korpus_xml failide elementidele id lisamine

addIds.py

+#/usr/bin/python
+# -*- coding: utf-8 -*-
+
+import os
+import sys
+import pathlib
+import re
+from bs4 import BeautifulSoup
+from pathlib import Path
+
+
+
+#lähtefailide kataloog
+#väljundis säilitame sama struktuuri
+
+indir = '/Users/rabauti/repos/tu/ut_veakorpus/korpus_xml'
+outdir = '/Users/rabauti/repos/tu/ut_veakorpus/korpus_ids'
+
+files = os.listdir(indir)
+
+#RP_algtekst_2004/A7.xml
+file_pattern_txt = re.compile('([^\/]+).txt$')
+
+for folder in files:
+	
+	folder = folder.strip()
+	files2 = os.listdir(indir+'/'+folder)
+	for name2 in files2:
+		if file_pattern_txt.search(name2):
+			
+			#avame faili lugemiseks
+			
+			
+			
+			
+			filename = indir + '/' + folder + '/' + name2
+			name2 = name2.replace('Veakorpus','').replace(', uus','').replace(' uus','')
+			filename2 = (outdir + '/' + folder + '/' + name2)#.replace('.txt', '.xml')
+			
+			foldername_components = folder.split('_')
+			
+			
+			name2_components = (name2.split('.'))
+			corrector = foldername_components[0]
+			year = foldername_components[2]
+			student = 	name2_components[0]
+			
+			docId =  corrector + year + '_' + student
+			
+			
+			#docId = docId.replace('_algtekst_', '_')
+			
+			
+			
+			soup = BeautifulSoup(open(filename, 'r').read(), "xml")
+				
+			path = Path(outdir + '/' + folder)
+			path.mkdir(parents=True, exist_ok=True)
+			
+			dokument = soup.find('dokument')
+			dokument['id'] = docId
+			
+			
+			for (i, eksimus) in enumerate(dokument.findAll('eksimus')):
+				eksimus['id'] = '%s_e%d' % (docId, i+1)
+				for (j, algne) in enumerate(eksimus.findAll('algne')):
+					algne['id'] = '%s_e%d_a%d' % (docId, i+1, j+1)
+				
+				for (j, parandus) in enumerate(eksimus.findAll('parandus')):
+					parandus['id'] = '%s_e%d_p%d' % (docId, i+1, j+1)
+				
+				for (j, kommentaar) in enumerate(eksimus.findAll('kommentaar')):
+					kommentaar['id'] = '%s_e%d_k%d' % (docId, i+1, j+1)
+				
+			
+					
+			file_out = open(filename2, 'w')
+			outXML = str(soup) 
+			
+			outXML = outXML.replace('<eksimus', '\n<eksimus')
+			
+			file_out.write(outXML)
+			file_out.close()
+			
\ No newline at end of file