jjg@1455: package tidystats;
jjg@1455: 
jjg@1455: import java.io.IOException;
jjg@1455: import java.nio.charset.Charset;
jjg@1455: import java.nio.file.FileSystem;
jjg@1455: import java.nio.file.FileSystems;
jjg@1455: import java.nio.file.Files;
jjg@1455: import java.nio.file.Path;
jjg@1455: import java.util.ArrayList;
jjg@1455: import java.util.Comparator;
jjg@1455: import java.util.HashMap;
jjg@1455: import java.util.List;
jjg@1455: import java.util.Map;
jjg@1455: import java.util.Set;
jjg@1455: import java.util.TreeMap;
jjg@1455: import java.util.TreeSet;
jjg@1455: import java.util.regex.Matcher;
jjg@1455: import java.util.regex.Pattern;
jjg@1455: 
jjg@1455: public class Main {
jjg@1455:     public static void main(String... args) throws IOException {
jjg@1455:         new Main().run(args);
jjg@1455:     }
jjg@1455: 
jjg@1455:     void run(String... args) throws IOException {
jjg@1455:         FileSystem fs = FileSystems.getDefault();
jjg@1455:         List<Path> paths = new ArrayList<>();
jjg@1455: 
jjg@1455:         int i;
jjg@1455:         for (i = 0; i < args.length; i++) {
jjg@1455:             String arg = args[i];
jjg@1455:             if (arg.startsWith("-"))
jjg@1455:                 throw new IllegalArgumentException(arg);
jjg@1455:             else
jjg@1455:                 break;
jjg@1455:         }
jjg@1455: 
jjg@1455:         for ( ; i < args.length; i++) {
jjg@1455:             Path p = fs.getPath(args[i]);
jjg@1455:             paths.add(p);
jjg@1455:         }
jjg@1455: 
jjg@1455:         for (Path p: paths) {
jjg@1455:             scan(p);
jjg@1455:         }
jjg@1455: 
jjg@1455:         print("%6d files read", files);
jjg@1455:         print("%6d files had no errors or warnings", ok);
jjg@1455:         print("%6d files reported \"Not all warnings/errors were shown.\"", overflow);
jjg@1455:         print("%6d errors found", errs);
jjg@1455:         print("%6d warnings found", warns);
jjg@1455:         print("%6d recommendations to use CSS", css);
jjg@1455:         print("");
jjg@1455: 
jjg@1455:         Map<Integer, Set<String>> sortedCounts = new TreeMap<>(
jjg@1455:                 new Comparator<Integer>() {
jjg@1455:                     @Override
jjg@1455:                     public int compare(Integer o1, Integer o2) {
jjg@1455:                         return o2.compareTo(o1);
jjg@1455:                     }
jjg@1455:                 });
jjg@1455: 
jjg@1455:         for (Map.Entry<Pattern, Integer> e: counts.entrySet()) {
jjg@1455:             Pattern p = e.getKey();
jjg@1455:             Integer n = e.getValue();
jjg@1455:             Set<String> set = sortedCounts.get(n);
jjg@1455:             if (set == null)
jjg@1455:                 sortedCounts.put(n, (set = new TreeSet<>()));
jjg@1455:             set.add(p.toString());
jjg@1455:         }
jjg@1455: 
jjg@1455:         for (Map.Entry<Integer, Set<String>> e: sortedCounts.entrySet()) {
jjg@1455:             for (String p: e.getValue()) {
jjg@1455:                 if (p.startsWith(".*")) p = p.substring(2);
jjg@1455:                 print("%6d: %s", e.getKey(), p);
jjg@1455:             }
jjg@1455:         }
jjg@1455:     }
jjg@1455: 
jjg@1455:     void scan(Path p) throws IOException {
jjg@1455:         if (Files.isDirectory(p)) {
jjg@1455:             for (Path c: Files.newDirectoryStream(p)) {
jjg@1455:                 scan(c);
jjg@1455:             }
jjg@1455:         } else if (isTidyFile(p)) {
jjg@1455:             scan(Files.readAllLines(p, Charset.defaultCharset()));
jjg@1455:         }
jjg@1455:     }
jjg@1455: 
jjg@1455:     boolean isTidyFile(Path p) {
jjg@1455:         return Files.isRegularFile(p) && p.getFileName().toString().endsWith(".tidy");
jjg@1455:     }
jjg@1455: 
jjg@1455:     void scan(List<String> lines) {
jjg@1455:         Matcher m;
jjg@1455:         files++;
jjg@1455:         for (String line: lines) {
jjg@1455:             if (okPattern.matcher(line).matches()) {
jjg@1455:                 ok++;
jjg@1455:             } else if ((m = countPattern.matcher(line)).matches()) {
jjg@1455:                 warns += Integer.valueOf(m.group(1));
jjg@1455:                 errs += Integer.valueOf(m.group(2));
jjg@1455:                 if (m.group(3) != null)
jjg@1455:                     overflow++;
jjg@1455:             } else if ((m = guardPattern.matcher(line)).matches()) {
jjg@1455:                 boolean found = false;
jjg@1455:                 for (Pattern p: patterns) {
