2 # -*- coding: utf-8 -*-
3 #Author: Pierre Ratinaud
4 #Copyright (c) 2008-2012 Pierre Ratinaud
9 from ConfigParser import ConfigParser
10 from subprocess import Popen, call, PIPE
22 from shutil import copyfile
24 #from dialog import BugDialog
27 log = logging.getLogger('iramuteq')
30 indices_simi = [u'cooccurrence' ,'pourcentage de cooccurrence',u'Russel',u'Jaccard', 'Kulczynski1', 'Kulczynski2', 'Mountford', 'Fager', 'simple matching', 'Hamman', 'Faith', 'Tanimoto', 'Dice', 'Phi', 'Stiles', 'Michael', 'Mozley', 'Yule', 'Yule2', 'Ochiai', 'Simpson', 'Braun-Blanquet','Chi-squared', 'Phi-squared', 'Tschuprow', 'Cramer', 'Pearson', 'binomial']
34 def __init__(self, filein, syscoding = 'utf8') :
36 self.syscoding = syscoding
38 self.openedcorpus = {}
46 d = shelve.open(self.filein)
47 self.history = d.get('history', [])
48 self.ordercorpus = dict([[corpus['uuid'], i] for i, corpus in enumerate(self.history)])
49 self.corpus = dict([[corpus['uuid'], corpus] for i, corpus in enumerate(self.history)])
50 self.analyses = dict([[analyse['uuid'], analyse] for corpus in self.history for analyse in corpus.get('analyses', [])])
54 d = shelve.open(self.filein)
55 d['history'] = self.history
58 def add(self, analyse) :
59 log.info('add to history %s' % analyse.get('corpus_name', 'pas un corpus'))
60 tosave = {'uuid' : analyse['uuid'], 'ira': analyse['ira'], 'type' : analyse['type']}
61 if analyse.get('corpus', False) :
62 if analyse['uuid'] in self.analyses :
64 tosave['corpus'] = analyse['corpus']
65 tosave['name'] = analyse['name']
66 acorpus_uuid = analyse['corpus']
67 if acorpus_uuid in self.ordercorpus :
68 if 'analyses' in self.history[self.ordercorpus[acorpus_uuid]] :
69 self.history[self.ordercorpus[acorpus_uuid]]['analyses'].append(tosave)
71 self.history[self.ordercorpus[acorpus_uuid]]['analyses'] = [tosave]
73 self.orph.append(tosave)
75 tosave['corpus_name'] = analyse['corpus_name']
76 self.history.append(tosave)
80 def delete(self, analyse, corpus = False) :
82 self.history.pop(self.ordercorpus[analyse['uuid']])
84 todel = [i for i, ana in enumerate(self.corpus[analyse['corpus']]['analyses']) if ana['uuid'] == analyse['uuid']][0]
85 self.history[self.ordercorpus[analyse['corpus']]]['analyses'].pop(todel)
89 def addtab(self, analyse) :
90 self.opened[analyse['uuid']] = analyse
92 def rmtab(self, analyse) :
93 del self.opened[analyse['uuid']]
96 return str(self.history)
99 def __init__(self, configfile=None, diff = None, parametres = None) :
100 self.configfile = configfile
101 self.conf = ConfigParser()
102 if configfile is not None :
103 self.conf.readfp(codecs.open(configfile, 'r', 'utf8'))
105 if parametres is not None :
106 self.doparametres(parametres)
108 def doparametres(self, parametres) :
111 def getsections(self) :
112 return self.conf.sections()
114 def getoptions(self, section = None, diff = None):
117 section = self.conf.sections()[0]
118 for option in self.conf.options(section) :
119 if self.conf.get(section, option).isdigit() :
120 parametres[option] = int(self.conf.get(section, option))
121 elif self.conf.get(section, option) == 'False' :
122 parametres[option] = False
123 elif self.conf.get(section, option) == 'True' :
124 parametres[option] = True
