알바천국 웹스크랩

Author: DeveloperYun

homework6.py

@@ -0,0 +1,64 @@
+import os
+import csv
+import requests
+from bs4 import BeautifulSoup
+
+os.system("clear")
+
+
+def write_company(company):
+    file = open(f"{company['name']}.csv", mode="w")
+    writer = csv.writer(file)
+    writer.writerow(["place", "title", "time", "pay", "date"])
+    for job in company["jobs"]:
+      writer.writerow(list(job.values()))
+    print(f"Completed....{company['name']}")
+
+
+alba_url = "http://www.alba.co.kr"
+
+alba_request = requests.get(alba_url)
+alba_soup = BeautifulSoup(alba_request.text, "html.parser")
+main = alba_soup.find("div", {"id": "MainSuperBrand"})
+brands = main.find_all("li", {"class": "impact"})
+for brand in brands:
+    link = brand.find("a", {"class": "goodsBox-info"})
+    name = brand.find("span", {"class": "company"})
+    if link and name:
+        link = link["href"]
+        name = name.text
+        if "/" in name :
+          name = name.replace("/"," ")
+        company = {'name': name, 'jobs': []}
+        jobs_request = requests.get(link)
+        jobs_soup = BeautifulSoup(jobs_request.text, "html.parser")
+        tbody = jobs_soup.find("div", {"id": "NormalInfo"}).find("tbody")
+        rows = tbody.find_all("tr", {"class": ""})
+        for row in rows:
+            local = row.find("td", {"class": "local"})
+            if local:
+                local = local.text
+            title = row.find("td", {"class": "title"})
+            if title:
+                title = title.find("a").find("span", {
+                    "class": "company"
+                }).text.strip()
+
+            time = row.find("td", {"class": "data"})
+            if time:
+                time = time.text
+            pay = row.find("td", {"class": "pay"})
+            if pay:
+                pay = pay.text
+            date = row.find("td", {"class": "regDate"})
+            if date:
+                date = date.text
+            job = {
+                "place": local,
+                "title": title,
+                "time": time,
+                "pay": pay,
+                "date": date
+            }
+            company['jobs'].append(job)
+        write_company(company)