start project

app/services/log_parser/apache.py

@@ -0,0 +1,57 @@
+"""Apache Common/Combined presets + the generic parser built from
+format_compiler (Chapter 07)."""
+from __future__ import annotations
+
+from app.services.log_parser import ParsedEntry
+from app.services.log_parser.format_compiler import (
+    CompiledFormat, compile_format, parse_apache_timestamp, parse_request_line, validate_ip,
+)
+
+COMMON_LOG_FORMAT = '%h %l %u %t "%r" %>s %b'
+COMBINED_LOG_FORMAT = '%h %l %u %t "%r" %>s %b "%{Referer}i" "%{User-agent}i"'
+
+
+class FormatCompiledParser:
+    """LogParser implementation driven by any CompiledFormat (Ch04/07 protocol)."""
+
+    def __init__(self, compiled: CompiledFormat):
+        self._compiled = compiled
+
+    def parse_line(self, line: str) -> ParsedEntry | None:
+        match = self._compiled.pattern.match(line.rstrip("\n"))
+        if match is None:
+            return None  # malformed line — skip, never raise (Ch07)
+        fields = match.groupdict()
+
+        ip = validate_ip(fields["ip"])
+        if ip is None:
+            return None
+
+        req = parse_request_line(fields["request"])
+        if req is None:
+            return None
+        method, path = req
+
+        try:
+            timestamp = parse_apache_timestamp(fields["timestamp"])
+            status_code = int(fields["status"])
+            bytes_sent = 0 if fields["bytes"] == "-" else int(fields["bytes"])
+        except (ValueError, KeyError):
+            return None
+
+        referrer = fields.get("referrer")
+        user_agent = fields.get("user_agent")
+        return ParsedEntry(
+            timestamp=timestamp, ip=ip, method=method, path=path,
+            status_code=status_code, bytes_sent=bytes_sent,
+            referrer=None if referrer in (None, "-") else referrer,
+            user_agent=None if user_agent in (None, "-") else user_agent,
+        )
+
+
+def common_parser() -> FormatCompiledParser:
+    return FormatCompiledParser(compile_format(COMMON_LOG_FORMAT))
+
+
+def combined_parser() -> FormatCompiledParser:
+    return FormatCompiledParser(compile_format(COMBINED_LOG_FORMAT))