#Copyright (c) 2008-2009 Pierre Ratinaud
#Lisense: GNU/GPL
-from chemins import ConstructPathOut, StatTxtPathOut, ffr, FFF
-from corpus import Corpus
-from guifunct import getPage, getCorpus
+from analysetxt import AnalyseText
+from guifunct import getPage, getCorpus, SelectColumn
from ConfigParser import RawConfigParser
from functions import sortedby, progressbar, CreateIraFile, exec_rcode, check_Rresult, MessageImage
from dialog import StatDialog, PrefWordCloud
-from openanalyse import OpenAnalyse
+from PrintRScript import WordCloudRScript
#from ttparser import *
import tempfile
from time import sleep
import wx
import os
+import logging
-class WordCloud():
- def __init__(self, parent, cmd = False, lem = True, exp = True):
-####################################################################
- self.conf = None
- self.parent = parent
- self.type = 'alceste'
- self.cmd = cmd
- self.ConfigPath = parent.ConfigPath
- self.DictPath = parent.DictPath
- self.KeyConf = RawConfigParser()
- self.KeyConf.read(self.ConfigPath['key'])
- 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'], 'WordCloud')
- self.dictpathout = StatTxtPathOut(self.pathout)
- self.val = wx.ID_OK
- else :
- self.corpus = Corpus(parent)
- self.corpus.parametre['encodage'] = parent.corpus_encodage
- self.corpus.parametre['lang'] = parent.corpus_lang
- self.corpus.parametre['filename'] = parent.filename
- self.pathout = ConstructPathOut(self.corpus.parametre['filename'], 'WordCloud')
- self.dictpathout = StatTxtPathOut(self.pathout)
- self.corpus.dictpathout = self.dictpathout
- if not self.cmd :
- dial = StatDialog(self,parent)
- dial.CenterOnParent()
- self.val = dial.ShowModal()
- else :
- self.val = wx.ID_OK
- if self.val == wx.ID_OK :
- if not self.cmd :
- if dial.radio_lem.GetSelection() == 0 : lem = True
- else : lem = False
- if dial.exp.GetSelection() == 0 : exp = True
- else : exp = False
- self.make_uce = dial.check_uce.GetValue()
- self.corpus.parametre['nbforme_uce'] = dial.spin_ctrl_4.GetValue()
- self.corpus.parametre['max_actives'] = dial.spin_max_actives.GetValue()
- self.corpus.parametre['eff_min_uce'] = self.corpus.parametre['nbforme_uce']
- else :
- lem = True
- exp = True
- self.make_uce = False
- self.corpus.parametre['nbforme_uce'] = None
- self.corpus.parametre['eff_min_uce'] = None
- self.corpus.parametre['lem'] = lem
- self.corpus.parametre['expressions'] = exp
- 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"]
- self.make_corpus()
+logger = logging.getLogger('iramuteq.textwordcloud')
- if self.val == wx.ID_OK :
- if 'supplementaires' not in dir(self.corpus) :
- self.corpus.supplementaires = [option for option in self.KeyConf.options('KEYS') if self.KeyConf.get('KEYS', option) == "2"]
- if 'typeactive' not in dir(self.corpus) :
- self.corpus.typeactive = [option for option in self.KeyConf.options('KEYS') if self.KeyConf.get('KEYS', option) == "1"]
- self.make_stats()
- def make_corpus(self) :
- if not self.cmd :
- self.dlg = progressbar(self, 7)
- else :
- self.dlg = None
- self.corpus.content = self.parent.content
- ucis_txt, ucis_paras_txt = self.corpus.start_analyse(self.parent, dlg = self.dlg, cmd = self.cmd, fromtt = False)
- del ucis_txt
-
- if not self.cmd :
- self.dlg.Update(5, '%i UCI...' % len(ucis_paras_txt))
- self.corpus.make_ucis_paras_uces(ucis_paras_txt, make_uce = self.make_uce)
- 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)
-
- if not self.cmd :
- self.dlg.Update(6, u'Dictionnaires')
- uces, orderuces = self.corpus.make_forms_and_uces()
- self.corpus.make_lems(self.parent.lexique)
-
- def make_stats(self):
- if not self.cmd :
- if not 'dlg' in dir(self) :
- self.dlg = progressbar(self, 7)
