X-Git-Url: http://iramuteq.org/git?p=iramuteq;a=blobdiff_plain;f=tabchi2.py;h=422b44735d1facb7f79f9156cf3068cfea561c47;hp=6a8cf550460d44717147356c65b3f72c52978625;hb=fd5ed5e9fe3c56891bcd819e6b66a173f44e2124;hpb=22cd27b2bbe9ab1ffa7ef06fa764b5147ae17dad

diff --git a/tabchi2.py b/tabchi2.py
index 6a8cf55..422b447 100755
--- a/tabchi2.py
+++ b/tabchi2.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 #Author: Pierre Ratinaud
 #Copyright (c) 2010 Pierre Ratinaud
-#Lisense: GNU/GPL
+#License: GNU/GPL
 
 import HTML
 import os
@@ -16,6 +16,7 @@ import wx.lib.sized_controls as sc
 from time import sleep
 from functions import exec_rcode, check_Rresult
 from dialog import ChiDialog, PrefChi
+from analysematrix import AnalyseMatrix
                     
 def make_res(line) :
     if float(line[5]) <= 0.05 and line[6] != 'warning':
@@ -45,11 +46,21 @@ def make_title(res, text) :
     return ['<a name=%i></a><br><font color=%s>%s</font><br><a href=#back_%i>retour</a><br>' % (i, val[-1], text[i], i) for i, val in enumerate(res)] 
 
 
-
+chioption = { 'valobs' : True,
+              'valtheo' : True,
+              'resi' : False,
+              'contrib' : True,
+              'pourcent' : False,
+              'pourcentl' : True,
+              'pourcentc' : True,
+              'graph' : True,
+              'bw' : False,
+              }
 
 
 class MakeChi2():
-    def __init__(self, parent, select1, select2, chioption):
+    def __init__(self, parent, select1, select2, chioption, tableau):
+        self.tableau = tableau
         self.OutFrame=tempfile.mktemp(dir=parent.TEMPDIR)
         print self.OutFrame
         self.parent=parent
@@ -59,7 +70,7 @@ class MakeChi2():
         self.TextCroise=[]
         for i in select1 :
             for j in select2 :
-                self.TextCroise.append(parent.tableau.colnames[i] + ' / ' + parent.tableau.colnames[j])
+                self.TextCroise.append(self.tableau.colnames[i] + ' / ' + self.tableau.colnames[j])
         rchioption = {}
         for val in chioption :
             if chioption[val]:
@@ -88,11 +99,11 @@ class MakeChi2():
         bw <- %s
         """ % (rchioption['valobs'], rchioption['valtheo'], rchioption['contrib'], rchioption['resi'], rchioption['pourcent'], rchioption['pourcentl'], rchioption['pourcentc'], rchioption['graph'], rchioption['bw'])
         txt+="""
-        datadm <- ReadData("%s", encoding="%s", header = TRUE, sep = "%s",quote = '%s', na.strings = "%s",rownames= 1)
+        datadm <- read.csv2("%s", encoding="%s", header = TRUE, row.names = 1, sep='\\t', quote = '"', na.string = '')
         listres<-list()
         listcol<-list()
         cont<-1
-        """%(ffr(parent.tableau.parametre['csvfile']),self.parent.encode, parent.tableau.parametre['colsep'], parent.tableau.parametre['txtsep'], self.parent.nastrings)
+        """%(ffr(self.tableau.parametres['csvfile']), self.tableau.parametres['syscoding'])
         if len(select1)==1:
             strsel1=str(select1).replace(',','')
         else:
@@ -148,9 +159,7 @@ class MakeChi2():
                 chi$prl <- round((chi$observed/sr)*100,2)
                 chi$prc <- t(round((t(chi$observed)/sc)*100,2))
             }
-            fileout<-paste('histo',%i,sep='')
-            fileout<-paste(fileout,'_',sep='')
-            fileout<-paste(fileout,count,sep='')
+            fileout<-paste('histo_',count,sep='')
             fileout<-paste(fileout,'.png',sep='')
             count<-count+1
             fileout<-file.path("%s",fileout)
@@ -258,7 +267,7 @@ class MakeChi2():
         li<-matrix('fin_analyse',1,maxcol)
         frameout<-rbind(frameout,li)
         write.csv2(frameout,file="%s")
-        """%(parent.FreqNum,ffr(parent.TEMPDIR),ffr(self.OutFrame))
+        """ % (ffr(parent.TEMPDIR),ffr(self.OutFrame))
         tmpfile=tempfile.mktemp(dir=self.TEMPDIR)
         print tmpfile
         tmpscript=open(tmpfile,'w')
@@ -350,77 +359,381 @@ class MakeChi2():
         txt = '<br><hr><br>\n'.join(['<br><br>'.join([tab[i] for tab in allhtml]) for i,val in enumerate(res)])
         txt = header + pretxt + txt + '\n</body></html>'
 
