fixing bugs with duplicate message analyzer (#84)

src/alogamous/__main__.py

@@ -41,7 +41,7 @@
             # echo_analyzer.EchoAnalyzer(),
             daily_count_analyzer.DailyCountAnalyzer(line_parser),
             error_counter_analyzer.ErrorCounterAnalyzer(line_parser),
-            flag_duplicate_log_messages.FlagDuplicateLogMessages(),
+            flag_duplicate_log_messages.FlagDuplicateLogMessages(line_parser),
             line_count_analyzer.LineCountAnalyzer(),
             format_analyzer.FormatAnalyzer(line_parser),
             loginfo_analyzer.InfoAnalyzer(line_parser),

src/alogamous/flag_duplicate_log_messages.py

@@ -1,21 +1,24 @@
-from alogamous import analyzer
+from alogamous import analyzer, log_line_parser
 
 
 class FlagDuplicateLogMessages(analyzer.Analyzer):
-    def __init__(self):
+    def __init__(self, line_parser):
+        self.parser = line_parser
         self.logMessages = set()
         self.duplicateMessages = set()
 
     def read_log_line(self, line):
-        message = line.split("-")[-1].strip()
-        if message in self.logMessages:
-            self.duplicateMessages.add(message)
-        else:
-            self.logMessages.add(message)
+        parsed_line = self.parser.parse(line)
+        if parsed_line["type"] == log_line_parser.LineType.LOG_LINE:
+            message = parsed_line["message"]
+            if message in self.logMessages:
+                self.duplicateMessages.add(message)
+            else:
+                self.logMessages.add(message)
 
     def report(self, out_stream):
         if len(self.duplicateMessages) > 0:
-            out_stream.write("Duplicate Log Messages:\n")
-            out_stream.write("\n".join(self.duplicateMessages))
+            out_stream.write("Duplicate Log Messages:\n- ")
+            out_stream.write("\n- ".join(self.duplicateMessages))
         else:
             out_stream.write("No duplicate log messages")

tests/flag_duplicate_log_messages_test.py

@@ -1,27 +1,54 @@
 import io
 
-from alogamous import flag_duplicate_log_messages
+from alogamous import flag_duplicate_log_messages, log_line_parser
 
 
 def test_flag_duplicate_log_messages():
-    flagger = flag_duplicate_log_messages.FlagDuplicateLogMessages()
+    line_parser = log_line_parser.LogLineParser(
+        ["datetime", "source", "level", "message"], " - ", "===================================================="
+    )
+    flagger = flag_duplicate_log_messages.FlagDuplicateLogMessages(line_parser)
     in_stream = io.StringIO("""Date - root - INFO - log message 1
         Date - root - WARNING - log message 2
         Date - root - WARNING - log message 1""")
     out_stream = io.StringIO()
     for line in in_stream:
-        flagger.read_log_line(line)
+        flagger.read_log_line(line.rstrip())
     flagger.report(out_stream)
-    assert out_stream.getvalue() == """Duplicate Log Messages:\nlog message 1"""
+    assert out_stream.getvalue() == """Duplicate Log Messages:\n- log message 1"""
 
 
 def test_flag_duplicate_log_messages_no_duplicates():
-    flagger = flag_duplicate_log_messages.FlagDuplicateLogMessages()
+    line_parser = log_line_parser.LogLineParser(
+        ["datetime", "source", "level", "message"], " - ", "===================================================="
+    )
+    flagger = flag_duplicate_log_messages.FlagDuplicateLogMessages(line_parser)
     in_stream = io.StringIO("""Date - root - INFO - log message 1
         Date - root - WARNING - log message 2
         Date - root - WARNING - log message 3""")
     out_stream = io.StringIO()
     for line in in_stream:
-        flagger.read_log_line(line)
+        flagger.read_log_line(line.rstrip())
+    flagger.report(out_stream)
+    assert out_stream.getvalue() == """No duplicate log messages"""
+
+
+def test_flag_duplicate_messages_with_header_and_dashes():
+    line_parser = log_line_parser.LogLineParser(
+        ["datetime", "source", "level", "message"], " - ", "===================================================="
+    )
+    flagger = flag_duplicate_log_messages.FlagDuplicateLogMessages(line_parser)
+    in_stream = io.StringIO("""====================================================
+STARTING Tracking service
+    Start time: 2024-06-20 09:00:00.001550+00:00
+    Version: 2729a
+    Command line: ['.venv/bin/python3', '-m', 'app.tracking_service', '--market', 'US', '--version', '2729a']
+====================================================
+2024-06-20 11:00:17,983 - root - INFO - Adding subscription for pid None
+2024-06-20 11:00:18,115 - root - INFO - Initialized Influx DB Client to host
+2024-06-20 11:00:18,185 - root - INFO - Kafka reading from start of day 2024-06-20 05:00:00+00:00 on topic internal""")
+    out_stream = io.StringIO()
+    for line in in_stream:
+        flagger.read_log_line(line.rstrip())
     flagger.report(out_stream)
     assert out_stream.getvalue() == """No duplicate log messages"""