From 1301403740fe3e9487f67a07870796f9e3dfb1f9 Mon Sep 17 00:00:00 2001
From: pierre <ratinaud@univ-tlse2.fr>
Date: Sun, 4 Feb 2024 22:58:12 +0100
Subject: [PATCH] ...

---
 PrintRScript.py | 14 ++++++++++++--
 corpus.py       |  6 ++++++
 2 files changed, 18 insertions(+), 2 deletions(-)

diff --git a/PrintRScript.py b/PrintRScript.py
index 27c4d27..0699b3f 100755
--- a/PrintRScript.py
+++ b/PrintRScript.py
@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 #Author: Pierre Ratinaud
-#Copyright (c) 2008-2020 Pierre Ratinaud
-#modification pour python 3 : Laurent Mérat, 6x7 - mai 2020
+#Copyright (c) 2008-2011 Pierre Ratinaud
 #License: GNU/GPL
 
 #------------------------------------
@@ -1385,6 +1384,17 @@ class LabbeScript(PrintRScript):
         tab <- read.csv2("%s", header=TRUE, sep=';', row.names=1)
         """ % (ffr(self.pathout['tableafcm.csv']))
         txt += """
+        cs <- colSums(tab)
+        if (min(cs) == 0) {
+            print('empty columns !!')
+            vide <- which(cs==0)
+            print(vide)
+            tab <- tab[,-vide]
+        }
+        #print('#### RcppIramuteq for C++ Labbe ####')
+        #library(RcppIramuteq)
+        #dist.mat <- labbe(as.matrix(tab))
+        #rownames(dist.mat) <- colnames(tab)
         dist.mat <- dist.labbe(tab)
         dist.mat <- as.dist(dist.mat, upper=F, diag=F)
         write.table(as.matrix(dist.mat), "%s", sep='\t')
diff --git a/corpus.py b/corpus.py
index 7ab0ebb..e6b0bf2 100644
--- a/corpus.py
+++ b/corpus.py
@@ -20,6 +20,9 @@ from operator import itemgetter
 from uuid import uuid4
 import datetime
 from copy import copy
+#------test spacy------------
+#import spacy
+#nlp = spacy.load("fr_core_news_lg")
 
 #------------------------------------
 # import des fichiers du projet
@@ -1546,6 +1549,7 @@ class BuildSubCorpus(BuildCorpus):
 class BuildFromAlceste(BuildCorpus) :
 
     def read_corpus(self, infile) :
+
         if self.dlg is not None :
             self.dlg.Pulse('textes : 0 - segments : 0')
         self.limitshow = 0
@@ -1565,6 +1569,8 @@ class BuildFromAlceste(BuildCorpus) :
                     if self.testuci(line) :
                         iduci += 1
                         if txt != [] :
+                            #doc = nlp(' '.join(txt))
+                            #print([[word, word.pos_, word.lemma_] for word in doc])
                             iduce, idpara = self.treattxt(txt, iduce, idpara, iduci - 1)
                             txt = []
                             self.corpus.ucis.append(Uci(iduci, line))
-- 
2.7.4