fix: add regex searching fucntion for amazon emails (#1079)

* fix: add regex searching fucntion for amazon emails * add capost tracking # and out for delivery emails * properly escape regex searches
moralmunky · Feb 27, 2025 · 5ca8db8 · 5ca8db8
1 parent 1efc3e1
commit 5ca8db8
Show file tree

Hide file tree

Showing 2 changed files with 42 additions and 9 deletions.
diff --git a/custom_components/mail_and_packages/const.py b/custom_components/mail_and_packages/const.py
@@ -150,6 +150,7 @@
     "Arrivée :",
     "Verwachte bezorgdatum:",
     "Votre date de livraison prévue est :",
+    "Arriving",
 ]
 AMAZON_TIME_PATTERN_END = [
     "Previously expected:",
@@ -166,6 +167,10 @@
     "Volg je pakket",
     "Je pakket volgen",
 ]
+AMAZON_TIME_PATTERN_REGEX = [
+    "Arriving (\\w+ \\d+) - (\\w+ \\d+)",
+    "Arriving (\\w+ \\d+)",
+]
 AMAZON_EXCEPTION_SUBJECT = "Delivery update:"
 AMAZON_EXCEPTION_BODY = "running late"
 AMAZON_EXCEPTION = "amazon_exception"
@@ -296,9 +301,16 @@
             "Delivery Notification",
         ],
     },
-    "capost_delivering": {},
+    "capost_delivering": {
+        "email": [
+            "donotreply-nepasrepondre@notifications.canadapost-postescanada.ca",
+        ],
+        "subject": [
+            "Your parcel is out for delivery",
+        ],
+    },
     "capost_packages": {},
-    "capost_tracking": {},
+    "capost_tracking": {"pattern": ["\\d{16}"]},
     "capost_mail": {
         "email": ["donotreply-nepasrepondre@communications.canadapost-postescanada.ca"],
         "subject": ["You have mail on the way"],

diff --git a/custom_components/mail_and_packages/helpers.py b/custom_components/mail_and_packages/helpers.py
@@ -56,6 +56,7 @@
     AMAZON_SHIPMENT_TRACKING,
     AMAZON_TIME_PATTERN,
     AMAZON_TIME_PATTERN_END,
+    AMAZON_TIME_PATTERN_REGEX,
     ATTR_AMAZON_IMAGE,
     ATTR_BODY,
     ATTR_BODY_COUNT,
@@ -1500,6 +1501,20 @@ def amazon_date_search(email_msg: str) -> int:
     return -1
 
 
+def amazon_date_regex(email_msg: str) -> str | None:
+    """Look for regex strings in email message and return them."""
+    for body_regex in AMAZON_TIME_PATTERN_REGEX:
+        pattern = re.compile(rf"{body_regex}")
+        search = pattern.search(email_msg)
+        if search is not None and len(search.groups() > 0):
+            _LOGGER.debug(
+                "Amazon Regex: %s Count: %s", body_regex, len(search.groups())
+            )
+            # return the first group match (first date from a date range)
+            return search.group(1)
+    return None
+
+
 def amazon_date_format(arrive_date: str, lang: str) -> tuple:
     """Return the date format."""
     if "de_" in lang:
@@ -1627,14 +1642,20 @@ def get_items(
                         if search not in email_msg:
                             continue
 
-                        start = email_msg.find(search) + len(search)
-                        end = amazon_date_search(email_msg)
+                        amazon_regex_result = amazon_date_regex(email_msg)
+                        if amazon_regex_result is not None:
+                            _LOGGER.debug("Found regex result: %s", amazon_regex_result)
+                            arrive_date = amazon_regex_result
 
-                        arrive_date = email_msg[start:end].replace(">", "").strip()
-                        _LOGGER.debug("First pass: %s", arrive_date)
-                        arrive_date = arrive_date.split(" ")
-                        arrive_date = arrive_date[0:3]
-                        arrive_date = " ".join(arrive_date).strip()
+                        else:
+                            start = email_msg.find(search) + len(search)
+                            end = amazon_date_search(email_msg)
+
+                            arrive_date = email_msg[start:end].replace(">", "").strip()
+                            _LOGGER.debug("First pass: %s", arrive_date)
+                            arrive_date = arrive_date.split(" ")
+                            arrive_date = arrive_date[0:3]
+                            arrive_date = " ".join(arrive_date).strip()
 
                         # Get the date object
                         dateobj = dateparser.parse(arrive_date)