Buildkite: Skip PR builds when there are no doc file changes

.buildkite/scripts/build_pr.sh

@@ -2,6 +2,9 @@
 set -euo pipefail
 set +x
 
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+DOCS_ROOT="$(cd "${SCRIPT_DIR}/../.." && pwd)"
+
 # This script should only be invoked by the Buildkite PR bot
 if [ -z ${GITHUB_PR_TARGET_BRANCH+set} ] || [ -z ${GITHUB_PR_NUMBER+set} ] || [ -z ${GITHUB_PR_BASE_REPO+set} ];then
   echo "One of the following env. variable GITHUB_PR_TARGET_BRANCH, GITHUB_PR_NUMBER, GITHUB_PR_BASE_REPO is missing - exiting."
@@ -54,7 +57,7 @@ if [[ "${GITHUB_PR_BASE_REPO}" != 'docs' ]]; then
   # The helper exits non-zero only when conf.yaml itself couldn't be read (e.g.
   # the perl YAML module is missing), which short-circuits the condition below
   # so we fail open and build as today, rather than silently going green.
-  if legacy_branches=$(perl "$(dirname "$0")/legacy_branches.pl" "$GITHUB_PR_BASE_REPO") \
+  if legacy_branches=$(perl "${SCRIPT_DIR}/legacy_branches.pl" "$GITHUB_PR_BASE_REPO") \
        && ! grep -qxF "${GITHUB_PR_TARGET_BRANCH}" <<< "${legacy_branches}"; then
     echo "Target branch '${GITHUB_PR_TARGET_BRANCH}' is not a legacy AsciiDoc branch in conf.yaml for ${GITHUB_PR_BASE_REPO} — skipping build (reporting success)."
     exit 0
@@ -218,9 +221,19 @@ if [[ "${GITHUB_PR_BASE_REPO}" != 'docs' ]]; then
       docs_diff=$(git diff --stat "origin/$GITHUB_PR_TARGET_BRANCH"...HEAD -- ./docs/en ./docs/kr ./docs/jp)
       ;;
 
-    # All other repos will always build
+    # Repos without a specialized arm: derive doc paths from conf.yaml
     *)
-      docs_diff="always build"
+      git fetch origin "$GITHUB_PR_TARGET_BRANCH"
+      if docs_paths=$(perl "${SCRIPT_DIR}/docs_paths.pl" "$GITHUB_PR_BASE_REPO" "${DOCS_ROOT}/conf.yaml" 2>/dev/null); then
+        if [[ -z "$docs_paths" ]]; then
+          docs_diff=""
+        else
+          mapfile -t paths <<< "$docs_paths"
+          docs_diff=$(git diff --stat "origin/$GITHUB_PR_TARGET_BRANCH"...HEAD -- "${paths[@]}")
+        fi
+      else
+        docs_diff="always build"
+      fi
       ;;
   esac
 

