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
24 ######################################
26 log = logging.getLogger('iramuteq')
27 formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
28 ch = logging.StreamHandler()
29 ch.setFormatter(formatter)
31 log.setLevel(logging.DEBUG)
32 #######################################
34 log.debug('----------TEST corpusNG-----------------')
35 from analysetxt import Alceste, gramact, gramsup
36 from corpusNG import *
38 #cmd = iracmd.CmdLine(args=['-f','/home/pierre/workspace/iramuteq/corpus/lru2.txt','-t','alceste'])
40 AppliPath = os.path.abspath(os.path.dirname(os.path.realpath(sys.argv[0])))
41 if os.getenv('HOME') != None:
42 user_home = os.getenv('HOME')
44 user_home = os.getenv('HOMEPATH')
45 UserConfigPath = os.path.abspath(os.path.join(user_home, '.iramuteq'))
48 def __init__(self, args = None) :
49 self.DictPath = ConstructDicoPath(AppliPath)
50 self.ConfigPath = ConstructConfigPath(UserConfigPath)
51 self.syscoding = sys.getdefaultencoding()
52 self.TEMPDIR = tempfile.mkdtemp('iramuteq')
53 self.RscriptsPath = ConstructRscriptsPath(AppliPath)
54 self.PathPath = ConfigParser()
55 self.PathPath.read(self.ConfigPath['path'])
56 self.RPath = self.PathPath.get('PATHS', 'rpath')
57 self.pref = RawConfigParser()
58 self.pref.read(self.ConfigPath['preferences'])
59 self.history = History(self.ConfigPath['history'])
61 parser = OptionParser()
63 parser.add_option("-f", "--file", dest="filename", help="chemin du corpus", metavar="FILE", default=False)
64 parser.add_option("-t", "--type", dest="type_analyse", help="type d'analyse", metavar="TYPE D'ANALYSE", default=False)
65 parser.add_option("-c", "--conf", dest="configfile", help="chemin du fichier de configuration pour l'analyse", metavar="CONF", default=None)
66 parser.add_option("-d", "--confcorp", dest="corpusconfigfile", help="chemin du fichier de configuration pour le corpus", metavar="CONF", default=None)
67 parser.add_option("-e", "--enc", dest="encodage", help="encodage du corpus", metavar="ENC", default=locale.getpreferredencoding())
68 parser.add_option("-l", "--lang", dest="language", help="langue du corpus", metavar="LANG", default='french')
69 parser.add_option("-r", "--read", dest="read", help="lire un corpus", metavar="READ", default = False)
70 parser.add_option("-b", "--build", action="store_true", dest="build", help = "construire un corpus", default = False)
73 (options, args) = parser.parse_args()
75 (options, args) = parser.parse_args(args)
79 if options.configfile is not None:
80 config = DoConf(os.path.abspath(options.configfile)).getoptions()
81 elif options.filename and options.type_analyse :
82 config = DoConf(self.ConfigPath[options.type_analyse]).getoptions()
83 #self.ConfigPath[options.type_analyse] = os.path.abspath(options.configfile)
84 elif options.read and options.type_analyse :
85 config = DoConf(self.ConfigPath[options.type_analyse]).getoptions()
88 elif options.filename and options.build :
95 if options.filename or options.read :#or options.build:
96 self.corpus_encodage = options.encodage
97 self.corpus_lang = options.language
100 #print 'PAS DE CODECS POUR CABLE'
101 ReadLexique(self, lang = options.language)
102 self.expressions = ReadDicoAsDico(self.DictPath.get(options.language + '_exp', 'french_exp'))
103 if options.filename :
104 self.filename = os.path.abspath(options.filename)
105 if options.corpusconfigfile is not None :
106 corpus_parametres = DoConf(options.corpusconfigfile).getoptions('corpus')
108 corpus_parametres = DoConf(self.ConfigPath['corpus']).getoptions()
109 dire, corpus_parametres['filename'] = os.path.split(self.filename)
110 corpus_parametres['originalpath'] = self.filename
111 corpus_parametres['encoding'] = self.corpus_encodage
112 corpus_parametres['syscoding'] = locale.getpreferredencoding()
113 corpus_parametres['pathout'] = PathOut(self.filename, 'corpus').mkdirout()
115 corpus = BuildFromAlceste(self.filename, corpus_parametres, self.lexique, self.expressions).corpus
116 except Exception, txt:
117 log.info('probleme lors de la construction: %s' %txt)
121 self.history.add(corpus.parametres)
122 corpus = copycorpus(corpus)
124 #with codecs.open(self.filename, 'r', self.corpus_encodage) as f:
126 corpus = Corpus(self, parametres = DoConf(options.read).getoptions('corpus'), read = options.read)
127 corpus.parametres['path'] = os.path.abspath(options.read)
128 pathout = os.path.dirname(os.path.dirname(os.path.abspath(options.read)))
131 if corpus is not None :
134 corpus.parse_active(gramact, gramsup)
135 # log.warning('ATTENTION gethapaxuces')
136 # MakeUciStat(corpus)
138 #corpus.gethapaxuces()
139 #ucisize = corpus.getucisize()
140 #ucisize = [`val` for val in ucisize]
141 #uciet = [uci.etoiles[1] for uci in corpus.ucis]
142 #res = zip(uciet, ucisize)
143 #with open('ucisize.csv', 'w') as f :
144 # f.write('\n'.join(['\t'.join(val) for val in res]))
145 # self.content = f.read()
146 #self.content = self.content.replace('\r','')
147 if options.type_analyse == 'alceste' :
148 log.debug('ATTENTION : ANALYSE NG')
149 #print corpus.make_etoiles()
151 #corpus.read_corpus()
152 #corpus.parse_active(gramact, gramsup)
153 config['type'] = 'alceste'
154 self.Text = Alceste(self, corpus, parametres = config)
155 # self.Text = AnalyseAlceste(self, cmd = True, big = True)
156 #self.Text = AnalyseAlceste(self, cmd = True)
157 elif options.type_analyse == 'pam' :
158 self.Text = PamTxt(self, cmd = True)
159 elif options.type_analyse == 'afcuci' :
160 self.Text = AfcUci(self, cmd = True)
161 elif options.type_analyse == 'stat' :
162 self.Text = Stat(self, corpus, parametres = config)
163 elif options.type_analyse == 'spec' :
164 self.Text = Lexico(self, corpus, config = {'type' : 'spec'})
165 #print self.Text.corpus.hours, 'h', self.Text.corpus.minutes,'min', self.Text.corpus.seconds, 's'
166 # self.Text.corpus.make_colored_corpus('colored.html')
168 if __name__ == '__main__':