-        fileout=os.path.join(self.TEMPDIR,'resultats-chi2_%s.html'%str(self.parent.FreqNum))
-        file=open(fileout,'w')
-        file.write(txt)
-        file.close()
+        fileout=os.path.join(self.parametres['pathout'],'resultats-chi2.html')
+        with open(fileout, 'w') as f :
+            f.write(txt)
         ListFile.append(fileout)         
         return ListFile
 
-class ChiSquare():
-    def __init__(self,parent):
-        chioption = { 'valobs' : True,
-                      'valtheo' : True,
-                      'resi' : False,
-                      'contrib' : True,
-                      'pourcent' : False,
-                      'pourcentl' : True,
-                      'pourcentc' : True,
-                      'graph' : True,
-                      'bw' : False,
-                      }
-        dlg = ChiDialog(parent, -1, u"Chi2", chioption, size=(400, 350),
+class ChiSquare(AnalyseMatrix):
+    def doparametres(self, dlg = None):
+        if dlg is None :
+            return
+        dial = ChiDialog(self.parent, -1, u"Chi2", chioption, self.tableau, size=(400, 350),
                      style = wx.DEFAULT_DIALOG_STYLE
                      )
-        dlg.CenterOnParent()
-        val = dlg.ShowModal()
-        if val==wx.ID_OK :
-            self.dlg=wx.ProgressDialog("Traitements",
-                       "Veuillez patienter...",
-                       maximum = 4,
-                       parent=parent,
-                       style = wx.PD_APP_MODAL|wx.PD_AUTO_HIDE|wx.PD_ELAPSED_TIME
-                        )
-            self.dlg.Center()
+        dial.CenterOnParent()
+        val = dial.ShowModal()
+        if val==wx.ID_OK : 
+            dlg.Center()
             self.count = 1
-            keepGoing = self.dlg.Update(self.count)
+            keepGoing = dlg.Update(self.count)
             
-            ColSel1 = dlg.list_box_1.GetSelections()
-            ColSel2 = dlg.list_box_2.GetSelections()
-            if dlg.chiopt :
-                chioption['valobs'] = dlg.dial.check1.GetValue()
-                chioption['valtheo'] = dlg.dial.check2.GetValue()
-                chioption['resi'] = dlg.dial.check3.GetValue()
-                chioption['contrib'] = dlg.dial.check4.GetValue()
-                chioption['pourcent'] = dlg.dial.check5.GetValue()
-                chioption['pourcentl'] = dlg.dial.check6.GetValue()
-                chioption['pourcentc'] = dlg.dial.check7.GetValue()
-                chioption['graph'] = dlg.dial.check8.GetValue()
-                chioption['bw'] = dlg.dial.checkbw.GetValue()
-                dlg.dial.Destroy()
+            self.colsel1 = dial.list_box_1.GetSelections()
+            self.colsel2 = dial.list_box_2.GetSelections()
+            if dial.chiopt :
+                chioption['valobs'] = dial.dial.check1.GetValue()
+                chioption['valtheo'] = dial.dial.check2.GetValue()
+                chioption['resi'] = dial.dial.check3.GetValue()
+                chioption['contrib'] = dial.dial.check4.GetValue()
+                chioption['pourcent'] = dial.dial.check5.GetValue()
+                chioption['pourcentl'] = dial.dial.check6.GetValue()
+                chioption['pourcentc'] = dial.dial.check7.GetValue()
+                chioption['graph'] = dial.dial.check8.GetValue()
+                chioption['bw'] = dial.dial.checkbw.GetValue()
+                dial.dial.Destroy()
+            dial.Destroy()
+            self.parametres.update(chioption)
+            self.chioption = chioption
+        else :
+            if dial.chiopt :
+                dial.dial.Destroy()
+            dial.Destroy()
+            self.parametres = None
             
