Merge pull request #252 from lalalaurentiu/main

Fixed Inetum scraper
peviitor-ro · Nov 12, 2023 · 4e96b8a · 4e96b8a
2 parents 1472127 + e86c066
commit 4e96b8a
Showing 1 changed file with 26 additions and 16 deletions.
diff --git a/sites/inetum.py b/sites/inetum.py
@@ -1,6 +1,7 @@
 from scraper_peviitor import Scraper, Rules, loadingData
-import uuid
 import json
+from getCounty import get_county
+from utils import translate_city, acurate_city_and_county
 
 url = "https://www.inetum.com/en/jobs?f%5B0%5D=region%3A1068"
 
@@ -21,32 +22,41 @@
     jobs = rules.getTags("div", {"class": "node node-job node-teaser"})
 
     for job in jobs:
-        id = uuid.uuid4()
         job_title = job.find("h3", {"class":"card-title"}).text.strip()
         job_link = "https://www.inetum.com" + job.find("a").get("href")
-        city = job.find("p", {"class": "card-text"}).text.split("-")[-1].strip()
+        city = translate_city(job.find("p", {"class": "card-text"}).text.split("-")[-1].split("/")[0].strip())
+        county = get_county(city)
+        remote = []
+
+        jobs_types = ["Remote", "Hybrid"]
+
+
+        for types in jobs_types:
+            if types in job.find("p", {"class": "card-text"}).text.split("-")[-1].strip():
+                remote.append(types) 
 
         finalJobs.append({
-            "id": str(id),
             "job_title": job_title,
             "job_link": job_link,
             "company": company.get("company"),
             "country": "Romania",
-            "city": city
+            "city": city,
+            "county": county,
+            "remote": remote,
         })
 
 print(json.dumps(finalJobs, indent=4))
 
-loadingData(finalJobs, company.get("company"))
+# loadingData(finalJobs, company.get("company"))
 
-logoUrl = "https://vtlogo.com/wp-content/uploads/2021/05/inetum-vector-logo-small.png"
+# logoUrl = "https://vtlogo.com/wp-content/uploads/2021/05/inetum-vector-logo-small.png"
 
-scraper.session.headers.update({
-    "Content-Type": "application/json",
-})
-scraper.post( "https://api.peviitor.ro/v1/logo/add/" ,json.dumps([
-    {
-        "id":company.get("company"),
-        "logo":logoUrl
-    }
-]))
+# scraper.session.headers.update({
+#     "Content-Type": "application/json",
+# })
+# scraper.post( "https://api.peviitor.ro/v1/logo/add/" ,json.dumps([
+#     {
+#         "id":company.get("company"),
+#         "logo":logoUrl
+#     }
+# ]))