)
-            hours, minutes = divmod(minutes, 60)            
-            self.parametres['time'] = '%.0fh %.0fm %.0fs' % (hours, minutes, seconds)
-            self.parametres['ira'] = self.pathout['ira']
-            DoConf().makeoptions([self.parametres['type']], [self.parametres], self.pathout['ira'])
-            self.ira.history.add(self.parametres)
-            if dlg :
-                dlg.Destroy()
-                OpenAnalyse(self.parent, self.parametres['ira'])
-                self.ira.tree.AddAnalyse(self.parametres)
-                self.val = 5100
+            if not self.parametres.get('dictionary', False) :
+                self.corpus.make_lems(lem=self.parametres['lem'])
+            else :
+                dico = ReadDicoAsDico(self.parametres['dictionary'])
+                self.corpus.make_lems_from_dict(dico, dolem=self.parametres['lem'])
+                dictname = os.path.basename(self.parametres['dictionary'])
+                dictpath = os.path.join(self.pathout.dirout, dictname)
+                copy(self.parametres['dictionary'], dictpath)
+                self.parametres['dictionary'] = dictpath
+            self.corpus.parse_active(gramact, gramsup)
+            result_analyse = self.doanalyse()
+            if result_analyse is None :
+                self.time = time() - self.t1
+                minutes, seconds = divmod(self.time, 60)
+                hours, minutes = divmod(minutes, 60)            
+                self.parametres['time'] = '%.0fh %.0fm %.0fs' % (hours, minutes, seconds)
+                self.parametres['ira'] = self.pathout['Analyse.ira']
+                DoConf().makeoptions([self.parametres['type']], [self.parametres], self.pathout['Analyse.ira'])
+                self.ira.history.add(self.parametres)
+                if dlg :
+                    dlg.Destroy()
+                    OpenAnalyse(self.parent, self.parametres['ira'])
+                    self.ira.tree.AddAnalyse(self.parametres)
+                    self.val = 5100
+            else :
+                self.val = False
+                if dlg :
+                    dlg.Destroy()
         else :
             if dlg :
                 dlg.Destroy()
@@ -94,165 +83,89 @@ class AnalyseText :
     def doanalyse(self) :
         pass
 
+    def lemparam(self) :
+        if self.dlg and self.lemdial:
+            dial = StatDialog(self.parent, self.keys)
+            dial.CenterOnParent()
+            val = dial.ShowModal()
+            if val == 5100 :
+                if dial.radio_lem.GetSelection() == 0 :
+                    lem = 1
+                else :
+                    lem = 0            
+                self.parametres['lem'] = lem
+                if dial.radio_dictchoice.GetSelection() == 1 :
+                    self.parametres['dictionary'] = dial.dictpath.GetValue() 
+                dial.Destroy()
+                return self.parametres
+            else :
+                dial.Destroy()
+                return None        
+        else :
+            return self.parametres
+
     def make_config(self, config) :
         if config is not None :
-            if isinstance(config, basestring) : 
-                return self.readconfig(config)
+            if not self.dlg : 
+                return config
             else :
                 return self.preferences()
+        else :
+            return None
 
     def readconfig(self, config) :
         return config
 
     def preferences(self) :
-        return {}
-
-    def doR(self):
-        pass
+        return self.parametres
 