.buildkite/scripts/docs_paths.pl

@@ -0,0 +1,76 @@
+#!/usr/bin/env perl
+use strict;
+use warnings;
+use YAML qw(LoadFile);
+
+# Usage: docs_paths.pl <github-repo-name> [conf.yaml]
+# Prints one git-diff path per line from conf.yaml sources for the repo.
+#
+# Exit 0: conf.yaml was read successfully. Output is empty if the repo is not
+#         in conf.yaml or has no source paths.
+# Exit 1: conf.yaml couldn't be loaded (e.g. missing YAML module, parse error)
+#         — caller should build as today (fail open), since we can't tell.
+
+my ( $github_repo, $conf_path ) = @ARGV;
+$conf_path //= 'conf.yaml';
+
+die "Usage: $0 <github-repo-name> [conf.yaml]\n" unless defined $github_repo;
+
+my $conf = eval { LoadFile($conf_path) };
+if ($@) {
+    warn "Failed to load $conf_path: $@\n";
+    exit 1;
+}
+
+# Sources in conf.yaml reference the conf key (e.g. "esf"), which for most repos
+# is already the GitHub repo name but for some (e.g. "elastic-serverless-forwarder")
+# differs from it. Try the GitHub name as a conf key directly first, then fall back
+# to matching it against each repo's URL basename.
+my $conf_key = exists $conf->{repos}{$github_repo} ? $github_repo : undef;
+unless ( defined $conf_key ) {
+    while ( my ( $key, $url ) = each %{ $conf->{repos} } ) {
+        ( my $name = $url ) =~ s{.*/|\.git$}{}g;    # URL -> repo name (strip path and .git)
+        if ( $name eq $github_repo ) {
+            $conf_key = $key;
+            last;
+        }
+    }
+}
+
+unless ( defined $conf_key ) {
+    exit 0;    # repo not in conf — no doc paths, caller should skip
+}
+
+my %paths;
+walk_entries( $conf->{contents}, $conf_key, \%paths );
+
+print "$_\n" for sort keys %paths;
+exit 0;
+
+
+sub walk_entries {
+    my ( $entries, $conf_key, $paths ) = @_;
+    for my $entry (@$entries) {
+        if ( $entry->{sections} ) {
+            walk_entries( $entry->{sections}, $conf_key, $paths );
+        } else {
+            collect_source_paths( $entry, $conf_key, $paths );
+        }
+    }
+}
+
+sub collect_source_paths {
+    my ( $book, $conf_key, $paths ) = @_;
+    for my $source ( @{ $book->{sources} // [] } ) {
+        next unless ( $source->{repo} // '' ) eq $conf_key;
+        my $path = normalize_path( $source->{path} );
+        $paths->{$path} = 1 if defined $path && length $path;
+    }
+}
+
+sub normalize_path {
+    my ($path) = @_;
+    return unless defined $path;
+    $path =~ s{^/}{};    # strip leading slash for git diff
+    return $path;
+}

.buildkite/scripts/docs_paths.t

@@ -0,0 +1,39 @@
+#!/usr/bin/env perl
+use strict;
+use warnings;
+use Test::More;
+use FindBin qw($RealBin);
+
+# Regression check for docs_paths.pl's repo-name resolution and path collection.
+# Run from the repo root since docs_paths.pl defaults to ./conf.yaml.
+
+chdir "$RealBin/../.." or die "Can't chdir to repo root: $!";
+
+my $conf = 'conf.yaml';
+
+sub run {
+    my ($repo) = @_;
+    my $out = `perl .buildkite/scripts/docs_paths.pl @{[quotemeta $repo]} $conf 2>/dev/null`;
+    return ( $? >> 8, $out );
+}
+
+{
+    my ( $exit, $out ) = run('kibana');
+    is( $exit, 0, 'kibana: exits 0' );
+    like( $out, qr{docs/}, 'kibana: includes docs/' );
+    like( $out, qr{:\(glob\).*asciidoc}, 'kibana: includes a glob asciidoc pathspec' );
+}
+
+{
+    my ( $exit, $out ) = run('elastic-serverless-forwarder');
+    is( $exit, 0, 'elastic-serverless-forwarder: exits 0' );
+    like( $out, qr{docs/en}, 'elastic-serverless-forwarder: resolves via URL basename to docs/en' );
+}
+
+{
+    my ( $exit, $out ) = run('some-repo-not-in-conf-yaml');
+    is( $exit, 0, 'unknown repo: exits 0' );
+    is( $out, '', 'unknown repo: no doc paths' );
+}
+
+done_testing();

.buildkite/scripts/legacy_branches.t

@@ -24,6 +24,14 @@ for my $repo (qw(kibana-cn swiftype esf elastic-serverless-forwarder)) {
     isnt( $out, '', "$repo: has at least one legacy branch (conf key resolved)" );
 }
 
+{
+    my ( $exit, $out ) = run('kibana');
+    is( $exit, 0, 'kibana: exits 0' );
+    isnt( $out, '', 'kibana: has at least one legacy branch' );
+    like( $out, qr{^8\.19$}m, 'kibana: includes 8.19 legacy branch' );
+    unlike( $out, qr{^9\.}m, 'kibana: has no 9.x legacy branches' );
+}
+
 my ( $exit, $out ) = run('some-repo-not-in-conf-yaml');
 is( $exit, 0, 'unknown repo: exits 0' );
 is( $out, '', 'unknown repo: no legacy branches' );