refactor: improve module source URL validation

- Use more specific regex pattern [a-zA-Z0-9-]+ instead of [^/]+ for namespace/module names
- Process all Terraform blocks instead of just the first one
- Report correct source if found in any block, only report incorrect sources if no correct source exists
- Add comprehensive test cases for multiple Terraform blocks

Co-authored-by: matifali <10648092+matifali@users.noreply.github.com>

cmd/readmevalidation/codermodules.go

@@ -12,7 +12,7 @@ import (
 var (
 	// Matches Terraform source lines with registry.coder.com URLs
 	// Pattern: source = "registry.coder.com/namespace/module/coder"
-	terraformSourceRe = regexp.MustCompile(`^\s*source\s*=\s*"` + registryDomain + `/([^/]+)/([^/]+)/coder"`)
+	terraformSourceRe = regexp.MustCompile(`^\s*source\s*=\s*"` + registryDomain + `/([a-zA-Z0-9-]+)/([a-zA-Z0-9-]+)/coder"`)
 )
 
 func validateModuleSourceURL(rm coderResourceReadme) []error {
@@ -27,6 +27,7 @@ func validateModuleSourceURL(rm coderResourceReadme) []error {
 
 	trimmed := strings.TrimSpace(rm.body)
 	foundCorrectSource := false
+	var incorrectSources []string
 	isInsideTerraform := false
 
 	lineScanner := bufio.NewScanner(strings.NewReader(trimmed))
@@ -39,7 +40,8 @@ func validateModuleSourceURL(rm coderResourceReadme) []error {
 				continue
 			}
 			if isInsideTerraform {
-				break
+				// End of current terraform block, continue to look for more
+				isInsideTerraform = false
 			}
 			continue
 		}
@@ -48,26 +50,34 @@ func validateModuleSourceURL(rm coderResourceReadme) []error {
 			continue
 		}
 
-		// Check for source line in the first terraform block
+		// Check for source line in terraform blocks
 		if matches := terraformSourceRe.FindStringSubmatch(nextLine); matches != nil {
 			actualNamespace := matches[1]
 			actualModule := matches[2]
 			actualSource := registryDomain + "/" + actualNamespace + "/" + actualModule + "/coder"
-			
+
 			if actualSource == expectedSource {
 				foundCorrectSource = true
-				break
+			} else {
+				// Collect incorrect sources but don't return immediately
+				incorrectSources = append(incorrectSources, actualSource)
 			}
-			// Found a registry.coder.com source but with wrong namespace/module
-			errs = append(errs, xerrors.Errorf("incorrect source URL format: found %q, expected %q", actualSource, expectedSource))
-			return errs
 		}
 	}
 
-	if !foundCorrectSource {
+	// If we found the correct source, ignore any incorrect ones
-		errs = append(errs, xerrors.Errorf("did not find correct source URL %q in first Terraform code block", expectedSource))
+	if foundCorrectSource {
+		return nil
 	}
 
+	// If we found incorrect sources but no correct one, report the first incorrect source
+	if len(incorrectSources) > 0 {
+		errs = append(errs, xerrors.Errorf("incorrect source URL format: found %q, expected %q", incorrectSources[0], expectedSource))
+		return errs
+	}
+
+	// If we found no sources at all
+	errs = append(errs, xerrors.Errorf("did not find correct source URL %q in any Terraform code block", expectedSource))
 	return errs
 }
 

cmd/readmevalidation/codermodules_test.go

@@ -35,6 +35,34 @@ var (
   version  = "1.0.0"
   agent_id = coder_agent.example.id
 }
+` + "```\n"
+
+	multipleBlocksValidBody = `# Test Module
+
+` + "```tf\n" + `module "other-module" {
+  source   = "registry.coder.com/other/module/coder"
+  version  = "1.0.0"
+}
+` + "```\n" + `
+` + "```tf\n" + `module "test-module" {
+  source   = "registry.coder.com/test-namespace/test-module/coder"
+  version  = "1.0.0"
+  agent_id = coder_agent.example.id
+}
+` + "```\n"
+
+	multipleBlocksInvalidBody = `# Test Module
+
+` + "```tf\n" + `module "test-module" {
+  source   = "registry.coder.com/wrong-namespace/test-module/coder"
+  version  = "1.0.0"
+}
+` + "```\n" + `
+` + "```tf\n" + `module "other-module" {
+  source   = "registry.coder.com/another-wrong/test-module/coder"
+  version  = "1.0.0"
+  agent_id = coder_agent.example.id
+}
 ` + "```\n"
 )
 
@@ -126,4 +154,39 @@ func TestValidateModuleSourceURL(t *testing.T) {
 			t.Errorf("Expected path format error, got: %s", errs[0].Error())
 		}
 	})
+
+	t.Run("Multiple blocks with valid source in second block", func(t *testing.T) {
+		t.Parallel()
+
+		rm := coderResourceReadme{
+			resourceType: "modules",
+			filePath:     "registry/test-namespace/modules/test-module/README.md",
+			namespace:    "test-namespace",
+			resourceName: "test-module",
+			body:         multipleBlocksValidBody,
+		}
+		errs := validateModuleSourceURL(rm)
+		if len(errs) != 0 {
+			t.Errorf("Expected no errors, got: %v", errs)
+		}
+	})
+
+	t.Run("Multiple blocks with incorrect source in second block", func(t *testing.T) {
+		t.Parallel()
+
+		rm := coderResourceReadme{
+			resourceType: "modules",
+			filePath:     "registry/test-namespace/modules/test-module/README.md",
+			namespace:    "test-namespace",
+			resourceName: "test-module",
+			body:         multipleBlocksInvalidBody,
+		}
+		errs := validateModuleSourceURL(rm)
+		if len(errs) != 1 {
+			t.Errorf("Expected 1 error, got %d: %v", len(errs), errs)
+		}
+		if !strings.Contains(errs[0].Error(), "incorrect source URL format") {
+			t.Errorf("Expected source URL format error, got: %s", errs[0].Error())
+		}
+	})
 }