fix pledge box; add generic parser; fix KeyError: tracking_number

Author: ljmerza

custom_components/email/const.py

@@ -16,3 +16,9 @@
 EMAIL_ATTR_FROM = 'from'
 EMAIL_ATTR_SUBJECT = 'subject'
 EMAIL_ATTR_BODY = 'body'
+
+USPS_TRACKING_NUMBER_REGEX = r"\b(94\d{20}|\d{4}\s\d{4}\s\d{4}\s\d{4}\s\d{4}\s\d{2})\b"
+UPS_TRACKING_NUMBER_REGEX = r"\b(1Z[A-HJ-NP-Z0-9]{16})\b"
+FEDEX_TRACKING_NUMBER_REGEX = r"\b(\d{12})\b"
+
+EMAIL_DOMAIN_REGEX = r"@([\w.-]+)"

custom_components/email/parsers/generic.py

@@ -0,0 +1,32 @@
+import logging
+import re
+
+from bs4 import BeautifulSoup
+from ..const import EMAIL_ATTR_BODY, USPS_TRACKING_NUMBER_REGEX, UPS_TRACKING_NUMBER_REGEX, FEDEX_TRACKING_NUMBER_REGEX
+
+_LOGGER = logging.getLogger(__name__)
+ATTR_GENERIC = 'generic'
+EMAIL_DOMAIN_GENERIC = ''
+
+def parse_generic(email):
+    """Tries to parse tracking numbers for any type of email."""
+    tracking_numbers = []
+
+    soup = BeautifulSoup(email[EMAIL_ATTR_BODY], 'html.parser')
+
+    matches = re.findall(UPS_TRACKING_NUMBER_REGEX, email[EMAIL_ATTR_BODY])
+    for tracking_number in matches:
+        if tracking_number not in tracking_numbers:
+            tracking_numbers.append(tracking_number)
+
+    matches = re.findall(USPS_TRACKING_NUMBER_REGEX, email[EMAIL_ATTR_BODY])
+    for tracking_number in matches:
+        if tracking_number not in tracking_numbers:
+            tracking_numbers.append(tracking_number)
+
+    matches = re.findall(FEDEX_TRACKING_NUMBER_REGEX, email[EMAIL_ATTR_BODY])
+    for tracking_number in matches:
+        if tracking_number not in tracking_numbers:
+            tracking_numbers.append(tracking_number)
+
+    return tracking_numbers,

custom_components/email/parsers/pledgebox.py

@@ -2,7 +2,7 @@
 import re
 
 from bs4 import BeautifulSoup
-from ..const import EMAIL_ATTR_BODY
+from ..const import EMAIL_ATTR_BODY, USPS_TRACKING_NUMBER_REGEX
 
 
 _LOGGER = logging.getLogger(__name__)
@@ -24,4 +24,9 @@ def parse_pledgebox(email):
         if match and match.group(1) not in tracking_numbers:
             tracking_numbers.append(match.group(1))
 
+    matches = re.findall(USPS_TRACKING_NUMBER_REGEX, email[EMAIL_ATTR_BODY])
+    for tracking_number in matches:
+        if tracking_number not in tracking_numbers:
+            tracking_numbers.append(tracking_number)
+
     return tracking_numbers

custom_components/email/sensor.py

@@ -61,6 +61,8 @@
 from .parsers.sylvane import ATTR_SYLVANE, EMAIL_DOMAIN_SYLVANE, parse_sylvane
 from .parsers.adafruit import ATTR_ADAFRUIT, EMAIL_DOMAIN_ADAFRUIT, parse_adafruit
 
+from .parsers.generic import ATTR_GENERIC, EMAIL_DOMAIN_GENERIC, parse_generic
+
 
 parsers = [
     (ATTR_UPS, EMAIL_DOMAIN_UPS, parse_ups),
@@ -106,12 +108,14 @@
     (ATTR_SONY, EMAIL_DOMAIN_SONY, parse_sony),
     (ATTR_SYLVANE, EMAIL_DOMAIN_SYLVANE, parse_sylvane),
     (ATTR_ADAFRUIT, EMAIL_DOMAIN_ADAFRUIT, parse_adafruit),
+
+    (ATTR_GENERIC, EMAIL_DOMAIN_GENERIC, parse_generic),
 ]
 
 _LOGGER = logging.getLogger(__name__)
 
 DOMAIN = 'email'
-SCAN_INTERVAL = timedelta(seconds=5*60)
+SCAN_INTERVAL = timedelta(seconds=30*60)
 
 PLATFORM_SCHEMA = PLATFORM_SCHEMA.extend({
     vol.Required(CONF_EMAIL): cv.string,
@@ -160,15 +164,15 @@
 ups_regex = "(" + ")|(".join(ups_pattern) + ")"
 
 def find_carrier(tracking_number, email_domain):
-    _LOGGER.debug(f'find_carrier email_domain: {email_domain}')
+    _LOGGER.debug(f'find_carrier email_domain: {email_domain} {tracking_number}')
 
     # we may have the carrier/link already parsed from parser
     if type(tracking_number) is dict:
         return {
-            'tracking_number': tracking_number['tracking_number'],
+            'tracking_number': tracking_number.get('tracking_number', ''),
             'carrier': email_domain,
             'origin': email_domain,
-            'link': tracking_number['link'],
+            'link': tracking_number.get('link', ''),
         }
 
     link = ""