Leo+Arriva dataframes

001_cd_scrapovani.py

@@ -148,7 +148,8 @@ def cd(odkud, kam, pocet_dni=0):
     ("České Budějovice", "Zürich"),
     ("Praha", "Záhřeb"),
     ("Praha", "Lublaň"),
-    ("Praha", "Amsterdam")
+    ("Praha", "Amsterdam"),
+    ("Praha","Krakow")
 ]
 
 random.shuffle(trasy_b)

004_rj_cisteni.ipynb

@@ -2762,7 +2762,7 @@
     "    kam = \"data\" \n",
     "    os.makedirs(kam, exist_ok=True)\n",
     "    den = []\n",
-    "    rj = [y for y in os.listdir(f\"downloads/{x}\") if \"rj_\" in y] \n",
+    "    rj = [y for y in os.listdir(f\"downloads/{x}\") if y[0:3] == \"rj_\"] \n",
     "    print(f\"{x}: {len(rj)}\")\n",
     "    for y in rj:\n",
     "        print(y)\n",

005_le_scrapovani.py

@@ -11,17 +11,19 @@
 from selenium.webdriver.common.keys import Keys
 from pyvirtualdisplay import Display
 
-def leo(odkud, kam, pocet_dni):
+def leo(odkud1, odkud2, kam1, kam2, pocet_dni):
 
     try:
 
         display = Display(visible=0, size=(1920, 1080))
         display.start()
-        driver = webdriver.Chrome()
-
+#        driver = webdriver.Chrome()
+
+        driver = webdriver.Firefox(executable_path="/usr/local/bin/geckodriver")
+
         sablona = "https://www.leoexpress.com/cs/vysledky-vyhledavani?date=KDY_POJEDEME&from=MISTO_ODKUD1&fromCountry=&fromName=MISTO_ODKUD2&persons=%5B%7B%22adult%22%3A%5B%5D%2C%22combine%22%3A1%2C%22count%22%3A1%2C%22parentTariffs%22%3Anull%7D%5D&returnDate=&services=%5B%7B%22service_id%22%3A%223%22%2C%22count%22%3A0%7D%2C%7B%22service_id%22%3A%224%22%2C%22count%22%3A0%7D%5D&to=MISTO_KAM1&toCountry=&toName=MISTO_KAM2&toggleDiscounts=false"""
         den = data(pocet_dni)
-        url = sablona.replace('KDY_POJEDEME',den).replace('MISTO_ODKUD1',odkud.upper()).replace('MISTO_ODKUD2',odkud).replace('MISTO_KAM1',kam.upper()).replace('MISTO_KAM2',kam)
+        url = sablona.replace('KDY_POJEDEME',den).replace('MISTO_ODKUD1',odkud1).replace('MISTO_ODKUD2',odkud2).replace('MISTO_KAM1',kam1).replace('MISTO_KAM2',kam2)
 
         slozka = f"downloads/{datetime.now().strftime('%Y-%m-%d')}"
         os.makedirs(slozka, exist_ok=True)
@@ -30,17 +32,41 @@ def leo(odkud, kam, pocet_dni):
         wait = WebDriverWait(driver, 10)
         sleep(random.randint(4, 6))
 
+        try:
+            driver.execute_script(
+            """document.getElementById('CybotCookiebotDialog').style.display = 'none';"""
+            )
+        except:
+            pass
+
+        element = WebDriverWait(driver, 10).until(
+            EC.presence_of_element_located((By.XPATH, f"//*[text()='Přímý spoj']"))
+        )
+        element.click()
+
+        try:
+
+            elements = driver.find_elements(By.XPATH, "//*[contains(text(), 'Přímý spoj')]")
+            for element in elements[1:]:
+                sleep(1)
+                element.click()
+
+        except:
+            pass
+
+        print("Dál už to nejde.")
+
         with open(
             os.path.join(
                 slozka,
-                f"le_{odkud}_{kam}_D{pocet_dni:02}_{datetime.now().strftime('%Y-%m-%d_%H-%M-%S')}.html",
+                f"le_{odkud2}_{kam2}_D{pocet_dni:02}_{datetime.now().strftime('%Y-%m-%d_%H-%M-%S')}.html",
             ),
             "w+",
             encoding="utf-8",
         ) as ven:
             ven.write(driver.page_source)
 
-        print(f"Uloženo: {odkud}-{kam} {pocet_dni} D")
+        print(f"Uloženo: {odkud2}-{kam2} {pocet_dni} D")
 
     except Exception as E:
         print(E)
@@ -59,25 +85,29 @@ def leo(odkud, kam, pocet_dni):
 
 def data(days):
     future_date = date.today() + timedelta(days=days)
-    formatted_date = future_date.strftime("%d.%m.%Y")  # Change "." to "-" for DD-MM-YYYY
+    formatted_date = future_date.strftime("%d.%m.%Y")
     return formatted_date
 
-odstup = [0, 1, 2, 3, 4]
-odstup.append(random.randint(5, 7))
-odstup.append(random.randint(8, 10))
-odstup.append(random.randint(11, 15))
+odstup = [0, 1, 2, 3]
+odstup.append(random.randint(4, 7))
+odstup.append(random.randint(8, 15))
 odstup.append(random.randint(16, 30))
 odstup.append(random.randint(31, 90))
 odstup
 
 trasy = [
-    ('Praha','Ostrava'),
-    ('Pardubice','Košice')
-]
+    (['5457076','Praha'],['OSTRAVA','Ostrava'])
+    ]
+
+trasy_b = [(['5457076','Praha'],['5100028','Krak%C3%B3w%20G%C5%82%C3%B3wny']),
+    (['PARDUBICE','Pardubice'],['KOSICE','Ko%C5%A1ice'])]
 
 random.shuffle(trasy)
+random.shuffle(trasy_b)
+
+trasy = trasy + [trasy_b[0]]
 
 for t in trasy:
     for o in odstup:
-        leo(t[0], t[1], o)
-        leo(t[1], t[0], o)
+        leo(t[0][0], t[0][1], t[1][0], t[1][1], o)
+        leo(t[1][0], t[1][1], t[0][0], t[0][1], o)