fix(strategy): sandbox sed for regex fields

scripts/validate-pkg.sh

@@ -85,8 +85,9 @@ validate_pkg() {
     if ! echo "$tag_version_regex" | grep -qE '\([^)]*\)'; then
       log_err "$file: upstream.tag_version_regex must contain a capture group (): $tag_version_regex"
       ((errors++))
-    # Then check valid regex syntax
-    elif ! echo "test" | sed -nE "s|${tag_version_regex}|\\1|p" >/dev/null 2>&1; then
+    # Then check valid regex syntax. The regex can close the s command;
+    # --sandbox refuses e, r and w
+    elif ! echo "test" | sed --sandbox -nE "s|${tag_version_regex}|\\1|p" >/dev/null 2>&1; then
       log_err "$file: upstream.tag_version_regex is not valid ERE regex: $tag_version_regex"
       ((errors++))
     fi
@@ -201,7 +202,7 @@ validate_pkg() {
       else
         # Validate regex syntax: grep exits 2 on invalid regex
         local grep_rc=0
-        echo "" | grep -qE "$pattern" 2>/dev/null || grep_rc=$?
+        echo "" | grep -qE -e "$pattern" 2>/dev/null || grep_rc=$?
         if [[ "$grep_rc" -eq 2 ]]; then
           log_err "$file: upstream.pattern is not valid ERE regex: $pattern"
           ((errors++))

strategies/github-nightly.sh

@@ -76,7 +76,8 @@ case "$version_source" in
   tag)
     version=$(echo "$release" | yq -r '.tag_name')
     if [[ -n "$tag_version_regex" ]]; then
-      extracted=$(echo "$version" | sed -nE "s|${tag_version_regex}|\\1|p" | head -1)
+      # The regex can close the s command; --sandbox refuses e, r and w
+      extracted=$(echo "$version" | sed --sandbox -nE "s|${tag_version_regex}|\\1|p" | head -1)
       if [[ -z "$extracted" ]]; then
         echo "tag_version_regex '${tag_version_regex}' did not match tag '${version}'" >&2
         exit 1
@@ -96,7 +97,7 @@ case "$version_source" in
       exit 1
     fi
     body=$(echo "$release" | yq -r '.body')
-    version=$(echo "$body" | grep -oP "$version_pattern" | head -1)
+    version=$(echo "$body" | grep -oP -e "$version_pattern" | head -1)
     # Clean up: replace - with + for pacman compatibility
     version="${version//-/+}"
     ;;

strategies/github-release.sh

@@ -34,7 +34,8 @@ fi
 
 # Extract version from tag
 if [[ -n "$tag_version_regex" ]]; then
-  extracted=$(echo "$version" | sed -nE "s|${tag_version_regex}|\\1|p" | head -1)
+  # The regex can close the s command; --sandbox refuses e, r and w
+  extracted=$(echo "$version" | sed --sandbox -nE "s|${tag_version_regex}|\\1|p" | head -1)
   if [[ -z "$extracted" ]]; then
     echo "tag_version_regex '${tag_version_regex}' did not match tag '${version}'" >&2
     exit 1

strategies/github-tag.sh

@@ -42,7 +42,8 @@ fi
 
 # Extract version from tags
 if [[ -n "$tag_version_regex" ]]; then
-  version=$(echo "$tags" | sed -nE "s|${tag_version_regex}|\\1|p" | sort -V | tail -1)
+  # The regex can close the s command; --sandbox refuses e, r and w
+  version=$(echo "$tags" | sed --sandbox -nE "s|${tag_version_regex}|\\1|p" | sort -V | tail -1)
 else
   # Default: strip v prefix, sort semver, take latest
   version=$(echo "$tags" | sed 's/^v//' | sort -V | tail -1)

strategies/webpage-scrape.sh

@@ -43,8 +43,9 @@ page=$(curl -sfL --max-time "$TIMEOUT" --max-redirs 5 --max-filesize "$MAX_SIZE"
   exit 1
 }
 
-# Extract versions using ERE, capture group via sed
-versions=$(echo "$page" | grep -oE "$pattern" | sed -E "s/$pattern/\\1/" | sort -Vu)
+# Extract versions using ERE, capture group via sed. The pattern can close
+# the s command; --sandbox refuses e, r and w
+versions=$(echo "$page" | grep -oE -e "$pattern" | sed --sandbox -E "s/$pattern/\\1/" | sort -Vu)
 
 if [[ -z "$versions" ]]; then
   echo "No versions matched pattern '$pattern' at $url" >&2