- if not self.corpus.parametre['lem'] :
- formes = self.corpus.formes
- else :
- formes = self.corpus.make_lem_eff()
- act = [[forme, formes[forme][0], formes[forme][2]] for forme in formes if formes[forme][2] in
- self.corpus.typeactive]
- act = sortedby(act, 2, 1)
- act = [[val[0], `val[1]`, val[2]] for val in act]
- with open(os.path.join(self.pathout, 'actives.csv'), 'w') as f:
- f.write('\n'.join([';'.join(line) for line in act]))
- #act = [[i, val] for i, val in enumerate(act)]
- self.corpus.dictpathout = self.dictpathout
- #self.corpus.make_type_tot()
-
- if not self.cmd :
- self.dlg.Destroy()
- pref = PrefWordCloud(self.parent)
- pref.CenterOnParent()
- res = pref.ShowModal()
- else :
- res = wx.ID_OK
+class WordCloud(AnalyseText):
+ def doanalyse(self) :
+ self.parametres['type'] = 'wordcloud'
+ #FIXME
+ limit = 3
+ self.make_option()
+ if self.parametres['mode'] == 2 :
+ self.actives = self.corpus.make_actives_limit(limit, 1)
+ self.actives += self.corpus.make_actives_limit(limit, 2)
+ elif self.parametres['mode'] == 0 :
+ self.actives = self.corpus.make_actives_limit(limit, 1)
+ elif self.parametres['mode'] == 1 :
+ self.actives = self.corpus.make_actives_limit(limit, 2)
+ dictcol = dict([[i, [act, self.corpus.getlemeff(act)]] for i, act in enumerate(self.actives)])
+ SelectColumn(self.ira, dictcol, self.actives, self.pathout['selected.csv'], dlg = True)
+ self.make_wordcloud()
+ script = WordCloudRScript(self)
+ script.make_script()
+ self.doR(script.scriptout)
+
+ def make_option(self) :
+ dial = PrefWordCloud(self.ira)
+ dial.CenterOnParent()
+ res = dial.ShowModal()
if res == wx.ID_OK :
- self.dlg = progressbar(self, 3)
- width = pref.spin_L.GetValue()
- height = pref.spin_H.GetValue()
- maxword = pref.spin_maxword.GetValue()
- mincex = float(pref.spin_mincex.GetValue())/float(10)
- maxcex = float(pref.spin_maxcex.GetValue())/float(10)
- col_text = pref.color_text.GetColour()
- col_bg = pref.color_bg.GetColour()
- col_text = str(col_text).replace(')', ', max=255)')
- col_bg = str(col_bg).replace(')', ', max=255)')
- outgraph = os.path.join(os.path.dirname(self.dictpathout['zipf']), 'nuage_')
+ self.parametres['width'] = dial.spin_L.GetValue()
+ self.parametres['height'] = dial.spin_H.GetValue()
+ self.parametres['maxword'] = dial.spin_maxword.GetValue()
+ self.parametres['mincex'] = float(dial.spin_mincex.GetValue())/float(10)
+ self.parametres['maxcex'] = float(dial.spin_maxcex.GetValue())/float(10)
+ self.parametres['col_text'] = dial.color_text.GetColour()
+ self.parametres['col_bg'] = dial.color_bg.GetColour()
+ self.parametres['mode'] = dial.typeformeschoice.GetSelection()
+ outgraph = os.path.join(os.path.dirname(self.pathout['zipf.png']), 'nuage_')
nb = 1
while os.path.exists(outgraph + str(nb) + '.png') :
nb += 1
- outgraph = outgraph + str(nb) + '.png'
+ self.parametres['graphout'] = outgraph + str(nb) + '.png'
+ dial.Destroy()
+
- txt = """
- source("%s")
- library(wordcloud)
- act <- read.csv2("%s", header = FALSE, row.names = 1)
- maxword <- %i
- maxtoprint <- ifelse(nrow(act) > maxword, maxword, nrow(act))
- toprint <- act[1:maxtoprint,]
- open_file_graph("%s", width = %i, height = %i)
- par(bg=rgb%s)
- wordcloud(row.names(toprint), toprint[,1], scale=c(%f,%f), random.order=FALSE, colors=rgb%s)
- dev.off()
- """ % (self.parent.RscriptsPath['Rgraph'], ffr(os.path.join(self.pathout, 'actives.csv')), maxword, ffr(outgraph), width, height, col_bg, maxcex, mincex, col_text)
- tmpscript = tempfile.mktemp(dir=self.parent.TEMPDIR)
- with open(tmpscript, 'w') as f :
- f.write(txt)
- pid = exec_rcode(self.parent.RPath, tmpscript, wait = False)
- while pid.poll() == None :
- self.dlg.Pulse('R...')
