|
|
|
@ -4,6 +4,7 @@ import os |
|
|
|
canto_dir = os.path.abspath(os.getcwd()) |
|
|
|
print('\n\n====================' + canto_dir + '====================') |
|
|
|
|
|
|
|
|
|
|
|
def add_tags(canto_txt_file): |
|
|
|
# Only grab 0XX-canticle from `*.txt` file names |
|
|
|
xmlfile = canto_txt_file.split('.')[0] + '.xml' |
|
|
|
@ -64,16 +65,19 @@ def add_tags(canto_txt_file): |
|
|
|
def add_space(match): |
|
|
|
match = match.group() |
|
|
|
return re.sub('<w>', ' <w>', match) |
|
|
|
j = 1 |
|
|
|
for line in Cxmltmp: |
|
|
|
# Swap double spaces from new_word with single spaces |
|
|
|
line_spaced = re.sub(' ',' ',line) |
|
|
|
# If there's a single space at the beginning of the line, delete it |
|
|
|
line_spaced = line_spaced.lstrip(' ') |
|
|
|
# Add three tabs and opening line tags at the beginning |
|
|
|
line_spaced = re.sub('^','\t\t\t\t\t<l> ',line_spaced) |
|
|
|
line_tag = '\t\t\t\t\t<l l_num="' + str(j) + '"> ' |
|
|
|
line_spaced = re.sub('^',line_tag,line_spaced) |
|
|
|
# Add closing line tags |
|
|
|
line_spaced = re.sub('\n',' </l>\n',line_spaced) |
|
|
|
Cxml.write(line_spaced) |
|
|
|
j += 1 |
|
|
|
Cxmltmp.close() |
|
|
|
os.remove(tmpxmlfile) |
|
|
|
Cxml.close() |
|
|
|
|