From 1b8f597c9b3dd9f8587cbb5034fd45a443292a31 Mon Sep 17 00:00:00 2001
From: Julian Squires <julian@cipht.net>
Date: Thu, 6 Mar 2025 14:12:58 -0330
Subject: [PATCH] Avoid using the manifest lexer on YAML

A YAML file like in the fixture added is parsed fine by Psych and
yamllint, but puppet-lint chokes on it because it gets fed through the
manifest lexer first, which can't cope with some things that are valid
YAML.

Since the one check that operates on YAML files doesn't support fixing
and doesn't use tokens, it seems reasonable to bypass the lexer; more
sophisticated YAML checks would need to use a separate YAML lexer
anyway.

Note that this can't be a unit test because those bypass the lexer
processing.
---
 lib/puppet-lint/checks.rb                   | 7 +++++--
 spec/acceptance/puppet_lint_spec.rb         | 7 +++++++
 spec/fixtures/test/manifests/parseable.yaml | 3 +++
 spec/unit/puppet-lint/checks_spec.rb        | 5 -----
 4 files changed, 15 insertions(+), 7 deletions(-)
 create mode 100644 spec/fixtures/test/manifests/parseable.yaml

diff --git a/lib/puppet-lint/checks.rb b/lib/puppet-lint/checks.rb
index 77b66e8e..f4c1fcff 100644
--- a/lib/puppet-lint/checks.rb
+++ b/lib/puppet-lint/checks.rb
@@ -54,10 +54,11 @@ def load_data(path, content)
   #
   # Returns an Array of problem Hashes.
   def run(fileinfo, data)
-    load_data(fileinfo, data)
-
     checks_run = []
     if File.extname(fileinfo).downcase.match?(%r{\.ya?ml$})
+      PuppetLint::Data.path = fileinfo
+      PuppetLint::Data.manifest_lines = data.split("\n", -1)
+
       enabled_checks.select { |check| YAML_COMPATIBLE_CHECKS.include?(check) }.each do |check|
         klass = PuppetLint.configuration.check_object[check].new
         # FIXME: shadowing #problems
@@ -65,6 +66,8 @@ def run(fileinfo, data)
         checks_run << [klass, problems]
       end
     else
+      load_data(fileinfo, data)
+
       enabled_checks.each do |check|
         klass = PuppetLint.configuration.check_object[check].new
         # FIXME: shadowing #problems
diff --git a/spec/acceptance/puppet_lint_spec.rb b/spec/acceptance/puppet_lint_spec.rb
index cd16fbcc..db620904 100644
--- a/spec/acceptance/puppet_lint_spec.rb
+++ b/spec/acceptance/puppet_lint_spec.rb
@@ -43,4 +43,11 @@
       expect(result[:stdout]).to have_warnings(2)
     end
   end
+
+  context 'with a YAML file provided' do
+    it 'returns zero errors' do
+      result = puppet_lint([File.join(manifest_root, 'parseable.yaml')])
+      expect(result[:stdout]).to have_errors(0)
+    end
+  end
 end
diff --git a/spec/fixtures/test/manifests/parseable.yaml b/spec/fixtures/test/manifests/parseable.yaml
new file mode 100644
index 00000000..c71c9402
--- /dev/null
+++ b/spec/fixtures/test/manifests/parseable.yaml
@@ -0,0 +1,3 @@
+---
+heredoc: |
+ contains $
diff --git a/spec/unit/puppet-lint/checks_spec.rb b/spec/unit/puppet-lint/checks_spec.rb
index 2ba1fa3c..c9a07eda 100644
--- a/spec/unit/puppet-lint/checks_spec.rb
+++ b/spec/unit/puppet-lint/checks_spec.rb
@@ -223,11 +223,6 @@
       allow(File).to receive(:extname).with(fileinfo).and_return('.yaml')
     end
 
-    it 'loads the yaml data' do
-      expect(instance).to receive(:load_data).with(fileinfo, data).and_call_original # rubocop: disable RSpec/SubjectStub
-      instance.run(fileinfo, data)
-    end
-
     context 'when there are checks enabled' do
       let(:enabled_checks) { [:legacy_facts] }
       let(:enabled_check_classes) { enabled_checks.map { |r| PuppetLint.configuration.check_object[r] } }