     def printRscript(self) :
         pass
 
-    def doR(self, Rscript, wait = False, dlg = None, message = '') :
+    def doR(self, Rscript, wait=False, dlg=None, message='') :
         log.info('R code...')
-        pid = exec_rcode(self.ira.RPath, Rscript, wait = wait)
+        pid = exec_rcode(self.ira.RPath, Rscript, wait=wait)
         while pid.poll() is None :
-            if dlg is not None :
+            if dlg :
                 self.dlg.Pulse(message)
                 sleep(0.2)
             else :
                 sleep(0.2)
-        check_Rresult(self.ira, pid)
-
-
-
-class Alceste(AnalyseText) :
-    def doanalyse(self) :
-        #self.pathout = PathOut(self.corpus.parametres['filename'], 'alceste')
-        self.parametres['type'] = 'alceste'
-        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 :
-            lenuc1, lenuc2 = self.corpus.make_and_write_sparse_matrix_from_uc(self.actives, self.parametres['tailleuc1'], self.parametres['tailleuc2'], self.pathout['TableUc1'], self.pathout['TableUc2'], self.pathout['listeuce1'], self.pathout['listeuce2'])
-            self.parametres['lenuc1'] = lenuc1
-            self.parametres['lenuc2'] = lenuc2
-        elif self.parametres['classif_mode'] == 1 :
-            self.corpus.make_and_write_sparse_matrix_from_uces(self.actives, self.pathout['TableUc1'], self.pathout['listeuce1'])
-        elif self.parametres['classif_mode'] == 2 :
-            self.corpus.make_and_write_sparse_matrix_from_uci(self.actives, self.pathout['TableUc1'], self.pathout['listeuce1'])
-        Rscript = self.printRscript()
-        self.doR(Rscript)
-        #self.lc = make_ucecl_from_R(self.pathout['uce'])
-        #self.lc0 = self.lc.pop(0)
-        self.corpus.make_ucecl_from_R(self.pathout['uce'])
-        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
-        Rscript = self.printRscript2()
-        self.doR(Rscript)
-        self.time = time() - self.t1
-        minutes, seconds = divmod(self.time, 60)
-        hours, minutes = divmod(minutes, 60)            
-        self.parametres['time'] = '%.0fh %.0fm %.0fs' % (hours, minutes, seconds)
-        self.print_graph_files()
-
-    def preferences(self) :
-        parametres = DoConf(self.parent.ConfigPath['alceste']).getoptions('ALCESTE')
-        parametres['corpus'] = self.corpus
-        parametres['pathout'] = self.pathout
-        self.dial = OptionAlc(self.parent, parametres)
-        self.dial.CenterOnParent()
-        self.dialok = self.dial.ShowModal()
-        if self.dialok == 5100 :
-            if self.dial.radio_1.GetSelection() == 0 :
-                lem = 1
-            else :
-                lem = 0
-            parametres['lem'] = lem
-            parametres['classif_mode'] = self.dial.radio_box_2.GetSelection()
-            parametres['tailleuc1'] = self.dial.spin_ctrl_1.GetValue()
-            parametres['tailleuc2'] = self.dial.spin_ctrl_2.GetValue()
-            parametres['mincl'] = self.dial.spin_ctrl_4.GetValue()
-            parametres['minforme'] = self.dial.spin_ctrl_5.GetValue()
-            parametres['nbcl_p1'] = self.dial.spin_nbcl.GetValue()
-            parametres['max_actives'] = self.dial.spin_max_actives.GetValue()
-            parametres['corpus'] = ''
-            parametres['pathout'] = self.pathout.dirout
-            for val in parametres :
-                print val, parametres[val]
-            DoConf(self.parent.ConfigPath['alceste']).makeoptions(['ALCESTE'], [parametres])
-            self.dial.Destroy()
-            return parametres
-        else :
-            self.dial.Destroy()
-            return None
-
-    def printRscript(self) :
-        RchdTxt(self.pathout, self.parent.RscriptsPath, self.parametres['mincl'], self.parametres['classif_mode'], nbt = self.parametres['nbcl_p1'] - 1, libsvdc = self.parent.pref.getboolean('iramuteq','libsvdc'), libsvdc_path = self.parent.pref.get('iramuteq','libsvdc_path'), R_max_mem = False)