- sleep(0.2)
- check_Rresult(self.parent, pid)
- self.corpus.save_corpus(self.dictpathout['db'])
- CreateIraFile(self.dictpathout, 0, corpname = os.path.basename(self.corpus.parametre['filename']), section =
- 'wordcloud')
- #win = MessageImage(self.parent, -1, u"Graphique", size=(700, 500),style = wx.DEFAULT_FRAME_STYLE)
- #win.imagename = "nuage.png"
- #win.addsaveimage(outgraph)
- #txt = "<img src='%s'>" % FFF(outgraph).replace('//','/')
- #win.HtmlPage.SetPage(txt)
- #win.Show(True)
- if not self.cmd :
- #OpenAnalyse(self.parent, self.dictpathout['ira'])
- #self.DoLayout(self.parent)
- OpenAnalyse(self.parent, self.dictpathout['ira'])
- self.dlg.Update(7, 'fini')
- self.dlg.Destroy()
+ def make_wordcloud(self) :
+ act = ['\t'.join([act, `self.corpus.getlemeff(act)`]) for act in self.actives]
+ with open(self.pathout['actives_eff.csv'], 'w') as f :
+ f.write('\n'.join(act).encode(self.ira.syscoding))
+ ####################################################################
+# self.conf = None
+# self.parent = parent
+# self.type = 'alceste'
+# self.cmd = cmd
+# self.ConfigPath = parent.ConfigPath
+# self.DictPath = parent.DictPath
+# self.KeyConf = RawConfigParser()
+# self.KeyConf.read(self.ConfigPath['key'])
+# 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'], 'WordCloud')
+# self.dictpathout = StatTxtPathOut(self.pathout)
+# self.val = wx.ID_OK
+# else :
+# self.corpus = Corpus(parent)
+# self.corpus.parametre['encodage'] = parent.corpus_encodage
+# self.corpus.parametre['lang'] = parent.corpus_lang
+# self.corpus.parametre['filename'] = parent.filename
+# self.pathout = ConstructPathOut(self.corpus.parametre['filename'], 'WordCloud')
+# self.dictpathout = StatTxtPathOut(self.pathout)
+# self.corpus.dictpathout = self.dictpathout
+# if not self.cmd :
+# dial = StatDialog(self,parent)
+# dial.CenterOnParent()
+# self.val = dial.ShowModal()
+# else :
+# self.val = wx.ID_OK
+# if self.val == wx.ID_OK :
+# if not self.cmd :
+# if dial.radio_lem.GetSelection() == 0 : lem = True
+# else : lem = False
+# if dial.exp.GetSelection() == 0 : exp = True
+# else : exp = False
+# self.make_uce = dial.check_uce.GetValue()
+# self.corpus.parametre['nbforme_uce'] = dial.spin_ctrl_4.GetValue()
+# self.corpus.parametre['max_actives'] = dial.spin_max_actives.GetValue()
+# self.corpus.parametre['eff_min_uce'] = self.corpus.parametre['nbforme_uce']
+# else :
+# lem = True
+# exp = True
+# self.make_uce = False
+# self.corpus.parametre['nbforme_uce'] = None
+# self.corpus.parametre['eff_min_uce'] = None
+# self.corpus.parametre['lem'] = lem
+# self.corpus.parametre['expressions'] = exp
+# 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"]
+# self.make_corpus()
+#
+# if self.val == wx.ID_OK :
+# if 'supplementaires' not in dir(self.corpus) :
+# self.corpus.supplementaires = [option for option in self.KeyConf.options('KEYS') if self.KeyConf.get('KEYS', option) == "2"]
+# if 'typeactive' not in dir(self.corpus) :
+# self.corpus.typeactive = [option for option in self.KeyConf.options('KEYS') if self.KeyConf.get('KEYS', option) == "1"]
+# self.make_stats()
+#
+# def make_corpus(self) :
+# if not self.cmd :
+# self.dlg = progressbar(self, 7)
+# else :
+# self.dlg = None
+# self.corpus.content = self.parent.content
+# ucis_txt, ucis_paras_txt = self.corpus.start_analyse(self.parent, dlg = self.dlg, cmd = self.cmd, fromtt = False)
+# del ucis_txt
+#
+# if not self.cmd :
