X-Git-Url: http://iramuteq.org/git?p=iramuteq;a=blobdiff_plain;f=PrintRScript.py;h=d925a4d84094e14fc9bb05126a755a8601239ec9;hp=be81a14f83f627d69df85a589b1aaec710f1ea3c;hb=5d10357688f47bf9ee5f7fdf356e146fc164bdc9;hpb=4c4a0f5c8a9c145bc81fa93df5d1b0af34399eaf

diff --git a/PrintRScript.py b/PrintRScript.py
index be81a14..d925a4d 100644
--- a/PrintRScript.py
+++ b/PrintRScript.py
@@ -164,15 +164,17 @@ def RchdTxt(DicoPath, RscriptPath, mincl, classif_mode, nbt = 9, svdmethod = 'sv
         """ % ffr(DicoPath['TableUc2'])
     txt += """
     log1 <- "%s"
-    chd1<-CHD(data1, x = nbt, mode.patate = mode.patate, svd.method =
-    svd.method, libsvdc.path = libsvdc.path, log.file = log1)
+    #print('FIXME : source newCHD')
+    #source('/home/pierre/workspace/iramuteq/Rscripts/newCHD.R')
+    #chd1<-CHD(data1, x = nbt, mode.patate = mode.patate, svd.method = svd.method, libsvdc.path = libsvdc.path, find='matrix', sample=20, amp=500)
+    chd1<-CHD(data1, x = nbt, mode.patate = mode.patate, svd.method = svd.method, libsvdc.path = libsvdc.path)#, log.file = log1)
     """ % ffr(DicoPath['log-chd1.txt'])
 
     if classif_mode == 0:
         txt += """
     log2 <- "%s"
     chd2<-CHD(data2, x = nbt, mode.patate = mode.patate, svd.method =
-    svd.method, libsvdc.path = libsvdc.path) log.file = log2)
+    svd.method, libsvdc.path = libsvdc.path)#, log.file = log2)
     """ % ffr(DicoPath['log-chd2.txt'])
 
     txt += """
@@ -196,21 +198,29 @@ def RchdTxt(DicoPath, RscriptPath, mincl, classif_mode, nbt = 9, svdmethod = 'sv
     txt += """
     classif_mode <- %i
     mincl <- %i
+    if (mincl == 0) {mincl <- round(nrow(chd1$n1)/(nbt+1))}
     uceout <- "%s"
+    write.csv2(chd1$n1, file="%s")
     if (classif_mode == 0) {
         chd.result <- Rchdtxt(uceout, chd1, chd2 = chd2, mincl = mincl,classif_mode = classif_mode, nbt = nbt)
+        classeuce1 <- chd.result$cuce1
+        tree.tot1 <- make_tree_tot(chd1)
+        tree.cut1 <- make_dendro_cut_tuple(tree.tot1$dendro_tuple, chd.result$coord_ok, classeuce1, 1, nbt)
+
     } else {
-        chd.result <- Rchdtxt(uceout, chd1, chd2 = chd1, mincl = mincl,classif_mode = classif_mode, nbt = nbt)
+        #chd.result <- Rchdtxt(uceout, chd1, chd2 = chd1, mincl = mincl,classif_mode = classif_mode, nbt = nbt)
+        tree.tot1 <- make_tree_tot(chd1)
+        terminales <- find.terminales(chd1$n1, chd1$list_mere, chd1$list_fille, mincl)
+        tree.cut1 <- make.classes(terminales, chd1$n1, tree.tot1$tree.cl, chd1$list_fille)
+        write.csv2(tree.cut1$n1, uceout)
+        chd.result <- tree.cut1
     }
-    n1 <- chd.result$n1
-    classeuce1 <- chd.result$cuce1
-    classes<-n1[,ncol(n1)]
-    write.csv2(n1, file="%s")
-    rm(n1)
-    """ % (classif_mode, mincl, ffr(DicoPath['uce']), ffr(DicoPath['n1.csv']))
+    classes<-chd.result$n1[,ncol(chd.result$n1)]
+    write.csv2(chd.result$n1, file="%s")
+    """ % (classif_mode, mincl, ffr(DicoPath['uce']), ffr(DicoPath['n1-1.csv']), ffr(DicoPath['n1.csv']))
 
