Move CName parser to a separate class

Signed-off-by: Tobias Wolf <wolf@b1-systems.de>

Author: NotTheEvilOne

src/gardenlinux/features/__init__.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 
+from .cname import CName
 from .parser import Parser
 
-__all__ = ["Parser"]
+__all__ = ["CName", "Parser"]

src/gardenlinux/features/__main__.py

@@ -1,6 +1,7 @@
 #!/usr/bin/env python3
 # -*- coding: utf-8 -*-
 
+from .cname import CName
 from .parser import Parser
 
 from functools import reduce
@@ -54,61 +55,43 @@ def main():
     ), "Please provide either `--features` or `--cname` argument"
 
     arch = None
-    cname_base = None
+    flavor = None
     commit_id = None
     gardenlinux_root = path.dirname(args.feature_dir)
     version = None
 
-    if args.cname:
-        re_match = re.match(
-            "([a-zA-Z0-9]+([\\_\\-][a-zA-Z0-9]+)*?)(-([a-z0-9]+)(-([a-z0-9.]+)-([a-z0-9]+))*)?$",
-            args.cname,
-        )
+    if args.arch is not None:
+        arch = args.arch
 
-        assert re_match, f"Not a valid GardenLinux canonical name {args.cname}"
+    if args.version is not None:
+        version = args.version
 
-        if re_match.lastindex == 1:
-            data_splitted = re_match[1].split("-", 1)
+    if arch is None or arch == "":
+        arch = args.default_arch
+
+    if version is None or version == "":
+        version_data = get_version_and_commit_id_from_files(gardenlinux_root)
+        version = f"{version_data[0]}-{version_data[1]}"
 
-            cname_base = data_splitted[0]
+    if args.cname:
+        cname = CName(args.cname, arch=arch, version=version)
 
-            if len(data_splitted) > 1:
-                if args.arch is None:
-                    arch = data_splitted[1]
-                else:
-                    cname_base += "-" + data_splitted[1]
-        else:
-            arch = re_match[4]
-            cname_base = re_match[1]
-            commit_id = re_match[7]
-            version = re_match[6]
+        arch = cname.arch
+        flavor = cname.flavor
+        commit_id = cname.commit_id
+        version = cname.version
 
-        input_features = Parser.get_cname_as_feature_set(cname_base)
+        input_features = Parser.get_cname_as_feature_set(flavor)
     else:
         input_features = args.features
 
-    if args.arch is not None:
-        arch = args.arch
-
-    if args.version is not None:
-        re_match = re.match("([a-z0-9.]+)(-([a-z0-9]+))?$", args.version)
-        assert re_match, f"Not a valid version {args.version}"
-
-        commit_id = re_match[3]
-        version = re_match[1]
-
     if arch is None or arch == "" and (args.type in ("cname", "arch")):
-        assert (
-            args.default_arch
-        ), "Architecture could not be determined and no default architecture set"
-        arch = args.default_arch
-
-    if not commit_id or not version:
-        version, commit_id = get_version_and_commit_id_from_files(gardenlinux_root)
+        raise RuntimeError(
+            "Architecture could not be determined and no default architecture set"
+        )
 
-    if not version and (args.type in ("cname", "version")):
-        assert args.default_version, "version not specified and no default version set"
-        version = args.default_version
+    if version is None or version == "" and (args.type in ("cname", "version")):
+        raise RuntimeError("Version not specified and no default version set")
 
     feature_dir_name = path.basename(args.feature_dir)
 
@@ -124,7 +107,7 @@ def main():
         print(arch)
     elif args.type in ("cname_base", "cname", "graph"):
         graph = Parser(gardenlinux_root, feature_dir_name).filter(
-            cname_base, additional_filter_func=additional_filter_func
+            flavor, additional_filter_func=additional_filter_func
         )
 
         sorted_features = Parser.sort_graph_nodes(graph)
@@ -137,7 +120,7 @@ def main():
         if args.type == "cname_base":
             print(cname_base)
         elif args.type == "cname":
-            cname = cname_base
+            cname = flavor
 
             if arch is not None:
                 cname += f"-{arch}"
@@ -147,16 +130,16 @@ def main():
 
