build: harden the release build scripts

anaconda/build.sh

@@ -1 +1,4 @@
-${PYTHON} setup.py install --single-version-externally-managed --record=record.txt
+# Run setup.py in isolated mode (-I) so the build does not pick up stray
+# modules from the working directory. -I is available on all supported
+# Pythons (>=3.4), unlike -P/PYTHONSAFEPATH which require Python >=3.11.
+${PYTHON} -I setup.py install --single-version-externally-managed --record=record.txt

build.sh

@@ -3,10 +3,12 @@
 # This script builds snowflake-telemetry-python for release. It is called from
 # a Jenkins job called SnowflakeTelemetryPythonPackageBuilder.
 #
-# Prequisites:
+# Prerequisites:
+# - run from a pristine git checkout (this script refuses to build from a
+#   workspace with modified, untracked, or ignored files; see
+#   scripts/release_lib.sh)
 # - activate a conda environment
 # - have conda-build installed, e.g. `conda install conda-build`
-# - have conda-verify installed, e.g. `conda install conda-verify`
 #
 # Then, run this script.
 #
@@ -18,12 +20,30 @@
 # conda activate my-conda-build-environment
 # # cd into the snowflake-telemetry-python git root dir
 # export SNOWFLAKE_TELEMETRY_DIR=$(pwd)
+#
+# For local test builds only, the workspace hygiene check can be bypassed with:
+# export SNOWFLAKE_TELEMETRY_ALLOW_DIRTY_WORKSPACE=1
+# Never set this in the Jenkins release job.
+
+REPO_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+source "${REPO_ROOT}/scripts/release_lib.sh"
+
+# Remove our own previous build output so it does not trip the hygiene gate.
+rm -rf ./anaconda/dist
+
+# Refuse to build unless the checkout is pristine (no modified, untracked, or
+# ignored files), so release artifacts contain only committed sources. The
+# explicit exit keeps this guard effective even if the script is invoked as
+# `bash build.sh` without -e.
+assert_clean_workspace "${REPO_ROOT}" || exit 1
 
 make_conda_build () {
   # set default build number to 0, if SNOWFLAKE_TELEMETRY_BUILD_NUMBER is not set
   echo "Start building snowflake-telemetry-python package with build_number: ${SNOWFLAKE_TELEMETRY_BUILD_NUMBER:=0}"
-  # conda build takes GIT_HASH as environment variable and embed it into the build package name
-  GIT_HASH=$(git rev-parse --short HEAD) SNOWFLAKE_TELEMETRY_BUILD_NUMBER=${SNOWFLAKE_TELEMETRY_BUILD_NUMBER:=0} conda build ./anaconda --output-folder ./anaconda/dist
+  # conda build takes GIT_HASH as environment variable and embed it into the build package name.
+  # Record the full HEAD commit so the artifact can be traced back to an exact,
+  # immutable revision (a short hash is ambiguous enough to collide).
+  GIT_HASH=$(git -C "${REPO_ROOT}" rev-parse HEAD) SNOWFLAKE_TELEMETRY_BUILD_NUMBER=${SNOWFLAKE_TELEMETRY_BUILD_NUMBER:=0} conda build ./anaconda --output-folder ./anaconda/dist
 }
 
 BUILD_CONDA_FORMAT=false
@@ -37,8 +57,8 @@ done
 # set up private channel to make opentelemetry dependencies available
 conda config --add channels https://repo.anaconda.com/pkgs/snowflake/
 
-# clean up the dist directory
-rm -rf ./anaconda/dist
+# create a clean dist directory (previous output was removed before the
+# workspace hygiene gate above)
 mkdir -p ./anaconda/dist
 
 if [ "$BUILD_CONDA_FORMAT" = true ] ; then
@@ -52,6 +72,10 @@ else
   make_conda_build
 fi
 
+# Bind the build outputs to SHA-256 digests so downstream release steps can
+# verify that what gets published is what this build produced.
+write_sha256_manifest ./anaconda/dist || exit 1
+
 # clean up the conda environment
 conda config --remove channels https://repo.anaconda.com/pkgs/snowflake/
 conda build purge

pypi-build.sh

@@ -18,6 +18,22 @@
 # source .venv/bin/activate
 # # cd into the snowflake-telemetry-python git root dir
 # export SNOWFLAKE_TELEMETRY_DIR=$(pwd)
+#
+# For local test builds only, the workspace hygiene check can be bypassed with:
+# export SNOWFLAKE_TELEMETRY_ALLOW_DIRTY_WORKSPACE=1
+# Never set this in the Jenkins release job.
+
+REPO_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+source "${REPO_ROOT}/scripts/release_lib.sh"
+
+# Remove our own previous build outputs so they do not trip the hygiene gate.
+rm -rf ./dist ./build venv_* src/*.egg-info
+
+# Refuse to build unless the checkout is pristine (no modified, untracked, or
+# ignored files), so release artifacts contain only committed sources. The
+# explicit exit keeps this guard effective even if the script is invoked as
+# `bash pypi-build.sh` without -e.
+assert_clean_workspace "${REPO_ROOT}" || exit 1
 
