Adding new parsing techinques

src/alogamous/__main__.py

@@ -31,7 +31,10 @@
     )
     seperator = str(log_line_parser.LOG_FILE_CONFIGS[sys.argv[1]][log_line_parser.ConfigParameters.SEPERATOR])
     header_line = str(log_line_parser.LOG_FILE_CONFIGS[sys.argv[1]][log_line_parser.ConfigParameters.HEADER_LINE])
-    line_parser = log_line_parser.LogLineParser(expected_fields, seperator, header_line)
+    seperator2 = None
+    if log_line_parser.ConfigParameters.SEPARATOR2 in log_line_parser.LOG_FILE_CONFIGS[sys.argv[1]]:
+        seperator2 = str(log_line_parser.LOG_FILE_CONFIGS[sys.argv[1]][log_line_parser.ConfigParameters.SEPARATOR2])
+    line_parser = log_line_parser.LogLineParser(expected_fields, seperator, header_line, seperator2)
     analyzer.analyze_log_stream(
         [
             # echo_analyzer.EchoAnalyzer(),

src/alogamous/log_line_parser.py

@@ -1,18 +1,27 @@
 from __future__ import annotations
 
+from typing import Any
+
 
 class ConfigParameters:
     EXPECTED_FIELDS = "expected fields"
-    SEPERATOR = "seperator"
+    SEPERATOR = "separator"
     HEADER_LINE = "header line"
+    SEPARATOR2 = "separator2"
 
 
-LOG_FILE_CONFIGS = {
-    "default": {
+LOG_FILE_CONFIGS: dict[str, dict[str, Any]] = {
+    "python_logs": {
         ConfigParameters.EXPECTED_FIELDS: ["datetime", "source", "level", "message"],
         ConfigParameters.SEPERATOR: " - ",
         ConfigParameters.HEADER_LINE: "====================================================",
-    }
+    },
+    "java_logs": {
+        ConfigParameters.EXPECTED_FIELDS: [["datetime", "thread", "level", "source"], ["message"]],
+        ConfigParameters.SEPERATOR: " - ",
+        ConfigParameters.HEADER_LINE: "====================================================",
+        ConfigParameters.SEPARATOR2: " ",
+    },
 }
 
 
@@ -23,19 +32,38 @@ class LineType:
 
 
 class LogLineParser:
-    def __init__(self, expected_fields: list[str], seperator: str, header_line: str):
+    def __init__(self, expected_fields: list, seperator: str, header_line: str, separator2=None):
         self.header_line = header_line
         self.expected_fields = expected_fields
         self.separator = seperator
+        self.separator2 = separator2
         self.separator_count = len(self.expected_fields) - 1
+        self.separator2_count = len(self.expected_fields[0]) - 1
 
     def parse(self, line):
         if line == self.header_line:
             return {"type": LineType.HEADER_LINE, "line": line}
         if line.count(self.separator) == self.separator_count:
-            parsed_line = {"type": LineType.LOG_LINE}
-            separated_line = line.split(self.separator)
-            for index in range(len(self.expected_fields)):
-                parsed_line[self.expected_fields[index]] = separated_line[index]
-            return parsed_line
+            if self.separator2 is None:
+                return self.parse_simple_line(line)
+            return self.parse_complex_line(line)
         return {"type": LineType.UNSTRUCTURED_LINE, "line": line}
+
+    def parse_simple_line(self, line):
+        parsed_line = {"type": LineType.LOG_LINE}
+        separated_line = line.split(self.separator)
+        for index in range(len(self.expected_fields)):
+            parsed_line[self.expected_fields[index]] = separated_line[index]
+        return parsed_line
+
+    def parse_complex_line(self, line):
+        parsed_line = {"type": LineType.LOG_LINE}
+        chunked_line = line.split(self.separator)
+        separated_line = chunked_line[0].split(self.separator2)
+        separated_line.append(chunked_line[1])
+        index = 0
+        for field_list in self.expected_fields:
+            for field in field_list:
+                parsed_line[field] = separated_line[index]
+                index += 1
+        return parsed_line

tests/log_line_parser_test.py

@@ -43,3 +43,21 @@ def test_parse_start_header_content():
         "type": "unstructured line",
         "line": "    Start time: 2024-06-20 09:00:00.001550+00:00",
     }
+
+
+def test_parse_complex_log_line():
+    parser = log_line_parser.LogLineParser(
+        [["datetime", "thread", "level", "source"], ["message"]],
+        " - ",
+        "====================================================",
+        " ",
+    )
+    line = "2024-06-28T12:00:00.460+0000 [main] INFO com.app.java_process.info - Starting with config"
+    assert parser.parse(line) == {
+        "type": "log line",
+        "datetime": "2024-06-28T12:00:00.460+0000",
+        "thread": "[main]",
+        "level": "INFO",
+        "source": "com.app.java_process.info",
+        "message": "Starting with config",
+    }