WIP 276 Add excludes to CLI and Maven plugin

ascheman · ascheman · commit 4f4ab8800e14 · 2025-02-24T23:31:57.000+01:00
diff --git a/htmlSanityCheck-cli/README.adoc b/htmlSanityCheck-cli/README.adoc
@@ -110,6 +110,16 @@ Default:
 include::../htmlSanityCheck-core/src/main/java/org/aim42/htmlsanitycheck/check/AllCheckers.java[tag=checker-classes,indent=0]
 ----
 
+`--excludes` (optional):: This is a set of regular expressions of URLs or even hosts that HSC should not check.
++
+Type: Set.
++
+Default: Empty list
++
+[source,groovy]
+----
+[]
+----
 
 `--httpWarningCodes` (optional):: HTTP response codes treated as warning.
 +
diff --git a/htmlSanityCheck-cli/src/main/groovy/org/aim42/htmlsanitycheck/cli/HscCommand.groovy b/htmlSanityCheck-cli/src/main/groovy/org/aim42/htmlsanitycheck/cli/HscCommand.groovy
@@ -88,6 +88,9 @@ class HscCommand implements Runnable {
     @Option(names = ["-s", "--suffix"], description = "File name suffixes to investigate", split = ",")
     String[] suffixes = ["html", "htm"]
 
+    @Option(names = ["-e", "--exclude"], description = "Exclude remote patterns to check", split = ',')
+    String[] excludes = []
+
     @Parameters(index = "0", arity = "0..1", description = "base directory (default: current directory)")
     File srcDir = new File(".").getAbsoluteFile()
 
@@ -172,6 +175,7 @@ class HscCommand implements Runnable {
                     .sourceDocuments(srcDocuments as Set)
                     .checkingResultsDir(resultsDirectory)
                     .checksToExecute(AllCheckers.CHECKER_CLASSES)
+                    .excludes(hscCommand.excludes as Set)
                     .build()
 
             // if we have no valid configuration, abort with exception
diff --git a/htmlSanityCheck-cli/src/test/groovy/org/aim42/htmlsanitycheck/cli/HscCommandSpec.groovy b/htmlSanityCheck-cli/src/test/groovy/org/aim42/htmlsanitycheck/cli/HscCommandSpec.groovy
@@ -8,7 +8,13 @@ import spock.lang.Unroll
 
 class HscCommandSpec extends Specification {
 
-    private final static VALID_HTML = """<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"><html><head></head><body></body><html>"""
+    private final static VALID_HTML = """<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN">
+        <html>
+            <head></head>
+            <body>
+                This <a href="https://tld.invalid/">Invalid TLD</a> should not make a problem!
+            </body>
+        <html>"""
     private final static INVALID_HTML = """<body><span id="id"/><span id="id"/></body> """
 
     @Rule
@@ -86,7 +92,7 @@ class HscCommandSpec extends Specification {
         if (args) {
             mainArgs.add(args)
         }
-        mainArgs.add(testProjectDir.root)
+        mainArgs.add(testProjectDir.root as String)
 
         when:
         HscCommand.main(mainArgs as String[])
@@ -112,7 +118,7 @@ class HscCommandSpec extends Specification {
     def "test with valid HTML file"() {
         given:
         htmlFile << VALID_HTML
-        String[] args = ["-r", testResultsDir.root, testProjectDir.root]
+        String[] args = ["-e", "^.*\\.invalid.*", "-r", testResultsDir.root, testProjectDir.root]
 
         when:
         HscCommand.main(args)
diff --git a/htmlSanityCheck-maven-plugin/README.adoc b/htmlSanityCheck-maven-plugin/README.adoc
@@ -130,6 +130,17 @@ Default: All available checker classes.
 include::../htmlSanityCheck-core/src/main/java/org/aim42/htmlsanitycheck/check/AllCheckers.java[tag=checker-classes,indent=0]
 ----
 
