working on table doc

main
jgenero 3 years ago
parent 9747858f27
commit 8f07e4ee52

@ -11,6 +11,7 @@ Update :
import os
from bs4 import BeautifulSoup
from collections import OrderedDict
from peewee import *
from tqdm import tqdm
@ -51,7 +52,42 @@ def _create_produc_place(xml_file: str, folder: str)-> None:
for data in tqdm(production_places, desc="Populating Place..."):
Production_place.create(**data)
def _create_doc(xml_file: str, folder: str)-> None:
details_doc = {}
infos_doc = []
for acte in os.listdir(folder):
soup = make_soup(os.path.join(folder, acte))
# doc_id = soup.TEI["xml:id"]
inst_doc = soup.repository
nb_doc_1 = soup.msIdentifier.find_all("idno", {"n": "1"})
diplo_state = soup.body.div["type"]
# details_doc.append([inst_doc, nb_doc_1, diplo_state])
details_doc[nb_doc_1]= [inst_doc, diplo_state]
print(details_doc)
"""
for item in details_doc:
inst_query = [t.id_institution for t in Institution.select().where(
Institution.full_label == item[0].text)]
state_query = [t.id_state for t in State.select().where(
State.state_label == item[2])]
if len(item[1]) > 0:
# if there is a collection
infos_doc.append({
"inst_doc": inst_query[0],
"collection_doc": item[1][0].text,
"state_doc": state_query[0],
})
else:
# if collection is missing
infos_doc.append({
"inst_doc": inst_query[0],
"collection_doc": "",
"state_doc": state_query[0],
})
for data in tqdm(infos_doc, desc="Populating Document..."):
print(data)
Document.create(**data)
"""
def init():
db.connect()
print("Dropping existing DB...")
@ -62,7 +98,11 @@ def init():
_create_state(state)
_create_diplo_type(diplomatic_type)
_create_produc_place(xml, "./static/xml/Bourbon/Brb_5_Charles_Ier")
print("ok")
_create_doc(xml, "./static/xml/Bourbon/Brb_5_Charles_Ier")
xml = "../bourbon-latex/charles-actes-latex.xml"
init()
# init()
_create_doc(xml, "./static/xml/Bourbon/Brb_5_Charles_Ier")

Loading…
Cancel
Save