Merge pull request #315 from lalalaurentiu/main

Fixed Allianx and BAT scraper
peviitor-ro · Nov 22, 2023 · 97221bd · 97221bd
2 parents 10e6270 + 2d457a9
commit 97221bd
Show file tree

Hide file tree

Showing 2 changed files with 32 additions and 17 deletions.
diff --git a/sites/allianz.py b/sites/allianz.py
@@ -32,14 +32,21 @@
         id = uuid.uuid4()
         job_title = element.find("a").text.strip()
         job_link = "https://careers.allianz.com" + element.find("a").get("href")
+        location = element.find("div", {"class": "location"}).find("div").text.strip()
+        remote = []
+
+        if "RO" in location:
+            remote.append("Hybrid")
 
         finaljobs.append({
             "id": str(id),
             "job_title": job_title,
             "job_link": job_link,
             "company": company.get("company"),
             "country": "Romania",
-            "city": "Romania",
+            "city": "Bucuresti",
+            "county": "Bucuresti",
+            "remote": remote
         })
 
 #Afisam numarul de total de joburi

diff --git a/sites/bat.py b/sites/bat.py
@@ -1,5 +1,6 @@
 from scraper_peviitor import Scraper, loadingData, Rules
-import uuid
+from utils import (translate_city)
+from getCounty import get_county, remove_diacritics
 import json
 
 url = "https://careers.bat.com/search-jobs/results?ActiveFacetID=0&CurrentPage=1&RecordsPerPage=1000&Distance=100&RadiusUnitType=0&Location=Romania&Latitude=46.00000&Longitude=25.00000&ShowRadius=False&IsPagination=False&FacetType=0&SearchResultsModuleName=Search+Results&SearchFiltersModuleName=Search+Filters&SortCriteria=0&SortDirection=0&SearchType=1&LocationType=2&LocationPath=798549&OrganizationIds=1045&ResultsType=0"
@@ -21,34 +22,41 @@
 finalJobs = list()
 
 for job in jobs:
-    id = uuid.uuid4()
     job_title = job.find("h3").text
     job_link = "https://careers.bat.com" + job.find("a").get("href")
-    city = job.find("span", {"class": "job-location"}).text.split(",")[0].strip()
-
-    finalJobs.append(
-        {
-            "id": str(id),
-            "job_title": job_title,
-            "job_link": job_link,
-            "company": company.get("company"),
-            "country": "Romania",
-            "city": city,
-        }
+    city = translate_city(
+        remove_diacritics(
+            job.find("span", {"class": "job-location"}
+                     ).text.split(",")[0].strip()
+        )
     )
+    county = get_county(city)
+
+    if county:
+        finalJobs.append(
+            {
+                "job_title": job_title,
+                "job_link": job_link,
+                "company": company.get("company"),
+                "country": "Romania",
+                "city": city,
+                "county": county,
+            }
+        )
 
 print(json.dumps(finalJobs, indent=4))
 
+
 loadingData(finalJobs, company.get("company"))
 
 logoUrl = "https://cdn.radancy.eu/company/1045/v2_0/img/temporary/shared/bat-logo.svg"
 
 scraper.session.headers.update({
     "Content-Type": "application/json",
 })
-scraper.post( "https://api.peviitor.ro/v1/logo/add" ,json.dumps([
+scraper.post("https://api.peviitor.ro/v1/logo/add", json.dumps([
     {
-        "id":company.get("company"),
-        "logo":logoUrl
+        "id": company.get("company"),
+        "logo": logoUrl
     }
 ]))