CFE-4559: cfbs convert, part 2

cfbs/commands.py

@@ -59,6 +59,7 @@ def search_command(terms: List[str]):
 
 from cfbs.cfbs_json import CFBSJson
 from cfbs.cfbs_types import CFBSCommandExitCode, CFBSCommandGitResult
+from cfbs.masterfiles.analyze import most_relevant_version
 from cfbs.updates import ModuleUpdates, update_module
 from cfbs.utils import (
     CFBSNetworkError,
@@ -1211,7 +1212,54 @@ def cfbs_convert_git_commit(
         % masterfiles_version
     )
     print(
-        "The next conversion step is to handle modified files and files from other versions of masterfiles."
+        "The next conversion step is to handle files from other versions of masterfiles."
+    )
+    if not prompt_user_yesno(non_interactive, "Do you want to continue?"):
+        raise CFBSExitError("User did not proceed, exiting.")
+    other_versions_files = analyzed_files.different
+    if len(other_versions_files) > 0:
+        print(
+            "The following files are taken from other versions of masterfiles (not %s):"
+            % masterfiles_version
+        )
+        other_versions_files = sorted(other_versions_files)
+        files_to_delete = []
+        for other_version_file, other_versions in other_versions_files:
+            # don't display all versions (which is an arbitrarily-shaped sequence), instead choose the most relevant one:
+            relevant_version_text = most_relevant_version(
+                other_versions, masterfiles_version
+            )
+            if len(other_versions) > 1:
+                relevant_version_text += ", ..."
+            print("-", other_version_file, "(%s)" % relevant_version_text)
+
+            files_to_delete.append(other_version_file)
+        print(
+            "This usually indicates that someone made mistakes during past upgrades and it's recommended to delete these."
+        )
+        print(
+            "Your policy set will include the versions from %s instead (if they exist)."
+            % masterfiles_version
+        )
+        print(
+            "(Deletions will be visible in Git history so you can review or revert later)."
+        )
+        if prompt_user_yesno(
+            non_interactive, "Delete files from other versions? (Recommended)"
+        ):
+            print("Deleting %s files." % len(files_to_delete))
+            for file_d in files_to_delete:
+                rm(os.path.join(dir_name, file_d))
+
+            print(
+                "Creating Git commit with deletion of policy files from other versions..."
+            )
+            cfbs_convert_git_commit("Deleted policy files from other versions")
+            print("Done.", end=" ")
+    else:
+        print("There are no files from other versions of masterfiles.")
+    print(
+        "The next conversion step is to handle files which have custom modifications."
     )
     print("This is not implemented yet.")
     return 0

cfbs/masterfiles/analyze.py

@@ -1,8 +1,11 @@
 from collections import OrderedDict
 import os
+from typing import Iterable, List
 
 from cfbs.utils import dict_sorted_by_key, file_sha256
 
+Version = str
+
 
 def initialize_vcf():
     versions_dict = {"versions": {}}
@@ -141,5 +144,47 @@ def sort_versions(versions: list, reverse: bool = True):
     )
 
 
-def highest_version(versions):
+def highest_version(versions: Iterable[Version]):
     return max(versions, key=version_as_comparable_list, default=None)
+
+
+def lowest_version(versions: Iterable[Version]):
+    return min(versions, key=version_as_comparable_list, default=None)
+
+
+def is_lower_version(version_a: Version, version_b: Version):
+    """Returns `True` if and only if `version_a` is lower than `version_b`."""
+
+    return version_as_comparable_list(version_a) < version_as_comparable_list(version_b)
+
+
+def most_relevant_version(
+    other_versions: List[Version], reference_version: Version
+) -> Version:
+    """
+    The most relevant version is the highest version among older other versions,
+    or if there are no older other versions, the lowest version among other versions.
+
+    `other_versions` is assumed to be non-empty and not contain `reference_version`."""
+    assert len(other_versions) > 0
+    assert reference_version not in other_versions
+
+    highest_other_version = highest_version(other_versions)
+    lowest_other_version = lowest_version(other_versions)
+    # Unfortunately, Pyright can not infer that these can't be `None` when `other_versions` is non-empty.
+    assert highest_other_version is not None
+    assert lowest_other_version is not None
+
+    if is_lower_version(highest_other_version, reference_version):
+        # all other versions are older
+        return highest_other_version
+    if is_lower_version(reference_version, lowest_other_version):
+        # all other versions are newer
+        return lowest_other_version
+    # there are both older and newer versions
+    lower_other_versions = [
+        o_v for o_v in other_versions if is_lower_version(o_v, reference_version)
+    ]
+    highest_lower = highest_version(lower_other_versions)
+    assert highest_lower is not None
+    return highest_lower