Mercurial > hg > IndexMetaContextualizer

comparison src/main/java/de/mpiwg/indexmeta/IndexMetaParser.java @ 8:9ce7979fd037

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Implementation Data Provider
author Jorge Urzua <jurzua@mpiwg-berlin.mpg.de>
date Wed, 24 Apr 2013 10:34:23 +0200
parents 7d231e4e86e5
children
comparison
equal deleted inserted replaced
7:bc57f2660b0f 8:9ce7979fd037
1 package de.mpiwg.indexmeta; 1 package de.mpiwg.indexmeta;
2 2
3 import java.io.File; 3 import java.io.File;
4 import java.io.FilenameFilter;
4 import java.util.ArrayList; 5 import java.util.ArrayList;
5 import java.util.List; 6 import java.util.List;
6 7
7 import javax.xml.parsers.DocumentBuilder; 8 import javax.xml.parsers.DocumentBuilder;
8 import javax.xml.parsers.DocumentBuilderFactory; 9 import javax.xml.parsers.DocumentBuilderFactory;
9 import javax.xml.xpath.XPath; 10 import javax.xml.xpath.XPath;
10 import javax.xml.xpath.XPathConstants; 11 import javax.xml.xpath.XPathConstants;
11 import javax.xml.xpath.XPathExpression; 12 import javax.xml.xpath.XPathExpression;
12 import javax.xml.xpath.XPathFactory; 13 import javax.xml.xpath.XPathFactory;
13 14
15 import org.apache.commons.lang.StringUtils;
14 import org.w3c.dom.Document; 16 import org.w3c.dom.Document;
15 import org.w3c.dom.Element; 17 import org.w3c.dom.Element;
16 import org.w3c.dom.Node; 18 import org.w3c.dom.Node;
17 import org.w3c.dom.NodeList; 19 import org.w3c.dom.NodeList;
18 20
19 import de.mpiwg.indexmeta.bo.Contextualization; 21 import de.mpiwg.indexmeta.bo.Contextualization;
20 22
21 public class IndexMetaParser { 23 public class IndexMetaParser {
22 24
23 25 public static String indexMetaFile = "index.meta";
26 public static String annotatedIndexMetaFile = "index.meta.annot";
27
28 /*
24 public static List<Contextualization> getCtxItems(String filePath){ 29 public static List<Contextualization> getCtxItems(String filePath){
25 List<Contextualization> rs = new ArrayList<Contextualization>(); 30 List<Contextualization> rs = new ArrayList<Contextualization>();
26 31
27 try { 32 try {
28 File file = new File("/Users/jurzua/Projects/workspace/contextualization/data/index.meta/01index.meta.anno.xml"); 33 File file = new File("/Users/jurzua/Projects/workspace/contextualization/data/index.meta/01index.meta.anno.xml");
29 34
35 if(file.isFile()){
36 rs = getCtxItemsFromFile(file, rs);
37 }else if(file.isDirectory()){
38 rs = getCtxItemsFromDirectory(file, rs);
39 }
40
41 } catch (Exception e) {
42 e.printStackTrace();
43 }
44
45 return rs;
46 }
47 */
48
49 public static List<Contextualization> getCtxFromDirectory(String directory){
50 List<Contextualization> rs = new ArrayList<Contextualization>();
51
52 try {
53 File file = new File(directory);
54 rs = getCtxItemsFromDirectory(file, rs);
55 } catch (Exception e) {
56 e.printStackTrace();
57 }
58
59 return rs;
60 }
61
62 public static List<Contextualization> getCtxFromResource(String indexMetaResource){
63 List<Contextualization> rs = new ArrayList<Contextualization>();
64
65 try {
66 File file = new File(indexMetaResource);
67 rs = getCtxItemsFromIndexMetaResource(file, rs);
68 } catch (Exception e) {
69 e.printStackTrace();
70 }
71
72 return rs;
73 }
74
75
76
77 private static List<Contextualization> getCtxItemsFromDirectory(File directory, List<Contextualization> rs) throws Exception{
78 String canonicalPath = directory.getCanonicalPath();
79 System.out.println(canonicalPath);
80
81 File[] files = directory.listFiles();
82
83 for(File indexMetaResource : files){
84
85 if(indexMetaResource.isDirectory()){
86 rs = getCtxItemsFromIndexMetaResource(indexMetaResource, rs);
87 }
88 }
89 return rs;
90 }
91
92 private static List<Contextualization> getCtxItemsFromIndexMetaResource(File indexMetaResource, List<Contextualization> rs){
93
94 System.out.print("Working on: " + indexMetaResource.getName());
95
96 File[] list0 = indexMetaResource.listFiles(indexMetaFilter);
97 File annotatedFile = null;
98
99 if(list0.length != 0){
100 System.out.print("\tFound: " + indexMetaFile);
