cmd-diff: Add new ability to generate rpm diffs from commitmeta.json

This adds two new arguments for generating rpm-ostree db diff compatible
output by just leveraging the information in the commitmeta.json files
in each build's directory.

This is a much more lightweight way to get diffs rather than having to
download and import the entire ociarchive, which we are now having to
do in the build-with-buildah world where the rpm information isn't
available to us in the OSTree commit object like it was in the past.

This was assisted by Google/Gemini and got me 75% of the way there.
The prompt provided:

```
Look at the files old-commitmeta.json and new-commitmeta.json and
generate a two functions in src/cmd-diff that will produce similar
diff output. The first function will produce human readable diff
output and should be in the form as shown in the human-diff.txt
file. The second function should produce JSON output and be in
the format as shown in the json-diff.txt file. The commitmeta.json
files will exist in the build directory for a build (so when diffing
two builds you'll need to look at the commitmeta.json for the first
build and then the commitmeta.json for the second build and then
produce the diff output based on the differences in those two files.
```

Assisted-By: <google/gemini-2.5-pro>

Author: dustymabe

src/cmd-diff

@@ -3,10 +3,12 @@
 import argparse
 import os
 import shutil
+import json
 import subprocess
 import sys
 import tempfile
 import time
+import rpm
 from multiprocessing import Process
 
 from dataclasses import dataclass
@@ -24,11 +26,79 @@ class DiffBuildTarget:
     id: str
     dir: str
     meta: dict
+    commitmeta: dict
 
     @staticmethod
     def from_build(builds, build, arch):
         return DiffBuildTarget(build, builds.get_build_dir(build, arch),
-                               builds.get_build_meta(build, arch))
+                               builds.get_build_meta(build, arch),
+                               builds.get_build_commitmeta(build, arch))
+
+
+class PackageDiffType(IntEnum):
+    ADD = 0
+    REMOVE = 1
+    UPGRADE = 2
+    DOWNGRADE = 3
+
+
+@dataclass
+class Package:
+    name: str
+    epoch: int
+    v: str # Version
+    r: str # Release
+    a: str # Architecture
+
+    def evr(self):
+        epoch_string = "" if self.epoch == "0" else f"{self.epoch}:"
+        return f"{epoch_string}{self.v}-{self.r}"
+
+    # Compare versions
+    # rc > 0 -> Newer; rc == 0 -> Same; rc < 0 -> Older
+    @staticmethod
+    def vercmp(from_pkg, to_pkg):
+        return rpm.labelCompare(
+            (to_pkg.epoch, to_pkg.v, to_pkg.r),
+            (from_pkg.epoch, from_pkg.v, from_pkg.r)
+        )
+
+    # Return a data sctructure representing a diff entry from the
+    # rpm-ostree db diff --json output.
+    # "pkgdiff" : [
+    #   [
+    #     "NetworkManager",
+    #     2,
+    #     {
+    #       "PreviousPackage" : [
+    #         "NetworkManager",
+    #         "1:1.52.0-5.el9_6",
+    #         "x86_64"
+    #       ],
+    #       "NewPackage" : [
+    #         "NetworkManager",
+    #         "1:1.52.0-7.el9_6",
+    #         "x86_64"
+    #       ]
+    #     }
+    #   ],
+    @staticmethod
+    def to_json_diff_entry(difftype: PackageDiffType, from_pkg, to_pkg):
+        name = from_pkg.name if from_pkg else to_pkg.name
+        change = ""
+        match difftype:
+            case PackageDiffType.ADD:
+                change = {"NewPackage": [to_pkg.name, to_pkg.evr(), to_pkg.a]}
+            case PackageDiffType.REMOVE:
+                change = {"PreviousPackage": [from_pkg.name, from_pkg.evr(), from_pkg.a]}
+            case PackageDiffType.UPGRADE | PackageDiffType.DOWNGRADE:
+                change = {
+                    "PreviousPackage": [from_pkg.name, from_pkg.evr(), from_pkg.a],
+                    "NewPackage": [to_pkg.name, to_pkg.evr(), to_pkg.a]
+                }
+            case _:
+                raise
+        return [name, difftype, change]
 
 
 class OSTreeImport(IntEnum):
@@ -167,6 +237,89 @@ def diff_rpms_rpm_ostree(diff_from, diff_to, json):
     runcmd(cmd)
 
