Fix error caused by unexpected links in HAFAS' connection output, improve date handling, add exception raiser to threads

workers/conn.py

@@ -4,6 +4,7 @@ import datetime
 import pytz
 import threading
 import queue
+import sys
 
 import workers.val
 from classes import *
@@ -11,7 +12,8 @@ from classes import *
 def getStation(name):
  return list(workers.val.validateName(name))[0]
 
-def getService(sid, lines, q):
+def getService(sid, lines, q, eq = None):
+ try:
   dep = lines[0]
   arr = lines[1]
   det = lines[2]
@@ -37,14 +39,20 @@ def getService(sid, lines, q):
    depdts = arrdts - datetime.timedelta(days=1)
    depdate = datetime.datetime.strftime(depdts, "%d.%m.%Y")
 
- if not (walk and depdate):
+  dest = None
+
+  if not (walk or depdate):
    purl = dep.find("td", { "class": "product" }).find("a").get("href")
    psource = requests.get(purl).text
 
    zuppa = BeautifulSoup(psource, "html5lib")
    depdate = zuppa.findAll("div", { "class": "block" })[1].text.strip()
-  arrdate = arrdate or depdate
-  dest = None if walk else list(workers.val.validateName(zuppa.findAll("div", { "class": "block" })[2].text.split(":")[1].strip()))[0]
+   arrdate = depdate
+   dest = list(workers.val.validateName(zuppa.findAll("div", { "class": "block" })[2].text.split(":")[1].strip()))[0]
+
+  elif not depdate:
+   depdate = "01.01.2000"
+   arrdate = depdate
 
   depts = datetime.datetime.strptime("%s %s" % (depdate, deptime), "%d.%m.%Y %H:%M")
   arrts = datetime.datetime.strptime("%s %s" % (arrdate, arrtime), "%d.%m.%Y %H:%M")
@@ -55,7 +63,12 @@ def getService(sid, lines, q):
   svc = Service(name, depst, depts, arrst, arrts, dest, depplat, depprog, arrplat, arrprog)
   q.put((sid, svc))
 
-def getDetails(cid, url, q, via = []):
+ except Exception as e:
+  if eq:
+   eq.put(sys.exc_info())
+
+def getDetails(cid, url, q, via = [], eq = None):
+ try:
   ssource = requests.get(url).text
   suppe = BeautifulSoup(ssource, "html5lib")
 
@@ -79,7 +92,7 @@ def getDetails(cid, url, q, via = []):
   iq = queue.PriorityQueue()
 
   for line in range(0, len(lines), 3):
-  t = threading.Thread(target=getService, args=(line, lines[line:line + 3], iq))
+   t = threading.Thread(target=getService, args=(line, lines[line:line + 3], iq, eq))
    t.start()
    threads += [t]
 
@@ -91,6 +104,10 @@ def getDetails(cid, url, q, via = []):
 
   q.put((cid, conn))
 
+ except:
+  if eq:
+   eq.put(sys.exc_info())
+
 def connRequest(frm, to, count = 3, time = datetime.datetime.now(), mode = False, details = False, via = []):
  outdate = datetime.datetime.strftime(time, "%d.%m.%Y")
  outtime = datetime.datetime.strftime(time, "%H:%M")
@@ -111,20 +128,25 @@ def connRequest(frm, to, count = 3, time = datetime.datetime.now(), mode = False
   conns = []
 
   for a in juha.findAll("a"):
-   if a.get("href") and "HWAI=CONNECTION$" in a.get("href"):
+   if a.get("href") and "GO_conViewMode" in a.get("href"):
     conns += [a.get("href")]
 
   threads = []
+  eq = queue.Queue()
   q = queue.PriorityQueue()
 
-  for i in range(len(conns[:-1])):
-   t = threading.Thread(target=getDetails, args=(i, conns[i], q, via))
+  for i in range(len(conns)):
+   t = threading.Thread(target=getDetails, args=(i, conns[i], q, via, eq))
    t.start()
    threads += [t]
 
   for t in threads:
    t.join()
 
+  if not eq.empty():
+   exc = eq.get()
+   raise exc[1].with_traceback(exc[2])
+
   while not q.empty():
    yield q.get()[1]