White reset at paragraphs not sentences + progress bar updates on paragraphs not sentences.
This commit is contained in:
		
							parent
							
								
									552f2e4bd0
								
							
						
					
					
						commit
						f1366548b6
					
				@ -197,12 +197,14 @@ def file_sentence_generator(et, args):
 | 
				
			|||||||
    previous_pc = False
 | 
					    previous_pc = False
 | 
				
			||||||
 | 
					
 | 
				
			||||||
    words = {}
 | 
					    words = {}
 | 
				
			||||||
    sentences = list(et.iter('s'))
 | 
					    paragraphs = list(et.iter('p'))
 | 
				
			||||||
    for sentence in progress(sentences, "load-text"):
 | 
					    for paragraph in progress(paragraphs, "load-text"):
 | 
				
			||||||
 | 
					        previous_glue = ''
 | 
				
			||||||
 | 
					        sentences = list(paragraph.iter('s'))
 | 
				
			||||||
 | 
					        for sentence in sentences:
 | 
				
			||||||
            # create fake root word
 | 
					            # create fake root word
 | 
				
			||||||
            words[sentence.get('id')] = Word.fake_root_word(sentence.get('id'))
 | 
					            words[sentence.get('id')] = Word.fake_root_word(sentence.get('id'))
 | 
				
			||||||
            last_word_id = None
 | 
					            last_word_id = None
 | 
				
			||||||
        previous_glue = ''
 | 
					 | 
				
			||||||
 | 
					
 | 
				
			||||||
            if args.new_tei:
 | 
					            if args.new_tei:
 | 
				
			||||||
                for w in sentence.iter():
 | 
					                for w in sentence.iter():
 | 
				
			||||||
@ -235,11 +237,6 @@ def file_sentence_generator(et, args):
 | 
				
			|||||||
                        if last_word_id:
 | 
					                        if last_word_id:
 | 
				
			||||||
                            words[last_word_id].glue += w.text
 | 
					                            words[last_word_id].glue += w.text
 | 
				
			||||||
 | 
					
 | 
				
			||||||
        # for w in sentence.iter("w"):
 | 
					 | 
				
			||||||
        #     words[w.get('id')] = Word.from_xml(w, do_msd_translate)
 | 
					 | 
				
			||||||
        # for pc in sentence.iter(pc_tag):
 | 
					 | 
				
			||||||
        #     words[pc.get('id')] = Word.pc_word(pc, do_msd_translate)
 | 
					 | 
				
			||||||
 | 
					 | 
				
			||||||
            for l in sentence.iter("link"):
 | 
					            for l in sentence.iter("link"):
 | 
				
			||||||
                if 'dep' in l.keys():
 | 
					                if 'dep' in l.keys():
 | 
				
			||||||
                    ana = l.get('afun')
 | 
					                    ana = l.get('afun')
 | 
				
			||||||
 | 
				
			|||||||
		Loading…
	
		Reference in New Issue
	
	Block a user