lkrsnik/svala-scripts
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Fixed some bugs.

Browse Source
This commit is contained in:
Luka
2022-05-30 07:07:11 +02:00
parent ecaaecb459
commit 5ec5cf3a28
2 changed files with 309 additions and 129 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
src/create_tei.py
View File

@@ -142,14 +142,6 @@ class TeiDocument:
text = etree.SubElement(root, 'text')
body = etree.SubElement(text, 'body')
for paras, bibl in self.divs:
div = etree.Element('div')
set_xml_attr(div, 'id', xml_id)
div.append(bibl)
for para in paras:
div.append(para.as_xml())
body.append(div)
encoding_desc = etree.SubElement(tei_header, 'encodingDesc')
tags_decl = etree.SubElement(encoding_desc, 'tagsDecl')
namespace = etree.SubElement(tags_decl, 'namespace')
@@ -159,6 +151,16 @@ class TeiDocument:
tag_usage = etree.SubElement(namespace, 'tagUsage')
tag_usage.set('gi', tag)
tag_usage.set('occurs', str(count))
for paras, bibl, div_id in self.divs:
div = etree.Element('div')
set_xml_attr(div, 'id', div_id)
div.append(bibl)
for para in paras:
div.append(para.as_xml())
body.append(div)
return root
def add_paragraph(self, paragraph):
@@ -245,7 +247,7 @@ def build_links(all_edges):
if len(token_edges['source_ids']) > 0:
random_source_id = token_edges['source_ids'][0]
sentence_id += '.'.join(random_source_id.split('.')[:3])
elif len(token_edges['target_ids']) > 0:
if len(token_edges['target_ids']) > 0:
random_target_id = token_edges['target_ids'][0]
if len(token_edges['source_ids']) > 0:
sentence_id += ' #'