     txt += """
-    tree.tot1 <- make_tree_tot(chd1)
+#    tree.tot1 <- make_tree_tot(chd1)
 #    open_file_graph("%s", widt = 600, height=400)
 #    plot(tree.tot1$tree.cl)
 #    dev.off()
@@ -226,8 +236,7 @@ def RchdTxt(DicoPath, RscriptPath, mincl, classif_mode, nbt = 9, svdmethod = 'sv
         """ % ffr(DicoPath['arbre2'] )
 
     txt += """
-    tree.cut1 <- make_dendro_cut_tuple(tree.tot1$dendro_tuple, chd.result$coord_ok, classeuce1, 1, nbt)
-    save(tree.cut1, file="%s")
+        save(tree.cut1, file="%s")
 
     open_file_graph("%s", width = 600, height=400)
     plot.dendropr(tree.cut1$tree.cl,classes, histo=TRUE)
@@ -337,9 +346,14 @@ datasup<-read.csv2("%s", header = FALSE, sep = ';',quote = '\"', row.names = 1,
 dataet<-read.csv2("%s", header = FALSE, sep = ';',quote = '\"', row.names = 1, na.strings = 'NA')
 """ % (ffr(DictChdTxtOut['Contout']), ffr(DictChdTxtOut['ContSupOut']), ffr(DictChdTxtOut['ContEtOut']))
     txt += """
-tablesqrpact<-BuildProf(as.matrix(dataact),n1,clnb)
-tablesqrpsup<-BuildProf(as.matrix(datasup),n1,clnb)
-tablesqrpet<-BuildProf(as.matrix(dataet),n1,clnb)
+print('ATTENTION NEW BUILD PROF')
+#tablesqrpact<-BuildProf(as.matrix(dataact),n1,clnb)
+#tablesqrpsup<-BuildProf(as.matrix(datasup),n1,clnb)
+#tablesqrpet<-BuildProf(as.matrix(dataet),n1,clnb)
+tablesqrpact<-new.build.prof(as.matrix(dataact),n1,clnb)
+tablesqrpsup<-new.build.prof(as.matrix(datasup),n1,clnb)
+tablesqrpet<-new.build.prof(as.matrix(dataet),n1,clnb)
+
 """
     txt += """
 PrintProfile(n1,tablesqrpact[4],tablesqrpet[4],tablesqrpact[5],tablesqrpet[5],clnb,"%s","%s",tablesqrpsup[4],tablesqrpsup[5])
@@ -723,6 +737,7 @@ class PrintSimiScript(PrintRScript) :
             else :
                 txt += """
                 word <- FALSE
+                index <- NULL
                 """
             txt += """
             dm <-readMM(dm.path)
@@ -847,10 +862,13 @@ class PrintSimiScript(PrintRScript) :
             txt += """
             mat <- graph.word(mat, index)
             cs <- colSums(mat)
-            if (length(cs)) mat <- mat[,-which(cs==0)]
+            if (length(which(cs==0))) mat <- mat[,-which(cs==0)]
             rs <- rowSums(mat)
-            if (length(rs)) mat <- mat[-which(rs==0),]
-            if (length(cs)) dm <- dm[, -which(cs==0)]
+            if (length(which(rs==0))) mat <- mat[-which(rs==0),]
+            if (length(which(cs==0))) dm <- dm[,-which(cs==0)]
+            if (word) {
+                index <- which(colnames(mat)==forme)
+            }
             """
 
         if self.parametres['layout'] == 0 : layout = 'random'
@@ -858,6 +876,8 @@ class PrintSimiScript(PrintRScript) :
         if self.parametres['layout'] == 2 : layout = 'frutch'
         if self.parametres['layout'] == 3 : layout = 'kawa'
         if self.parametres['layout'] == 4 : layout = 'graphopt'
+        if self.parametres['layout'] == 5 : layout = 'spirale'
+        if self.parametres['layout'] == 6 : layout = 'spirale3D'
 
