scrapují se pouze neoscrapované dny

DataRozhlas · Nov 10, 2024 · e997ad9 · e997ad9
1 parent 05a2f2a
commit e997ad9
Show file tree

Hide file tree

Showing 15 changed files with 611,484 additions and 176,251 deletions.
diff --git a/.gitignore b/.gitignore
@@ -2,4 +2,8 @@ downloads/
 data_raw/
 .ipynb_checkpoints/
 pokus.txt
-pokusy
+pokusy
+cd_*.parquet
+ar_*.parquet
+le_*.parquet
+rj_*.parquet
diff --git a/001_cd_scrapovani.py b/001_cd_scrapovani.py
@@ -124,6 +124,7 @@ def cd(odkud, kam, pocet_dni=0):
                     f"Projáníčka, propáníčka, pro ztracené korále – máme všichni namále: {e}."
                 )
 
+
 trasy_a = [("Praha", "Brno"), ("Praha", "Ostrava")]
 
 trasy_b = [
@@ -132,7 +133,7 @@ def cd(odkud, kam, pocet_dni=0):
     ("Praha", "Budapešť"),
     ("Praha", "Vídeň"),
     ("Praha", "Bratislava"),
-    ("Praha", "Varšava")
+    ("Praha", "Varšava"),
 ]
 
 trasy_c = [
@@ -149,18 +150,18 @@ def cd(odkud, kam, pocet_dni=0):
     ("Praha", "Záhřeb"),
     ("Praha", "Lublaň"),
     ("Praha", "Amsterdam"),
-    ("Praha","Krakow")
+    ("Praha", "Krakow"),
 ]
 
 random.shuffle(trasy_b)
 random.shuffle(trasy_c)
 
 trasy = trasy_a + trasy_b[0:2] + [trasy_c[0]]
 
-dny = [0, 1, 2, 3, 4]
-dny.append(random.randint(5, 7))
-dny.append(random.randint(8, 10))
-dny.append(random.randint(11, 15))
+dny = [0, 1, 2, 3]
+dny.append(random.randint(4, 5))
+dny.append(random.randint(6, 7))
+dny.append(random.randint(8, 15))
 dny.append(random.randint(16, 30))
 dny.append(random.randint(31, 90))
 

diff --git a/002_cd_cisteni.ipynb b/002_cd_cisteni.ipynb
diff --git a/003_rj_scrapovani.py b/003_rj_scrapovani.py
@@ -60,25 +60,25 @@ def uloz_stranku(pocet_dni):
 
         pole1.send_keys(Keys.ENTER)
 
-        sleep(random.randint(4, 6))
+        sleep(random.randint(4, 5))
         pole2 = wait.until(
         EC.presence_of_element_located((By.CSS_SELECTOR, "[aria-label='Kam']"))
         )
 
         print("Vyťukávám KAM.")
 
         pole2.send_keys(kam)
-        sleep(random.randint(3, 5))
+        sleep(random.randint(3, 4))
         pole2.send_keys(Keys.ENTER)
-        sleep(random.randint(3, 5))
+        sleep(random.randint(3, 4))
 
         print("Posílám druhý ENTER.")
 
         pole2.send_keys(Keys.ENTER)
 
         wait.until(EC.presence_of_element_located((By.TAG_NAME, "body")))
 
-        sleep(random.randint(3, 5))
+        sleep(random.randint(3, 4))
 
         uloz_stranku(pocet_dni=0)
 
@@ -110,19 +110,25 @@ def uloz_stranku(pocet_dni):
             except Exception as e:
                 print(e)
 
-trasy = [
-     ('Praha','Brno'),
+trasy_a = [('Praha','Brno'),
+           ('Praha','Ostrava')]
+
+trasy_b = [
      ('Olomouc','Przemysl'),
      ('Kolín','Ústí nad Labem'),
      ('Ostrava','Brno'),
      ('Brno','Vídeň'),
      ('Praha','Budapešť'),
      ('Praha','Košice'),
      ('Praha','Čop'),
+     ('Praha','Vídeň'),
+     ('Praha','Bratislava'),
+     ('Praha','Krakov')
 ]
 
+random.shuffle(trasy_b)
+trasy = trasy_a + trasy_b[:3]
 random.shuffle(trasy)
-trasy = trasy[:-3]
 
 for t in trasy:
     regiojet(t[0],t[1])