From 8bc3e9190f9122997066c4358a4d57d5b72f98ef Mon Sep 17 00:00:00 2001
From: DeltaDaniel <139119540+DeltaDaniel@users.noreply.github.com>
Date: Mon, 7 Oct 2024 18:52:21 +0200
Subject: [PATCH] changed sorting key including major and minor versions,
 adapted tests

---
 src/migmose/__main__.py                       |  2 +-
 src/migmose/parsing.py                        | 17 ++++-----
 unittests/__snapshots__/test_parsing.ambr     | 35 ++++++++++++++++---
 ...test_reduced_nested_nachrichtenstruktur.py |  2 +-
 4 files changed, 41 insertions(+), 15 deletions(-)

diff --git a/src/migmose/__main__.py b/src/migmose/__main__.py
index af5d94e..b0498eb 100644
--- a/src/migmose/__main__.py
+++ b/src/migmose/__main__.py
@@ -142,7 +142,7 @@ def main(
                 format_version,
                 output_dir_for_format,
             )
-            document_version = _extract_document_version(file)
+            document_version, *_ = _extract_document_version(file)
             reduced_nested_nachrichtenstruktur.output_tree(m_format, output_dir_for_format, document_version)
 
 
diff --git a/src/migmose/parsing.py b/src/migmose/parsing.py
index 7cd5711..58b65cd 100644
--- a/src/migmose/parsing.py
+++ b/src/migmose/parsing.py
@@ -160,28 +160,29 @@ def parse_raw_nachrichtenstrukturzeile(input_path: Path) -> list[str]:
 
 
 _pattern = re.compile(
-    r"MIG(?:Strom|Gas)?-?informatorischeLesefassung?(.*?)"
+    r"MIG(?:Strom|Gas)?-?informatorischeLesefassung?((\d+)\.(\d+)([a-z]?))"
     r"(?:_|KonsolidierteLesefassung|-AußerordentlicheVeröffentlichung)",
     re.IGNORECASE,
 )
 
 
-def _extract_document_version(path: Path | str) -> str:
+def _extract_document_version(path: Path | str) -> tuple[str, int | None, int | None, str]:
+    """Returns the document version, major, minor, and suffix from the given file path."""
     if isinstance(path, str):
         document_str = path
     else:
         document_str = str(path)
     matches = _pattern.search(document_str)
     if matches:
-        document_version = matches.group(1)
+        document_version, major, minor, suffix = matches.groups()
         if document_version == "":
             logger.warning(f"❌ No document version found in {path}.", fg="red")
-        return document_version
+        return document_version or "", int(major) or 0, int(minor) or 0, suffix or ""
     logger.error(f"❌ Unexpected document name in {path}.", fg="red")
-    return ""
+    return "", None, None, ""
 
 
-def _get_sort_key(path: Path) -> tuple[int, int, str]:
+def _get_sort_key(path: Path) -> tuple[int, int, int | None, int | None, str]:
     """
     Extracts the sort key from the given path.
 
@@ -194,5 +195,5 @@ def _get_sort_key(path: Path) -> tuple[int, int, str]:
     parts = path.stem.split("_")
     gueltig_von_date = int(parts[-1])
     gueltig_bis_date = int(parts[-2])
-    version_number = _extract_document_version(parts[-3])
-    return gueltig_von_date, gueltig_bis_date, version_number
+    _, major, minor, suffix = _extract_document_version(parts[-3])
+    return gueltig_von_date, gueltig_bis_date, major, minor, suffix
diff --git a/unittests/__snapshots__/test_parsing.ambr b/unittests/__snapshots__/test_parsing.ambr
index a3ea520..42f5164 100644
--- a/unittests/__snapshots__/test_parsing.ambr
+++ b/unittests/__snapshots__/test_parsing.ambr
@@ -1,16 +1,41 @@
 # serializer version: 1
 # name: TestParsing.test_extract_document_version[IFTSTA]
-  ''
+  tuple(
+    '',
+    None,
+    None,
+    '',
+  )
 # ---
 # name: TestParsing.test_extract_document_version[REMADV]
-  '2.9b'
+  tuple(
+    '2.9b',
+    2,
+    9,
+    'b',
+  )
 # ---
 # name: TestParsing.test_extract_document_version[REQOTE]
-  '1.3'
+  tuple(
+    '1.3',
+    1,
+    3,
+    '',
+  )
 # ---
 # name: TestParsing.test_extract_document_version[UTILMDG]
-  'G1.0a'
+  tuple(
+    '',
+    None,
+    None,
+    '',
+  )
 # ---
 # name: TestParsing.test_extract_document_version[UTILMDS]
-  'S1.1'
+  tuple(
+    '',
+    None,
+    None,
+    '',
+  )
 # ---
diff --git a/unittests/test_reduced_nested_nachrichtenstruktur.py b/unittests/test_reduced_nested_nachrichtenstruktur.py
index b11dcf5..3dec7c5 100644
--- a/unittests/test_reduced_nested_nachrichtenstruktur.py
+++ b/unittests/test_reduced_nested_nachrichtenstruktur.py
@@ -92,7 +92,7 @@ def test_output_tree(self, message_format: EdifactFormat, tmp_path, snapshot):
         reduced_nested_nachrichtenstruktur = ReducedNestedNachrichtenstruktur.create_reduced_nested_nachrichtenstruktur(
             nested_nachrichtenstruktur
         )
-        document_version = _extract_document_version(file_path)
+        document_version, *_ = _extract_document_version(file_path)
         reduced_nested_nachrichtenstruktur.output_tree(message_format, tmp_path, document_version)
         with open(tmp_path / f"{message_format}{document_version}.tree", "r", encoding="utf-8") as actual_file:
             assert actual_file.read() == snapshot