 
         self.filename=''
@@ -877,7 +897,7 @@ class PrintSimiScript(PrintRScript) :
             while os.path.exists(os.path.join(dirout,'web_'+str(graphnb))):
                 graphnb +=1
             self.filename = ffr(os.path.join(dirout,'web_'+str(graphnb)))
-            os.mkdir(self.filename)        
+            os.mkdir(self.filename)
             self.filename = os.path.join(self.filename, 'gexf.gexf')
         if self.parametres['type_graph'] == 4 : 
             graphnb = 1
@@ -940,7 +960,7 @@ class PrintSimiScript(PrintRScript) :
         if self.parametres['film'] : 
             txt += """
             film <- "%s"
-            """ % self.pathout['film']
+            """ % ffr(self.pathout['film'])
         else : 
             txt += """
             film <- NULL
@@ -1064,7 +1084,7 @@ class PrintSimiScript(PrintRScript) :
         txt += """
         eff <- colSums(dm)
         x <- list(mat = mat, eff = eff)
-        graph.simi <- do.simi(x, method='%s', seuil = seuil, p.type = '%s', layout.type = '%s', max.tree = %s, coeff.vertex=%s, coeff.edge = %s, minmaxeff = minmaxeff, vcexminmax = vcexminmax, cex = cex, coords = coords, communities = communities, halo = halo)
+        graph.simi <- do.simi(x, method='%s', seuil = seuil, p.type = '%s', layout.type = '%s', max.tree = %s, coeff.vertex=%s, coeff.edge = %s, minmaxeff = minmaxeff, vcexminmax = vcexminmax, cex = cex, coords = coords, communities = communities, halo = halo, index.word=index)
         """ % (method, type, layout, arbremax, coeff_tv, coeff_te)
             
         if self.parametres.get('bystar',False) :
@@ -1207,11 +1227,12 @@ class ExportAfc(PrintRScript) :
         """
 
 class MergeGraphes(PrintRScript) :
-    def __init__(self, parametres):
+    def __init__(self, analyse):
         self.script = u"#Script genere par IRaMuTeQ - %s\n" % datetime.now().ctime()
         self.pathout = PathOut()
-        self.parametres = parametres
+        self.parametres = analyse.parametres
         self.scriptout = self.pathout['temp']
+        self.analyse = analyse 
 
