Camelot reading

This commit is contained in:
2021-11-02 15:37:36 -03:00
parent 5ee267568a
commit 0c44554375
18 changed files with 121 additions and 22 deletions

View File

@ -5,6 +5,7 @@ from flask import Flask, request
import contabilidad.pdf as pdf
import contabilidad.passwords as passwords
import contabilidad.log as log
import contabilidad.text_handler as th
app = Flask(__name__)
@ -18,17 +19,21 @@ def pdf_parse():
data['files'] = [data['files']]
password_file = '/app/config/.passwords.yml'
pwds = passwords.get_passwords(password_file)
texts = []
output = []
for file in data['files']:
filename = os.path.realpath(os.path.join('/app/data', file['filename']))
texts = []
for p in pwds:
obj = pdf.get_data(filename, p)
print(obj)
obj = pdf.get_text(filename, p)
if obj is None:
continue
print(obj)
texts.append(json.dumps(obj))
return json.dumps(texts)
text = th.text_cleanup(obj, file['filename'])
texts.append(text)
output.append({'filename': file['filename'], 'text': texts})
return json.dumps(output)
if __name__ == '__main__':
app.run(host='0.0.0.0')
app.run(host='0.0.0.0', debug=True)