125 elif self.conf.get(section, option).startswith('(') and self.conf.get(section, option).endswith(')') :
126 parametres[option] = ast.literal_eval(self.conf.get(section, option))
127 elif self.conf.get(section, option).startswith('[') and self.conf.get(section, option).endswith(']') :
128 parametres[option] = ast.literal_eval(self.conf.get(section, option))
130 parametres[option] = self.conf.get(section, option)
131 if 'type' not in parametres :
132 parametres['type'] = section
135 def makeoptions(self, sections, parametres, outfile = None) :
137 for i, section in enumerate(sections) :
138 txt += '[%s]\n' % section
139 if not self.conf.has_section(section) :
140 self.conf.add_section(section)
141 for option in parametres[i] :
142 if isinstance(parametres[i][option], int) :
143 self.conf.set(section, option, `parametres[i][option]`)
144 txt += '%s = %i\n' % (option, parametres[i][option])
145 elif isinstance(parametres[i][option], basestring) :
146 self.conf.set(section, option, parametres[i][option].encode('utf8'))
147 txt += '%s = %s\n' % (option, parametres[i][option])
148 elif isinstance(parametres[i][option], wx.Colour) :
149 self.conf.set(section, option, str(parametres[i][option]))
150 txt += '%s = %s\n' % (option, str(parametres[i][option]))
151 elif option == 'analyses' :
154 self.conf.set(section, option, `parametres[i][option]`)
155 txt += '%s = %s\n' % (option, `parametres[i][option]`)
157 outfile = self.configfile
158 with codecs.open(outfile, 'w', 'utf8') as f :
162 def totext(self, parametres) :
165 for val in parametres :
166 if isinstance(parametres[val], int) :
167 txt.append(' \t\t: '.join([val, `parametres[val]`]))
168 elif isinstance(parametres[val], basestring) :
169 txt.append(' \t\t: '.join([val, parametres[val]]))
170 elif val in ['listet', 'stars'] :
173 txt.append(' \t\t: '.join([val, `parametres[val]`]))
174 return '\n'.join(txt)
177 def write_tab(tab, fileout) :
178 writer = csv.writer(open(fileout, 'wb'), delimiter=';', quoting = csv.QUOTE_NONNUMERIC)
179 writer.writerows(tab)
181 class BugDialog(wx.Dialog):
182 def __init__(self, *args, **kwds):
183 # begin wxGlade: MyDialog.__init__
184 kwds["style"] = wx.DEFAULT_DIALOG_STYLE
185 kwds["size"] = wx.Size(500, 200)
186 wx.Dialog.__init__(self, *args, **kwds)
187 self.text_ctrl_1 = wx.TextCtrl(self, -1, "", style=wx.TE_MULTILINE)
188 self.button_1 = wx.Button(self, wx.ID_OK, "")
190 self.__set_properties()
194 def __set_properties(self):
195 # begin wxGlade: MyDialog.__set_properties
197 self.SetMinSize(wx.Size(500, 200))
198 self.text_ctrl_1.SetMinSize(wx.Size(500, 200))
202 def __do_layout(self):
203 # begin wxGlade: MyDialog.__do_layout
204 sizer_1 = wx.BoxSizer(wx.VERTICAL)
205 sizer_1.Add(self.text_ctrl_1, 1, wx.EXPAND, 0)
206 sizer_1.Add(self.button_1, 0, wx.ALIGN_CENTER_HORIZONTAL, 0)
207 self.SetSizer(sizer_1)
212 def CreateIraFile(DictPathOut, clusternb, corpname='corpus_name', section = 'analyse'):
213 AnalyseConf = ConfigParser()
214 AnalyseConf.read(DictPathOut['ira'])
215 AnalyseConf.add_section(section)
216 date = datetime.datetime.now().ctime()
217 AnalyseConf.set(section, 'date', str(date))
218 AnalyseConf.set(section, 'clusternb', clusternb)
219 AnalyseConf.set(section, 'corpus_name', corpname)
