2 # -*- coding: utf-8 -*-
3 #Author: Pierre Ratinaud
4 #Copyright (c) 2010 Pierre Ratinaud
8 from optparse import OptionParser
13 sys.setdefaultencoding(locale.getpreferredencoding())
14 from chemins import ConstructConfigPath, ConstructDicoPath, ConstructRscriptsPath
15 from functions import ReadLexique, DoConf, History
16 from ConfigParser import *
17 #######################################
18 #from textchdalc import AnalyseAlceste
19 #from textdist import PamTxt
20 #from textafcuci import AfcUci
21 from textaslexico import Lexico
22 from textstat import Stat
23 from tools import SubCorpus
25 ######################################
27 log = logging.getLogger('iramuteq')
28 formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
29 ch = logging.StreamHandler()
30 ch.setFormatter(formatter)
32 log.setLevel(logging.DEBUG)
33 #######################################
35 log.debug('----------TEST corpusNG-----------------')
36 from analysetxt import Alceste, gramact, gramsup
37 from corpusNG import *
39 #cmd = iracmd.CmdLine(args=['-f','/home/pierre/workspace/iramuteq/corpus/lru2.txt','-t','alceste'])
41 AppliPath = os.path.abspath(os.path.dirname(os.path.realpath(sys.argv[0])))
42 if os.getenv('HOME') != None:
43 user_home = os.getenv('HOME')
45 user_home = os.getenv('HOMEPATH')
46 UserConfigPath = os.path.abspath(os.path.join(user_home, '.iramuteq'))
49 def __init__(self, args = None) :
50 self.DictPath = ConstructDicoPath(AppliPath)
51 self.ConfigPath = ConstructConfigPath(UserConfigPath)
52 self.syscoding = sys.getdefaultencoding()
53 self.TEMPDIR = tempfile.mkdtemp('iramuteq')
54 self.RscriptsPath = ConstructRscriptsPath(AppliPath)
55 self.PathPath = ConfigParser()
56 self.PathPath.read(self.ConfigPath['path'])
57 self.RPath = self.PathPath.get('PATHS', 'rpath')
58 self.pref = RawConfigParser()
59 self.pref.read(self.ConfigPath['preferences'])
60 self.history = History(os.path.join(UserConfigPath, 'history.db'))
62 parser = OptionParser()
64 parser.add_option("-f", "--file", dest="filename", help="chemin du corpus", metavar="FILE", default=False)
65 parser.add_option("-t", "--type", dest="type_analyse", help="type d'analyse", metavar="TYPE D'ANALYSE", default=False)
66 parser.add_option("-c", "--conf", dest="configfile", help="chemin du fichier de configuration pour l'analyse", metavar="CONF", default=None)
67 parser.add_option("-d", "--confcorp", dest="corpusconfigfile", help="chemin du fichier de configuration pour le corpus", metavar="CONF", default=None)
68 parser.add_option("-e", "--enc", dest="encodage", help="encodage du corpus", metavar="ENC", default=locale.getpreferredencoding())
69 parser.add_option("-l", "--lang", dest="language", help="langue du corpus", metavar="LANG", default='french')
70 parser.add_option("-r", "--read", dest="read", help="lire un corpus", metavar="READ", default = False)
71 parser.add_option("-b", "--build", action="store_true", dest="build", help = "construire un corpus", default = False)
74 (options, args) = parser.parse_args()
76 (options, args) = parser.parse_args(args)
80 if options.configfile is not None:
81 config = DoConf(os.path.abspath(options.configfile)).getoptions()
82 elif options.filename and options.type_analyse :
83 config = DoConf(self.ConfigPath[options.type_analyse]).getoptions()
84 #self.ConfigPath[options.type_analyse] = os.path.abspath(options.configfile)
85 elif options.read and options.type_analyse :
86 config = DoConf(self.ConfigPath[options.type_analyse]).getoptions()
89 elif options.filename and options.build :
96 if options.filename or options.read :#or options.build:
97 self.corpus_encodage = options.encodage
98 self.corpus_lang = options.language
101 #print 'PAS DE CODECS POUR CABLE'
102 ReadLexique(self, lang = options.language)
103 self.expressions = ReadDicoAsDico(self.DictPath.get(options.language + '_exp', 'french_exp'))
104 if options.filename :
105 self.filename = os.path.abspath(options.filename)
106 if options.corpusconfigfile is not None :
107 corpus_parametres = DoConf(options.corpusconfigfile).getoptions('corpus')
109 corpus_parametres = DoConf(self.ConfigPath['corpus']).getoptions()
110 dire, corpus_parametres['filename'] = os.path.split(self.filename)
111 corpus_parametres['originalpath'] = self.filename
112 corpus_parametres['encoding'] = self.corpus_encodage
113 corpus_parametres['syscoding'] = locale.getpreferredencoding()
114 corpus_parametres['pathout'] = PathOut(self.filename, 'corpus').mkdirout()
116 corpus = BuildFromAlceste(self.filename, corpus_parametres, self.lexique, self.expressions).corpus
117 except Exception, txt:
118 log.info('probleme lors de la construction: %s' %txt)
122 self.history.add(corpus.parametres)
123 corpus = copycorpus(corpus)
125 #with codecs.open(self.filename, 'r', self.corpus_encodage) as f:
127 corpus = Corpus(self, parametres = DoConf(options.read).getoptions('corpus'), read = options.read)
128 corpus.parametres['pathout'] = os.path.dirname(os.path.abspath(options.read))
129 pathout = os.path.dirname(os.path.dirname(os.path.abspath(options.read)))
132 if corpus is not None :
134 corpus = SubCorpus(self, corpus, [0,1,2,3,4,5,6,7])
137 corpus.parse_active(gramact, gramsup)
138 print corpus.getlemconcorde('de').fetchall()
139 # log.warning('ATTENTION gethapaxuces')
140 # MakeUciStat(corpus)
142 #corpus.gethapaxuces()
143 #ucisize = corpus.getucisize()
144 #ucisize = [`val` for val in ucisize]
145 #uciet = [uci.etoiles[1] for uci in corpus.ucis]
146 #res = zip(uciet, ucisize)
147 #with open('ucisize.csv', 'w') as f :
148 # f.write('\n'.join(['\t'.join(val) for val in res]))
149 # self.content = f.read()
150 #self.content = self.content.replace('\r','')
151 if options.type_analyse == 'alceste' :
152 log.debug('ATTENTION : ANALYSE NG')
153 #print corpus.make_etoiles()
155 #corpus.read_corpus()
156 #corpus.parse_active(gramact, gramsup)
157 config['type'] = 'alceste'
158 self.Text = Alceste(self, corpus, parametres = config)
159 # self.Text = AnalyseAlceste(self, cmd = True, big = True)
160 #self.Text = AnalyseAlceste(self, cmd = True)
161 elif options.type_analyse == 'pam' :
162 self.Text = PamTxt(self, cmd = True)
163 elif options.type_analyse == 'afcuci' :
164 self.Text = AfcUci(self, cmd = True)
165 elif options.type_analyse == 'stat' :
166 self.Text = Stat(self, corpus, parametres = {'type':'stat'})
167 elif options.type_analyse == 'spec' :
168 self.Text = Lexico(self, corpus, config = {'type' : 'spec'})
169 #print self.Text.corpus.hours, 'h', self.Text.corpus.minutes,'min', self.Text.corpus.seconds, 's'
170 # self.Text.corpus.make_colored_corpus('colored.html')
172 if __name__ == '__main__':