-        return self.pathout['Rchdtxt']
-
-    def printRscript2(self) :
-        AlcesteTxtProf(self.pathout, self.parent.RscriptsPath, self.clnb, 0.9)
-        return self.pathout['RTxtProfGraph']
-
-    def print_graph_files(self) :
-        afc_graph_list = [[os.path.basename(self.pathout['AFC2DL_OUT']), u'Variables actives - coordonnées - facteurs 1 / 2'],
-                      [os.path.basename(self.pathout['AFC2DSL_OUT']), u'variables supplémentaires - coordonnées - facteurs 1 / 2'],
-                      [os.path.basename(self.pathout['AFC2DEL_OUT']), u'Variables illustratives - Coordonnées - facteur 1 / 2'],
-                      [os.path.basename(self.pathout['AFC2DCL_OUT']), u'Classes - Coordonnées - facteur 1 / 2'],
-                      [os.path.basename(self.pathout['AFC2DCoul']), u'Variables actives - Corrélation - facteur 1 / 2'],
-                      [os.path.basename(self.pathout['AFC2DCoulSup']), u'Variables supplémentaires - Corrélation - facteur 1 / 2'],
-                      [os.path.basename(self.pathout['AFC2DCoulEt']), u'Variables illustratives - Corrélations - facteur 1 / 2'],
-                      [os.path.basename(self.pathout['AFC2DCoulCl']), u'Classes - Corrélations - facteurs 1 / 2'],]
-        chd_graph_list = [[os.path.basename(self.pathout['dendro1']), u'dendrogramme à partir de chd1']]
-        if self.parametres['classif_mode'] == 0 :
-            chd_graph_list.append([os.path.basename(self.pathout['dendro2']), u'dendrogramme à partir de chd2'])
-        chd_graph_list.append([os.path.basename(self.pathout['arbre1']), u'chd1'])
-        if self.parametres['classif_mode'] == 0 :
-            chd_graph_list.append([os.path.basename(self.pathout['arbre2']), u'chd2'])       
-        print_liste(self.pathout['liste_graph_afc'],afc_graph_list)
-        print_liste(self.pathout['liste_graph_chd'],chd_graph_list)
-        PrintRapport(self, self.corpus, self.parametres)
-
-
-keys = {'art_def' : 2,
-        'pre' : 2,
-        'adj_dem' : 2,
-        'ono' : 2,
-        'pro_per' : 2,
-        'ver_sup' : 2,
-        'adv' : 1,
-        'ver' : 1,
-        'adj_ind' : 2,
-        'adj_pos' : 2,
-        'aux' : 2,
-        'adj_int' : 2,
-        'pro_ind' : 2,
-        'adj' : 1,
-        'pro_dem' : 2,
-        'nom' : 1,
-        'art_ind' : 2,
-        'pro_pos' : 2,
-        'nom_sup' : 2,
-        'adv_sup' : 2,
-        'adj_sup' : 2,
-        'adj_num' : 2,
-        'pro_rel' : 2,
-        'con' : 2,
-        'num' : 2,
-        'nr' : 1,
-        'sw' : 2,
-}
-
-gramact = [k for k in keys if keys[k] == 1]
-gramsup = [k for k in keys if keys[k] == 2]
-
-#corpus = Corpus('', {'filename': '/home/pierre/workspace/iramuteq/dev/testcorpus.txt','formesdb':'formes.db', 'ucesdb': 'uces.db', 'corpusdb' : 'corpus.db', 'syscoding' : 'utf-8'})
-#corpus.read_corpus()
-#corpus.parse_active(gramact, gramsup)
-#Alceste(corpus).doanalyse()
+        return check_Rresult(self.ira, pid)
+
+
+
+
+
+
+# keys = {'art_def' : 2,
+#        'pre' : 2,
+#        'adj_dem' : 2,
+#        'ono' : 2,
+#        'pro_per' : 2,
+#        'ver_sup' : 2,
+#        'adv' : 1,
+#        'ver' : 1,
+#        'adj_ind' : 2,
+#        'adj_pos' : 2,
+#        'aux' : 2,
+#        'adj_int' : 2,
+#        'pro_ind' : 2,
+#        'adj' : 1,
+#        'pro_dem' : 2,
+#        'nom' : 1,
+#        'art_ind' : 2,
+#        'pro_pos' : 2,
+#        'nom_sup' : 2,
+#        'adv_sup' : 2,
+#        'adj_sup' : 2,
+#        'adj_num' : 2,
+#        'pro_rel' : 2,
+#        'con' : 2,
+#        'num' : 2,
+#        'nr' : 1,
+#        'sw' : 2,
+# }
+#
+# gramact = [k for k in keys if keys[k] == 1]
+# gramsup = [k for k in keys if keys[k] == 2]