221 fileout = open(DictPathOut['ira'], 'w')
222 AnalyseConf.write(fileout)
225 def sortedby(list, direct, *indices):
228 sortedby: sort a list of lists (e.g. a table) by one or more indices
229 (columns of the table) and return the sorted list
232 for list = [[2,3],[1,2],[3,1]]:
233 sortedby(list,1) will return [[3, 1], [1, 2], [2, 3]],
234 sortedby(list,0) will return [[1, 2], [2, 3], [3, 1]]
237 nlist = map(lambda x, indices=indices:
238 map(lambda i, x=x: x[i], indices) + [x],
243 nlist.sort(reverse=True)
244 return map(lambda l: l[-1], nlist)
246 def add_type(line, dictlem):
247 if line[4] in dictlem:
248 line.append(dictlem[line[4]])
253 def treat_line_alceste(i, line) :
254 if line[0] == '*' or line[0] == '*****' :
259 elif float(line[5].replace(',', '.')) < 0.0001:
261 elif float(line[5].replace(',', '.')) > 0.05:
262 line[5] = 'NS (%s)' % str(float(line[5].replace(',', '.')))[0:7]
264 line[5] = str(float(line[5].replace(',', '.')))[0:7]
265 return [i, int(line[0]), int(line[1]), float(line[2]), float(line[3]), line[6], line[4], line[5]]
267 def ReadProfileAsDico(File, Alceste=False, encoding = sys.getdefaultencoding()):
268 #print 'lecture des profils : ReadProfileAsDico'
270 # print 'lecture du dictionnaire de type'
272 # for line in parent.corpus.lem_type_list :
273 # dictlem[line[0]] = line[1]
275 print 'lecture des profiles'
276 #encoding = sys.getdefaultencoding()
277 FileReader = codecs.open(File, 'r', encoding)
278 Filecontent = FileReader.readlines()
282 rows = [row.replace('\n', '').replace("'", '').replace('\"', '').replace(',', '.').replace('\r','').split(';') for row in Filecontent]
284 ClusterNb = rows[0][2]
286 clusters = [row[2] for row in rows if row[0] == u'**']
287 valclusters = [row[1:4] for row in rows if row[0] == u'****']
288 lp = [i for i, line in enumerate(rows) if line[0] == u'****']
289 prof = [rows[lp[i] + 1:lp[i+1] - 1] for i in range(0, len(lp)-1)] + [rows[lp[-1] + 1:len(rows)]]
291 prof = [[add_type(row, dictlem) for row in pr] for pr in prof]
292 prof = [[treat_line_alceste(i,line) for i, line in enumerate(pr)] for pr in prof]
294 prof = [[line + [''] for line in pr] for pr in prof]
295 prof = [[treat_line_alceste(i,line) for i, line in enumerate(pr)] for pr in prof]
296 for i, cluster in enumerate(clusters):
297 DictProfile[cluster] = [valclusters[i]] + prof[i]
300 def GetTxtProfile(dictprofile, cluster_size) :
302 for classe in range(0, len(dictprofile)) :
303 prof = dictprofile[str(classe + 1)]
304 clinfo = cluster_size[classe]
305 proflist.append('\n'.join([' '.join(['classe %i' % (classe + 1), '-', '%s uce sur %s - %s%%' % (clinfo[0], clinfo[1], clinfo[2])]), '\n'.join(['%5s|%5s|%6s|%6s|%8s|%8s|%20s\t%10s' % tuple([str(val) for val in line]) for line in prof if len(line)==8])]))
306 return '\n\n'.join(proflist)
308 def formatExceptionInfo(maxTBlevel=5):
309 cla, exc, trbk = sys.exc_info()
310 excName = cla.__name__
312 excArgs = exc.args[0]
314 excArgs = "<no args>"
315 excTb = traceback.format_tb(trbk, maxTBlevel)
316 return (excName, excArgs, excTb)
319 #fonction des etudiants de l'iut
320 def decoupercharact(chaine, longueur, longueurOptimale, separateurs = None) :
322 on part du dernier caractère, et on recule jusqu'au début de la chaîne.
