fix(scripts): the SPDX recipe also repairs the first form of the template rule

campaigns/TEMPLATE-DRIFT-CAMPAIGN.adoc

@@ -34,8 +34,11 @@ No rule in `AGENTIC.a2ml` assigns these recipes; that grant is the owner's.
 | `workflow-linter.yml` with the rule `head -1 "$file" \| grep -q "^# SPDX-License-Identifier:"`.
   The tool `gh actions-lock` writes its comment on line 1, so each lock run
   makes each managed workflow fail the lint.
+  Also a `workflow-linter.yml` with the first form of the template rule,
+  which stops only at a line in column 1 (`/^[^#[:space:]]/`). A workflow
+  with an indented body then passes with a licence line after the body.
 | The rule of the template: the licence line can be at any position in the
-  leading comment block.
+  leading comment block, and the block ends at the first line of the body.
 
 | `recipe-sonarqube-token-guard`
 | `fix-sonarqube-token-guard.sh`
@@ -73,6 +76,9 @@ Each recipe was also run on the files of `system-et` at `main` (`04a8b6e`):
 * SPDX rule: the step is identical to the step of the template branch
   (`rsr-template-repo` pull request 323, `915acb2`). That pull request makes
   the rule stop at an indented workflow body also.
+* SPDX rule, first form of the template rule: run on the linter of
+  `system-et` at `main` (`bbe1d13`), which has that form. 1 line changes, and
+  the step is identical to the step of the template at `main` (`0208563`).
 * Token guard: the job is identical to the job in pull request 92
   (`9f529ef`). The guard step is identical to the guard step of the template
   branch (`rsr-template-repo` pull request 321, `806dac0`).
@@ -102,10 +108,10 @@ and `fix-lock-self-sha-refs.sh` already cover these.
   inactive file (`e2e.yml.template`).
 * The token guard recipe edits only the shape of the old template. It does
   not rewrite a workflow that a person changed.
-* The SPDX recipe has 1 target: the `head -1` rule. A linter that has the
-  first form of the template rule (the form before `rsr-template-repo` pull
-  request 323, for example in `system-et`) is not a target, and keeps that
-  form.
+* The SPDX recipe has 2 targets: the `head -1` rule, and the first form of
+  the template rule (the form before `rsr-template-repo` pull request 323,
+  for example in `system-et`). A linter with any other rule is not a target,
+  and keeps its rule.
 * The number of repositories with each target is not known. No estate scan
   was made.
 

scripts/fix-workflow-linter-spdx-rule.sh

@@ -3,18 +3,30 @@
 #
 # fix-workflow-linter-spdx-rule.sh: accept the SPDX line in the leading comment block
 #
-# Older copies of the estate template have this rule in
-# .github/workflows/workflow-linter.yml:
+# Older copies of the estate template have 1 of 2 rules in
+# .github/workflows/workflow-linter.yml.
+#
+# Form 1 reads line 1 only:
 #
 #     if ! head -1 "$file" | grep -q "^# SPDX-License-Identifier:"; then
 #
 # The tool gh actions-lock writes its own comment on line 1 of each workflow
 # that it manages. Each lock run thus makes every managed workflow fail the
-# lint ("missing SPDX header"). The current template (rsr-template-repo)
-# accepts the licence line at any position in the leading comment block.
-# This recipe replaces the old rule with the rule of the template.
+# lint ("missing SPDX header").
+#
+# Form 2 is the first template rule that reads the leading comment block:
+#
+#     if ! awk '... /^[^#[:space:]]/ { exit } ...' "$file"; then
+#
+# It stops only at a line that starts in column 1. A workflow with an
+# indented body thus passes with a licence line after the body
+# (rsr-template-repo pull request 323).
+#
+# The current template (rsr-template-repo) accepts the licence line at any
+# position in the leading comment block, and stops at the first line of the
+# body. This recipe replaces each form with the rule of the template.
 #
-# Only the exact old rule is a target. A linter with a different rule (for
+# Only the 2 exact forms are targets. A linter with a different rule (for
 # example `head -5`) is not changed.
 #
 # Idempotent: a file that has the new rule is not changed.
@@ -27,7 +39,23 @@ set -euo pipefail
 
