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.corpus :
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 addmultiple(self, analyses) :
81 for analyse in analyses :
82 tosave = {'uuid' : analyse['uuid'], 'ira': analyse['ira'], 'type' : analyse['type']}
83 corpus = analyse['uuid']
84 tosave['corpus'] = corpus
85 tosave['name'] = analyse['name']
86 if corpus in self.corpus :
87 if 'analyses' in self.history[self.ordercorpus[corpus]] :
88 self.history[self.ordercorpus[corpus]]['analyses'].append(tosave)
90 self.history[self.ordercorpus[corpus]]['analyses'] = [tosave]
94 def delete(self, analyse, corpus = False) :
96 self.history.pop(self.ordercorpus[analyse['uuid']])
97 if analyse['uuid'] in self.openedcorpus :
98 del self.openedcorpus[analyse['uuid']]
100 todel = [i for i, ana in enumerate(self.corpus[analyse['corpus']]['analyses']) if ana['uuid'] == analyse['uuid']][0]
101 self.history[self.ordercorpus[analyse['corpus']]]['analyses'].pop(todel)
105 def addtab(self, analyse) :
106 self.opened[analyse['uuid']] = analyse
108 def rmtab(self, analyse) :
109 del self.opened[analyse['uuid']]
112 return str(self.history)
115 def __init__(self, configfile=None, diff = None, parametres = None) :
116 self.configfile = configfile
117 self.conf = ConfigParser()
118 if configfile is not None :
119 self.conf.readfp(codecs.open(configfile, 'r', 'utf8'))
121 if parametres is not None :
122 self.doparametres(parametres)
124 def doparametres(self, parametres) :
127 def getsections(self) :
128 return self.conf.sections()
130 def getoptions(self, section = None, diff = None):
133 section = self.conf.sections()[0]
134 for option in self.conf.options(section) :
135 if self.conf.get(section, option).isdigit() :
136 parametres[option] = int(self.conf.get(section, option))
137 elif self.conf.get(section, option) == 'False' :
138 parametres[option] = False
139 elif self.conf.get(section, option) == 'True' :
140 parametres[option] = True
141 elif self.conf.get(section, option).startswith('(') and self.conf.get(section, option).endswith(')') :
142 parametres[option] = ast.literal_eval(self.conf.get(section, option))
143 elif self.conf.get(section, option).startswith('[') and self.conf.get(section, option).endswith(']') :
144 parametres[option] = ast.literal_eval(self.conf.get(section, option))
146 parametres[option] = self.conf.get(section, option)
147 if 'type' not in parametres :
148 parametres['type'] = section
151 def makeoptions(self, sections, parametres, outfile = None) :
153 for i, section in enumerate(sections) :
154 txt += '[%s]\n' % section
155 if not self.conf.has_section(section) :
156 self.conf.add_section(section)
157 for option in parametres[i] :
158 if isinstance(parametres[i][option], int) :
159 self.conf.set(section, option, `parametres[i][option]`)
160 txt += '%s = %i\n' % (option, parametres[i][option])
161 elif isinstance(parametres[i][option], basestring) :
162 self.conf.set(section, option, parametres[i][option].encode('utf8'))
163 txt += '%s = %s\n' % (option, parametres[i][option])
164 elif isinstance(parametres[i][option], wx.Colour) :
165 self.conf.set(section, option, str(parametres[i][option]))
166 txt += '%s = %s\n' % (option, str(parametres[i][option]))
167 elif option == 'analyses' :
170 self.conf.set(section, option, `parametres[i][option]`)
171 txt += '%s = %s\n' % (option, `parametres[i][option]`)
173 outfile = self.configfile
174 with codecs.open(outfile, 'w', 'utf8') as f :
178 def totext(self, parametres) :
181 for val in parametres :
182 if isinstance(parametres[val], int) :
183 txt.append(' \t\t: '.join([val, `parametres[val]`]))
184 elif isinstance(parametres[val], basestring) :
185 txt.append(' \t\t: '.join([val, parametres[val]]))
186 elif val in ['listet', 'stars'] :
189 txt.append(' \t\t: '.join([val, `parametres[val]`]))
190 return '\n'.join(txt)
193 def write_tab(tab, fileout) :
194 writer = csv.writer(open(fileout, 'wb'), delimiter=';', quoting = csv.QUOTE_NONNUMERIC)
195 writer.writerows(tab)
197 class BugDialog(wx.Dialog):
198 def __init__(self, *args, **kwds):
199 # begin wxGlade: MyDialog.__init__