 
+def diff_rpms_commitmeta_json(diff_from, diff_to):
+    diff_rpms_commitmeta(diff_from, diff_to, json_output=True)
+
+
+def diff_rpms_commitmeta_no_json(diff_from, diff_to):
+    diff_rpms_commitmeta(diff_from, diff_to, json_output=False)
+
+
+def diff_rpms_commitmeta(diff_from, diff_to, json_output=False):
+    """
+    Diff the commitmeta.json files between two builds.
+    """
+    from_meta = diff_from.meta
+    to_meta = diff_to.meta
+    from_commitmeta = diff_from.commitmeta
+    to_commitmeta = diff_to.commitmeta
+
+    # Gather info about upgraded, downgraded, added, removed RPMS
+    from_rpms = {pkg[0]: Package(*pkg) for pkg in from_commitmeta.get('rpmostree.rpmdb.pkglist', [])}
+    to_rpms = {pkg[0]: Package(*pkg) for pkg in to_commitmeta.get('rpmostree.rpmdb.pkglist', [])}
+
+    added_rpms = set(to_rpms.keys()) - set(from_rpms.keys())
+    removed_rpms = set(from_rpms.keys()) - set(to_rpms.keys())
+    common_rpms = set(from_rpms.keys()) & set(to_rpms.keys())
+
+    upgraded_rpms = []
+    downgraded_rpms = []
+    for name in common_rpms:
+        from_pkg = from_rpms[name]
+        to_pkg = to_rpms[name]
+        rc = Package.vercmp(from_pkg, to_pkg)
+        if rc > 0:
+            upgraded_rpms.append((name, from_pkg, to_pkg))
+        elif rc == 0:
+            pass # They are equal versions
+        elif rc < 0:
+            downgraded_rpms.append((name, from_pkg, to_pkg))
+
+    if json_output:
+        diff = {
+            "ostree-commit-from": from_meta.get('ostree-commit'),
+            "ostree-commit-to": to_meta.get('ostree-commit'),
+            "pkgdiff": [],
+            "advisories": []
+        }
+        for name in sorted(added_rpms):
+            diff['pkgdiff'].append(Package.to_json_diff_entry(PackageDiffType.ADD, from_pkg=None, to_pkg=to_rpms[name]))
+        for name in sorted(removed_rpms):
+            diff['pkgdiff'].append(Package.to_json_diff_entry(PackageDiffType.REMOVE, from_pkg=from_rpms[name], to_pkg=None))
+        for name, from_pkg, to_pkg in sorted(upgraded_rpms):
+            diff['pkgdiff'].append(Package.to_json_diff_entry(PackageDiffType.UPGRADE, from_pkg=from_pkg, to_pkg=to_pkg))
+        for name, from_pkg, to_pkg in sorted(downgraded_rpms):
+            diff['pkgdiff'].append(Package.to_json_diff_entry(PackageDiffType.DOWNGRADE, from_pkg=from_pkg, to_pkg=to_pkg))
+
+        # Only add advisory info for the JSON output
+        from_advisories = {adv[0]: adv for adv in from_commitmeta.get('rpmostree.advisories', [])}
+        to_advisories = {adv[0]: adv for adv in to_commitmeta.get('rpmostree.advisories', [])}
+        added_advisories = set(to_advisories.keys()) - set(from_advisories.keys())
+        for name in added_advisories:
+            diff['advisories'].append(to_advisories[name])
+        # Dump the JSON to stdout
+        print(json.dumps(diff, indent=2))
+    else:
+        print(f"ostree diff commit from: {diff_from.id} ({from_meta.get('ostree-commit')})")
+        print(f"ostree diff commit to:   {diff_to.id} ({to_meta.get('ostree-commit')})")
+        if upgraded_rpms:
+            print("Upgraded:")
+            for name, from_pkg, to_pkg in sorted(upgraded_rpms):
+                print(f"  {name} {from_pkg.evr()} -> {to_pkg.evr()}")
+        if downgraded_rpms:
+            print("Downgraded:")
+            for name, from_pkg, to_pkg in sorted(downgraded_rpms):
+                print(f"  {name} {from_pkg.evr()} -> {to_pkg.evr()}")
+        if removed_rpms:
+            print("Removed:")
+            for name in sorted(removed_rpms):
+                print(f"  {name}-{from_rpms[name].evr()}")
+        if added_rpms:
+            print("Added:")
+            for name in sorted(added_rpms):
+                print(f"  {name}-{to_rpms[name].evr()}")
+
+
 def diff_ostree_ls(diff_from, diff_to):
     commit_from = diff_from.meta['ostree-commit']
     commit_to = diff_to.meta['ostree-commit']
@@ -476,6 +629,10 @@ DIFFERS = [
            needs_ostree=OSTreeImport.FULL, function=diff_rpms_rpm_ostree_no_json),
     Differ("rpms-json", "Diff RPMs, output JSON",
            needs_ostree=OSTreeImport.FULL, function=diff_rpms_rpm_ostree_json),
+    Differ("rpms-commitmeta", "Diff rpms from commitmeta.json",
+           needs_ostree=OSTreeImport.NO, function=diff_rpms_commitmeta_no_json),
+    Differ("rpms-json-commitmeta", "Diff RPMS & Advisories from commitmeta.json, output JSON",
+           needs_ostree=OSTreeImport.NO, function=diff_rpms_commitmeta_json),
     Differ("source-control", "Diff config and COSA input commits",
            needs_ostree=OSTreeImport.NO, function=diff_source_control),
     Differ("ostree-ls", "Diff OSTree contents using 'ostree diff'",