 REL=".github/workflows/workflow-linter.yml"
 OLD_RULE='^[[:space:]]*if ! head -1 "\$[A-Za-z_][A-Za-z0-9_]*" \| grep -q "\^# SPDX-License-Identifier:"; then[[:space:]]*$'
+OLD_AWK_RULE='/^# SPDX-License-Identifier:/ { found=1 } /^[^#[:space:]]/ { exit } END { exit !found }'
 NEW_RULE='/^# SPDX-License-Identifier:/ { found=1 } /^[[:space:]]*[^#[:space:]]/ { exit } END { exit !found }'
+export OLD_AWK_RULE NEW_RULE
+
+# awk function: the position of form 2 in a line, or 0. A line is a target
+# only when the whole line is `if ! awk '<form 2>' "$name"; then`.
+OLD_AWK_AT=$(cat <<'AWK'
+function old_awk_at(s,    lit, p) {
+    lit = "awk '" ENVIRON["OLD_AWK_RULE"] "' \"$"
+    p = index(s, lit)
+    if (p == 0) return 0
+    if (substr(s, 1, p - 1) !~ /^[[:space:]]*if ! $/) return 0
+    if (substr(s, p + length(lit)) !~ /^[A-Za-z_][A-Za-z0-9_]*"; then[[:space:]]*$/) return 0
+    return p
+}
+AWK
+)
 
 # Print the path of the linter when it is a regular file inside the repository.
 # A missing file, a symlink, or a path that leaves the repository is no target.
@@ -41,9 +69,11 @@ target_file() {
 
 # Print the number of old-rule lines in the repository (0 when no target file).
 count_targets() {
-    local file
+    local file form1 form2
     file=$(target_file "$1") || { echo 0; return 0; }
-    grep -cE -- "$OLD_RULE" "$file" || true
+    form1=$(grep -cE -- "$OLD_RULE" "$file" || true)
+    form2=$(awk "$OLD_AWK_AT"' old_awk_at($0) { n++ } END { print n + 0 }' "$file")
+    echo $((form1 + form2))
 }
 
 if [[ "${1:-}" == "--check" ]]; then
@@ -76,11 +106,19 @@ PROGRAM=$(cat <<'AWK'
 /echo "Add '# SPDX-License-Identifier: [A-Za-z0-9.+-]+' as first line"/ {
     sub(/ as first line"/, " to the leading comment block\"")
 }
+{
+    # Form 2: only the rule text changes. The 2 comment lines are there.
+    p = old_awk_at($0)
+    if (p) {
+        rest = substr($0, p + length("awk '" ENVIRON["OLD_AWK_RULE"] "' \"$"))
+        $0 = substr($0, 1, p - 1) "awk '" ENVIRON["NEW_RULE"] "' \"$" rest
+    }
+}
 { print }
 AWK
 )
 
-NEW_RULE="$NEW_RULE" awk "$PROGRAM" "$FILE" > "$TMP"
+awk "$OLD_AWK_AT"$'\n'"$PROGRAM" "$FILE" > "$TMP"
 [[ -s "$TMP" ]] || { echo "fix-workflow-linter-spdx-rule: the edit gave no text; $REL is not changed" >&2; exit 1; }
 cat "$TMP" > "$FILE"
 

tests/template-drift-recipes-test.sh

@@ -107,6 +107,23 @@ printf '# SPDX-License-Identifier: MPL-2.0\n  name: a\n  jobs: {}\n' > "$TMP/s5"
 if awk "$RULE" "$TMP/s4"; then fail "$R: the new rule accepts a licence after an indented body"; fi
 awk "$RULE" "$TMP/s5" || fail "$R: the new rule refuses a licence before an indented body"
 
+# Form 2, the first template rule that reads the leading comment block, is a
+# target too. Only the rule text changes: the result is the same file.
+REPO=$(new_repo spdx-form2); F="$REPO/.github/workflows/workflow-linter.yml"
+OLD_STOP='/^[^#[:space:]]/ { exit }'
+NEW_STOP='/^[[:space:]]*[^#[:space:]]/ { exit }'
+awk -v old="$OLD_STOP" -v new="$NEW_STOP" '{ p = index($0, new); if (p) $0 = substr($0, 1, p - 1) old substr($0, p + length(new)); print }' "$TMP/expected" > "$F"
+grep -qF -- "$OLD_STOP" "$F" || fail "$R: the test file does not have form 2"
+# Form 2 accepts a licence after an indented body. That is its defect.
+FORM2=$(sed -n "s/^ *if ! awk '\(.*\)' \"\$file\"; then\$/\1/p" "$F")
+awk "$FORM2" "$TMP/s4" || fail "$R: the test file does not have the defect of form 2"
+repair "$R" "$REPO" "$F" 1 "$TMP/expected"
+
+# The rule text of form 2 in a line of a different shape is not a target.
+REPO=$(new_repo spdx-form2-other); F="$REPO/.github/workflows/workflow-linter.yml"
+printf '%s\n' "            awk '/^# SPDX-License-Identifier:/ { found=1 } $OLD_STOP END { exit !found }' \"\$file\" || failed=1" > "$F"
+expect_untouched "$R" "$REPO" "$F"
+
 # A linter with a different rule is not a target.
 REPO=$(new_repo spdx-other); F="$REPO/.github/workflows/workflow-linter.yml"
 printf '%s\n' '            if ! head -5 "$f" | grep -q "SPDX-License-Identifier"; then' > "$F"