Dodana tika & format (v swagger definiciji se ni nic novega v tem commitu)

This commit is contained in:
Kikimanox
2022-08-31 14:49:47 +02:00
parent 5a87193490
commit c30b014a68
4 changed files with 316 additions and 294 deletions
+1 -1
View File
@@ -179,7 +179,7 @@
"tags": [ "tags": [
"doc-2text" "doc-2text"
], ],
"summary": "Pretvori datoteko formata pdf, doc, docx, ppt, xls,... v besedilo", "summary": "Pretvori datoteko formata pdf, doc, docx, ppt, xls,... vraca besedilo",
"requestBody": { "requestBody": {
"content": { "content": {
"multipart/form-data": { "multipart/form-data": {
@@ -1,8 +1,10 @@
from tempfile import TemporaryFile
import connexion import connexion
import six import six
# import tika # import tika
import requests import requests
import json
from swagger_server import util from swagger_server import util
import pandas as pd import pandas as pd
import docx import docx
@@ -10,6 +12,11 @@ import codecs
import xml.etree.ElementTree as ET import xml.etree.ElementTree as ET
from PyPDF2 import PdfReader from PyPDF2 import PdfReader
tika_server = "http://tika:9998/tika"
tika_server = "http://rsdo.lhrs.feri.um.si:9998/tika"
def datoteka_v_besedilo_post(file=None): # noqa: E501 def datoteka_v_besedilo_post(file=None): # noqa: E501
"""Pretvori datoteko formata pdf, doc, docx, ppt, xls,... v besedilo """Pretvori datoteko formata pdf, doc, docx, ppt, xls,... v besedilo
@@ -21,25 +28,33 @@ def datoteka_v_besedilo_post(file=None): # noqa: E501
:rtype: str :rtype: str
""" """
#zakaj ? tika server se požene zraven v dodatnem containerju in se datoteka samo pošlje tja, nazaj pa se dobi čisto besedilo...
#tika ima rest api
if file is None: if file is None:
return "No file provided", 400 return "No file provided", 400
try:
tika_server="http://tika:9998/tika" response = requests.put(tika_server, data=file)
response=requests.post(tika_server, data=file)
return response.text, 200 return response.text, 200
except Exception as e:
return str(e), 500
def get_text_ocr(file=None): # noqa: E501 def get_text_ocr(file=None):
"""Pretvori datoteko formata pdf, doc, docx, ppt, xls,... v besedilo s pomočjo ocr razpoznavanja
# noqa: E501
:param file:
:type file: strstr
:rtype: str
"""
if file is None: if file is None:
return "No file provided", 400 return "No file provided", 400
try:
tika_server="http://tika:9998/tika" response = requests.put(tika_server, data=file,
#za ocr se zraven samo doda header headers={"X-Tika-PDFOcrStrategy": "ocr_only", "X-Tika-OCRLanguage": "slv+eng"})
response=requests.post(tika_server, data=file,headers={"X-Tika-PDFOcrStrategy":"ocr_only","X-Tika-OCRLanguage":"slv+eng"})
return response.text, 200 return response.text, 200
except Exception as e:
return str(e), 500
@@ -19,9 +19,8 @@ def get_candidates(body): # noqa: E501
if connexion.request.is_json: if connexion.request.is_json:
body = IzlusciBody.from_dict(connexion.request.get_json()) # noqa: E501 body = IzlusciBody.from_dict(connexion.request.get_json()) # noqa: E501
data = { data = {
terminoloski_kandidati:[ "terminoloski_kandidati": [
{ {
"POSoznake": "Ncmsn", "POSoznake": "Ncmsn",
"kandidat": "vpliv", "kandidat": "vpliv",
+8
View File
@@ -569,4 +569,12 @@ components:
file: file:
type: string type: string
format: binary format: binary
datotekaVBesediloSync_body:
required:
- file
type: object
properties:
file:
type: string
format: binary