200 kwds["style"] = wx.DEFAULT_DIALOG_STYLE
201 kwds["size"] = wx.Size(500, 200)
202 wx.Dialog.__init__(self, *args, **kwds)
203 self.text_ctrl_1 = wx.TextCtrl(self, -1, "", style=wx.TE_MULTILINE)
204 self.button_1 = wx.Button(self, wx.ID_OK, "")
206 self.__set_properties()
210 def __set_properties(self):
211 # begin wxGlade: MyDialog.__set_properties
213 self.SetMinSize(wx.Size(500, 200))
214 self.text_ctrl_1.SetMinSize(wx.Size(500, 200))
218 def __do_layout(self):
219 # begin wxGlade: MyDialog.__do_layout
220 sizer_1 = wx.BoxSizer(wx.VERTICAL)
221 sizer_1.Add(self.text_ctrl_1, 1, wx.EXPAND, 0)
222 sizer_1.Add(self.button_1, 0, wx.ALIGN_CENTER_HORIZONTAL, 0)
223 self.SetSizer(sizer_1)
228 def CreateIraFile(DictPathOut, clusternb, corpname='corpus_name', section = 'analyse'):
229 AnalyseConf = ConfigParser()
230 AnalyseConf.read(DictPathOut['ira'])
231 AnalyseConf.add_section(section)
232 date = datetime.datetime.now().ctime()
233 AnalyseConf.set(section, 'date', str(date))
234 AnalyseConf.set(section, 'clusternb', clusternb)
235 AnalyseConf.set(section, 'corpus_name', corpname)
237 fileout = open(DictPathOut['ira'], 'w')
238 AnalyseConf.write(fileout)
241 def sortedby(list, direct, *indices):
244 sortedby: sort a list of lists (e.g. a table) by one or more indices
245 (columns of the table) and return the sorted list
248 for list = [[2,3],[1,2],[3,1]]:
249 sortedby(list,1) will return [[3, 1], [1, 2], [2, 3]],
250 sortedby(list,0) will return [[1, 2], [2, 3], [3, 1]]
253 nlist = map(lambda x, indices=indices:
254 map(lambda i, x=x: x[i], indices) + [x],
259 nlist.sort(reverse=True)
260 return map(lambda l: l[-1], nlist)
262 def add_type(line, dictlem):
263 if line[4] in dictlem:
264 line.append(dictlem[line[4]])
269 def treat_line_alceste(i, line) :
270 if line[0] == '*' or line[0] == '*****' :
275 elif float(line[5].replace(',', '.')) < 0.0001:
277 elif float(line[5].replace(',', '.')) > 0.05:
278 line[5] = 'NS (%s)' % str(float(line[5].replace(',', '.')))[0:7]
280 line[5] = str(float(line[5].replace(',', '.')))[0:7]
281 return [i, int(line[0]), int(line[1]), float(line[2]), float(line[3]), line[6], line[4], line[5]]
283 def ReadProfileAsDico(File, Alceste=False, encoding = sys.getdefaultencoding()):
284 #print 'lecture des profils : ReadProfileAsDico'
286 # print 'lecture du dictionnaire de type'
288 # for line in parent.corpus.lem_type_list :
289 # dictlem[line[0]] = line[1]
291 print 'lecture des profiles'
292 #encoding = sys.getdefaultencoding()
293 FileReader = codecs.open(File, 'r', encoding)
294 Filecontent = FileReader.readlines()
298 rows = [row.replace('\n', '').replace("'", '').replace('\"', '').replace(',', '.').replace('\r','').split(';') for row in Filecontent]
300 ClusterNb = rows[0][2]
302 clusters = [row[2] for row in rows if row[0] == u'**']
303 valclusters = [row[1:4] for row in rows if row[0] == u'****']
304 lp = [i for i, line in enumerate(rows) if line[0] == u'****']
305 prof = [rows[lp[i] + 1:lp[i+1] - 1] for i in range(0, len(lp)-1)] + [rows[lp[-1] + 1:len(rows)]]
307 prof = [[add_type(row, dictlem) for row in pr] for pr in prof]
308 prof = [[treat_line_alceste(i,line) for i, line in enumerate(pr)] for pr in prof]
310 prof = [[line + [''] for line in pr] for pr in prof]
311 prof = [[treat_line_alceste(i,line) for i, line in enumerate(pr)] for pr in prof]
312 for i, cluster in enumerate(clusters):
313 DictProfile[cluster] = [valclusters[i]] + prof[i]
316 def GetTxtProfile(dictprofile, cluster_size) :
318 for classe in range(0, len(dictprofile)) :
319 prof = dictprofile[str(classe + 1)]
320 clinfo = cluster_size[classe]
321 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])]))
322 return '\n\n'.join(proflist)
324 def formatExceptionInfo(maxTBlevel=5):
325 cla, exc, trbk = sys.exc_info()
326 excName = cla.__name__
328 excArgs = exc.args[0]
330 excArgs = "<no args>"
331 excTb = traceback.format_tb(trbk, maxTBlevel)
332 return (excName, excArgs, excTb)
335 #fonction des etudiants de l'iut
336 def decoupercharact(chaine, longueur, longueurOptimale, separateurs = None) :
338 on part du dernier caractère, et on recule jusqu'au début de la chaîne.
