X-Git-Url: http://iramuteq.org/git?p=iramuteq;a=blobdiff_plain;f=textsimi.py;h=efd3a67f3201f75f92ece5a2e2204a9e87ad9ced;hp=1340276e9caf9507e420192515b5c4b2caf58b22;hb=6eb30f429cf7aa6d7e84e7b64b8b64afbeaf3435;hpb=22cd27b2bbe9ab1ffa7ef06fa764b5147ae17dad diff --git a/textsimi.py b/textsimi.py index 1340276..efd3a67 100644 --- a/textsimi.py +++ b/textsimi.py @@ -6,24 +6,50 @@ from chemins import ConstructPathOut, construct_simipath from corpus import Corpus import os +from analysetxt import AnalyseText from ConfigParser import RawConfigParser from guifunct import getPage, getCorpus from dialog import StatDialog from functions import indices_simi, progressbar, treat_var_mod from tableau import Tableau from tabsimi import DoSimi +from PrintRScript import PrintRScript import wx from copy import copy -class SimiTxt : - def __init__(self, parent, cmd = False, param = None): - self.parent = parent - self.cmd = cmd - self.ConfigPath = parent.ConfigPath - self.DictPath = parent.DictPath - self.KeyConf = RawConfigParser() - self.KeyConf.read(self.ConfigPath['key']) +import logging + +logger = logging.getLogger('iramuteq.textsimi') + + + +class SimiTxt(AnalyseText): + def doanalyse(self) : self.indices = indices_simi + self.makesimiparam() + self.makefiles() + prep = PrepSimi(self.ira, self.parametres, indices_simi) + self.parametres = prep.parametres + script = PrintSimScript(self) + + + def preferences(self) : + dial = StatDialog(self, self.parent) + dial.CenterOnParent() + val = dial.ShowModal() + if val == 5100 : + if dial.radio_lem.GetSelection() == 0 : + lem = 1 + else : + lem = 0 + self.parametres['lem'] = lem + dial.Destroy() + return self.parametres + else : + dial.Destroy() + return None + + def makesimiparam(self) : self.paramsimi = {'coeff' : 0, 'layout' : 2, 'type' : 1, @@ -54,117 +80,313 @@ class SimiTxt : 'keep_coord' : True, 'alpha' : 20, 'film': False, + #'ira' : self.pathout['Analyse.ira'] } - page = getPage(self.parent) - if page is not None : - self.corpus = getCorpus(page) - if self.corpus is not None : - self.pathout = ConstructPathOut(self.corpus.parametre['openpath'], 'simitxt') - self.dictpathout = construct_simipath(self.pathout) - self.val = wx.ID_OK - self.make_table() - self.make_simi() + self.parametres.update(self.paramsimi) + + def makefiles(self) : + self.actives, lim = self.corpus.make_actives_nb(self.parametres.get('max_actives',1500), 1) + self.parametres['eff_min_forme'] = lim + self.parametres['nbactives'] = len(self.actives) + self.parametres['fromprof'] = True + self.corpus.make_and_write_sparse_matrix_from_uces(self.actives, self.pathout['mat01.csv'], self.pathout['listeuce1.csv']) + with open(self.pathout['actives.csv'], 'w') as f : + f.write('\n'.join(self.actives).encode(self.ira.syscoding)) + + self.listet = self.corpus.make_etoiles() + self.listet.sort() + self.parametres['stars'] = copy(self.listet) + self.parametres['sfromchi'] = False + +class PrepSimi : + def _init_(self, parent, parametres, indices_simi) : + self.parametres = parametres + self.dial = PrefSimi(parent, -1, self.parametres, indices_simi) + self.dial.CenterOnParent() + self.val = self.dial.ShowModal() + if self.val == wx.ID_OK : + self.make_param() + + def make_param(self) : + self.select = self.dial.check_colch.GetValue() + param = {'coeff' : self.dial.choice1.GetSelection(), + 'layout' : self.dial.choice2.GetSelection(), + 'type' : self.dial.choice3.GetSelection(), + 'arbremax' : self.dial.check1.GetValue(), + 