Parser/OVI/EVT/prepare_main_evt_file.py

# %%
# Imports
import xml.etree.ElementTree as ET
import json
import os

basedir = '../../DATA/'
baseindir = basedir + 'OVI/datiniXML/xmlgat/'
baseoutdir = basedir + 'OVI/datiniXML/xmlevt/'
# %%
# This is to handle the xmnls attribute in the TEI element in the templates
uri1 = "{http://www.tei-c.org/ns/1.0}"
namespaces = {
    '':         "http://www.tei-c.org/ns/1.0",
}
for prefix, uri in namespaces.items():
    ET.register_namespace(prefix, uri)
# Reference directories
basedir = '../../DATA/'
baseindir = basedir + 'OVI/datiniXML/xmlgat/'
baseoutdir = basedir + 'OVI/datiniXML/xmlevt/'
# %%
# Import lems list + xml info file

# get lem list as a json object
lemfile = basedir + 'OVI/datiniXML/power_lemmarioB.json'
lems = json.load(open(lemfile, 'r'))
# %%
persNames = [lem for lem in lems if lem['lemma']['categoria']=='antr.']
placeNames = [lem for lem in lems if lem['lemma']['categoria']=='n.g.']

fileNames = os.listdir(baseoutdir + "lettere/")
# %%
print(len(persNames), len(lems), len(placeNames))
print(len(os.listdir(baseindir)), len(os.listdir(baseoutdir + 'lettere/')))
# %%
placeTemplate ='''
                    <place xml:id="PLACE_ID">
                        <settlement type="">PLACE_NAME</settlement>
                        <note></note>
                    </place>'''
personTemplate ='''
                    <person xml:id="PERSON_ID">
                        <persName>PERSON_NAME</persName>
                        <sex></sex>
                    </person>'''

fileTemplate = '<xi:include href="lettere/LETTER_FILENAME" xmlns:xi="http://www.w3.org/2001/XInclude" xpointer="LETTER_TEXTPOINT"/>'
fileIndent = '            '
# %%
persNames[10]
# %%
persListString = ""
for person in persNames:
    persListString = persListString + personTemplate.replace('PERSON_ID', ''+str(person['id'])).replace('PERSON_NAME', person['lemma']['forma_standard'])

placeListString = ""
for place in placeNames:
    placeListString = placeListString + placeTemplate.replace('PLACE_ID', ''+str(place['id'])).replace('PLACE_NAME', '"'+place['lemma']['forma_standard']+'"')

fileListString = ""
for ii, f1 in enumerate(fileNames):
    fileListString = fileListString + fileTemplate.replace('LETTER_FILENAME', f1).replace('LETTER_TEXTPOINT', f1.replace('.xml', '_text'))
    if ii<len(fileNames)-1:
        fileListString = fileListString + '\n' + fileIndent
# %%
with open('pre_main_template.xml', 'r') as f1:
    mainTemplateString = f1.read()
# %%
mainFileString = mainTemplateString.replace('<PERSON_LIST>', persListString).replace('<PLACE_LIST', placeListString).replace('<FILE_LIST>', fileListString)
# %%
with open('test_main.xml', 'w') as f2:
    f2.write(mainFileString)
# %%
with open(baseoutdir + 'main.xml', 'w') as f3:
    f3.write(mainFileString)
# %%

# %%