improve diff script to show versions diff for a package and its dependencies

Author: rosbo

diff

@@ -11,6 +11,7 @@ Options:
     -b, --base      The base image to diff against.
     -t, --target    The image to diff against the base image.
                     Default is the locally built image.
+    -p, --package   Only show diff for this package and its dependencies.
 EOF
 }
 
@@ -19,6 +20,7 @@ BASE_IMAGE_TAG='gcr.io/kaggle-images/python:latest'
 BASE_IMAGE_TAG_OVERRIDE=''
 TARGET_IMAGE_TAG='kaggle/python-build'
 TARGET_IMAGE_TAG_OVERRIDE=''
+PACKAGE_NAME=''
 
 while :; do
     case "$1" in 
@@ -48,6 +50,15 @@ while :; do
             TARGET_IMAGE_TAG_OVERRIDE="$2"
             shift # skip the flag value
             ;;
+        -p|--package)
+            if [[ -z "$2" ]]; then
+                usage
+                printf 'ERROR: No PACKAGE specified after the %s flag.\n' "$1" >&2
+                exit
+            fi
+            PACKAGE_NAME="$2"
+            shift # skip the flag value
+            ;;
         -?*)
             usage
             printf 'ERROR: Unknown option: %s\n' "$1" >&2
@@ -78,11 +89,18 @@ if [[ "$BASE_IMAGE_TAG" == "gcr.io/"* ]]; then
     docker pull "$BASE_IMAGE_TAG"
 fi
 
-CMDS=('dpkg-query --show -f "${Package}==${Version}\n"' 'pip freeze')
+
+if [[ -n "$PACKAGE_NAME" ]]; then
+    echo "Package: $PACKAGE_NAME"
+    CMDS=("python /tools/pip_list_versions.py $PACKAGE_NAME | sort")
+else
+    CMDS=('dpkg-query --show -f "${Package}==${Version}\n"' 'pip freeze')
+fi
+
 for cmd in "${CMDS[@]}"; do
     echo "== Comparing $cmd =="
     diff --suppress-common-lines --side-by-side \
-        <(docker run --rm "$BASE_IMAGE_TAG" $cmd) \
-        <(docker run --rm "$TARGET_IMAGE_TAG" $cmd) \
+        <(docker run -v $PWD/tools:/tools --rm "$BASE_IMAGE_TAG" /bin/bash -c "$cmd") \
+        <(docker run -v $PWD/tools:/tools --rm "$TARGET_IMAGE_TAG" /bin/bash -c "$cmd") \
         && echo 'No diff' || true
 done

tools/pip_list_versions.py

@@ -0,0 +1,42 @@
+import argparse
+import logging
+import re
+import subprocess
+
+
+def pip_show(package_name, packages=[]):
+    if package_name in packages:
+        return # avoid checking the same package twice if multiple packages depends on it.
+    packages.append(package_name)
+
+    result = subprocess.run(['pip', 'show', package_name], stdout=subprocess.PIPE)
+    if result.returncode != 0:
+        logging.error("pip show %s failed", package_name)
+    
+    show_stdout = result.stdout.decode("utf-8")
+    print(package_name + "==" + get_version(show_stdout))
+
+    for dependency in get_dependencies(show_stdout):
+        pip_show(dependency, packages=packages)
+
+def get_version(show_stdout):
+    for line in show_stdout.split("\n"):
+        m = re.match(r"^Version:\s(?P<version>.+)$", line)
+        if m:
+            return m.group('version')
+    return "not found"
+
+def get_dependencies(show_stdout):
+    for line in show_stdout.split("\n"):
+        m = re.match(r"^Requires:\s(?P<requires>.+)$", line)
+        if m:
+            return m.group('requires').split(', ')
+    return []
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument('package', type=str, help='package name')
+    args = parser.parse_args()
+    
+    pip_show(args.package)