     def make_script(self) :
         #FIXME
@@ -1227,7 +1248,7 @@ class MergeGraphes(PrintRScript) :
         V(g)$weight <- (graph.simi$mat.eff/nrow(dm))*100
         graphs[['%s']] <- g
         """
-        for i, graph in enumerate(self.parametres['lgraphes']) :
+        for i, graph in enumerate(self.parametres['graphs']) :
             path = os.path.dirname(graph)
             gname = ''.join(['g', `i`])
             RData = os.path.join(path,'RData.RData')
@@ -1315,7 +1336,7 @@ class LabbeScript(PrintRScript) :
                       self.analyse.parent.RscriptsPath['Rgraph']])
         txt = """
         tab <- read.csv2("%s", header=TRUE, sep=';', row.names=1)
-        """ % (self.pathout['tableafcm.csv'])
+        """ % (ffr(self.pathout['tableafcm.csv']))
         txt += """
         dist.mat <- dist.labbe(tab)
         dist.mat <- as.dist(dist.mat, upper=F, diag=F)
@@ -1327,8 +1348,171 @@ class LabbeScript(PrintRScript) :
         par(cex=1.2)
         plot.phylo(as.phylo(chd), type='unrooted', lab4ut="axial")
         dev.off()
-        """ % (self.pathout['distmat.csv'], self.pathout['dist-labbe.png'])
+        """ % (ffr(self.pathout['distmat.csv']), ffr(self.pathout['labbe-tree.png']))
+        txt +="""
+        open_file_graph("%s", width=1000, height=1000, svg=F)
+        par(mar=c(10,1,1,10))
+        heatmap(as.matrix(dist.mat), symm = T, distfun=function(x) as.dist(x), margins=c(10,10))
+        dev.off()
+        """ % ffr(self.pathout['labbe-heatmap.png'])
+        txt += """
+        #http://stackoverflow.com/questions/3081066/what-techniques-exists-in-r-to-visualize-a-distance-matrix
+        dst <- data.matrix(dist.mat)
+        dim <- ncol(dst)
+        rn <- row.names(as.matrix(dist.mat))
+        open_file_graph("%s", width=1500, height=1000, svg=F)
+        par(mar=c(10,10,3,3))
+        image(1:dim, 1:dim, dst, axes = FALSE, xlab="", ylab="")
+        axis(1, 1:dim, rn, cex.axis = 0.9, las=3)
+        axis(2, 1:dim, rn, cex.axis = 0.9, las=1)
+        text(expand.grid(1:dim, 1:dim), sprintf("%%0.2f", dst), cex=0.6)
+        dev.off()
+        """  % ffr(self.pathout['labbe-matrix.png'])
+        self.add(txt)
+        self.write()
+
+class ChronoChi2Script(PrintRScript) :
+    def make_script(self) :
+        self.sources([self.analyse.parent.RscriptsPath['Rgraph']])
+        print self.parametres
+        txt = """
+        inRData <- "%s"
+        dendrof <- "%s"
+        load(inRData)
+        load(dendrof)
+        """ % (ffr(self.pathout['RData.RData']), ffr(self.pathout['dendrogramme.RData']))
+        txt += """
+        svg <- %s
+        """ % self.parametres['svg']
+        txt += """
+        tc <- which(grepl("%s",rownames(chistabletot)))
+        rn <- rownames(chistabletot)[tc]
+        tc <- tc[order(rn)]
+        dpt <- chistabletot[tc,]
+        tot <- afctable[tc,]
+        tcp <- rowSums(tot)
+        ptc <- tcp/sum(tcp)
+        dpt <- t(dpt)
+        dd <- dpt
+        """ % self.parametres['var'].replace(u'*', u"\\\\*")
+        txt += """
+        classes <- n1[,ncol(n1)]
+        tcl <- table(classes)
+        if ('0' %in% names(tcl)) {
+            to.vire <- which(names(tcl) == '0')
+            tcl <- tcl[-to.vire]
+        }
+        tclp <- tcl/sum(tcl)
+
+        #chi2 colors
+        library(ape)
+        k <- 1e-02
+        lcol <- NULL
+        lk <- k
+        for (i in 1:5) {
+            lcol <- c(lcol, qchisq(1-k,1))
+            k <- k/10
+            lk <- c(lk,k)
+        }
+        lcol <- c(3.84, lcol)
+        lcol <- c(-Inf,lcol)
+        lcol <- c(lcol, Inf)
+        lk <- c(0.05,lk)
+        breaks <- lcol
+        alphas <- seq(0,1, length.out=length(breaks))
+        clod <- rev(as.numeric(tree.cut1$tree.cl$tip.label))
+        #end
+        """