'coeff_tv' : self.dial.check_s_size.GetValue(), + 'coeff_tv_nb' : self.dial.spin_tv.GetValue(), + 'tvprop' : self.dial.check2.GetValue(), + 'tvmin' : self.dial.spin_tvmin.GetValue(), + 'tvmax' : self.dial.spin_tvmax.GetValue(), + 'coeff_te' : self.dial.check3.GetValue(), + 'coeff_temin' : self.dial.spin_temin.GetValue(), + 'coeff_temax' : self.dial.spin_temax.GetValue(), + 'label_e' : self.dial.check_elab.GetValue(), + 'label_v' : self.dial.check_vlab.GetValue(), + 'vcex' : self.dial.check_vcex.GetValue(), + 'vcexmin' : self.dial.spin_vcexmin.GetValue(), + 'vcexmax' : self.dial.spin_vcexmax.GetValue(), + 'cex' : self.dial.spin_cex.GetValue(), + 'seuil_ok' : self.dial.check_seuil.GetValue(), + 'seuil' : self.dial.spin_seuil.GetValue(), + 'cols' : self.dial.cols.GetColour(), + 'cola' : self.dial.cola.GetColour(), + 'width' : self.dial.spin_width.GetValue(), + 'height' : self.dial.spin_height.GetValue(), + 'first' : False, + 'keep_coord' : keep_coord, + 'alpha' : self.dial.slider_sphere.GetValue(), + 'film' : self.dial.film.GetValue() + } + if 'cexfromchi' in self.parametres : + param['cexfromchi'] = self.dial.checkit.GetValue() + if 'sfromchi' in self.parametres : + param['sfromchi'] = self.dial.checki.GetValue() + if 'vlabcolor' in self.parametres : + param['vlabcolor'] = self.parametres['vlabcolor'] + if 'check_bystar' in dir(self.dial) : + param['bystar'] = self.dial.check_bystar.GetValue() + param['stars'] = self.parametres['stars'] + self.parametres.update(param) + +class PrintSimiScript(PrintRScript) : + def make_script(self) : + self.load(['igraph', 'proxy', 'Matrix']) + self.source([self.analyse.parent.RscriptsPath['simi'], self.analyse.parent.RscriptsPath['Rgraph']]) + txt = """ + dm.path <- "%s" + cn.path <- "%s" + selected.col <- "%s" + """ % (self.pathout['mat01.csv'], self.pathout['actives.csv'], self.pathout['selected.csv']) + + txt += """ + dm <- dm[, selected.col+1] + """ + if self.parametres['coeff'] == 0 : + method = 'cooc' + txt += """ + method <- 'cooc' + mat <- make.a(dm) + """ else : - self.corpus = Corpus(parent) - self.corpus.content = self.parent.content - self.corpus.parametre['encodage'] = parent.corpus_encodage - self.corpus.parametre['lang'] = parent.corpus_lang - self.corpus.parametre['filename'] = parent.filename - self.corpus.parametre['eff_min_uce'] = None - self.pathout = ConstructPathOut(self.corpus.parametre['filename'], 'simitxt') - self.dictpathout = construct_simipath(self.pathout) - dial = StatDialog(self, self.parent) - dial.check_uce.SetValue(True) - dial.check_uce.Enable(False) - dial.OnCheckUce(wx.EVT_MENU) - self.val = dial.ShowModal() - if self.val == wx.ID_OK : - with open(self.parent.ConfigPath['key'], 'w') as f: - self.KeyConf.write(f) - if dial.radio_lem.GetSelection() == 0 : lem = True - else : lem = False - if dial.exp.GetSelection() == 0 : exp = True - else : exp = False - dial.Destroy() - self.corpus.parametre['lem'] = lem - self.corpus.parametre['expressions'] = exp - self.corpus.parametre['nbforme_uce'] = dial.spin_ctrl_4.GetValue() - self.corpus.parametre['eff_min_forme'] = dial.spin_ctrl_5.GetValue() - self.corpus.parametre['max_actives'] = dial.spin_max_actives.GetValue() - self.make_corpus() - self.make_table() - self.make_simi() - else : - dial.Destroy() - - def make_corpus(self) : - print 'make corpus' - if not self.cmd : - dlg = progressbar(self, maxi = 6) - self.corpus.supplementaires = [option for option in self.KeyConf.options('KEYS') if self.KeyConf.get('KEYS', option) == "2"] - self.corpus.typeactive = [option for option in self.KeyConf.options('KEYS') if self.KeyConf.get('KEYS', option) == "1"] - ucis_txt, ucis_paras_txt = self.corpus.start_analyse(self.parent, dlg = dlg, cmd = self.cmd) - del ucis_txt - - if not self.cmd : - dlg.Update(5, '%i ucis - Construction des uces' % len(ucis_paras_txt)) - self.corpus.make_ucis_paras_uces(ucis_paras_txt, make_uce = True) - del ucis_paras_txt - - if self.corpus.para_coords != [[] for val in self.corpus.para_coords] : - self.corpus.parametre['para'] = True + txt += """ + dm <- as.matrix(dm) + """ + if self.parametres['coeff'] == 1 : + method = 'prcooc' + txt += """ + method <- 'Russel' + mat <- simil(dm, method = 'Russel', diag = TRUE, upper = TRUE, by_rows = FALSE) + """ + elif self.analyses.indices[self.parametres['coeff']] == 'binomial' : + method = 'binomial' + txt += """ + method <- 'binomial' + mat <- binom.sim(dm) + """ else : - self.corpus.parametre['para'] = False - self.corpus.make_etoiles(self.corpus.para_coords) - print 'len(ucis_paras_uces)', len(self.corpus.ucis_paras_uces) - - if not self.cmd : - dlg.Update(6, u'Dictionnaires') - uces, self.orderuces = self.corpus.make_forms_and_uces() - self.corpus.ucenb = len(uces) - self.corpus.make_lems(self.parent.lexique) - - self.corpus.make_var_actives() - self.corpus.make_var_supp() - self.corpus.lems_eff = self.corpus.make_lem_eff() - - #variables = treat_var_mod(listet) - #print(variables) - #self.corpus.write_etoiles(self.dictpathout['etoiles']) - if not self.cmd : - dlg.Destroy() - - def make_table(self) : - if 'orderuces' not in dir(self) : - self.orderuces = [(i,j,k) for i, uci in enumerate(self.corpus.ucis_paras_uces) for j, para in enumerate(uci) for k, uce in enumerate(para)] - self.orderuces = dict([[val, i] for i, val in enumerate(self.orderuces)]) - self.corpus.ucenb = len(self.orderuces) - #tabuc1 = self.corpus.make_table_with_uce(self.orderuces) - #tabuc1.insert(0,self.corpus.actives) - #tabuc1 = self.corpus.make_sparse_matrix_with_uce(self.orderuces) - #self.corpus.write_sparse_matrix(self.dictpathout['mat01'], tabuc1, self.corpus.ucenb, len(self.corpus.actives)) - if self.corpus.actives is None : - self.corpus.typeactive = [option for option in self.KeyConf.options('KEYS') if self.KeyConf.get('KEYS', option) == "1"] - self.corpus.min_eff_formes() - self.corpus.make_var_actives() - self.corpus.make_and_write_sparse_matrix_from_uce(self.orderuces, self.dictpathout['mat01']) - #self.corpus.write_tab(tabuc1,self.dictpathout['mat01']) - - def make_simi(self) : + method = self.types[self.paramsimi['coeff']] + txt += """ + method <-"%s" + mat <- simil(dm, method = method, diag = TRUE, upper = TRUE, by_rows = FALSE) + """ % self.analyse.indices[self.parametres['coeff']] + txt += """ + mat <- as.matrix(stats::as.dist(mat,diag=TRUE,upper=TRUE)) + mat[is.na(mat)] <- 0 + mat[is.infinite(mat)] <- 0 + """ + if self.parametres['layout'] == 0 : layout = 'random' + if self.parametres['layout'] == 1 : layout = 'circle' + if self.parametres['layout'] == 2 : layout = 'frutch' + if self.parametres['layout'] == 3 : layout = 'kawa' + if self.parametres['layout'] == 4 : layout = 'graphopt' + + txt += """ + eff <- colSums(dm) + g.ori <- graph.adjacency(mat, mode='lower', weighted = TRUE) + w.ori <- E(g.ori)$weight + if (max.tree) { + if (method == 'cooc') { + E(g.ori)$weight <- 1 / w.ori + } else { + E(g.ori)$weigth <- 1 - w.ori + } + g.max <- minimum.spanning.tree(g.ori) + if (method == 'cooc') { + E(g.max)$weight <- 1 / E(g.max)$weight + } else { + E(g.max)$weight <- 1 - E(g.max)$weight + } + g.toplot <- g.max + } else { + g.toplot <- g.ori + } + """ + + + self.tableau = Tableau(self.parent, '') - self.tableau.listactives = self.corpus.actives + self.tableau.listactives = self.actives self.tableau.parametre['fromtxt'] = True - if 'lems_eff' not in dir(self.corpus) : - self.corpus.lems_eff = self.corpus.make_lem_eff() + self.corpus.lems_eff = dict([[lem,[self.corpus.lems[lem].freq]] for lem in self.actives]) #print('ATTENTION ETOILES') #self.paramsimi['bystar'] = True - self.listet = self.corpus.get_unique_etoiles() - self.listet.sort() self.tableau.listet = copy(self.listet) - self.paramsimi['stars'] = copy(self.listet) #self.paramsimi['cexfromchi'] = True - self.paramsimi['sfromchi'] = False #self.paramsimi['vlabcolor'] = True - self.tableau.actives = dict([[lem, self.corpus.lems_eff[lem]] for lem in self.corpus.actives]) - self.corpus.save_corpus(self.dictpathout['corpus']) - DoSimi(self, fromprof = self.dictpathout['mat01'], param = self.paramsimi, pathout = self.pathout) + self.tableau.actives = copy(self.corpus.lems_eff) + DoSimi(self, fromprof = self.pathout['mat01.csv'], param = self.paramsimi, pathout = self.pathout.dirout) + +#class SimiTxt : +# def __init__(self, parent, cmd = False, param = None): +# self.parent = parent +# self.cmd = cmd +# self.ConfigPath = parent.ConfigPath +# self.DictPath = parent.DictPath +# self.KeyConf = RawConfigParser() +# self.KeyConf.read(self.ConfigPath['key']) +# self.indices = indices_simi +# self.paramsimi = {'coeff' : 0, +# 'layout' : 2, +# 'type' : 1, +# 'arbremax' : 1, +# 'coeff_tv' : 1, +# 'coeff_tv_nb' : 0, +# 'tvprop' : 0, +# 'tvmin' : 5, +# 'tvmax' : 30, +# 'coeff_te' : 1, +# 'coeff_temin' : 1, +# 'coeff_temax' : 10, +# 'label_v': 1, +# 'label_e': 0, +# 'vcex' : 1, +# 'cexfromchi' : False, +# 'vcexmin' : 10, +# 'vcexmax' : 25, +# 'cex' : 10, +# 'seuil_ok' : 0, +# 'seuil' : 1, +# 'cols' : (255,0,0), +# 'cola' : (200,200,200), +# 'width' : 1000, +# 'height' : 1000, +# 'bystar' : False, +# 'first' : True, +# 'keep_coord' : True, +# 'alpha' : 20, +# 'film': False, +# } +# page = getPage(self.parent) +# if page is not None : +# self.corpus = getCorpus(page) +# if self.corpus is not None : +# self.pathout = ConstructPathOut(self.corpus.parametre['openpath'], 'simitxt') +# self.dictpathout = construct_simipath(self.pathout) +# self.val = wx.ID_OK +# self.make_table() +# self.make_simi() +# else : +# self.corpus = Corpus(parent) +# self.corpus.content = self.parent.content +# self.corpus.parametre['encodage'] = parent.corpus_encodage +# self.corpus.parametre['lang'] = parent.corpus_lang +# self.corpus.parametre['filename'] = parent.filename +# self.corpus.parametre['eff_min_uce'] = None +# self.pathout = ConstructPathOut(self.corpus.parametre['filename'], 'simitxt') +# self.dictpathout = construct_simipath(self.pathout) +# dial = StatDialog(self, self.parent) +# dial.check_uce.SetValue(True) +# dial.check_uce.Enable(False) +# dial.OnCheckUce(wx.EVT_MENU) +# self.val = dial.ShowModal() +# if self.val == wx.ID_OK : +# with open(self.parent.ConfigPath['key'], 'w') as f: +# self.KeyConf.write(f) +# if dial.radio_lem.GetSelection() == 0 : lem = True +# else : lem = False +# if dial.exp.GetSelection() == 0 : exp = True +# else : exp = False +# dial.Destroy() +# self.corpus.parametre['lem'] = lem +# self.corpus.parametre['expressions'] = exp +# self.corpus.parametre['nbforme_uce'] = dial.spin_ctrl_4.GetValue() +# self.corpus.parametre['eff_min_forme'] = dial.spin_ctrl_5.GetValue() +# self.corpus.parametre['max_actives'] = dial.spin_max_actives.GetValue() +# self.make_corpus() +# self.make_table() +# self.make_simi() +# else : +# dial.Destroy() +# +# def make_corpus(self) : +# print 'make corpus' +# if not self.cmd : +# dlg = progressbar(self, maxi = 6) +# self.corpus.supplementaires = [option for option in self.KeyConf.options('KEYS') if self.KeyConf.get('KEYS', option) == "2"] +# self.corpus.typeactive = [option for option in self.KeyConf.options('KEYS') if self.KeyConf.get('KEYS', option) == "1"] +# ucis_txt, ucis_paras_txt = self.corpus.start_analyse(self.parent, dlg = dlg, cmd = self.cmd) +# del ucis_txt +# +# if not self.cmd : +# dlg.Update(5, '%i ucis - Construction des uces' % len(ucis_paras_txt)) +# self.corpus.make_ucis_paras_uces(ucis_paras_txt, make_uce = True) +# del ucis_paras_txt +# +# if self.corpus.para_coords != [[] for val in self.corpus.para_coords] : +# self.corpus.parametre['para'] = True +# else : +# self.corpus.parametre['para'] = False +# self.corpus.make_etoiles(self.corpus.para_coords) +# print 'len(ucis_paras_uces)', len(self.corpus.ucis_paras_uces) +# +# if not self.cmd : +# dlg.Update(6, u'Dictionnaires') +# uces, self.orderuces = self.corpus.make_forms_and_uces() +# self.corpus.ucenb = len(uces) +# self.corpus.make_lems(self.parent.lexique) +# +# self.corpus.make_var_actives() +# self.corpus.make_var_supp() +# self.corpus.lems_eff = self.corpus.make_lem_eff() +# +# #variables = treat_var_mod(listet) +# #print(variables) +# #self.corpus.write_etoiles(self.dictpathout['etoiles']) +# if not self.cmd : +# dlg.Destroy() +# +# def make_table(self) : +# if 'orderuces' not in dir(self) : +# self.orderuces = [(i,j,k) for i, uci in enumerate(self.corpus.ucis_paras_uces) for j, para in enumerate(uci) for k, uce in enumerate(para)] +# self.orderuces = dict([[val, i] for i, val in enumerate(self.orderuces)]) +# self.corpus.ucenb = len(self.orderuces) +# #tabuc1 = self.corpus.make_table_with_uce(self.orderuces) +# #tabuc1.insert(0,self.corpus.actives) +# #tabuc1 = self.corpus.make_sparse_matrix_with_uce(self.orderuces) +# #self.corpus.write_sparse_matrix(self.dictpathout['mat01'], tabuc1, self.corpus.ucenb, len(self.corpus.actives)) +# if self.corpus.actives is None : +# self.corpus.typeactive = [option for option in self.KeyConf.options('KEYS') if self.KeyConf.get('KEYS', option) == "1"] +# self.corpus.min_eff_formes() +# self.corpus.make_var_actives() +# self.corpus.make_and_write_sparse_matrix_from_uce(self.orderuces, self.dictpathout['mat01']) +# #self.corpus.write_tab(tabuc1,self.dictpathout['mat01']) +# +# def make_simi(self) : +# self.tableau = Tableau(self.parent, '') +# self.tableau.listactives = self.corpus.actives +# self.tableau.parametre['fromtxt'] = True +# if 'lems_eff' not in dir(self.corpus) : +# self.corpus.lems_eff = self.corpus.make_lem_eff() +# #print('ATTENTION ETOILES') +# #self.paramsimi['bystar'] = True +# self.listet = self.corpus.get_unique_etoiles() +# self.listet.sort() +# self.tableau.listet = copy(self.listet) +# self.paramsimi['stars'] = copy(self.listet) +# #self.paramsimi['cexfromchi'] = True +# self.paramsimi['sfromchi'] = False +# #self.paramsimi['vlabcolor'] = True +# self.tableau.actives = dict([[lem, self.corpus.lems_eff[lem]] for lem in self.corpus.actives]) +# self.corpus.save_corpus(self.dictpathout['corpus']) +# DoSimi(self, fromprof = self.dictpathout['mat01'], param = self.paramsimi, pathout = self.pathout)