From 699aa6203c445e63df260134e7c94ec0bfa2f322 Mon Sep 17 00:00:00 2001
From: Itxaka <itxaka.garcia@spectrocloud.com>
Date: Fri, 15 Sep 2023 20:59:31 +0200
Subject: [PATCH] Read the first ten lines of a config to check for headers
 (#51)

There could be comments before the cloud-config header or jinja
expressions as we found already[0] so our current header check its a bit
strict.

This patch makes it so er read the first ten lines of the config source,
check for # at the start and then check for the headers as we did
before, so we are a bit more letinent in case the cloud-config header is
not the first thing in the file

Signed-off-by: Itxaka <itxaka@kairos.io>
---
 collector/collector.go      | 26 +++++++++++++-----
 collector/collector_test.go | 54 +++++++++++++++++++++++++++++++++++++
 2 files changed, 73 insertions(+), 7 deletions(-)

diff --git a/collector/collector.go b/collector/collector.go
index 28858b51..1d0ac494 100644
--- a/collector/collector.go
+++ b/collector/collector.go
@@ -400,14 +400,26 @@ func fetchRemoteConfig(url string) (*Config, error) {
 }
 
 func HasValidHeader(data string) bool {
-	header := strings.SplitN(data, "\n", 2)[0]
-
-	// Trim trailing whitespaces
-	header = strings.TrimRightFunc(header, unicode.IsSpace)
+	// Get the first 10 lines
+	headers := strings.SplitN(data, "\n", 10)
+
+	// iterate over them as there could be comments or the jinja template info:
+	// https://cloudinit.readthedocs.io/en/latest/explanation/instancedata.html#example-cloud-config-with-instance-data
+
+	for _, line := range headers {
+		// Trim trailing whitespaces
+		header := strings.TrimRightFunc(line, unicode.IsSpace)
+		// If it starts with a hash check it, in case its a huge line, we dont want to waste time
+		if strings.HasPrefix(header, "#") {
+			// NOTE: we also allow "legacy" headers. Should only allow #cloud-config at
+			// some point.
+			if (header == DefaultHeader) || (header == "#kairos-config") || (header == "#node-config") {
+				return true
+			}
+		}
+	}
 
-	// NOTE: we also allow "legacy" headers. Should only allow #cloud-config at
-	// some point.
-	return (header == DefaultHeader) || (header == "#kairos-config") || (header == "#node-config")
+	return false
 }
 
 func (c Config) Query(s string) (res string, err error) {
diff --git a/collector/collector_test.go b/collector/collector_test.go
index cfc3d961..246b6445 100644
--- a/collector/collector_test.go
+++ b/collector/collector_test.go
@@ -833,6 +833,60 @@ remote_key_2: remote_value_2`), os.ModePerm)
 				Expect(v).To(Equal("remote_value_1"))
 			})
 		})
+		Context("when files have comments before the headers or jinja declarations", func() {
+			var tmpDir string
+			var err error
+
+			BeforeEach(func() {
+				tmpDir, err = os.MkdirTemp("", "config")
+				Expect(err).ToNot(HaveOccurred())
+
+				// Local configs
+				err = os.WriteFile(path.Join(tmpDir, "local_config.yaml"), []byte(`## template: jinja
+#cloud-config
+local_key_1: local_value_1
+`), os.ModePerm)
+				Expect(err).ToNot(HaveOccurred())
+
+				// comments before the header
+				err = os.WriteFile(path.Join(tmpDir, "local_config_2.yaml"),
+					[]byte(`
+# this is a comment
+## then another comment
+#and the last one
+
+#cloud-config
+local_key_2: local_value_2
+`), os.ModePerm)
+				Expect(err).ToNot(HaveOccurred())
+
+			})
+
+			AfterEach(func() {
+				err = os.RemoveAll(tmpDir)
+				Expect(err).ToNot(HaveOccurred())
+			})
+
+			It("reads them", func() {
+				o := &Options{}
+				err := o.Apply(Directories(tmpDir), NoLogs)
+				Expect(err).ToNot(HaveOccurred())
+
+				c, err := Scan(o, FilterKeysTest)
+				Expect(err).ToNot(HaveOccurred())
+
+				Expect((*c)["local_key_1"]).ToNot(BeNil())
+				Expect((*c)["local_key_2"]).ToNot(BeNil())
+
+				v, ok := (*c)["local_key_1"].(string)
+				Expect(ok).To(BeTrue())
+				Expect(v).To(Equal("local_value_1"))
+
+				v, ok = (*c)["local_key_2"].(string)
+				Expect(ok).To(BeTrue())
+				Expect(v).To(Equal("local_value_2"))
+			})
+		})
 	})
 
 	Describe("String", func() {