101 File[] list1 = indexMetaResource.listFiles(annotatedIndexMetaFilter);
102 if(list1.length == 0){
103 //create annotated file
104 System.out.print("\tAnnotated no found");
105 try{
106 annotatedFile = AnnotateIndexMeta.xmlParse(list0[0].getAbsolutePath(), null);
107 }catch (Exception e) {
108 e.printStackTrace();
109 }
110
111 }else{
112 annotatedFile = list1[0];
113 }
114 }
115 rs = getCtxItemsFromFile(annotatedFile, rs, indexMetaResource.getName());
116
117 System.out.println();
118
119 return rs;
120 }
121
122 private static List<Contextualization> getCtxItemsFromFile(File file, List<Contextualization> rs, String indexMetaId){
123
124 try {
30 DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance(); 125 DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
31 DocumentBuilder builder = factory.newDocumentBuilder(); 126 DocumentBuilder builder = factory.newDocumentBuilder();
32 Document doc = builder.parse(file); 127 Document doc = builder.parse(file);
33 128
129
130 for(String ctxElement : Contextualization.contextualizableList){}
131
34 List<Node> nodes = getNodeList(doc, Contextualization.AUTHOR); 132 List<Node> nodes = getNodeList(doc, Contextualization.AUTHOR);
35 for(Node node : nodes){ 133 for(Node node : nodes){
36 Contextualization ctx = getCtx(node, "XX"); 134 Contextualization ctx = getCtx(node, indexMetaId);
37 if(ctx != null){ 135 if(ctx != null){
38 rs.add(ctx); 136 rs.add(ctx);
39 } 137 }
40 } 138 }
41 139
42 nodes = getNodeList(doc, Contextualization.CITY); 140 nodes = getNodeList(doc, Contextualization.CITY);
43 for(Node node : nodes){ 141 for(Node node : nodes){
44 Contextualization ctx = getCtx(node, "XX"); 142 Contextualization ctx = getCtx(node, indexMetaId);
45 if(ctx != null){ 143 if(ctx != null){
46 rs.add(ctx); 144 rs.add(ctx);
47 } 145 }
48 } 146 }
49 147 } catch (Exception e) {
50 for(Contextualization ctx : rs){ 148 e.printStackTrace();
51 System.out.println(ctx.toString()); 149 }
52 } 150
53 151
54 } catch (Exception e) { 152 return rs;
55 e.printStackTrace(); 153 }
56 } 154
57
58 return rs;
59 }
60 155
61 public static List<Node> getNodeList(Document doc, String tagName){ 156 public static List<Node> getNodeList(Document doc, String tagName){
62 List<Node> rs = new ArrayList<Node>(); 157 List<Node> rs = new ArrayList<Node>();
63 158
64 try { 159 try {
104 return null; 199 return null;
105 } 200 }
106 201
107 public static void main(String[] args){ 202 public static void main(String[] args){
108 203
109 getCtxItems("/Users/jurzua/Projects/workspace/contextualization/data/index.meta/01index.meta"); 204 //--/Volumes/online_permanent/library
205 //getCtxItems("/Users/jurzua/Projects/workspace/contextualization/data/index.meta/01index.meta");
206 List<Contextualization> rs = getCtxFromDirectory("/Users/jurzua/Projects/max-planck/index_meta/library");
207 //List<Contextualization> rs = getCtxFromResource("/Users/jurzua/Projects/max-planck/index_meta/library/BB1RH90M");
208
209 for(Contextualization ctx : rs){
210 System.out.println(ctx.toString());
211 }
212
110 } 213 }
111 214
112 public static void printXpathResult(Object result){ 215 public static void printXpathResult(Object result){
113 NodeList nodes = (NodeList) result; 216 NodeList nodes = (NodeList) result;
114 for (int i = 0; i < nodes.getLength(); i++) { 217 for (int i = 0; i < nodes.getLength(); i++) {
122 System.out.println("getFirstChild value= " + node.getFirstChild().getNodeValue()); 225 System.out.println("getFirstChild value= " + node.getFirstChild().getNodeValue());
123 System.out.println(node); 226 System.out.println(node);
124 } 227 }
125 } 228 }
126 } 229 }
230
231 private static FilenameFilter indexMetaFilter = new FilenameFilter() {
232 public boolean accept(File directory, String fileName) {
233 return StringUtils.equals(fileName, indexMetaFile);
234 }
235 };
236
237 private static FilenameFilter annotatedIndexMetaFilter = new FilenameFilter() {
238 public boolean accept(File directory, String fileName) {
239 return StringUtils.equals(fileName, annotatedIndexMetaFile);
240 }
241 };
127 } 242 }