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

rsdo5.json

@@ -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": {

swagger_server/controllers/doc2text_controller.py

@@ -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
 

swagger_server/controllers/extract_controller.py

@@ -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:[
+        "terminoloski_kandidati": [
             {
                 "POSoznake": "Ncmsn",
                 "kandidat": "vpliv",

swagger_server/swagger/swagger.yaml

@@ -569,4 +569,12 @@ components:
         file:
           type: string
           format: binary
+    datotekaVBesediloSync_body:
+      required:
+        - file
+      type: object
+      properties:
+        file:
+          type: string
+          format: binary