+`excludes` (optional):: This is a set of regular expressions of URLs or even hosts that HSC should not check.
++
+Type: Set.
++
+Default: Empty list
++
+[source,groovy]
+----
+[]
+----
+
 `httpWarningCodes` (optional):: Additional HTTP response codes treated as warning.
 +
 Type: List.
diff --git a/htmlSanityCheck-maven-plugin/src/main/java/org/aim42/htmlsanitycheck/maven/HtmlSanityCheckMojo.java b/htmlSanityCheck-maven-plugin/src/main/java/org/aim42/htmlsanitycheck/maven/HtmlSanityCheckMojo.java
@@ -16,6 +16,7 @@
 import java.io.File;
 import java.io.IOException;
 import java.util.Collections;
+import java.util.HashSet;
 import java.util.List;
 import java.util.Set;
 
@@ -183,6 +184,20 @@ public class HtmlSanityCheckMojo extends AbstractMojo {
     @Parameter
     private List<Class<? extends Checker>> checkerClasses = AllCheckers.CHECKER_CLASSES;
 
+    /**
+     * (optional)
+     * Provides a set of patterns to exclude certain URLs (hosts) from being processed.
+     * <p>
+     * Patterns can include wildcards and will be matched against the hrefs.
+     * Useful for excluding hosts that frequently run into problems (temporarily not reachable etc.).
+     * <p>
+     * Type: Set of Strings.
+     * <p></p>
+     * Default: Empty set (no exclusions).
+     */
+    @Parameter
+    private Set<String> excludes = new HashSet<>();
+
     static PerRunResults performChecks(Configuration myConfig) throws MojoExecutionException {
         try {
             AllChecksRunner allChecksRunner = new AllChecksRunner(myConfig);
@@ -249,10 +264,12 @@ void logBuildParameter(Configuration myConfig) {
         getLog().info("Results dir     : " + myConfig.getCheckingResultsDir());
         getLog().info("JUnit dir       : " + myConfig.getJunitResultsDir());
         getLog().info("Fail on errors  : " + myConfig.getFailOnErrors());
+        getLog().info("Excludes        : " + myConfig.getExcludes());
     }
 
     protected Configuration setupConfiguration() {
         Configuration result = Configuration.builder()
+                .excludes(excludes)
                 .sourceDocuments(sourceDocuments)
                 .sourceDir(sourceDir)
                 .checkingResultsDir(checkingResultsDir)
diff --git a/htmlSanityCheck-maven-plugin/src/test/java/org/aim42/htmlsanitycheck/maven/HtmlSanityCheckMojoTest.java b/htmlSanityCheck-maven-plugin/src/test/java/org/aim42/htmlsanitycheck/maven/HtmlSanityCheckMojoTest.java
@@ -19,7 +19,13 @@
 
 class HtmlSanityCheckMojoTest {
 
-    static final String VALID_HTML = "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01//EN\"><html><head></head><body></body><html>";
+    static final String VALID_HTML = "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01//EN\">  " +
+            "<html> " +
+            "  <head></head> " +
+            "  <body>" +
+            "    This <a href=\"https://tld.invalid/\">Invalid TLD</a> should not make a problem! " +
+            "  </body> " +
+            "<html>";
 
     @Test
     void setupConfiguration() {
@@ -162,13 +168,16 @@ void execuuserte() throws IOException, MojoExecutionException {
         Files.write(sourceFile.toPath(), VALID_HTML.getBytes(StandardCharsets.UTF_8));
         Set<File> fileset = new HashSet<>();
         fileset.add(sourceFile);
+        Set<String> excludes = new HashSet<>();
+        excludes.add("^.*\\.invalid.*");
 
         Configuration myConfig = Configuration.builder()
                 .checksToExecute(AllCheckers.CHECKER_CLASSES)
                 .junitResultsDir(junitDir.toFile())
                 .checkingResultsDir(resultDir.toFile())
                 .sourceDir(sourceDir.toFile())
                 .sourceDocuments(fileset)
+                .excludes(excludes)
                 .build();
         HtmlSanityCheckMojo mojo = new HtmlSanityCheckMojo();
 