 VENV_DIR=venv_$(date +%s)
 python3 -m venv ${VENV_DIR}
@@ -35,5 +51,9 @@ mkdir ./dist
 echo "Start building snowflake-telemetry-python package with build_number: ${SNOWFLAKE_TELEMETRY_BUILD_NUMBER:=0}"
 SNOWFLAKE_TELEMETRY_BUILD_NUMBER=${SNOWFLAKE_TELEMETRY_BUILD_NUMBER:=0} python3 -m build
 
+# Bind the build outputs to SHA-256 digests so downstream release steps can
+# verify that what gets published is what this build produced.
+write_sha256_manifest ./dist || exit 1
+
 deactivate
 rm -rf ${VENV_DIR}

scripts/release_lib.sh

@@ -0,0 +1,81 @@
+# Guards for the Jenkins release build scripts (build.sh and pypi-build.sh).
+#
+# This file is sourced by those scripts. It must only define functions and
+# have no side effects when sourced.
+
+# assert_clean_workspace <repo_root>
+#
+# Return 0 only when the git working tree at <repo_root> is pristine: no
+# modified, staged, untracked, or ignored files. Otherwise print the offending
+# entries to stderr and return 1.
+#
+# Release builds must produce artifacts from committed sources only, so the
+# build refuses to run from a checkout with any leftover or unreviewed files.
+#
+# For local test builds only, set SNOWFLAKE_TELEMETRY_ALLOW_DIRTY_WORKSPACE=1
+# to bypass this check. Never set it in the Jenkins release job.
+assert_clean_workspace() {
+    local repo_root="${1:-}"
+
+    if [ -z "$repo_root" ]; then
+        echo "assert_clean_workspace: no repository path given; refusing to build." >&2
+        return 1
+    fi
+
+    if [ "${SNOWFLAKE_TELEMETRY_ALLOW_DIRTY_WORKSPACE:-0}" = "1" ]; then
+        echo "WARNING: SNOWFLAKE_TELEMETRY_ALLOW_DIRTY_WORKSPACE=1 is set; skipping the workspace hygiene check." >&2
+        echo "WARNING: never set this for official release builds." >&2
+        return 0
+    fi
+
+    local status
+    if ! status="$(git -C "$repo_root" status --porcelain --untracked-files=all --ignored=matching 2>&1)"; then
+        echo "Unable to determine the git state of $repo_root; refusing to build." >&2
+        echo "$status" >&2
+        return 1
+    fi
+
+    if [ -n "$status" ]; then
+        echo "Refusing to build release artifacts from a non-pristine workspace: $repo_root" >&2
+        echo "The following modified, untracked, or ignored (!!) files are present:" >&2
+        echo "$status" >&2
+        echo "Release builds must run from a pristine checkout so that artifacts" >&2
+        echo "contain only committed sources." >&2
+        echo "For local builds, set SNOWFLAKE_TELEMETRY_ALLOW_DIRTY_WORKSPACE=1 to skip this check." >&2
+        return 1
+    fi
+}
+
+# write_sha256_manifest <dist_dir>
+#
+# Write <dist_dir>/SHA256SUMS containing the SHA-256 digest of every file
+# under <dist_dir>, so downstream release steps can verify that the artifacts
+# they publish are bit-for-bit what this build produced. Fails if no artifacts
+# are present.
+write_sha256_manifest() {
+    local dist_dir="${1:-}"
+
+    if [ -z "$dist_dir" ] || [ ! -d "$dist_dir" ]; then
+        echo "write_sha256_manifest: '$dist_dir' is not a directory." >&2
+        return 1
+    fi
+
+    local sha256_cmd
+    if command -v sha256sum >/dev/null 2>&1; then
+        sha256_cmd="sha256sum"
+    else
+        # macOS (local test builds) ships shasum instead of sha256sum.
+        sha256_cmd="shasum -a 256"
+    fi
+
+    local files
+    files="$(cd "$dist_dir" && find . -type f ! -name SHA256SUMS -print | sort)"
+    if [ -z "$files" ]; then
+        echo "No build artifacts found under $dist_dir" >&2
+        return 1
+    fi
+
+    (cd "$dist_dir" && printf '%s\n' "$files" | while IFS= read -r artifact; do
+        $sha256_cmd "$artifact"
+    done > SHA256SUMS)
+}