-            self.count += 1
-            keepGoing = self.dlg.Update(self.count,u"Analyse dans R...")
-            analyse=MakeChi2(parent,ColSel1,ColSel2, chioption)
+    def doanalyse(self):
+
+
+            #self.dlg=wx.ProgressDialog("Traitements",
+            #           "Veuillez patienter...",
+            #           maximum = 4,
+            #           parent=parent,
+            #           style = wx.PD_APP_MODAL|wx.PD_AUTO_HIDE|wx.PD_ELAPSED_TIME
+            #            )
+
+
             
-            self.count += 1
-            keepGoing = self.dlg.Update(self.count,u"Ecriture des résultats")
+        self.count += 1
+        keepGoing = self.dlg.Update(self.count,u"Analyse dans R...")
+        #analyse=MakeChi2(self.parent, self.colsel1, self.colsel2, self.chioption, self.tableau)
+        #self.tableau = tableau
+        self.OutFrame=tempfile.mktemp(dir=self.parent.TEMPDIR)
+        print self.OutFrame
+        #self.parent=parent
+        self.encode=self.parent.encode
+        self.TEMPDIR=self.parent.TEMPDIR
+        self.RPath=self.parent.PathPath.get('PATHS','rpath')
+        self.TextCroise=[]
+        for i in self.colsel1 :
+            for j in self.colsel2 :
+                self.TextCroise.append(self.tableau.colnames[i] + ' / ' + self.tableau.colnames[j])
+        rchioption = {}
+        for val in self.chioption :
+            if self.chioption[val]:
+                rchioption[val] = 'TRUE'
+            else :
+                rchioption[val] = 'FALSE'
+        txt="""
+        source("%s")
+        """%self.parent.RscriptsPath['Rfunct']
+#        if parent.tableau.: rownames=1
+#        else : rownames='NULL'
+#        if parent.g_header : header = 'TRUE'
+#        else : header = 'FALSE'
+        txt += """
+        source("%s")
+        """ % ffr(self.parent.RscriptsPath['Rgraph'])
+        txt += """
+        doobs <- %s
+        doexp <- %s
+        docontrib <- %s
+        doresi <- %s
+        dopr <- %s
+        doprl <- %s
+        doprc <- %s
+        dograph <- %s
+        bw <- %s
+        """ % (rchioption['valobs'], rchioption['valtheo'], rchioption['contrib'], rchioption['resi'], rchioption['pourcent'], rchioption['pourcentl'], rchioption['pourcentc'], rchioption['graph'], rchioption['bw'])
+        txt+="""
+        datadm <- read.csv2("%s", encoding="%s", header = TRUE, row.names = 1, sep='\\t', quote = '"', na.string = '')
+        listres<-list()
+        listcol<-list()
+        cont<-1
+        """%(ffr(self.tableau.parametres['csvfile']), self.tableau.parametres['syscoding'])
+        if len(self.colsel1)==1:
+            strsel1=str(self.colsel1).replace(',','')
+        else:
+            strsel1=str(self.colsel1)
+        if len(self.colsel2)==1:
+            strsel2=str(self.colsel2).replace(',','')
+        else:
+            strsel2=str(self.colsel2)
+        txt+="""
+        for (i in c%s) {""" % strsel1
+        txt+="""
+            for (j in c%s) {""" % strsel2
+        txt+="""
+                tab<-table(datadm[,i+1],datadm[,j+1])
+                if (min(dim(tab)) != 1) {
+                    chi<-chisq.test(tab)
+                    CS<-colSums(tab)
+                    RS<-rowSums(tab)
+                    GT<-sum(tab)
+                    chi$contrib<-(tab-chi$expected)/sqrt(chi$expected * ((1 - RS/GT) %%*%% t(1 - CS/GT)))
+                    listres[[cont]]<-chi
+                    listcol[[cont]]<-ncol(tab)
