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": [
"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": {
"content": {
"multipart/form-data": {
@@ -1,8 +1,10 @@
from tempfile import TemporaryFile
import connexion
import six
# import tika
import requests
import json
from swagger_server import util
import pandas as pd
import docx
@@ -10,6 +12,11 @@ import codecs
import xml.etree.ElementTree as ET
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
"""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
"""
#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:
return "No file provided", 400
tika_server="http://tika:9998/tika"
response=requests.post(tika_server, data=file)
try:
response = requests.put(tika_server, data=file)
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:
return "No file provided", 400
tika_server="http://tika:9998/tika"
#za ocr se zraven samo doda header
response=requests.post(tika_server, data=file,headers={"X-Tika-PDFOcrStrategy":"ocr_only","X-Tika-OCRLanguage":"slv+eng"})
try:
response = requests.put(tika_server, data=file,
headers={"X-Tika-PDFOcrStrategy": "ocr_only", "X-Tika-OCRLanguage": "slv+eng"})
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:
body = IzlusciBody.from_dict(connexion.request.get_json()) # noqa: E501
data={
terminoloski_kandidati:[
data = {
"terminoloski_kandidati": [
{
"POSoznake": "Ncmsn",
"kandidat": "vpliv",
@@ -31,7 +30,7 @@ def get_candidates(body): # noqa: E501
6.0274563,
6.0274563
],
"pogostostpojavljanja": [101,71]
"pogostostpojavljanja": [101, 71]
},
{
"POSoznake": "Agpnsg Ncnsg",
@@ -42,7 +41,7 @@ def get_candidates(body): # noqa: E501
3.263,
2.134
],
"pogostostpojavljanja": [27,11]
"pogostostpojavljanja": [27, 11]
},
{
"POSoznake": "Agpmsny Ncmsn Ncmpg",
@@ -53,7 +52,7 @@ def get_candidates(body): # noqa: E501
1.221,
3.323
],
"pogostostpojavljanja": [31,51]
"pogostostpojavljanja": [31, 51]
},
{
"POSoznake": "Agpnsn Ncnsn",
@@ -64,7 +63,7 @@ def get_candidates(body): # noqa: E501
0.883,
1.02
],
"pogostostpojavljanja": [1241,111]
"pogostostpojavljanja": [1241, 111]
}
]
}
+9 -1
View File
@@ -3,7 +3,7 @@ info:
title: OpenAPI definition
version: v0
servers:
- url: http://localhost:8089
- url: http://localhost:8089
description: Generated server url
paths:
/job/{job_id}:
@@ -569,4 +569,12 @@ components:
file:
type: string
format: binary
datotekaVBesediloSync_body:
required:
- file
type: object
properties:
file:
type: string
format: binary