CI: add "report-image-diff" job

This is a solution for [YT-55] CI: Add a 'test' job to grab information on the build result: Changed packages, change in image size.

Added a python script with corresponding GitLab CI job to launch this script.

1. The script retrieves build artifacts for two branches:

   • for the default/main branch (currently kirkstone) in the manifest repo;
   • for the integration branch from merge request.

   After that, the script compares those two sets of artifacts and creates a summary about following differences between these two sets:

   • the size of compressed artifacts (artifacts.zip);
   • the size of built image;
   • the difference between two manifest files.

   Herewith this comparison is done only for the machines and distributions hard coded in the script:

   machines = (
       "seco-genio510",
       "seco-genio700",
       "seco-mx6",
       "seco-mx6ull",
       "seco-mx8mm",
       "seco-mx8mp",
   )
   pipelines_and_distros = {
       "fngsystem-pipeline": "seconorth-fngsystem",
       "yocto-pipeline": "seconorth-wayland",
   }

   The produced summary is reported both to stdout and as a comment to the merge request.

   Notes:

   1. there are a lot of lines in diff file for manifests of linux kernel when it’s updated. So we:
      a) remove all these lines from the diff to keep summary readable
      b) print a message about kernel update.
      The result looks like:
      The kernel was updated from 304b502415 to 0691f0a31a commit. 1586 lines were removed from diff.

   So the report for non kernel diff looks like as follows:

   ⮘ seconorth-wayland | seco-mx6 ⮚
   ├── artifacts.zip size: [ 342.08 MiB → 388.44 MiB ] | +46.36 MiB
   ├── image size: [ 332 MiB → 378 MiB ]
   └── manifest diff:

   --- seconorth-image-seco-mx6-20240405130924.rootfs.manifest
   +++ seconorth-image-seco-mx6-20240410145347.rootfs.manifest
   +autoconf cortexa9t2hf_neon 2.71
   +automake cortexa9t2hf_neon 1.16.5
   +binutils cortexa9t2hf_neon 2.38
   +cpp cortexa9t2hf_neon 11.3.0
   +cpp-symlinks cortexa9t2hf_neon 11.3.0
   +g++ cortexa9t2hf_neon 11.3.0
   +g++-symlinks cortexa9t2hf_neon 11.3.0
   +gcc cortexa9t2hf_neon 11.3.0
   +gcc-symlinks cortexa9t2hf_neon 11.3.0
   +gettext cortexa9t2hf_neon 0.21
   +gnu-config cortexa9t2hf_neon 20211108+git0+191bcb948f
   +libatomic-dev cortexa9t2hf_neon 11.3.0
   +libc-malloc-debug0 cortexa9t2hf_neon 2.35
   +libc6-dev cortexa9t2hf_neon 2.35
   +libc6-extra-nss cortexa9t2hf_neon 2.35
   +libcroco cortexa9t2hf_neon 0.6.13
   +libgcc-s-dev cortexa9t2hf_neon 11.3.0
   +libgettextlib cortexa9t2hf_neon 0.21
   +libgettextsrc cortexa9t2hf_neon 0.21
   +libmpc3 cortexa9t2hf_neon 1.2.1
   +libmpfr6 cortexa9t2hf_neon 4.1.0
   +libnss-db2 cortexa9t2hf_neon 2.35
   +libstdc++-dev cortexa9t2hf_neon 11.3.0
   +libtool cortexa9t2hf_neon 2.4.7
   +linux-libc-headers-dev cortexa9t2hf_neon 5.16
   +m4 cortexa9t2hf_neon 1.4.19
   +make cortexa9t2hf_neon 4.3
   +packagegroup-core-buildessential cortexa9t2hf_neon 1.0
   +pkgconfig cortexa9t2hf_neon 0.29.2+git0+d97db4fae4

2. The CI job is triggered when the build is complete.

   Each run of the pipeline causes:

   • the removal of the old summary comment;
   • the creation of a new comment with summar a new one when the build is finished.

Demo MR: seco-ne/yocto/layers/meta-seconorth-distro!750 (closed)

scripts/report_image_diff.py

@@ -3,7 +3,7 @@ import argparse
 import fnmatch
 import logging
 from difflib import unified_diff
-from pathlib import Path
+from typing import Optional
 
 import requests
 from bs4 import BeautifulSoup
@@ -122,20 +122,27 @@ class BuildArtifacts:
 
         return ret
 
-    def download_file(self, path: str, outfile: str) -> None:
+    def get_artifact(self, path: str, outfile: Optional[str] = None) -> Optional[bytes]:
         """
-        Get the artifact file from GitLab and save it to the specified location.
+        Get a single artifact file from GitLab.
+        Save it to the file if "outfile" arg is specified.
 
         Args:
             path: str — relative path to artifact file.
             outfile: str — path to save the output file.
+
+        Returns:
+            None if "outfile" arg is given. Otherwise, returns the artifact file as bytes.
         """
 
-        url = (
-            f"{self.project.web_url}/-/jobs/{self.pipelinejob.id}/artifacts/raw/{path}"
-        )
-        r = requests.get(url, allow_redirects=True)
-        open(rf"{outfile}", "wb").write(r.content)
+        job = self.project.jobs.get(self.pipelinejob.id, lazy=True)
+
+        if not outfile:
+            return job.artifact(path)
+
+        with open(outfile, "wb") as f:
+            job.artifact(path, streamed=True, action=f.write)
+        return None
 
     def get_archive_size(self) -> int:
         """
@@ -240,15 +247,15 @@ def main():
         if comment.body[0:4] == "**`⮘":
             comment.delete()
 