+                    cont<-cont+1
+                } else {
+                    chi <- list(observed = tab, residuals = tab, contrib = tab, statistic = 0, p.value = 1, expected = tab, message = 'pas de calcul')
+                    listres[[cont]] <- chi
+                    listcol[[cont]]<-ncol(tab)
+                    cont <- cont + 1
+                }
+            }
+        }
+        maxcol<-max(unlist(listcol))+1
+        if (maxcol<7) {maxcol<-7}
+        frameout<-matrix('*',1,maxcol)
+        count<-0
+        for (chi in listres) {
+            if (min(chi$expected)<5) {
+                att<-"warning"
+            } else {
+                att<-""
+            }
+            if ('message' %%in%% attributes(chi)$names) {
+                att <- "Ce chi2 n\'a pas été calculé"
+                nom_colresi<-colnames(chi$observed)
+                chi$prl <- chi$expected
+                chi$prc <- chi$expected
+                st <- sum(chi$observed)
+            } else {
+                nom_colresi<-colnames(chi$observed)
+                st <- sum(chi$observed)
+                sc <- colSums(chi$observed)
+                sr <- rowSums(chi$observed)
+                chi$prl <- round((chi$observed/sr)*100,2)
+                chi$prc <- t(round((t(chi$observed)/sc)*100,2))
+            }
+            fileout<-paste('histo_',count,sep='')
+            fileout<-paste(fileout,'.png',sep='')
+            count<-count+1
+            fileout<-file.path("%s",fileout)
+            if (max(nchar(colnames(chi$observed)))>15) {
+                leg <- 1:length(colnames(chi$observed))
+            } else {
+                leg <- colnames(chi$observed)
+            }
+            if (dograph) {
+                width<-ncol(chi$observed)*100
+                if (width < 350) {width <- 350}
+                open_file_graph(fileout,width = width, height = 300)
+                par(mar=c(0,0,0,0))
+                layout(matrix(c(1,2),1,2, byrow=TRUE),widths=c(3,1))
+                par(mar=c(2,2,1,0))
+                par(cex=0.8)
+                if (!bw) colors <- rainbow(length(rownames(chi$observed)))
+                else colors <- gray.colors(length(rownames(chi$observed)))
+                barplot(chi$prl,names.arg = leg, beside=TRUE,border=NA, col=colors)
+                par(mar=c(0,0,0,0))
+                par(cex=0.8)
+                plot(0, axes = FALSE, pch = '')
+                legend(x = 'center' , rownames(chi$observed), fill = colors)
+                dev.off()
+            }
+            chi$prl <- cbind(chi$prl, total = rowSums(chi$prl))
+            chi$prc <- rbind(chi$prc, total = colSums(chi$prc))
+            chi$observed<-rbind(chi$observed,total=colSums(chi$observed))
+            chi$observed<-cbind(chi$observed,total=rowSums(chi$observed))
+            chi$pr <- round((chi$observed/st)*100,2)
+            chi$expected<-rbind(chi$expected,total=colSums(chi$expected))
+            chi$expected<-cbind(chi$expected,total=rowSums(chi$expected))
+            chi$expected<-round(chi$expected,digits=2)
+            chi$residuals<-round(chi$residuals,digits=2)
+            chi$contrib<-round(chi$contrib, digits=2)
+            nom_col<-colnames(chi$observed)
+           
+            if (ncol(chi$observed)<maxcol) {
+                for (i in 1:(maxcol-ncol(chi$observed))) {
+                    chi$observed<-cbind(chi$observed,'**')