+        txt += """
+        open_file_graph("%s", w=%i, h=%i, svg=svg)
+        """ % (ffr(self.parametres['tmpgraph']), self.parametres['width'], self.parametres['height'])
+        txt += """
+        par(mar=c(3,3,3,3))
+        mat.graphic <- matrix(c(rep(1,nrow(dd)),c(2:(nrow(dd)+1))), ncol=2)
+        mat.graphic <- rbind(mat.graphic, c(max(mat.graphic) + 1 , max(mat.graphic) + 2))
+        hauteur <- tclp[clod] * 0.9
+        heights.graphic <- append(hauteur, 0.1)
+        layout(mat.graphic, heights=heights.graphic, widths=c(0.15,0.85))
+        par(mar=c(0,0,0,0))
+        tree.toplot <- tree.cut1$tree.cl
+        num.label <- as.numeric(tree.cut1$tree.cl$tip.label)
+        col.tree <- rainbow(length(num.label))[num.label]
+        #tree.toplot$tip.label <- paste('classe ', tree.toplot$tip.label)
+        plot.phylo(tree.toplot,label.offset=0.1, cex=1.1, no.margin=T, tip.color = col.tree)
+        for (i in clod) {
+            print(i)
+            par(mar=c(0,0,0,0))
+            lcol <- cut(dd[i,], breaks, include.lowest=TRUE)
+            ulcol <- names(table(lcol))
+            lcol <- as.character(lcol)
+            for (j in 1:length(ulcol)) {
+                lcol[which(lcol==ulcol[j])] <- j
+            }
+            lcol <- as.numeric(lcol)
+            mcol <- rainbow(nrow(dd))[i]
+            last.col <- NULL
+            for (k in alphas) {
+                last.col <- c(last.col, rgb(r=col2rgb(mcol)[1]/255, g=col2rgb(mcol)[2]/255, b=col2rgb(mcol)[3]/255, a=k))
+            }
+            #print(last.col)
+
+            barplot(rep(1,ncol(dd)), width=ptc, names.arg=FALSE, axes=FALSE, col=last.col[lcol], border=rgb(r=0, g=0, b=0, a=0.3))
+        }
+        plot(0,type='n',axes=FALSE,ann=FALSE)
+        label.coords <- barplot(rep(1, ncol(dd)), width=ptc, names.arg = F, las=2, axes=F, ylim=c(0,1), plot=T, col='white')
+        text(x=label.coords, y=0.5, labels=rn[order(rn)], srt=90)
+        dev.off()
+        """
+        self.add(txt)
+        self.write()
+
+class ChronoPropScript(PrintRScript) :
+    def make_script(self) :
+        self.sources([self.analyse.parent.RscriptsPath['Rgraph']])
+        print self.parametres
+        txt = """
+        inRData <- "%s"
+        dendrof <- "%s"
+        load(inRData)
+        load(dendrof)
+        """ % (ffr(self.pathout['RData.RData']), ffr(self.pathout['dendrogramme.RData']))
+        txt += """
+        svg <- %s
+        """ % self.parametres['svg']
+        txt += """
+        tc <- which(grepl("%s",rownames(chistabletot)))
+        rn <- rownames(chistabletot)[tc]
+        tc <- tc[order(rn)]
+        dpt <- chistabletot[tc,]
+        tot <- afctable[tc,]
+        tcp <- rowSums(tot)
+        ptc <- tcp/sum(tcp)
+        dpt <- t(dpt)
+        dd <- dpt
+        """ % self.parametres['var'].replace(u'*', u"\\\\*")
+        txt += """
+        classes <- n1[,ncol(n1)]
+        tcl <- table(classes)
+        if ('0' %in% names(tcl)) {
+            to.vire <- which(names(tcl) == '0')
+            tcl <- tcl[-to.vire]
+        }
+        tclp <- tcl/sum(tcl)
+        """
+        txt += """
+        open_file_graph("%s", w=%i, h=%i, svg=svg)
+        """ % (ffr(self.parametres['tmpgraph']), self.parametres['width'], self.parametres['height'])
+        txt+= """
+        ptt <- prop.table(as.matrix(tot), 1)
+        par(mar=c(10,2,2,2))
+        barplot(t(ptt)[as.numeric(tree.cut1$tree.cl$tip.label),], col=rainbow(ncol(ptt))[as.numeric(tree.cut1$tree.cl$tip.label)], width=ptc, las=3, space=0.05, cex.axis=0.7, border=NA)
+        dev.off()
+        """
         self.add(txt)
         self.write()
 
+class ReDoProfScript(PrintRScript) :
+    def make_script(self) :
+        self.sources([self.analyse.parent.RscriptsPath['chdfunct.R']])
+        print self.parametres