tests/strategies.bats

@@ -273,6 +273,75 @@ teardown() {
   [ "$output" = "3.5.0" ]
 }
 
+# === YAML regex fields ===
+
+# A regex field closes the s command of sed and adds the e flag, which
+# runs the pattern space as a command
+@test "github-release: tag_version_regex runs no command" {
+  create_test_package "test-regex-exec" "github-release" "upstream:
+  project: custom-prefix/repo
+  tag_version_regex: '.*|touch $TEST_TMPDIR/pwned|e;s|(x)|y|;#'"
+
+  run "$TEST_TMPDIR/strategies/github-release.sh" "test-regex-exec"
+  [ "$status" -ne 0 ]
+  [ ! -e "$TEST_TMPDIR/pwned" ]
+}
+
+@test "github-tag: tag_version_regex runs no command" {
+  create_test_package "test-regex-exec" "github-tag" "upstream:
+  project: custom-prefix/repo
+  tag_version_regex: '.*|touch $TEST_TMPDIR/pwned|e;s|(x)|y|;#'"
+
+  run "$TEST_TMPDIR/strategies/github-tag.sh" "test-regex-exec"
+  [ "$status" -ne 0 ]
+  [ ! -e "$TEST_TMPDIR/pwned" ]
+}
+
+@test "github-nightly: tag_version_regex runs no command" {
+  create_test_package "test-regex-exec" "github-nightly" "upstream:
+  project: custom-nightly/repo
+  nightly_tag: nightly
+  version_source: tag
+  tag_version_regex: '.*|touch $TEST_TMPDIR/pwned|e;s|(x)|y|;#'"
+
+  run "$TEST_TMPDIR/strategies/github-nightly.sh" "test-regex-exec"
+  [ "$status" -ne 0 ]
+  [ ! -e "$TEST_TMPDIR/pwned" ]
+}
+
+@test "webpage-scrape: pattern runs no command" {
+  create_test_package "test-scrape-exec" "webpage-scrape" "upstream:
+  url: https://example.com/releases
+  pattern: 'test-scrape-(7)|/touch pwned/e;s/(x)'"
+
+  cd "$TEST_TMPDIR"
+  run "$TEST_TMPDIR/strategies/webpage-scrape.sh" "test-scrape-exec"
+  [ "$status" -ne 0 ]
+  [ ! -e "$TEST_TMPDIR/pwned" ]
+}
+
+@test "webpage-scrape: a pattern may start with a dash" {
+  create_test_package "test-scrape-dash" "webpage-scrape" "upstream:
+  url: https://example.com/releases
+  pattern: '-([0-9]+\\.[0-9]+\\.[0-9]+)\\.tar\\.gz'"
+
+  run "$TEST_TMPDIR/strategies/webpage-scrape.sh" "test-scrape-dash"
+  [ "$status" -eq 0 ]
+  [ "$output" = "7.0.0" ]
+}
+
+@test "github-nightly: a version_pattern may start with a dash" {
+  create_test_package "test-nightly-dash" "github-nightly" "upstream:
+  project: owner/nightly-repo
+  nightly_tag: nightly
+  version_source: release_body
+  version_pattern: '-\\K[a-z]+'"
+
+  run "$TEST_TMPDIR/strategies/github-nightly.sh" "test-nightly-dash"
+  [ "$status" -eq 0 ]
+  [ "$output" = "dev" ]
+}
+
 # === github-asset ===
 
 @test "github-asset: single asset, version is its updated_at as YYYYMMDD" {

tests/validate-pkg.bats

@@ -381,6 +381,14 @@ EOF
   [ "$status" -eq 0 ]
 }
 
+@test "validate: webpage-scrape accepts a pattern that starts with a dash" {
+  create_test_package "mypkg" "webpage-scrape" "upstream:
+  url: https://example.com/releases
+  pattern: '-([0-9]+\\.[0-9]+)\\.tar\\.gz'"
+  run "$TEST_TMPDIR/scripts/validate-pkg.sh" "mypkg"
+  [ "$status" -eq 0 ]
+}
+
 @test "validate: webpage-scrape rejects invalid regex pattern" {
   create_test_package "mypkg" "webpage-scrape" "upstream:
   url: https://example.com/releases
@@ -426,6 +434,17 @@ EOF
   [[ "$output" == *"not valid ERE regex"* ]]
 }
 
+# The field closes the s command of sed and adds the e flag, which runs the
+# pattern space as a command
+@test "validate: tag_version_regex runs no command" {
+  create_test_package "mypkg" "github-release" "upstream:
+  project: owner/repo
+  tag_version_regex: '.*|touch $TEST_TMPDIR/pwned|e;s|(x)|y|;#'"
+  run "$TEST_TMPDIR/scripts/validate-pkg.sh" "mypkg"
+  [ "$status" -ne 0 ]
+  [ ! -e "$TEST_TMPDIR/pwned" ]
+}
+
 @test "validate: tag_version_regex without capture group rejected" {
   create_test_package "mypkg" "github-release" "upstream:
   project: owner/repo