Merge pull request #375 from AndriusV4/feature/improve-google-parser

Improvements to Google Parser

Author: jvanderaa

changes/375.added

@@ -0,0 +1 @@
+Added Google subject parser with support for multiple status types and notifications without end times

circuit_maintenance_parser/parsers/google.py

@@ -4,7 +4,7 @@
 import re
 from datetime import datetime
 
-from circuit_maintenance_parser.parser import CircuitImpact, Html, Impact, Status
+from circuit_maintenance_parser.parser import CircuitImpact, EmailSubjectParser, Html, Impact, Status
 
 # pylint: disable=too-many-nested-blocks, too-many-branches
 
@@ -18,7 +18,7 @@ def parse_html(self, soup):
         """Execute parsing."""
         data = {}
         data["circuits"] = []
-        data["status"] = Status.CONFIRMED
+        end_time_explicit = False
 
         for span in soup.find_all("span"):
             if span.string is None:
@@ -29,6 +29,7 @@ def parse_html(self, soup):
             elif span.string.strip() == "End Time:":
                 dt_str = span.next_sibling.string.strip()
                 data["end"] = self.dt2ts(datetime.strptime(dt_str, "%Y-%m-%d %H:%M:%S %z UTC"))
+                end_time_explicit = True
             elif span.string.strip() == "Peer ASN:":
                 data["account"] = span.parent.next_sibling.string.strip()
             elif span.string.strip() == "Google Neighbor Address(es):":
@@ -37,9 +38,42 @@ def parse_html(self, soup):
                 cid = googleaddr + "-" + span.parent.next_sibling.string.strip()
                 data["circuits"].append(CircuitImpact(circuit_id=cid, impact=Impact.OUTAGE))
 
-        summary = list(soup.find("div").find("div").strings)[-1].strip()
-        match = re.search(r" - Reference (.*)$", summary)
-        data["summary"] = summary
-        data["maintenance_id"] = match[1]
+        # Google sometimes send notifications without End Time specified
+        if not end_time_explicit and data["start"]:
+            # Since start and end times cannot be equal, manufacturing end date by adding 1hr to start date
+            end_time_delta = 3600
+            data["end"] = data["start"] + end_time_delta
+
+        return [data]
+
+
+class SubjectParserGoogle1(EmailSubjectParser):
+    """Subject Parser for Google notifications."""
+
+    def parse_subject(self, subject):
+        """Parse the subject line."""
+        data = {}
+
+        # Example subject format - "[Scheduled] Google Planned Network Maintenance Notification - Reference PCR/123456"
+        # Group 1: Status (e.g., Scheduled, Completed, Canceled)
+        # Group 2: Maintenance ID (e.g., PCR/123456)
+        match = re.search(r"(\[\S+\]).*Reference\s+(\S+)", subject, re.IGNORECASE | re.DOTALL)
+        match_2 = re.search(r"\[\S+\]\s+(.*)", subject, re.IGNORECASE | re.DOTALL)
+
+        if match:
+            status_str = match.group(1).upper()
+            data["maintenance_id"] = match.group(2).strip()
+            if "COMPLETED" in status_str:
+                data["status"] = Status.COMPLETED
+            # To handle both Cancelled and Canceled spelling options just in case
+            elif "CANCEL" in status_str:
+                data["status"] = Status.CANCELLED
+            elif "SCHEDULED" in status_str:
+                data["status"] = Status.CONFIRMED
+            # If unable to match, we fallback to default confirmed
+            else:
+                data["status"] = Status.CONFIRMED
+        if match_2:
+            data["summary"] = match_2.group(1)
 
         return [data]

circuit_maintenance_parser/provider.py

@@ -24,7 +24,7 @@
 from circuit_maintenance_parser.parsers.crowncastle import HtmlParserCrownCastle1
 from circuit_maintenance_parser.parsers.equinix import HtmlParserEquinix, SubjectParserEquinix
 from circuit_maintenance_parser.parsers.globalcloudxchange import HtmlParserGcx1, SubjectParserGcx1
-from circuit_maintenance_parser.parsers.google import HtmlParserGoogle1
+from circuit_maintenance_parser.parsers.google import HtmlParserGoogle1, SubjectParserGoogle1
 from circuit_maintenance_parser.parsers.gtt import HtmlParserGTT1
 from circuit_maintenance_parser.parsers.hgc import HtmlParserHGC1, HtmlParserHGC2, SubjectParserHGC1
 from circuit_maintenance_parser.parsers.lumen import HtmlParserLumen1
@@ -374,7 +374,7 @@ class Google(GenericProvider):
 
     _processors: List[GenericProcessor] = PrivateAttr(
         [
-            CombinedProcessor(data_parsers=[EmailDateParser, HtmlParserGoogle1]),
+            CombinedProcessor(data_parsers=[EmailDateParser, HtmlParserGoogle1, SubjectParserGoogle1]),
         ]
     )
     _default_organizer = PrivateAttr("noc-noreply@google.com")

…t/data/google/google1_parser_result.json …a/google/google1_html_parser_result.jsontests/unit/data/google/google1_parser_result.json renamed to tests/unit/data/google/google1_html_parser_result.json tests/unit/data/google/google1_parser_result.json renamed to tests/unit/data/google/google1_html_parser_result.json

@@ -12,9 +12,6 @@
       }
     ],
     "end": 1701839100,
-    "maintenance_id": "PCR/123456",
-    "start": 1701809700,
-    "status": "CONFIRMED",
-    "summary": "Network Maintenance Notification - Reference PCR/123456"
+    "start": 1701809700
   }
 ]

tests/unit/data/google/google2_scheduled_subject

@@ -0,0 +1 @@
+[Scheduled] Google Planned Network Maintenance Notification - Reference PCR/123456

tests/unit/data/google/google2_scheduled_subject_parser_result.json

@@ -0,0 +1,7 @@
+[
+  {
+    "maintenance_id": "PCR/123456",
+    "status": "CONFIRMED",
+    "summary": "Google Planned Network Maintenance Notification - Reference PCR/123456"
+  }
+]

tests/unit/data/google/google3_completed_subject

@@ -0,0 +1 @@
+[Completed] Google Planned Network Maintenance Notification - Reference PCR/123456

tests/unit/data/google/google3_completed_subject_parser_result.json

@@ -0,0 +1,7 @@
+[
+  {
+    "maintenance_id": "PCR/123456",
+    "status": "COMPLETED",
+    "summary": "Google Planned Network Maintenance Notification - Reference PCR/123456"
+  }
+]

tests/unit/data/google/google4_canceled_subject

@@ -0,0 +1 @@
+[Canceled] Google Planned Network Maintenance Notification - Reference PCR/123456

tests/unit/data/google/google4_canceled_subject_parser_result.json

@@ -0,0 +1,7 @@
+[
+  {
+    "maintenance_id": "PCR/123456",
+    "status": "CANCELLED",
+    "summary": "Google Planned Network Maintenance Notification - Reference PCR/123456"
+  }
+]