Bugfix for 429 (Maven sourceDocuments configuration is optional)

htmlSanityCheck-core/src/main/java/org/aim42/htmlsanitycheck/Configuration.java

@@ -51,7 +51,7 @@ public class Configuration {
     @Builder.Default
     Set<Pattern> excludes = new HashSet<>();
     @Builder.Default
-    Set<String> indexFilenames = defaultIndeFilenames();
+    Set<String> indexFilenames = defaultIndexFilenames();
 
     /*
      * Explanation for configuring http status codes:
@@ -77,13 +77,13 @@ public Configuration() {
         this.ignoreIPAddresses = false;// warning if numerical IP addresses
         this.ignoreLocalhost = false;// warning if localhost-URLs
         this.indexFilenames
-                = defaultIndeFilenames();
+                = defaultIndexFilenames();
         this.prefixOnlyHrefExtensions = Web.POSSIBLE_EXTENSIONS;
 
         this.checksToExecute = AllCheckers.CHECKER_CLASSES;
     }
 
-    private static Set<String> defaultIndeFilenames() {
+    private static Set<String> defaultIndexFilenames() {
         return Arrays.stream("index.html,index.htm".split(",")).collect(Collectors.toSet());
     }
 

htmlSanityCheck-gradle-plugin/README.adoc

@@ -92,7 +92,7 @@ Type: Directory.
 +
 Type: `org.gradle.api.file.FileCollection`.
 +
-Default: All files in `+{sourceDir}+` whose names end with `.html`.
+Default: All files in `+{sourceDir}+` whose names end with `.html` or `.htm`.
 
 `checkingResultsDir` (optional):: Directory where the checking results are written to.
 +

htmlSanityCheck-gradle-plugin/src/main/groovy/org/aim42/htmlsanitycheck/gradle/HtmlSanityCheckTask.groovy

@@ -117,6 +117,7 @@ class HtmlSanityCheckTask extends DefaultTask {
         if (sourceDocuments == null) {
             sourceDocuments = project.fileTree(sourceDir)
             sourceDocuments.include('**/*.html')
+            sourceDocuments.include('**/*.htm')
         }
     }
 

htmlSanityCheck-gradle-plugin/src/test/groovy/org/aim42/htmlsanitycheck/gradle/HtmlSanityCheckTaskSpec.groovy

@@ -62,4 +62,32 @@ class HtmlSanityCheckTaskSpec extends HtmlSanityCheckBaseSpec {
         e.message.contains("Your build configuration included 'failOnErrors=true', and 1 error(s) were found on all checked pages.")
     }
 
+    def "should include .htm files when auto-populating sourceDocuments"() {
+        given:
+        // Create .htm file
+        def htmFile = new File(sourceDir, "document.htm")
+        htmFile << VALID_HTML
+
+        // Create .html file
+        htmlFile << VALID_HTML
+
+        // Create subdirectory with .htm file
+        def subDir = new File(sourceDir, "docs")
+        subDir.mkdirs()
+        def nestedHtmFile = new File(subDir, "nested.htm")
+        nestedHtmFile << VALID_HTML
+
+        when:
+        task.setSourceDir(testProjectDir.root)
+
+        then:
+        task.sourceDocuments != null
+        // Verify that sourceDocuments includes both .html and .htm files
+        def fileNames = task.sourceDocuments.files.collect { it.name }
+        fileNames.contains("test.html")
+        fileNames.contains("document.htm")
+        fileNames.contains("nested.htm")
+        task.sourceDocuments.files.size() == 3
+    }
+
 }

htmlSanityCheck-maven-plugin/README.adoc

@@ -41,7 +41,7 @@ Use the following snippet inside a Maven build file:
     </executions>
     <configuration>
         <sourceDocuments>
-            <file>src/file-to-test.html</file> <2>
+            <sourceDocument>src/file-to-test.html</sourceDocument> <2>
         </sourceDocuments>
         <sourceDir>src</sourceDir> <3>
     </configuration>
@@ -79,7 +79,7 @@ Type: Directory.
 +
 Type: `org.maven.api.file.FileCollection`.
 +
-Default: All files in `+{sourceDir}+` whose names end with `.html`.
+Default: All files in `+{sourceDir}+` whose names end with `.html` or `.htm`.
 
 `checkingResultsDir` (optional):: Directory where the checking results are written to.
 +

htmlSanityCheck-maven-plugin/src/main/java/org/aim42/htmlsanitycheck/maven/HtmlSanityCheckMojo.java

@@ -270,9 +270,16 @@ void logBuildParameter(Configuration myConfig) {
     }
 
     protected Configuration setupConfiguration() {
+        // Auto-populate sourceDocuments from sourceDir if not specified
+        // (same pattern as Gradle plugin and CLI)
+        Set<File> documentsToCheck = sourceDocuments;
+        if (documentsToCheck == null && sourceDir != null) {
+            documentsToCheck = findHtmlFiles(sourceDir);
+        }
+
         Configuration result = Configuration.builder()
                 .excludes(excludes.stream().map(Pattern::compile).collect(Collectors.toSet()))
-                .sourceDocuments(sourceDocuments)
+                .sourceDocuments(documentsToCheck)
                 .sourceDir(sourceDir)
                 .checkingResultsDir(checkingResultsDir)
                 .junitResultsDir(junitResultsDir)
@@ -301,6 +308,33 @@ protected Configuration setupConfiguration() {
 
         return result;
     }
+
+    /**
+     * Recursively finds all HTML files in the given directory.
+     * Searches for files with .html and .htm extensions.
+     *
+     * @param directory the directory to search
+     * @return set of HTML files found
+     */
+    private Set<File> findHtmlFiles(File directory) {
+        Set<File> htmlFiles = new HashSet<>();
+        if (directory == null || !directory.exists() || !directory.isDirectory()) {
+            return htmlFiles;
+        }
+
+        File[] files = directory.listFiles();
+        if (files != null) {
+            for (File file : files) {
+                if (file.isDirectory()) {
+                    // Recursively search subdirectories
+                    htmlFiles.addAll(findHtmlFiles(file));
+                } else if (file.isFile() && (file.getName().endsWith(".html") || file.getName().endsWith(".htm"))) {
+                    htmlFiles.add(file);
+                }
+            }
+        }
+        return htmlFiles;
+    }
 }
 
 /*========================================================================