             print(cname)
         elif args.type == "graph":
-            print(graph_as_mermaid_markup(cname_base, graph))
+            print(graph_as_mermaid_markup(flavor, graph))
     elif args.type == "features":
         print(
             Parser(gardenlinux_root, feature_dir_name).filter_as_string(
-                cname_base, additional_filter_func=additional_filter_func
+                flavor, additional_filter_func=additional_filter_func
             )
         )
     elif args.type in ("flags", "elements", "platforms"):
         features_by_type = Parser(gardenlinux_root, feature_dir_name).filter_as_dict(
-            cname_base, additional_filter_func=additional_filter_func
+            flavor, additional_filter_func=additional_filter_func
         )
 
         if args.type == "platforms":
@@ -194,15 +177,15 @@ def get_minimal_feature_set(graph):
     return set([node for (node, degree) in graph.in_degree() if degree == 0])
 
 
-def graph_as_mermaid_markup(cname_base, graph):
+def graph_as_mermaid_markup(flavor, graph):
     """
     Generates a mermaid.js representation of the graph.
     This is helpful to identify dependencies between features.
 
     Syntax docs:
     https://mermaid.js.org/syntax/flowchart.html?id=flowcharts-basic-syntax
     """
-    markup = f"---\ntitle: Dependency Graph for Feature {cname_base}\n---\ngraph TD;\n"
+    markup = f"---\ntitle: Dependency Graph for Feature {flavor}\n---\ngraph TD;\n"
     for u, v in graph.edges:
         markup += f"    {u}-->{v};\n"
     return markup

src/gardenlinux/features/cname.py

@@ -0,0 +1,76 @@
+# -*- coding: utf-8 -*-
+
+from typing import Optional
+import re
+
+
+class CName(object):
+    def __init__(self, cname, arch=None, version=None):
+        self._arch = None
+        self._flavor = None
+        self._commit_id = None
+        self._version = None
+
+        re_match = re.match(
+            "([a-zA-Z0-9]+([\\_\\-][a-zA-Z0-9]+)*?)(-([a-z0-9]+)(-([a-z0-9.]+)-([a-z0-9]+))*)?$",
+            cname,
+        )
+
+        assert re_match, f"Not a valid GardenLinux canonical name {cname}"
+
+        if re_match.lastindex == 1:
+            self._flavor = re_match[1]
+        else:
+            self._commit_id = re_match[7]
+            self._flavor = re_match[1]
+            self._version = re_match[6]
+
+            if re_match[2] is None:
+                self._flavor += re_match[3]
+            else:
+                self._arch = re_match[4]
+
+        if self._arch is None and arch is not None:
+            self._arch = arch
+
+        if self._version is None and version is not None:
+            re_match = re.match("([a-z0-9.]+)(-([a-z0-9]+))?$", version)
+            assert re_match, f"Not a valid version {version}"
+
+            self._commit_id = re_match[3]
+            self._version = re_match[1]
+
+    @property
+    def arch(self) -> Optional[str]:
+        return self._arch
+
+    @property
+    def cname(self) -> str:
+        cname = self._flavor
+
+        if self._arch is not None:
+            cname += f"-{self._arch}"
+
+        if self._commit_id is not None:
+            cname += f"-{self.version_and_commit_id}"
+
+        return cname
+
+    @property
+    def commit_id(self) -> Optional[str]:
+        return self._commit_id
+
+    @property
+    def flavor(self) -> str:
+        return self._flavor
+
+    @property
+    def version(self) -> Optional[str]:
+        return self._version
+
+    @property
+    def version_and_commit_id(self) -> Optional[str]:
+        if self._commit_id is None:
+            return None
+
+        return f"{self._version}-{self._commit_id}"

src/gardenlinux/features/cname_main.py

@@ -12,6 +12,7 @@
     get_version_and_commit_id_from_files,
     sort_subset,
 )
+from .cname import CName
 from .parser import Parser
 
 
@@ -33,62 +34,44 @@ def main():
     assert re_match, f"Not a valid GardenLinux canonical name {args.cname}"
 
     arch = None
-    commit_id = None
     gardenlinux_root = dirname(args.feature_dir)
-    version = None
-
-    if re_match.lastindex == 1:
-        data_splitted = re_match[1].split("-", 1)
-
-        cname_base = data_splitted[0]
-
-        if len(data_splitted) > 1:
-            if args.arch is None:
-                arch = data_splitted[1]
-            else:
-                cname_base += "-" + data_splitted[1]
-    else:
-        arch = re_match[4]
-        cname_base = re_match[1]
-        commit_id = re_match[7]
-        version = re_match[6]
+    version = args.version
 
     if args.arch is not None:
         arch = args.arch
 
-    assert arch is not None and arch != "", "Architecture could not be determined"
+    if args.version is not None:
+        version = args.version
 