+# self.dlg.Update(5, '%i UCI...' % len(ucis_paras_txt))
+# self.corpus.make_ucis_paras_uces(ucis_paras_txt, make_uce = self.make_uce)
+# 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)
+#
+# if not self.cmd :
+# self.dlg.Update(6, u'Dictionnaires')
+# uces, orderuces = self.corpus.make_forms_and_uces()
+# self.corpus.make_lems(self.parent.lexique)
+#
+# def make_stats(self):
+# if not self.cmd :
+# if not 'dlg' in dir(self) :
+# self.dlg = progressbar(self, 7)
+# if not self.corpus.parametre['lem'] :
+# formes = self.corpus.formes
+# else :
+# formes = self.corpus.make_lem_eff()
+# act = [[forme, formes[forme][0], formes[forme][2]] for forme in formes if formes[forme][2] in
+# self.corpus.typeactive]
+# act = sortedby(act, 2, 1)
+# act = [[val[0], `val[1]`, val[2]] for val in act]
+# with open(os.path.join(self.pathout, 'actives.csv'), 'w') as f:
+# f.write('\n'.join([';'.join(line) for line in act]))
+# #act = [[i, val] for i, val in enumerate(act)]
+# self.corpus.dictpathout = self.dictpathout
+# #self.corpus.make_type_tot()
+#
+# if not self.cmd :
+# self.dlg.Destroy()
+# pref = PrefWordCloud(self.parent)
+# pref.CenterOnParent()
+# res = pref.ShowModal()
+# else :
+# res = wx.ID_OK
+# if res == wx.ID_OK :
+# self.dlg = progressbar(self, 3)
+# width = pref.spin_L.GetValue()
+# height = pref.spin_H.GetValue()
+# maxword = pref.spin_maxword.GetValue()
+# mincex = float(pref.spin_mincex.GetValue())/float(10)
+# maxcex = float(pref.spin_maxcex.GetValue())/float(10)
+# col_text = pref.color_text.GetColour()
+# col_bg = pref.color_bg.GetColour()
+# col_text = str(col_text).replace(')', ', max=255)')
+# col_bg = str(col_bg).replace(')', ', max=255)')
+# outgraph = os.path.join(os.path.dirname(self.dictpathout['zipf']), 'nuage_')
+# nb = 1
+# while os.path.exists(outgraph + str(nb) + '.png') :
+# nb += 1
+# outgraph = outgraph + str(nb) + '.png'
+#
+# txt = """
+# source("%s")
+# library(wordcloud)
+# act <- read.csv2("%s", header = FALSE, row.names = 1)
+# maxword <- %i
+# maxtoprint <- ifelse(nrow(act) > maxword, maxword, nrow(act))
+# toprint <- act[1:maxtoprint,]
+# open_file_graph("%s", width = %i, height = %i)
+# par(bg=rgb%s)
+# wordcloud(row.names(toprint), toprint[,1], scale=c(%f,%f), random.order=FALSE, colors=rgb%s)
+# dev.off()
+# """ % (self.parent.RscriptsPath['Rgraph'], ffr(os.path.join(self.pathout, 'actives.csv')), maxword, ffr(outgraph), width, height, col_bg, maxcex, mincex, col_text)
+# tmpscript = tempfile.mktemp(dir=self.parent.TEMPDIR)
+# with open(tmpscript, 'w') as f :
+# f.write(txt)
+# pid = exec_rcode(self.parent.RPath, tmpscript, wait = False)
+# while pid.poll() == None :
+# self.dlg.Pulse('R...')
+# sleep(0.2)
+# check_Rresult(self.parent, pid)
+# self.corpus.save_corpus(self.dictpathout['db'])
+# CreateIraFile(self.dictpathout, 0, corpname = os.path.basename(self.corpus.parametre['filename']), section =
+# 'wordcloud')
+# #win = MessageImage(self.parent, -1, u"Graphique", size=(700, 500),style = wx.DEFAULT_FRAME_STYLE)
+# #win.imagename = "nuage.png"
+# #win.addsaveimage(outgraph)
+# #txt = "<img src='%s'>" % FFF(outgraph).replace('//','/')
+# #win.HtmlPage.SetPage(txt)
+# #win.Show(True)
+# if not self.cmd :
+# #OpenAnalyse(self.parent, self.dictpathout['ira'])
+# #self.DoLayout(self.parent)
+# OpenAnalyse(self.parent, self.dictpathout['ira'])
+# self.dlg.Update(7, 'fini')
+# self.dlg.Destroy()
projects / iramuteq / blobdiff