W22

Medevio5_ReadNamesFromKartoteka_html.py

@@ -0,0 +1,42 @@
+import mysql.connector
+from bs4 import BeautifulSoup
+import re
+
+
+# ---------- CONFIG ----------
+# MySQL connection settings (fill in)
+MYSQL_CFG = dict(
+    host="192.168.1.76",
+    port=3307,
+    user="root",
+    password="Vlado9674+",
+    database="medevio",
+)
+
+conn=mysql.connector.connect(**MYSQL_CFG)
+cur=conn.cursor()
+cur.execute("select html from kartoteka_html where 'fetched-at'=(SELECT MAX('fetched-at') FROM kartoteka_html)")
+html=cur.fetchone()
+html=html[0]
+
+
+# html is the string containing the entire web page
+soup = BeautifulSoup(html, "html.parser")
+
+# Find every <button> that has that specific class sequence
+# (space-separated class names → match as a set)
+buttons = soup.find_all(
+    "button",
+    class_="MuiTypography-root MuiTypography-body2 "
+           "MuiLink-root MuiLink-underlineAlways "
+           "MuiLink-button css-xf7pf8"
+)
+names = []
+for btn in buttons:
+    text = btn.get_text(strip=True)
+    print(text)
+    names.append(text)
+
+print(names)
+# names = [btn.get_text(strip=True) for btn in buttons]
+print(names)