-    if not commit_id or not version:
-        version, commit_id = get_version_and_commit_id_from_files(gardenlinux_root)
+    if not version:
+        version_data = get_version_and_commit_id_from_files(gardenlinux_root)
+        version = f"{version_data[0]}-{version_data[1]}"
 
-    if args.version is not None:
-        re_match = re.match("([a-z0-9.]+)(-([a-z0-9]+))?$", args.version)
-        assert re_match, f"Not a valid version {args.version}"
+    cname = CName(args.cname, arch=arch, version=version)
 
-        commit_id = re_match[3]
-        version = re_match[1]
+    assert cname.arch, "Architecture could not be determined"
 
     feature_dir_name = basename(args.feature_dir)
 
     if gardenlinux_root == "":
         gardenlinux_root = "."
 
-    graph = Parser(gardenlinux_root, feature_dir_name).filter(cname_base)
+    graph = Parser(gardenlinux_root, feature_dir_name).filter(cname.flavor)
 
     sorted_features = Parser.sort_graph_nodes(graph)
     minimal_feature_set = get_minimal_feature_set(graph)
 
     sorted_minimal_features = sort_subset(minimal_feature_set, sorted_features)
 
-    cname = get_cname_base(sorted_minimal_features)
+    generated_cname = get_cname_base(sorted_minimal_features)
 
-    if arch is not None:
-        cname += f"-{arch}"
+    if cname.arch is not None:
+        generated_cname += f"-{cname.arch}"
 
-    if commit_id is not None:
-        cname += f"-{version}-{commit_id}"
+    if cname.version_and_commit_id is not None:
+        generated_cname += f"-{cname.version_and_commit_id}"
 
-    print(cname)
+    print(generated_cname)
 
 
 if __name__ == "__main__":

src/gardenlinux/features/parser.py

@@ -221,57 +221,6 @@ def get_cname_as_feature_set(cname):
         cname = cname.replace("_", "-_")
         return set(cname.split("-"))
 
-    @staticmethod
-    def get_flavor_from_cname(cname: str, get_arch: bool = True) -> str:
-        """
-        Extracts the flavor from a canonical name.
-
-        This method parses a Garden Linux canonical name (cname) and extracts
-        the flavor component, with or without the architecture suffix.
-
-        Example canonical names:
-        - "aws-gardener_prod-amd64"
-        - "azure-gardener_prod_tpm2_trustedboot-amd64-1312.2-80ffcc87"
-
-        The flavor is the platform plus feature string (e.g., "aws-gardener_prod")
-
-        Args:
-            cname (str): Canonical name of an image
-            get_arch (bool): Whether to include the architecture in the returned flavor
-                            If True: returns "aws-gardener_prod-amd64"
-                            If False: returns "aws-gardener_prod"
-
-        Returns:
-            str: The extracted flavor string, with or without architecture
-        """
-        # Use regex to extract components from the canonical name
-        # This handles complex cnames with version and commit hash
-        re_match = re.match(
-            "([a-zA-Z0-9]+([\\_\\-][a-zA-Z0-9]+)*?)(-([a-z0-9]+)(-([a-z0-9.]+)-([a-z0-9]+))*)?$",
-            cname,
-        )
-
-        assert re_match, f"Not a valid GardenLinux canonical name {cname}"
-
-        if re_match.lastindex == 1:
-            data_splitted = re_match[1].split("-", 1)
-
-            flavor = data_splitted[0]
-
-            if len(data_splitted) > 1:
-                if get_arch is True:
-                    arch = data_splitted[1]
-                else:
-                    flavor += "-" + data_splitted[1]
-        else:
-            arch = re_match[4]
-            flavor = re_match[1]
-        # Add architecture if requested
-        if get_arch and arch:
-            return f"{flavor}-{arch}"
-        else:
-            return flavor
-
     @staticmethod
     def _get_filter_set_callable(filter_set, additional_filter_func):
         def filter_func(node):