+                    chi$pr<-cbind(chi$pr,'**')
+                    chi$prl<-cbind(chi$prl,'**')
+                    chi$prc<-cbind(chi$prc,'**')
+                    chi$expected<-cbind(chi$expected,'**')
+                    chi$residuals<-cbind(chi$residuals,'**')
+                    chi$contrib<-cbind(chi$contrib,'**')
+                    nom_col<-append(nom_col,'**')
+                    nom_colresi<-append(nom_colresi,'**')
+                }
+                chi$residuals<-cbind(chi$residuals,'**')
+                chi$contrib<-cbind(chi$contrib,'**')
+                nom_colresi<-append(nom_colresi,'**')
+                chi$prc<-cbind(chi$prc,'**')
+            } else if (ncol(chi$observed)==maxcol) {
+                chi$residuals<-cbind(chi$residuals,'**')
+                chi$contrib<-cbind(chi$contrib,'**')
+                nom_colresi<-append(nom_colresi,'**')
+                chi$prc<-cbind(chi$prc,'**')
+            }
+            if (doobs) {
+                li<-matrix('*obs*',1,maxcol)
+                frameout<-rbind(frameout,li)
+                frameout<-rbind(frameout,nom_col)
+                frameout<-rbind(frameout,chi$observed)
+            }
+            if (doexp) {
+                li<-matrix('*exp*',1,maxcol)
+                frameout<-rbind(frameout,li)
+                frameout<-rbind(frameout,nom_col)
+                frameout<-rbind(frameout,chi$expected)
+            }
+            if (doresi) {
+                li<-matrix('*resi*',1,maxcol)
+                frameout<-rbind(frameout,li)
+                frameout<-rbind(frameout,nom_colresi)
+                frameout<-rbind(frameout,chi$residuals)
+            }
+            if (docontrib) {
+                li<-matrix('*contrib*',1,maxcol)
+                frameout<-rbind(frameout,li)
+                frameout<-rbind(frameout,nom_colresi)
+                frameout<-rbind(frameout,chi$contrib)
+            }
+            if (dopr) {
+                li<-matrix('*pr*', 1, maxcol)
+                frameout<-rbind(frameout,li)
+                frameout<-rbind(frameout,nom_col)
+                frameout<-rbind(frameout,chi$pr)
+            }
+            if (doprl) {
+                li<-matrix('*prl*', 1, maxcol)
+                frameout<-rbind(frameout,li)
+                frameout<-rbind(frameout,nom_col)
+                frameout<-rbind(frameout,chi$prl)
+            }
+            if (doprc) {
+                li<-matrix('*prc*', 1, maxcol)
+                frameout<-rbind(frameout,li)
+                frameout<-rbind(frameout,nom_colresi)
+                frameout<-rbind(frameout,chi$prc)
+            }
+            res<-c('****','chi',chi$statistic,'p',chi$p.value,att,fileout)
+            frameout<-rbind(frameout,res)
+        }
+        li<-matrix('fin_analyse',1,maxcol)
+        frameout<-rbind(frameout,li)
+        write.csv2(frameout,file="%s")
+        """ % (ffr(self.parametres['pathout']),ffr(self.OutFrame))
+        tmpfile=tempfile.mktemp(dir=self.TEMPDIR)
+        print tmpfile
+        tmpscript=open(tmpfile,'w')
+        tmpscript.write(txt)
+        tmpscript.close()
+        pid = exec_rcode(self.RPath, tmpfile, wait = False)
+        while pid.poll() == None :
+            sleep(0.2)
+        check_Rresult(self.parent, pid)            
+        self.count += 1
+        keepGoing = self.dlg.Update(self.count,u"Ecriture des résultats")
          