-    manifest_main_commit = manifest_project.commits.list(
+    manifest_commit__main = manifest_project.commits.list(
         all=False, ref_name=args.target_branch
     )[0]
-    main_build = BuildPipelines(manifest_project, manifest_main_commit.id)
+    build__main = BuildPipelines(manifest_project, manifest_commit__main.id)
 
-    manifest_mr_commit = manifest_project.commits.list(
+    manifest_commit__mr = manifest_project.commits.list(
         all=False, ref_name=args.source_branch
     )[0]
-    mr_build = BuildPipelines(manifest_project, manifest_mr_commit.id)
+    build__mr = BuildPipelines(manifest_project, manifest_commit__mr.id)
 
     machines = (
         "seco-genio510",
@@ -271,28 +278,31 @@ def main():
 
             deploy_img_dir = f"build-{distro}-{machine}/tmp/deploy/images/{machine}/"
 
-            artifacts_main = BuildArtifacts(
-                manifest_project, main_build.get_jobs(pipeline, f"build-{machine}")[0]
+            artifacts__main = BuildArtifacts(
+                manifest_project, build__main.get_jobs(pipeline, f"build-{machine}")[0]
             )
-            artifacts_mr = BuildArtifacts(
-                manifest_project, mr_build.get_jobs(pipeline, f"build-{machine}")[0]
+            artifacts__mr = BuildArtifacts(
+                manifest_project, build__mr.get_jobs(pipeline, f"build-{machine}")[0]
             )
 
+            deploy_files__main = artifacts__main.list_dir(deploy_img_dir)
+            deploy_files__mr = artifacts__mr.list_dir(deploy_img_dir)
+
             # The difference in size of artifacts.zip for main and MR builds
-            zip_size_main = artifacts_main.get_archive_size()
-            zip_size_mr = artifacts_mr.get_archive_size()
+            zip_size__main = artifacts__main.get_archive_size()
+            zip_size__mr = artifacts__mr.get_archive_size()
 
-            zip_size_diff = sizeof_fmt(abs(zip_size_main - zip_size_mr))
-            sign = "+" if zip_size_main < zip_size_mr else "-"
+            zip_size_diff = sizeof_fmt(abs(zip_size__main - zip_size__mr))
+            sign = "+" if zip_size__main < zip_size__mr else "-"
 
-            zip_size_main = sizeof_fmt(zip_size_main)
-            zip_size_mr = sizeof_fmt(zip_size_mr)
+            zip_size__main = sizeof_fmt(zip_size__main)
+            zip_size__mr = sizeof_fmt(zip_size__mr)
 
-            summary += f"    ├── artifacts.zip size: [ {zip_size_main} → {zip_size_mr} ] | {sign}{zip_size_diff}\\\n"
+            summary += f"    ├── artifacts.zip size: [ {zip_size__main} → {zip_size__mr} ] | {sign}{zip_size_diff}\\\n"
 
             # The dirrerence in size of image for main and MR build
-            image_size_main = 0
-            image_size_mr = 0
+            image_size__main = 0
+            image_size__mr = 0
 
             # The image file format may vary depending on machine and distribution.
             if distro == "seconorth-fngsystem":
@@ -304,34 +314,39 @@ def main():
             else:
                 img_pattern = ""
 
-            for name, size in artifacts_main.list_dir(deploy_img_dir).items():
+            for name, size in deploy_files__main.items():
                 if fnmatch.fnmatch(name, img_pattern):
-                    image_size_main = size
-            for name, size in artifacts_mr.list_dir(deploy_img_dir).items():
+                    image_size__main = size
+            for name, size in deploy_files__mr.items():
                 if fnmatch.fnmatch(name, img_pattern):
-                    image_size_mr = size
+                    image_size__mr = size
             summary += (
-                f"    ├── image size: [ {image_size_main} → {image_size_mr} ]\\\n"
+                f"    ├── image size: [ {image_size__main} → {image_size__mr} ]\\\n"
             )
 
             # Comparision of manifest files for main in MR builds
-            for file in artifacts_main.list_dir(deploy_img_dir).keys():
+            manifestfile_lines__main = []
+            manifestfile_lines__mr = []
+
+            for file in deploy_files__main.keys():
                 if fnmatch.fnmatch(file, "*.rootfs.manifest"):
-                    artifacts_main.download_file(deploy_img_dir + file, "main.manifest")
+                    manifestfile_lines__main = (
+                        artifacts__main.get_artifact(deploy_img_dir + file)
+                        .decode("utf-8")
+                        .splitlines()
+                    )
                     diff_fromfile = file
-            for file in artifacts_mr.list_dir(deploy_img_dir).keys():
+
+            for file in deploy_files__mr.keys():
                 if fnmatch.fnmatch(file, "*.rootfs.manifest"):
-                    artifacts_mr.download_file(deploy_img_dir + file, "mr.manifest")
+                    manifestfile_lines__mr = (
+                        artifacts__mr.get_artifact(deploy_img_dir + file)
+                        .decode("utf-8")
+                        .splitlines()
+                    )
                     diff_tofile = file
 
-            manifestfile_main_lines = (
-                Path("main.manifest").read_text(encoding="UTF-8").splitlines()
-            )
-            manifestfile_mr_lines = (
-                Path("mr.manifest").read_text(encoding="UTF-8").splitlines()
-            )
-
-            if not manifestfile_main_lines or not manifestfile_mr_lines:
+            if not manifestfile_lines__main or not manifestfile_lines__mr:
                 continue
 
             summary += "    └── manifest diff:\n"
@@ -340,8 +355,8 @@ def main():
             kernel_rev_old = ""
             kernel_rev_new = ""
             for l in unified_diff(
-                manifestfile_main_lines,
-                manifestfile_mr_lines,
+                manifestfile_lines__main,
+                manifestfile_lines__mr,
                 fromfile=diff_fromfile,
                 tofile=diff_tofile,
                 n=0,