339 Si on trouve un '$', c'est fini.
340 Sinon, on cherche le meilleur candidat. C'est-Ã -dire le rapport poids/distance le plus important.
342 separateurs = [[u'.', 60.0], [u'?', 60.0], [u'!', 60.0], [u'£$£', 60], [u':', 50.0], [u';', 40.0], [u',', 10.0], [u' ', 0.1]]
343 trouve = False # si on a trouvé un bon séparateur
344 iDecoupe = 0 # indice du caractere ou il faut decouper
346 # on découpe la chaine pour avoir au maximum 240 caractères
347 longueur = min(longueur, len(chaine) - 1)
348 chaineTravail = chaine[:longueur + 1]
350 meilleur = ['', 0, 0] # type, poids et position du meilleur separateur
352 # on vérifie si on ne trouve pas un '$'
353 indice = chaineTravail.find(u'$')
358 # si on ne trouve rien, on cherche le meilleur séparateur
361 caractere = chaineTravail[nbCar]
362 distance = abs(longueurOptimale - nbCar) + 1
363 meilleureDistance = abs(longueurOptimale - meilleur[2]) + 1
365 # on vérifie si le caractére courant est une marque de ponctuation
366 for s in separateurs:
367 if caractere == s[0]:
368 # si c'est une ponctuation
370 if s[1] / distance > float(meilleur[1]) / meilleureDistance:
378 # et on termine la recherche
381 # on passe au caractère précédant
386 fin = chaine[iDecoupe + 1:]
387 retour = chaineTravail[:iDecoupe]
388 return len(retour) > 0, retour.split(), fin
389 # si on a rien trouvé
390 return False, chaine.split(), ''
392 def BugReport(parent, error = None):
393 for ch in parent.GetChildren():
394 if "<class 'wx._windows.ProgressDialog'>" == str(type(ch)):
396 excName, exc, excTb = formatExceptionInfo()
397 if excName == 'Exception' :
400 txt = u' !== BUG ==! \n'
401 txt += u'*************************************\n'
402 txt += '\n'.join(excTb).replace(' ', ' ')
403 txt += excName + '\n'
406 dial = BugDialog(parent)
407 if 'Rerror' in dir(parent) :
411 dial.text_ctrl_1.write(txt)
412 dial.CenterOnParent()
416 def PlaySound(parent):
417 if parent.pref.getboolean('iramuteq', 'sound') :
419 if "gtk2" in wx.PlatformInfo:
420 error = Popen(['aplay','-q',os.path.join(parent.AppliPath,'son_fin.wav')])
422 sound = wx.Sound(os.path.join(parent.AppliPath, 'son_fin.wav'))
423 sound.Play(wx.SOUND_SYNC)
427 def ReadDicoAsDico(dicopath):
428 with codecs.open(dicopath, 'r', 'UTF8') as f:
429 content = f.readlines()
431 for line in content :
433 line = line.replace(u'\n', '').replace('"', '').split('\t')
434 dico[line[0]] = line[1:]
437 def ReadLexique(parent, lang = 'french', filein = None):
440 parent.lexique = ReadDicoAsDico(parent.DictPath.get(lang, 'french'))
442 parent.lexique = ReadDicoAsDico(filein)
446 def ReadList(filein, encoding = sys.getdefaultencoding()):
448 file = codecs.open(filein, 'r', encoding)
449 content = file.readlines()
451 first = content.pop(0)
452 first = first.replace('\n', '').replace('\r','').replace('\"', '').split(';')
456 line = line.replace('\n', '').replace('\r','').replace('\"', '').replace(',', '.')
457 line = line.split(';')
466 don = float('%.5f' % float(val))
472 def exec_RCMD(rpath, command) :
473 log.info('R CMD INSTALL %s' % command)
474 rpath = rpath.replace('\\','\\\\')
475 error = call(["%s" % rpath, 'CMD', 'INSTALL', "%s" % command])
478 def exec_rcode(rpath, rcode, wait = True, graph = False):
479 log.info("R Script : %s" % rcode)
481 if sys.platform == 'darwin' :
483 macversion = platform.mac_ver()[0].split('.')
