src/org/marc4j/samples/PermissiveReaderExample.java

   1 package org.marc4j.samples;\r
   2 \r
   3 import java.io.File;\r
   4 import java.io.FileInputStream;\r
   5 import java.io.FileNotFoundException;\r
   6 import java.io.FileOutputStream;\r
   7 import java.io.InputStream;\r
   8 import java.io.OutputStream;\r
   9 import java.io.PrintStream;\r
  10 import java.util.Iterator;\r
  11 import java.util.List;\r
  12 \r
  13 import org.marc4j.ErrorHandler;\r
  14 import org.marc4j.MarcException;\r
  15 import org.marc4j.MarcPermissiveStreamReader;\r
  16 import org.marc4j.MarcReader;\r
  17 import org.marc4j.MarcStreamWriter;\r
  18 import org.marc4j.MarcWriter;\r
  19 import org.marc4j.marc.Record;\r
  20 \r
  21 public class PermissiveReaderExample\r
  22 {\r
  23 \r
  24     /**\r
  25      * This test program demonstrates the use of the MarcPermissiveStreamReader \r
  26      * to read Marc records, with the permissive setting turned on.  It also \r
  27      * demonstrates the capability of printing out the error messages that are\r
  28      * generated when the MarcPermissiveStreamReader encounters records with \r
  29      * structural error or encoding errors.\r
  30      * \r
  31      *  When run in verbose mode, (by passing -v as the first parameter) the \r
  32      *  program will display the entire record highlighting the lines in the \r
  33      *  record that have errors that the permissive reader was able to detect \r
  34      *  and make an attempt at correcting.  Following that the program will \r
  35      *  list all of the errors that it found in the record.\r
  36      *  \r
  37      *  When run in verbose mode as described above, the program is useful for\r
  38      *  validating records.\r
  39      *  \r
  40      *  Shown below is the output generated when the program is run on the file\r
  41      *  error.mrc found in the resources sub-directory in the samples directory:\r
  42      *  \r
  43      *  Fatal Exception: error parsing data field for tag: 250 with data:    \1fa1st ed.\r
  44      *  Typo         : Record terminator character not found at end of record length --- [ n/a : n/a ]\r
  45      *  Typo         : Record terminator appears after stated record length, reading extra bytes --- [ n/a : n/a ]\r
  46      *  Minor Error  : Field length found in record different from length stated in the directory. --- [ n/a : n/a ]\r
  47      *     LEADER 00715cam a2200205 a 4500\r
  48      *     001 12883376\r
  49      *     005 20030616111422.0\r
  50      *     008 020805s2002    nyu    j      000 1 eng  \r
  51      *     020   $a0786808772\r
  52      *     020   $a0786816155 (pbk.)\r
  53      *     040   $aDLC$cDLC$dDLC\r
  54      *     100 1 $aChabon, Michael.\r
  55      *     245 10$aSummerland /$cMichael Chabon.\r
  56      *     250   $a1st ed.\r
  57      *     260   $aNew York :$bMiramax Books/Hyperion Books for Children,$cc2002.\r
  58      *     300   $a500 p. ;$c22 cm.\r
  59      *     520   $aEthan Feld, the worst baseball player in the history of the game, finds himself recruited by a 100-year-old scout to help a band of fairies triumph over an ancient enemy.\r
  60      *     650  1$aFantasy.\r
  61      *     650  1$aBaseball$vFiction.\r
  62      *     650  1$aMagic$vFiction.\r
  63      */\r
  64     public static void main(String[] args)\r
  65     {\r
  66         PrintStream out = System.out;\r
  67         boolean verbose = Boolean.parseBoolean(System.getProperty("marc.verbose"));\r
  68         boolean veryverbose = Boolean.parseBoolean(System.getProperty("marc.verbose"));\r
  69         if (args[0].equals("-v")) \r
  70         {\r
  71             verbose = true;\r
  72             String newArgs[] = new String[args.length-1];\r
  73             System.arraycopy(args, 1, newArgs, 0, args.length-1);\r
  74             args = newArgs;\r
  75         }\r
  76         if (args[0].equals("-vv")) \r
  77         {\r
  78             verbose = true;\r
  79             veryverbose = true;\r
  80             String newArgs[] = new String[args.length-1];\r
  81             System.arraycopy(args, 1, newArgs, 0, args.length-1);\r
  82             args = newArgs;\r
  83         }\r
  84         String fileStr = args[0];\r
  85         File file = new File(fileStr);\r
  86         MarcReader readerNormal = null;\r
  87         MarcReader readerPermissive = null;\r
  88         boolean to_utf_8 = true;\r
  89        \r
  90         InputStream inNorm;\r
  91         InputStream inPerm;\r
  92         OutputStream patchedRecStream = null;\r
  93         MarcWriter patchedRecs = null;\r
  94         ErrorHandler errorHandler = new ErrorHandler();\r
  95         try\r
  96         {\r
  97             inNorm = new FileInputStream(file);\r
  98             readerNormal = new MarcPermissiveStreamReader(inNorm, false, to_utf_8);\r
  99             inPerm = new FileInputStream(file);\r
 100             readerPermissive = new MarcPermissiveStreamReader(inPerm, errorHandler, to_utf_8, "BESTGUESS");\r
 101         }\r
 102         catch (FileNotFoundException e)\r
 103         {\r
 104             // TODO Auto-generated catch block\r
 105             e.printStackTrace();\r
 106         }\r
 107         if (args.length > 1)\r
 108         {\r
 109             try\r
 110             {\r
 111                 patchedRecStream = new FileOutputStream(new File(args[1]));\r
 112                 patchedRecs = new MarcStreamWriter(patchedRecStream);\r
 113             }\r
 114             catch (FileNotFoundException e)\r
 115             {\r
 116                 // TODO Auto-generated catch block\r
 117                 e.printStackTrace();\r
 118             }\r
 119         }\r
 120         while (readerNormal.hasNext() && readerPermissive.hasNext())\r
 121         {\r
 122             Record recNorm;\r
 123             Record recPerm;\r
 124             recPerm = readerPermissive.next();\r
 125             String strPerm = recPerm.toString();\r
 126             try {\r
 127                 recNorm = readerNormal.next();\r
 128             }\r
 129             catch (MarcException me)\r
 130             {\r
 131                 if (verbose)\r
 132                 {\r
 133                     out.println("Fatal Exception: "+ me.getMessage());\r
 134                     dumpErrors(out, errorHandler);\r
 135                     showDiffs(out, null, strPerm);\r
 136                     out.println("-------------------------------------------------------------------------------------");\r
 137                 }\r
 138                 continue;\r
 139             }\r
 140             String strNorm = recNorm.toString();\r
 141             if (!strNorm.equals(strPerm))\r
 142             {\r
 143                 if (verbose)\r
 144                 {\r
 145                     dumpErrors(out, errorHandler);\r
 146                     showDiffs(out, strNorm, strPerm);\r
 147                     out.println("-------------------------------------------------------------------------------------");\r
 148                     \r
 149                 }\r
 150                 if (patchedRecs != null)\r
 151                 {\r
 152                     patchedRecs.write(recPerm);\r
 153                 }\r
 154             }\r
 155             else if (errorHandler.hasErrors())\r
 156             {\r
 157                 if (verbose)\r
 158                 {\r
 159                     out.println("Results identical, but errors reported");\r
 160                     dumpErrors(out, errorHandler);\r
 161                     showDiffs(out, strNorm, strPerm);\r
 162                     out.println("-------------------------------------------------------------------------------------");\r
 163                 }\r
 164                 if (patchedRecs != null)\r
 165                 {\r
 166                     patchedRecs.write(recPerm);\r
 167                 }\r
 168             }\r
 169             else if (veryverbose)\r
 170             {\r
 171                 showDiffs(out, strNorm, strPerm);                \r
 172             }\r
 173                 \r
 174         }\r
 175     }\r
 176 \r
 177     public static void showDiffs(PrintStream out, String strNorm, String strPerm)\r
 178     {\r
 179         if (strNorm != null)\r
 180         {\r
 181             String normLines[] = strNorm.split("\n");\r
 182             String permLines[] = strPerm.split("\n");\r
 183             if (normLines.length == permLines.length)\r
 184             {\r
 185                 for (int i = 0; i < normLines.length; i++)\r
 186                 {\r
 187                     if (normLines[i].equals(permLines[i]))\r
 188                     {\r
 189                         out.println("   " + normLines[i]);\r
 190                     }\r
 191                     else\r
 192                     {\r
 193                         out.println(" < " + normLines[i]);\r
 194                         out.println(" > " + permLines[i]);                    \r
 195                     }\r
 196                 }\r
 197             }\r
 198         }\r
 199         else\r
 200         {\r
 201             String permLines[] = strPerm.split("\n");\r
 202             for (int i = 0; i < permLines.length; i++)\r
 203             {\r
 204                 out.println("   " + permLines[i]);\r
 205             }\r
 206         }\r
 207 \r
 208     }\r
 209     \r
 210     @SuppressWarnings("unchecked")\r
 211         public static void dumpErrors(PrintStream out, ErrorHandler errorHandler)\r
 212     {\r
 213         List<Object> errors = errorHandler.getErrors();\r
 214         if (errors != null) \r
 215         {\r
 216             Iterator<Object> iter = errors.iterator();\r
 217             while (iter.hasNext())\r
 218             {\r
 219                 Object error = iter.next();\r
 220                 out.println(error.toString());\r
 221             }\r
 222         }\r
 223     }\r
 224 }\r