test(parser): ✅ refactor initial test, add new tests and fixtures

tests/test_parser.py

@@ -1,29 +1,64 @@
-import re
+from pathlib import Path
 
 import pytest
 
 from logwatcher.config import FIXTURE_PATH
-from logwatcher.parser import parse_log_line
+from logwatcher.models import LogEntry
+from logwatcher.parser import parse_log_file, parse_log_line
+
+ORIGINAL_LOGS_DIR = FIXTURE_PATH / "original_logs"
+INVALID_LOGS_DIR = FIXTURE_PATH / "invalid_logs"
+SPECIFIC_LOGS_DIR = FIXTURE_PATH / "valid_logs"
+
+@pytest.fixture(name="original_log_files")
+def original_log_files_fixture() -> tuple[Path, ...]:
+    """
+        All real log files provided by N2 technicians.
+        Used for integration tests: these files contain both valid lines
+        and lines to be ignored (system logs, empty lines, directory headers).
+    """
+    return tuple(ORIGINAL_LOGS_DIR.iterdir())
+
+@pytest.fixture(name="empty_log_file")
+def empty_log_file_fixture() -> Path:
+    """
+        Path of an empty log file.
+        Ensures the parser does not raise an exception
+        and returns an empty list.
+    """
+    return Path(INVALID_LOGS_DIR / "empty_file.txt")
+
+@pytest.fixture(name="bad_log_file")
+def bad_log_file_fixture() -> Path:
+    """
+        Path of a file containing only incorrectly formatted lines.
+        Each line fails at a different point in LOG_PATTERN
+        (invalid IP, missing timestamp, missing DOSSIER EN COURS, etc.).
+        No LogEntry should be produced.
+    """
+    return Path(INVALID_LOGS_DIR / "bad_log_file.txt")
+
+@pytest.fixture(name="specific_logs")
+def specific_logs_fixture() -> list[str]:
+    """
+        Single log lines chosen to cover specific cases.
+        Each line is structurally valid and must be parsed
+        successfully. Used for unit tests of parse_log_line().
+    """
+    log_file = Path(SPECIFIC_LOGS_DIR / "specific_logs.txt")
+    with open(log_file, 'r') as file:
+        return file.readlines()
 
 
-@pytest.fixture(name="log_data")
-def load_logs_fixture() -> dict[str, list[str]]:
-    file_content_list: dict[str, list[str]] = {}
-
-    for file_path in FIXTURE_PATH.iterdir():
-        with open(file_path, 'r') as file:
-            file_content_list[file_path.name] = []
-            for line in file:
-                if re.match(r"\\\\", line):
-                    file_content_list[file_path.name].append(line)
-    
-    return file_content_list
-
-def test_create_log_entry(log_data: dict[str, list[str]]):
-    for log_list in log_data.values():
-        for log in log_list:
-            log_entry = parse_log_line(log)
-            if log_entry:
+def test_parse_log_line(specific_logs: list[str]):
+    """
+        Every structurally valid line produces a complete LogEntry.
+        Ensures no required field is empty or None after parsing
+        a line conforming to the LAME MDC format.
+    """
+    for log_line in specific_logs:
+        log_entry = parse_log_line(log_line)
+        assert log_entry is not None
         assert log_entry.server_ip is not None
         assert log_entry.mdc_server_name is not None
         assert log_entry.start_time is not None
@@ -31,3 +66,50 @@ def test_create_log_entry(log_data: dict[str, list[str]]):
         assert log_entry.error_time is not None
         assert log_entry.error_message is not None
         assert log_entry.raw_line != ''
+
+
+
+def test_parse_empty_log_line():
+    """
+        An empty or structurally invalid line returns None.
+        Covered cases:
+        - Empty string.
+        - Directory header (\"Répertoire scanné : \\\\...\").
+    """
+    log_entry = parse_log_line("")
+    assert not log_entry
+
+    log_entry = parse_log_line("Répertoire scanné : \\\\192.168.60.48\\e$\\MDC_9110\\Logs")
+    assert not log_entry
+
+def test_parse_valid_log_file(original_log_files: tuple[Path, ...]):
+    """
+        Each real log file produces at least one valid LogEntry.
+        Integration test: files provided by technicians contain
+        a mix of valid lines and lines to be ignored.
+        Ensures the parser extracts at least one entry per file.
+    """
+    for log_file in original_log_files:
+        log_list = parse_log_file(log_file)
+        assert len(log_list) > 0
+
+        for log_entry in log_list:
+            assert isinstance(log_entry, LogEntry)
+
+def test_parse_empty_log_file(empty_log_file):
+    """
+        An empty log file can be parsed and must returns an 
+        empty LogEntry list.
+    """
+    log_list = parse_log_file(empty_log_file)
+    assert len(log_list) == 0
+
+def test_parse_bad_log_file(bad_log_file):
+    """
+        A file containing only invalid lines produces nothing.
+        Any line that does not match the LAME MDC structure
+        (IP, timestamps, DOSSIER EN COURS, error message)
+        must be silently ignored.
+    """
+    log_list = parse_log_file(bad_log_file)
+    assert len(log_list) == 0