jjg@1455:                     if ((m = p.matcher(line)).matches()) {
jjg@1455:                         found = true;
jjg@1455:                         count(p);
jjg@1455:                         break;
jjg@1455:                     }
jjg@1455:                 }
jjg@1455:                 if (!found)
jjg@1455:                     System.err.println("Unrecognized line: " + line);
jjg@1455:             } else if (cssPattern.matcher(line).matches()) {
jjg@1455:                 css++;
jjg@1455:             }
jjg@1455:         }
jjg@1455:     }
jjg@1455: 
jjg@1455:     Map<Pattern, Integer> counts = new HashMap<>();
jjg@1455:     void count(Pattern p) {
jjg@1455:         Integer i = counts.get(p);
jjg@1455:         counts.put(p, (i == null) ? 1 : i + 1);
jjg@1455:     }
jjg@1455: 
jjg@1455:     void print(String format, Object... args) {
jjg@1455:         System.out.println(String.format(format, args));
jjg@1455:     }
jjg@1455: 
jjg@1455:     Pattern okPattern = Pattern.compile("No warnings or errors were found.");
jjg@1455:     Pattern countPattern = Pattern.compile("([0-9]+) warnings, ([0-9]+) errors were found!.*?(Not all warnings/errors were shown.)?");
jjg@1455:     Pattern cssPattern = Pattern.compile("You are recommended to use CSS.*");
jjg@1455:     Pattern guardPattern = Pattern.compile("line [0-9]+ column [0-9]+ - (Error|Warning):.*");
jjg@1455: 
jjg@1455:     Pattern[] patterns = {
jjg@1455:         Pattern.compile(".*Error: <.*> is not recognized!"),
jjg@1455:         Pattern.compile(".*Error: missing quote mark for attribute value"),
jjg@1455:         Pattern.compile(".*Warning: <.*> anchor \".*\" already defined"),
jjg@1455:         Pattern.compile(".*Warning: <.*> attribute \".*\" has invalid value \".*\""),
jjg@1455:         Pattern.compile(".*Warning: <.*> attribute \".*\" lacks value"),
jjg@1455:         Pattern.compile(".*Warning: <.*> attribute \".*\" lacks value"),
jjg@1455:         Pattern.compile(".*Warning: <.*> attribute with missing trailing quote mark"),
jjg@1455:         Pattern.compile(".*Warning: <.*> dropping value \".*\" for repeated attribute \".*\""),
jjg@1455:         Pattern.compile(".*Warning: <.*> inserting \".*\" attribute"),
jjg@1455:         Pattern.compile(".*Warning: <.*> is probably intended as </.*>"),
jjg@1455:         Pattern.compile(".*Warning: <.*> isn't allowed in <.*> elements"),
jjg@1455:         Pattern.compile(".*Warning: <.*> lacks \".*\" attribute"),
jjg@1455:         Pattern.compile(".*Warning: <.*> missing '>' for end of tag"),
jjg@1455:         Pattern.compile(".*Warning: <.*> proprietary attribute \".*\""),
jjg@1455:         Pattern.compile(".*Warning: <.*> unexpected or duplicate quote mark"),
jjg@1455:         Pattern.compile(".*Warning: <a> cannot copy name attribute to id"),
jjg@1455:         Pattern.compile(".*Warning: <a> escaping malformed URI reference"),
jjg@1455:         Pattern.compile(".*Warning: <blockquote> proprietary attribute \"pre\""),
jjg@1455:         Pattern.compile(".*Warning: discarding unexpected <.*>"),
jjg@1455:         Pattern.compile(".*Warning: discarding unexpected </.*>"),
jjg@1455:         Pattern.compile(".*Warning: entity \".*\" doesn't end in ';'"),
jjg@1455:         Pattern.compile(".*Warning: inserting implicit <.*>"),
jjg@1455:         Pattern.compile(".*Warning: inserting missing 'title' element"),
jjg@1455:         Pattern.compile(".*Warning: missing <!DOCTYPE> declaration"),
jjg@1455:         Pattern.compile(".*Warning: missing <.*>"),
jjg@1455:         Pattern.compile(".*Warning: missing </.*> before <.*>"),
jjg@1455:         Pattern.compile(".*Warning: nested emphasis <.*>"),
jjg@1455:         Pattern.compile(".*Warning: plain text isn't allowed in <.*> elements"),
jjg@1455:         Pattern.compile(".*Warning: replacing <p> by <br>"),
jjg@1455:         Pattern.compile(".*Warning: replacing invalid numeric character reference .*"),
jjg@1455:         Pattern.compile(".*Warning: replacing unexpected .* by </.*>"),
jjg@1455:         Pattern.compile(".*Warning: trimming empty <.*>"),
jjg@1455:         Pattern.compile(".*Warning: unescaped & or unknown entity \".*\""),
jjg@1455:         Pattern.compile(".*Warning: unescaped & which should be written as &amp;"),
jjg@1455:         Pattern.compile(".*Warning: using <br> in place of <p>")
jjg@1455:     };
jjg@1455: 
jjg@1455:     int files;
jjg@1455:     int ok;
jjg@1455:     int warns;
jjg@1455:     int errs;
jjg@1455:     int css;
jjg@1455:     int overflow;
jjg@1455: }
jjg@1455: