RAT-325 : Added check to skip expensive processing in SPDX

apache-rat-core/src/main/java/org/apache/rat/Report.java

@@ -126,13 +126,12 @@ public class Report {
      */
     public static void main(String[] args) throws Exception {
         Options opts = buildOptions();
-
         CommandLine cl;
         try {
             cl = new DefaultParser().parse(opts, args);
         } catch (ParseException e) {
-            System.err.println(e.getMessage());
-            System.err.println("Please use the \"--help\" option to see a list of valid commands and options");
+            DefaultLog.INSTANCE.error(e.getMessage());
+            DefaultLog.INSTANCE.error("Please use the \"--help\" option to see a list of valid commands and options");
             System.exit(1);
             return; // dummy return (won't be reached) to avoid Eclipse complaint about possible NPE
                     // for "cl"
@@ -147,7 +146,7 @@ public static void main(String[] args) throws Exception {
             printUsage(opts);
         } else {
             ReportConfiguration configuration = createConfiguration(args[0], cl);
-            configuration.validate(System.err::println);
+            configuration.validate(DefaultLog.INSTANCE::error);
 
             boolean dryRun = false;
 
@@ -210,7 +209,7 @@ static ReportConfiguration createConfiguration(String baseDirectory, CommandLine
             if (cl.hasOption(STYLESHEET_CLI)) {
                 String[] style = cl.getOptionValues(STYLESHEET_CLI);
                 if (style.length != 1) {
-                    System.err.println("Please specify a single stylesheet");
+                    DefaultLog.INSTANCE.error("Please specify a single stylesheet");
                     System.exit(1);
                 }
                 configuration.setStyleSheet(() -> Files.newInputStream(Paths.get(style[0])));
@@ -256,10 +255,10 @@ static FilenameFilter parseExclusions(List<String> excludes) {
                 orFilter.addFileFilter(new NameFileFilter(exclusion));
                 orFilter.addFileFilter(WildcardFileFilter.builder().setWildcards(exclusion).get());
             } catch (PatternSyntaxException e) {
-                System.err.println("Will skip given exclusion '" + exclude + "' due to " + e);
+                DefaultLog.INSTANCE.error("Will skip given exclusion '" + exclude + "' due to " + e);
             }
         }
-        System.err.println("Ignored " + ignoredLines + " lines in your exclusion files as comments or empty lines.");
+        DefaultLog.INSTANCE.error("Ignored " + ignoredLines + " lines in your exclusion files as comments or empty lines.");
         return new NotFileFilter(orFilter);
     }
 

apache-rat-core/src/main/java/org/apache/rat/ReportConfiguration.java

@@ -263,7 +263,7 @@ public void setStyleReport(boolean styleReport) {
      * closed consider wrapping it in a {@code NoCloseOutputStream}
      * 
      * @param out The OutputStream supplier that provides the output stream to write
-     * the report to. (may be null)
+     * the report to. A null value will use System.out.
      * @see NoCloseOutputStream
      */
     public void setOut(IOSupplier<OutputStream> out) {

apache-rat-core/src/main/java/org/apache/rat/analysis/DefaultAnalyserFactory.java

@@ -29,6 +29,8 @@
 import org.apache.rat.document.impl.guesser.BinaryGuesser;
 import org.apache.rat.document.impl.guesser.NoteGuesser;
 import org.apache.rat.license.ILicense;
+import org.apache.rat.utils.Log;
+import org.apache.rat.utils.Log.Level;
 
 /**
  * Creates default analysers.
@@ -40,11 +42,13 @@ public class DefaultAnalyserFactory {
      * @param licenses The licenses to use in  the Analyser.
      * @return A document analyser that uses the provides licenses.
      */
-    public static IDocumentAnalyser createDefaultAnalyser(Collection<ILicense> licenses) {
+    public static IDocumentAnalyser createDefaultAnalyser(Log log, Collection<ILicense> licenses) {
         if (licenses.isEmpty()) {
             throw new ConfigurationException("At least one license must be defined");
         }
-        return new DefaultAnalyser(new LicenseCollection(licenses));
+        log.debug("Licenses in Test");
+        licenses.forEach(log::debug);
+        return new DefaultAnalyser(log, new LicenseCollection(licenses));
     }
 
     /**
@@ -56,13 +60,16 @@ private final static class DefaultAnalyser implements IDocumentAnalyser {
          * The license to analyze
          */
         private final ILicense license;
+        /** The log to use */
+        private final Log log;
 
         /**
          * Constructs a DocumentAnalyser for the specified license.
          * @param license The license to analyse
          */
-        public DefaultAnalyser(final ILicense license) {
+        public DefaultAnalyser(final Log log, final ILicense license) {
             this.license = license;
+            this.log = log;
         }
 
         @Override
@@ -76,7 +83,7 @@ public void analyse(Document document) throws RatDocumentAnalysisException {
                 documentCategory = MetaData.RAT_DOCUMENT_CATEGORY_DATUM_BINARY;
             } else {
                 documentCategory = MetaData.RAT_DOCUMENT_CATEGORY_DATUM_STANDARD;
-                final DocumentHeaderAnalyser headerAnalyser = new DocumentHeaderAnalyser(license);
+                final DocumentHeaderAnalyser headerAnalyser = new DocumentHeaderAnalyser(log, license);
                 headerAnalyser.analyse(document);
             }
             document.getMetaData().set(documentCategory);

apache-rat-core/src/main/java/org/apache/rat/analysis/DocumentHeaderAnalyser.java

@@ -25,6 +25,9 @@
 import org.apache.rat.document.IDocumentAnalyser;
 import org.apache.rat.document.RatDocumentAnalysisException;
 import org.apache.rat.license.ILicense;
+import org.apache.rat.utils.Log;
+
+import static java.lang.String.format;
 
 /**
  * A Document analyzer that analyses document headers for a license.
@@ -35,21 +38,24 @@ class DocumentHeaderAnalyser implements IDocumentAnalyser {
      * The license to analyse
      */
     private final ILicense license;
+    /** the logger to use */
+    private final Log log;
 
     /**
      * Constructs the HeaderAnalyser for the specific license.
      * 
      * @param license The license to analyse
      */
-    public DocumentHeaderAnalyser(final ILicense license) {
+    public DocumentHeaderAnalyser(final Log log, final ILicense license) {
         super();
         this.license = license;
+        this.log = log;
     }
 
     @Override
     public void analyse(Document document) throws RatDocumentAnalysisException {
         try (Reader reader = document.reader()) {
-            // TODO: worker function should be moved into this class
+            log.info(format("Processing: %s", document));
             HeaderCheckWorker worker = new HeaderCheckWorker(reader, license, document);
             worker.read();
         } catch (IOException e) {

apache-rat-core/src/main/java/org/apache/rat/analysis/LicenseCollection.java

@@ -19,6 +19,7 @@
 package org.apache.rat.analysis;
 
 import java.util.Collection;
+import java.util.Collections;
 
 import org.apache.rat.analysis.matchers.AbstractMatcherContainer;
 import org.apache.rat.license.ILicense;
@@ -45,7 +46,7 @@ class LicenseCollection extends AbstractMatcherContainer implements ILicense {
      */
     public LicenseCollection(Collection<ILicense> enclosed) {
         super(enclosed);
-        this.enclosed = enclosed;
+        this.enclosed = Collections.unmodifiableCollection(enclosed);
         this.matchingLicense = null;
         this.lastState = State.i;
     }

apache-rat-core/src/main/java/org/apache/rat/analysis/matchers/CopyrightMatcher.java

@@ -106,16 +106,19 @@ public CopyrightMatcher(String id, String start, String stop, String owner) {
 
     @Override
     protected boolean doMatch(String line) {
-        Matcher matcher = COPYRIGHT_PATTERN.matcher(line);
-        if (matcher.find()) {
-            String buffer = line.substring(matcher.end());
-            matcher = dateOwnerPattern.matcher(buffer);
-            if (matcher.find() && matcher.start() == 0) {
-                return true;
-            }
-            if (ownerDatePattern != null) {
-                matcher = ownerDatePattern.matcher(buffer);
-                return matcher.find() && matcher.start() == 0;
+        String lowerLine = line.toLowerCase();
+        if (lowerLine.contains("copyright") || lowerLine.contains("(c)") || line.contains("©")) {
+            Matcher matcher = COPYRIGHT_PATTERN.matcher(line);
+            if (matcher.find()) {
+                String buffer = line.substring(matcher.end());
+                matcher = dateOwnerPattern.matcher(buffer);
+                if (matcher.find() && matcher.start() == 0) {
+                    return true;
+                }
+                if (ownerDatePattern != null) {
+                    matcher = ownerDatePattern.matcher(buffer);
+                    return matcher.find() && matcher.start() == 0;
+                }
             }
         }
         return false;

apache-rat-core/src/main/java/org/apache/rat/analysis/matchers/SPDXMatcherFactory.java

@@ -96,7 +96,7 @@ private boolean check(String line, Match caller) {
         // if the line has not been seen yet see if we can extract the SPDX id from the line.
         // if so then see if that name has been registered.  If so then we have a match and set 
         // lastMatch.
-        if (lastLine == null || !lastLine.equals(line)) {
+        if ((lastLine == null || !lastLine.equals(line)) && line.contains("SPDX-License-Identifier")) {
             Matcher matcher = groupSelector.matcher(line);
             if (matcher.find()) {
                 lastMatch = matchers.get(matcher.group(1));
@@ -127,5 +127,12 @@ public class Match extends AbstractSimpleMatcher {
         protected boolean doMatch(String line) {
             return SPDXMatcherFactory.this.check(line, this);
         }
+
+        @Override
+        public void reset() {
+            super.reset();
+            SPDXMatcherFactory.this.lastMatch = null;
+
+        }
     }
 }

apache-rat-core/src/main/java/org/apache/rat/annotation/AbstractLicenseAppender.java

@@ -19,6 +19,7 @@
 package org.apache.rat.annotation;
 
 import org.apache.commons.io.IOUtils;
+import org.apache.rat.utils.Log;
 
 import java.io.BufferedReader;
 import java.io.File;
@@ -196,9 +197,16 @@ public abstract class AbstractLicenseAppender {
     }
 
     private boolean isForced;
+    /** The log to use */
+    private final Log log;
 
-    public AbstractLicenseAppender() {
+    /**
+     * Constructor
+     * @param log The log to use.
+     */
+    public AbstractLicenseAppender(final Log log) {
         super();
+        this.log = log;
     }
 
     /**
@@ -247,11 +255,11 @@ public void append(File document) throws IOException {
         if (isForced) {
             boolean deleted = document.delete();
             if (!deleted) {
-                System.err.println("Could not delete original file to prepare renaming.");
+                log.error("Could not delete original file to prepare renaming.");
             }
             boolean renamed = newDocument.renameTo(document.getAbsoluteFile());
             if (!renamed) {
-                System.err.println("Failed to rename new file, original file remains unchanged.");
+                log.error("Failed to rename new file, original file remains unchanged.");
             }
         }
     }

apache-rat-core/src/main/java/org/apache/rat/annotation/ApacheV2LicenseAppender.java

@@ -20,6 +20,8 @@
 
 import java.io.File;
 
+import org.apache.rat.utils.Log;
+
 
 /**
  * Add an Apache License V2 license header to a
@@ -35,8 +37,8 @@ public class ApacheV2LicenseAppender extends AbstractLicenseAppender {
     /**
      * Create a license appender with the standard ASF license header.
      */
-    public ApacheV2LicenseAppender() {
-        super();
+    public ApacheV2LicenseAppender(final Log log) {
+        super(log);
     }
 
     /**
@@ -45,8 +47,8 @@ public ApacheV2LicenseAppender() {
      *
      * @param copyright copyright line.
      */
-    public ApacheV2LicenseAppender(String copyright) {
-        super();
+    public ApacheV2LicenseAppender(final Log log, String copyright) {
+        super(log);
         this.copyright = copyright;
     }
 

apache-rat-core/src/main/java/org/apache/rat/api/MetaData.java

@@ -205,6 +205,7 @@ public MetaData() {
      * @param license the license to add metadata for.
      */
     public void reportOnLicense(ILicense license) {
+
         if (StringUtils.isNotBlank(license.getNotes())) {
             set(new MetaData.Datum(MetaData.RAT_URL_HEADER_SAMPLE, license.getNotes()));
         }

apache-rat-core/src/main/java/org/apache/rat/report/claim/impl/xml/SimpleXmlClaimReporter.java

@@ -24,6 +24,7 @@
 import org.apache.rat.api.RatException;
 import org.apache.rat.report.AbstractReport;
 import org.apache.rat.report.xml.writer.IXmlWriter;
+import org.apache.rat.report.xml.writer.impl.base.XmlWriter;
 
 import java.io.IOException;
 import java.util.Calendar;

apache-rat-core/src/main/java/org/apache/rat/report/claim/util/LicenseAddingReport.java

@@ -27,13 +27,14 @@
 import org.apache.rat.api.MetaData.Datum;
 import org.apache.rat.api.RatException;
 import org.apache.rat.report.AbstractReport;
+import org.apache.rat.utils.Log;
 
 
 public class LicenseAddingReport extends AbstractReport {
     private final AbstractLicenseAppender appender;
 
-    public LicenseAddingReport(String pCopyrightMsg, boolean pForced) {
-        appender = pCopyrightMsg == null ? new ApacheV2LicenseAppender() : new ApacheV2LicenseAppender(pCopyrightMsg);
+    public LicenseAddingReport(final Log log, String pCopyrightMsg, boolean pForced) {
+        appender = pCopyrightMsg == null ? new ApacheV2LicenseAppender(log) : new ApacheV2LicenseAppender(log,pCopyrightMsg);
         appender.setForce(pForced);
     }
 

apache-rat-core/src/main/java/org/apache/rat/report/xml/XmlReportFactory.java

@@ -56,12 +56,12 @@ public static RatReport createStandardReport(IXmlWriter writer,
             reporters.add(new ClaimAggregator(statistic));
         }
         if (configuration.isAddingLicenses()) {
-            reporters.add(new LicenseAddingReport(configuration.getCopyrightMessage(), configuration.isAddingLicensesForced()));
+            reporters.add(new LicenseAddingReport(configuration.getLog(), configuration.getCopyrightMessage(), configuration.isAddingLicensesForced()));
         }
         reporters.add(new SimpleXmlClaimReporter(writer));
 
         final IDocumentAnalyser analyser =
-            DefaultAnalyserFactory.createDefaultAnalyser(configuration.getLicenses(LicenseFilter.all));
+            DefaultAnalyserFactory.createDefaultAnalyser(configuration.getLog(), configuration.getLicenses(LicenseFilter.all));
         final DefaultPolicy policy = new DefaultPolicy(configuration.getLicenseFamilies(LicenseFilter.approved));
 
         final IDocumentAnalyser[] analysers = {analyser, policy};

apache-rat-core/src/main/java/org/apache/rat/utils/DefaultLog.java

@@ -27,25 +27,32 @@ public class DefaultLog implements Log {
      * The instance of the default log.
      */
     public static final DefaultLog INSTANCE = new DefaultLog();
+
+    private Level level;
 
     private DefaultLog() {
+        level = Level.INFO;
     }
 
+    public void setLevel(Level level) {
+        this.level = level;
+    }
     @Override
     public void log(Level level, String msg) {
-        switch (level) {
-        case DEBUG:
-        case INFO:
-        case WARN:
-            System.out.format("%s: %s%n", level, msg);
-            break;
-        case ERROR:
-            System.err.format("%s: %s%n", level, msg);
-            break;
-		case OFF:
-			break;
-		default:
-			break;
-        }
+        if (this.level.ordinal() <= level.ordinal())
+            switch (level) {
+            case DEBUG:
+            case INFO:
+            case WARN:
+                System.out.format("%s: %s%n", level, msg);
+                break;
+            case ERROR:
+                System.err.format("%s: %s%n", level, msg);
+                break;
+            case OFF:
+                break;
+            default:
+                break;
+            }
     }
 }