323 Si on trouve un '$', c'est fini.
324 Sinon, on cherche le meilleur candidat. C'est-à-dire le rapport poids/distance le plus important.
326 separateurs = [[u'.', 60.0], [u'?', 60.0], [u'!', 60.0], [u'£$£', 60], [u':', 50.0], [u';', 40.0], [u',', 10.0], [u' ', 0.1]]
327 trouve = False # si on a trouvé un bon séparateur
328 iDecoupe = 0 # indice du caractere ou il faut decouper
330 # on découpe la chaine pour avoir au maximum 240 caractères
331 longueur = min(longueur, len(chaine) - 1)
332 chaineTravail = chaine[:longueur + 1]
334 meilleur = ['', 0, 0] # type, poids et position du meilleur separateur
336 # on vérifie si on ne trouve pas un '$'
337 indice = chaineTravail.find(u'$')
342 # si on ne trouve rien, on cherche le meilleur séparateur
345 caractere = chaineTravail[nbCar]
346 distance = abs(longueurOptimale - nbCar) + 1
347 meilleureDistance = abs(longueurOptimale - meilleur[2]) + 1
349 # on vérifie si le caractére courant est une marque de ponctuation
350 for s in separateurs:
351 if caractere == s[0]:
352 # si c'est une ponctuation
354 if s[1] / distance > float(meilleur[1]) / meilleureDistance:
362 # et on termine la recherche
365 # on passe au caractère précédant
370 fin = chaine[iDecoupe + 1:]
371 retour = chaineTravail[:iDecoupe]
372 return len(retour) > 0, retour.split(), fin
373 # si on a rien trouvé
374 return False, chaine.split(), ''
376 def BugReport(parent, error = None):
377 for ch in parent.GetChildren():
378 if "<class 'wx._windows.ProgressDialog'>" == str(type(ch)):
380 excName, exc, excTb = formatExceptionInfo()
381 if excName == 'Exception' :
384 txt = u' !== BUG ==! \n'
385 txt += u'*************************************\n'
386 txt += '\n'.join(excTb).replace(' ', ' ')
387 txt += excName + '\n'
390 dial = BugDialog(parent)
391 #for line in formatExceptionInfo():
392 # if type(line) == type([]):
394 # txt += don.replace(' ', ' ')
397 if 'Rerror' in dir(parent) :
400 #if error is not None :
401 # txt += '\n%s\n' %error
402 print formatExceptionInfo()
404 dial.text_ctrl_1.write(txt)
405 dial.CenterOnParent()
407 #raise Exception('Bug')
409 def PlaySound(parent):
410 if parent.pref.getboolean('iramuteq', 'sound') :
412 if "gtk2" in wx.PlatformInfo:
413 error = Popen(['aplay','-q',os.path.join(parent.AppliPath,'son_fin.wav')])
415 sound = wx.Sound(os.path.join(parent.AppliPath, 'son_fin.wav'))
416 sound.Play(wx.SOUND_SYNC)
420 def ReadDicoAsDico(dicopath):
421 with codecs.open(dicopath, 'r', 'UTF8') as f:
422 content = f.readlines()
424 for line in content :
426 line = line.replace(u'\n', '').replace('"', '').split('\t')
427 dico[line[0]] = line[1:]
430 def ReadLexique(parent, lang = 'french', filein = None):
433 parent.lexique = ReadDicoAsDico(parent.DictPath.get(lang, 'french'))
435 parent.lexique = ReadDicoAsDico(filein)
439 def ReadList(filein, encoding = sys.getdefaultencoding()):
441 file = codecs.open(filein, 'r', encoding)
442 content = file.readlines()
444 first = content.pop(0)
445 first = first.replace('\n', '').replace('\r','').replace('\"', '').split(';')
449 line = line.replace('\n', '').replace('\r','').replace('\"', '').replace(',', '.')
450 line = line.split(';')
459 don = float('%.5f' % float(val))
465 def exec_rcode(rpath, rcode, wait = True, graph = False):
466 log.info("R Script : %s" % rcode)
468 if sys.platform == 'darwin' :
470 macversion = platform.mac_ver()[0].split('.')