-            listfileout = analyse.dolayout(chioption)
+        listfileout = self.dolayout(self.chioption)
             #listfileout=dlg.ShowChi2(ColSel1,ColSel2)
-            parent.FreqNum += 1
-            parent.DictTab[u"Chi2_%s*"%parent.FreqNum]=listfileout
-            parent.newtab = wx.html.HtmlWindow(parent.nb, -1)
-            if "gtk2" in wx.PlatformInfo:
-               parent.newtab.SetStandardFonts()
-            parent.newtab.LoadPage(listfileout[len(listfileout)-1])
-            parent.nb.AddPage(parent.newtab,u"Chi2_%s*"%parent.FreqNum)
-            parent.nb.SetSelection(parent.nb.GetPageCount()-1)
-            parent.ShowTab(wx.EVT_BUTTON)
-            parent.DisEnSaveTabAs(True)
-            self.count += 1
-            keepGoing = self.dlg.Update(self.count,u"Fini")
-        else :
-            if dlg.chiopt :
-                dlg.dial.Destroy()
-            dlg.Destroy()
+            #parent.FreqNum += 1
+            #parent.DictTab[u"Chi2_%s*"%parent.FreqNum]=listfileout
+#             parent.newtab = wx.html.HtmlWindow(parent.nb, -1)
+#             if "gtk2" in wx.PlatformInfo:
+#                 parent.newtab.SetStandardFonts()
+#             parent.newtab.LoadPage(listfileout[len(listfileout)-1])
+#             parent.nb.AddPage(parent.newtab,u"Chi2_%s*"%parent.FreqNum)
+#             parent.nb.SetSelection(parent.nb.GetPageCount()-1)
+#             parent.ShowTab(wx.EVT_BUTTON)
+#             parent.DisEnSaveTabAs(True)
+        self.count += 1
+        keepGoing = self.dlg.Update(self.count,u"Fini")
+
+    def dolayout(self, option):
+        ListFile=[False]
+        file=open(self.OutFrame,'rU')
+        content=file.readlines()
+        file.close()
+        lcont = [line.replace('"','').replace('\n','').split(';') for line in content]
+    
+        lcont.pop(0)
+        lcont.pop(0)
+        
+        allcoord = []
+        names = []
+
+        res = [chi for chi in lcont if chi[0]=='res']
+        res = [make_res(line) for line in res]
+        coord_res = [i for i,chi in enumerate(lcont) if chi[0]=='res']
+        if option['valobs']:
+            allcoord.append([i for i,chi in enumerate(lcont) if chi[1]=='*obs*'])
+            names.append(u'Valeurs observées')
+        if option['valtheo'] :
+            allcoord.append([i for i,chi in enumerate(lcont) if chi[1]=='*exp*'])
+            names.append(u'Valeurs théoriques')
+        if option['resi'] :
+            allcoord.append([i for i,chi in enumerate(lcont) if chi[1]=='*resi*'])
+            names.append(u'Residuals')
+        if option['contrib'] :
+            allcoord.append([i for i,chi in enumerate(lcont) if chi[1]=='*contrib*'])
+            names.append(u'Contributions a posteriori')
+        if option['pourcent'] : 
+            allcoord.append([i for i,chi in enumerate(lcont) if chi[1]=='*pr*'])
+            names.append(u'Pourcentages')
+        if option['pourcentl'] :
+            allcoord.append([i for i,chi in enumerate(lcont) if chi[1]=='*prl*'])
+            names.append(u'Pourcentages en ligne')
+        if option['pourcentc'] :
+            allcoord.append([i for i,chi in enumerate(lcont) if chi[1]=='*prc*'])
+            names.append(u'Pourcentages en colonne')
+    
+        allcoord.append(coord_res)
+        allhtml = [[clean_line(lcont[allcoord[i][j]+1:allcoord[i+1][j]]) for j, line in enumerate(allcoord[i])] for i, tab in enumerate(allcoord) if i!=len(allcoord)-1]
+    
+        allhtml = [make_table(val,names[i],res) for i,val in enumerate(allhtml)]
+    
+        links = make_link_list(res, self.TextCroise)
+        #colors = [line[-1] for line in res]
+    
+    #    good = [i for i,chi in enumerate(res) if chi[-1] == 'green' or chi[-1] == 'blue']
+        #select_good = [[val[i] for i in good] for val in tout]
+    
+    
+        html_res = make_restab(res)
+        allhtml.insert(0,html_res)
+
+        titles = make_title(res, self.TextCroise)
+        allhtml.insert(0,titles)
+        
+        if option['graph'] :
+            graphs = [line[7] for line in res]
+            ListFile += graphs
+            html_graphs = make_htmlgraphs(graphs)
+            allhtml.append(html_graphs)
+
+        header=u"""
+        <html>\n
+        <meta http-equiv="content-Type" content="text/html; charset=%s" />\n
+        <body>\n
+        <h1>Test du Chi2</h1>\n
+        <br>
+        <table border=1><tr><td>
+        Légende : <br>
+        <font color=green>p &lt;= 0.05</font><br>
+        <font color=blue>p &lt;= 0.05 mais il y a des valeurs théoriques &lt; 5</font><br>
+        <font color=red>p &gt; 0.05</font>
+        </td></tr></table><br><br>
+        """%self.parent.SysEncoding
+
+    
+        pretxt = '<br>\n'.join(links)+'<br><hr><br>\n'
+        txt = '<br><hr><br>\n'.join(['<br><br>'.join([tab[i] for tab in allhtml]) for i,val in enumerate(res)])
+        txt = header + pretxt + txt + '\n</body></html>'
+
+        fileout=os.path.join(self.parametres['pathout'],'resultats-chi2.html')
+        with open(fileout, 'w') as f :
+            f.write(txt)
+        ListFile.append(fileout)         
+        return ListFile
\ No newline at end of file