485 if int(macversion[1]) < 5 :
492 rpath = rpath.replace('\\','\\\\')
495 if sys.platform == 'win32':
496 error = call(["%s" % rpath, "--vanilla","--slave","-f", "%s" % rcode])
498 error = call([rpath, '--vanilla','--slave',"-f %s" % rcode])
501 if sys.platform == 'win32':
502 pid = Popen(["%s" % rpath, '--vanilla','--slave','-f', "%s" % rcode])
504 pid = Popen([rpath, '--vanilla','--slave',"-f %s" % rcode], stderr = PIPE)
508 if sys.platform == 'win32':
509 error = call(["%s" % rpath, '--vanilla','--slave','-f', "%s" % rcode])
510 elif sys.platform == 'darwin' and needX11:
511 os.environ['DISPLAY'] = ':0.0'
512 error = call([rpath, '--vanilla','--slave',"-f %s" % rcode])
514 error = call([rpath, '--vanilla','--slave',"-f %s" % rcode])
517 if sys.platform == 'win32':
518 pid = Popen(["%s" % rpath, '--vanilla','--slave','-f', "%s" % rcode])
519 elif sys.platform == 'darwin' and needX11:
520 os.environ['DISPLAY'] = ':0.0'
521 pid = Popen([rpath, '--vanilla','--slave',"-f %s" % rcode], stderr = PIPE)
523 pid = Popen([rpath, '--vanilla','--slave',"-f %s" % rcode], stderr = PIPE)
526 def check_Rresult(parent, pid) :
527 if isinstance(pid, Popen) :
528 if pid.returncode != 0 :
529 error = pid.communicate()
531 error = [str(error[0]), error[1]]
532 if error[1] is None :
534 parent.Rerror = '\n'.join([str(pid.returncode), '\n'.join(error)])
536 raise Exception('\n'.join([u'Erreur R', '\n'.join(error[1:])]))
545 raise Exception(u'Erreur R')
552 def print_liste(filename,liste):
553 with open(filename,'w') as f :
555 f.write(';'.join(graph)+'\n')
557 def read_list_file(filename, encoding = sys.getdefaultencoding()):
558 with codecs.open(filename,'rU', encoding) as f :
559 content=f.readlines()
560 ncontent=[line.replace('\n','').split(';') for line in content if line.strip() != '']
563 class MessageImage(wx.Frame):
564 def __init__(self, parent,title, size):
565 wx.Frame.__init__ ( self, parent, id = wx.ID_ANY, title = title, pos = wx.DefaultPosition, size = size, style = wx.DEFAULT_FRAME_STYLE )
566 self.SetSizeHintsSz( wx.DefaultSize, wx.DefaultSize )
567 self.imageFile = False
568 self.imagename = u"chi_classe.png"
569 self.HtmlPage = wx.html.HtmlWindow(self, -1)
570 self.HtmlPage.SetMinSize(size)
571 if "gtk2" in wx.PlatformInfo:
572 self.HtmlPage.SetStandardFonts()
573 self.HtmlPage.SetFonts('Courier', 'Courier')
575 self.button_1 = wx.Button(self, wx.ID_CANCEL)
576 self.button_2 = wx.Button(self, wx.ID_SAVE)
577 self.Bind(wx.EVT_BUTTON, self.OnCloseMe, self.button_1)
578 self.Bind(wx.EVT_BUTTON, self.OnSaveImage, self.button_2)
582 self.sizer_1 = wx.BoxSizer(wx.VERTICAL)
583 self.sizer_2 = wx.BoxSizer(wx.HORIZONTAL)
584 self.sizer_1.Add(self.HtmlPage, 2, wx.EXPAND, 0)
586 self.m_sdbSizer1 = wx.StdDialogButtonSizer()
587 self.m_sdbSizer1.AddButton( self.button_2 )
588 self.m_sdbSizer1.AddButton( self.button_1 )
589 self.m_sdbSizer1.Realize()
590 self.sizer_1.Add(self.m_sdbSizer1, 0, wx.EXPAND, 5)
591 self.SetSizer(self.sizer_1)
593 self.sizer_1.Fit( self )
595 def addsaveimage(self, imageFile) :
596 self.imageFile = imageFile
598 def OnCloseMe(self, event):
601 def OnSaveImage(self, event) :
603 self, message="Enregistrer sous...", defaultDir=os.getcwd(),
604 defaultFile= self.imagename, wildcard="png|*.png", style=wx.SAVE | wx.OVERWRITE_PROMPT
606 dlg.SetFilterIndex(2)
608 if dlg.ShowModal() == wx.ID_OK:
610 copyfile(self.imageFile, path)
613 def progressbar(self, maxi) :
614 if 'parent' in dir(self) :
618 return wx.ProgressDialog("Traitements",
619 "Veuillez patienter...",
622 style=wx.PD_APP_MODAL | wx.PD_AUTO_HIDE | wx.PD_ELAPSED_TIME | wx.PD_CAN_ABORT
626 def treat_var_mod(variables) :
628 for variable in variables :
629 if u'_' in variable :
630 forme = variable.split(u'_')
633 if not var in var_mod :
634 var_mod[var] = [variable]
636 if not mod in var_mod[var] :
637 var_mod[var].append(variable)