From 0eb942456239b2708a7b3b9fb628b0d0a4f89e5d Mon Sep 17 00:00:00 2001
From: "James E. Blair" <corvus@gnu.org>
Date: Sat, 21 Sep 2019 15:52:51 -0700
Subject: Update United parsing

---
 email_assistant/assistant.py      |  1 +
 email_assistant/plugins/united.py | 64 +++++++++++++++++++++------------------
 2 files changed, 35 insertions(+), 30 deletions(-)

diff --git a/email_assistant/assistant.py b/email_assistant/assistant.py
index d39b016..e6bfc66 100644
--- a/email_assistant/assistant.py
+++ b/email_assistant/assistant.py
@@ -198,6 +198,7 @@ class Assistant:
             try:
                 for p in self.plugins:
                     if p.match(msg):
+                        self.log.debug("Matched %s", p.name)
                         events = p.get_events(msg)
                         break
             except Exception:
diff --git a/email_assistant/plugins/united.py b/email_assistant/plugins/united.py
index 38de3e2..0ed5dd8 100644
--- a/email_assistant/plugins/united.py
+++ b/email_assistant/plugins/united.py
@@ -31,28 +31,18 @@ import vobject
 from email_assistant import iata
 from email_assistant import plugin
 
-def parse_dep_arr(flight_date, dep_arr):
-    flight_year = dateutil.parser.parse(flight_date).year
-    city, br, code, flight_time = dep_arr.span.children
-    city = city.strip()
-    code = code.strip()[1:-1]
-    code = code.split()[0]
+def parse_dep_arr(flight_date, flight_time, flight_loc):
+    code = re.compile('^.*\((...)\)$').match(flight_loc).group(1)
     tz = iata.tzmap[code]
-    flight_time = flight_time.get_text().strip()
-    m = re.match(r'(.*) \((\d+[A-Z]+)\)', flight_time)
-    if m:
-        s = '%s%s %s' % (m.group(2), flight_year, m.group(1))
-        flight_time = dateutil.parser.parse(s)
-    else:
-        flight_time = dateutil.parser.parse(flight_date +' '+ flight_time)
+    flight_time = dateutil.parser.parse(flight_date + ' ' + flight_time)
     flight_time = flight_time.replace(tzinfo=dateutil.tz.gettz(tz))
-    return (city, code, flight_time)
+    return (code, flight_time)
 
 class Plugin(plugin.Plugin):
     name = 'united'
 
     def match(self, msg):
-        if ('unitedairlines@united.com' in msg['From'] and
+        if ('Receipts@united.com' in msg['From'] and
             'Itinerary and Receipt' in msg['Subject']):
             return True
 
@@ -61,32 +51,47 @@ class Plugin(plugin.Plugin):
         for part in msg.walk():
             if part.get_content_type() == 'text/html':
                 soup = BeautifulSoup(part.get_payload(decode=True), 'html.parser')
-                # confirmation_number = soup.find(class_="eTicketConfirmation").string
 
                 index = 0
                 while True:
-                    info = soup.find(id="ShowSegments_ShowSegment_ctl%02i_Flight" % index)
+                    index += 1
+                    info = soup.find(string=re.compile('Flight %s of' % index))
                     if not info:
                         break
-                    for row in info.parents:
-                        if row.name == 'tr':
-                            break
+                    while info.name != 'table': info = info.parent
+
+                    row = info.find('tr')
+                    cols = row.find_all('td')
+                    cols = [x.strip() for x in row.strings if x.strip()]
+                    flight_num, flight_class = cols
+
+                    row = row.nextSibling
+                    cols = row.find_all('td')
+                    cols = [x.strip() for x in row.strings if x.strip()]
+                    dep_date, arr_date = cols
+
+                    row = row.nextSibling
+                    cols = row.find_all('td')
+                    cols = [x.strip() for x in row.strings if x.strip()]
+                    dep_time, arr_time = cols
+
+                    row = row.nextSibling
                     cols = row.find_all('td')
-                    flight_date, flight_num, flight_class, dep, arr, ac, meal = cols
-                    flight_date = flight_date.get_text().strip()
-                    flight_num = flight_num.get_text().strip()
-                    flight_class = flight_class.get_text().strip()
+                    cols = [x.strip() for x in row.strings if x.strip()]
+                    dep_loc, arr_loc = cols
 
-                    dep_city, dep_code, dep_time = parse_dep_arr(flight_date, dep)
-                    arr_city, arr_code, arr_time = parse_dep_arr(flight_date, arr)
-                    self.log.debug("dep: %s %s %s", dep_city, dep_code, dep_time)
-                    self.log.debug("arr: %s %s %s", arr_city, arr_code, arr_time)
+                    flight_num = flight_num.split()[-1]
+                    dep_code, dep_time = parse_dep_arr(dep_date, dep_time, dep_loc)
+                    arr_code, arr_time = parse_dep_arr(arr_date, arr_time, arr_loc)
+
+                    self.log.debug("dep: %s %s", dep_code, dep_time)
+                    self.log.debug("arr: %s %s", arr_code, arr_time)
 
                     cal = vobject.iCalendar()
                     event = cal.add('vevent')
                     event.add('dtstart').value = dep_time
                     event.add('dtend').value = arr_time
-                    summary = "Flight from %s to %s" % (dep_code, arr_code)
+                    summary = "Flight %s from %s to %s" % (flight_num, dep_code, arr_code)
                     event.add('summary').value = summary
                     text = inscriptis.get_text(str(soup))
                     text = re.sub(r'([^ ]+)\s*\n', '\\1\n', text)
@@ -95,5 +100,4 @@ class Plugin(plugin.Plugin):
                     uid = hashlib.sha1((str(dep_time) + summary).encode('utf8')).hexdigest()
                     event.add('uid').value = uid
                     events.append(cal)
-                    index += 1
         return events
-- 
cgit v1.2.3