diff options
| author | Jinesh K J <jinsbond007@jinesh.cvit> | 2009-04-07 03:10:02 +0530 |
|---|---|---|
| committer | Jinesh K J <jinsbond007@jinesh.cvit> | 2009-04-07 03:10:02 +0530 |
| commit | 1b62cfff2a910765b700bbea15786a1d62d800ef (patch) | |
| tree | 459ff9e5ea25fe93a6114060d6e65f0ab1601559 /silpa/modules/ngram/visualizer.py | |
| parent | 97f2231e942f67450b27f161da11e863334921ba (diff) | |
| download | AnjaliOldLipi.git-1b62cfff2a910765b700bbea15786a1d62d800ef.tar.gz AnjaliOldLipi.git-1b62cfff2a910765b700bbea15786a1d62d800ef.tar.xz AnjaliOldLipi.git-1b62cfff2a910765b700bbea15786a1d62d800ef.zip | |
Updated the ngram module with option parser
Diffstat (limited to 'silpa/modules/ngram/visualizer.py')
| -rw-r--r-- | silpa/modules/ngram/visualizer.py | 19 |
1 files changed, 1 insertions, 18 deletions
diff --git a/silpa/modules/ngram/visualizer.py b/silpa/modules/ngram/visualizer.py index e2c1136..585c11f 100644 --- a/silpa/modules/ngram/visualizer.py +++ b/silpa/modules/ngram/visualizer.py @@ -26,7 +26,7 @@ import pickle class NGramVisualizer: depth=0 - def loadCorpus(self,new_file_name,corpus_file_name): + def loadCorpus(self,corpus,corpus_file_name): limiters = [".","!","?",",",";"] try: corpusfile = open(corpus_file_name) @@ -35,23 +35,6 @@ class NGramVisualizer: else: graph_dict = pickle.load(corpusfile) # graph_dict = dict() - line = [] - line_number = 0 - rule_number = 0 - corpus="" - data_file = codecs. open(new_file_name,encoding='utf-8', errors='ignore') - while 1: - line_number = line_number +1 - text = unicode( data_file.readline()) - if text == "": - break - if text[0] == '#': - continue - line_number = line_number +1 - line = text.strip() - if(line == ""): - continue - corpus=corpus+" "+line sentences=[] sentence = "" start = 0 |