webapps/term_extractor
18
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

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

Browse Source
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
Download Patch File Download Diff File Expand all files Collapse all files
swagger_server/controllers/doc2text_controller.py
+30 -15
View File
@@ -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)
return response.text, 200
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"})
return response.text, 200
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
swagger_server/controllers/extract_controller.py
+24 -25
View File
@@ -19,53 +19,52 @@ 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",
"kanonicnaoblika": "vpliv",
"nosilnautez": 0.8008282,
"podporneutezi": [
6.0274563,
6.0274563
6.0274563,
6.0274563
],
"pogostostpojavljanja": [101,71]
},
{
"pogostostpojavljanja": [101, 71]
},
{
"POSoznake": "Agpnsg Ncnsg",
"kandidat": "bivalen okolje",
"kanonicnaoblika": "bivalno okolje",
"nosilnautez": 0.60254,
"podporneutezi": [
3.263,
2.134
3.263,
2.134
],
"pogostostpojavljanja": [27,11]
},
{
"pogostostpojavljanja": [27, 11]
},
{
"POSoznake": "Agpmsny Ncmsn Ncmpg",
"kandidat": "motorični status otrok",
"kanonicnaoblika": "motorični status otrok",
"nosilnautez": 0.3324,
"podporneutezi": [
1.221,
3.323
1.221,
3.323
],
"pogostostpojavljanja": [31,51]
},
{
"pogostostpojavljanja": [31, 51]
},
{
"POSoznake": "Agpnsn Ncnsn",
"kandidat": "diplomsko delo",
"kanonicnaoblika": "diplomsko delo",
"nosilnautez": 0.2008282,
"podporneutezi": [
0.883,
1.02
0.883,
1.02
],
"pogostostpojavljanja": [1241,111]
}
]
}
"pogostostpojavljanja": [1241, 111]
}
]
}
return data