472 if int(macversion[1]) < 5 :
479 rpath = rpath.replace('\\','\\\\')
482 if sys.platform == 'win32':
483 error = call(["%s" % rpath, "--vanilla","--slave","-f", "%s" % rcode])
485 error = call([rpath, '--vanilla','--slave',"-f %s" % rcode])
488 if sys.platform == 'win32':
489 pid = Popen(["%s" % rpath, '--vanilla','--slave','-f', "%s" % rcode])
491 pid = Popen([rpath, '--vanilla','--slave',"-f %s" % rcode], stderr = PIPE)
495 if sys.platform == 'win32':
496 error = call(["%s" % rpath, '--vanilla','--slave','-f', "%s" % rcode])
497 elif sys.platform == 'darwin' and needX11:
498 os.environ['DISPLAY'] = ':0.0'
499 error = call([rpath, '--vanilla','--slave',"-f %s" % rcode])
501 error = call([rpath, '--vanilla','--slave',"-f %s" % rcode])
504 if sys.platform == 'win32':
505 pid = Popen(["%s" % rpath, '--vanilla','--slave','-f', "%s" % rcode])
506 elif sys.platform == 'darwin' and needX11:
507 os.environ['DISPLAY'] = ':0.0'
508 pid = Popen([rpath, '--vanilla','--slave',"-f %s" % rcode], stderr = PIPE)
510 pid = Popen([rpath, '--vanilla','--slave',"-f %s" % rcode], stderr = PIPE)
513 def check_Rresult(parent, pid) :
514 if isinstance(pid, Popen) :
515 if pid.returncode != 0 :
516 error = pid.communicate()
517 error = [str(error[0]), error[1]]
518 if error[1] is None :
520 parent.Rerror = '\n'.join([str(pid.returncode), '\n'.join(error)])
522 raise Exception('\n'.join(u'Erreur R', '\n'.join(error[1:])))
528 raise Exception(u'Erreur R')
532 def print_liste(filename,liste):
533 with open(filename,'w') as f :
535 f.write(';'.join(graph)+'\n')
537 def read_list_file(filename, encoding = sys.getdefaultencoding()):
538 with codecs.open(filename,'rU', encoding) as f :
539 content=f.readlines()
540 ncontent=[line.replace('\n','').split(';') for line in content if line.strip() != '']
543 class MessageImage(wx.Frame):
544 def __init__(self, parent,title, size):
545 wx.Frame.__init__ ( self, parent, id = wx.ID_ANY, title = title, pos = wx.DefaultPosition, size = size, style = wx.DEFAULT_FRAME_STYLE )
546 self.SetSizeHintsSz( wx.DefaultSize, wx.DefaultSize )
547 self.imageFile = False
548 self.imagename = u"chi_classe.png"
549 self.HtmlPage = wx.html.HtmlWindow(self, -1)
550 self.HtmlPage.SetMinSize(size)
551 if "gtk2" in wx.PlatformInfo:
552 self.HtmlPage.SetStandardFonts()
553 self.HtmlPage.SetFonts('Courier', 'Courier')
555 self.button_1 = wx.Button(self, wx.ID_CANCEL)
556 self.button_2 = wx.Button(self, wx.ID_SAVE)
557 self.Bind(wx.EVT_BUTTON, self.OnCloseMe, self.button_1)
558 self.Bind(wx.EVT_BUTTON, self.OnSaveImage, self.button_2)
562 self.sizer_1 = wx.BoxSizer(wx.VERTICAL)
563 self.sizer_2 = wx.BoxSizer(wx.HORIZONTAL)
564 self.sizer_1.Add(self.HtmlPage, 2, wx.EXPAND, 0)
566 self.m_sdbSizer1 = wx.StdDialogButtonSizer()
567 self.m_sdbSizer1.AddButton( self.button_2 )
568 self.m_sdbSizer1.AddButton( self.button_1 )
569 self.m_sdbSizer1.Realize()
570 self.sizer_1.Add(self.m_sdbSizer1, 0, wx.EXPAND, 5)
571 self.SetSizer(self.sizer_1)
573 self.sizer_1.Fit( self )
575 def addsaveimage(self, imageFile) :
576 self.imageFile = imageFile
578 def OnCloseMe(self, event):
581 def OnSaveImage(self, event) :
583 self, message="Enregistrer sous...", defaultDir=os.getcwd(),
584 defaultFile= self.imagename, wildcard="png|*.png", style=wx.SAVE | wx.OVERWRITE_PROMPT
586 dlg.SetFilterIndex(2)
588 if dlg.ShowModal() == wx.ID_OK:
590 copyfile(self.imageFile, path)
593 def progressbar(self, maxi) :
594 if 'parent' in dir(self) :
598 return wx.ProgressDialog("Traitements",
599 "Veuillez patienter...",
602 style=wx.PD_APP_MODAL | wx.PD_AUTO_HIDE | wx.PD_ELAPSED_TIME | wx.PD_CAN_ABORT
606 def treat_var_mod(variables) :
608 for variable in variables :
609 if u'_' in variable :
610 forme = variable.split(u'_')
613 if not var in var_mod :
614 var_mod[var] = [variable]
616 if not mod in var_mod[var] :
617 var_mod[var].append(variable)