1 | package eu.clarin.cmdi.vlo.importer; |
---|
2 | |
---|
3 | import com.ximpleware.AutoPilot; |
---|
4 | import com.ximpleware.NavException; |
---|
5 | import com.ximpleware.VTDException; |
---|
6 | import com.ximpleware.VTDGen; |
---|
7 | import com.ximpleware.VTDNav; |
---|
8 | import com.ximpleware.XPathEvalException; |
---|
9 | import com.ximpleware.XPathParseException; |
---|
10 | import eu.clarin.cmdi.vlo.config.VloConfig; |
---|
11 | import java.io.File; |
---|
12 | import java.io.FileInputStream; |
---|
13 | import java.io.IOException; |
---|
14 | import java.util.List; |
---|
15 | import java.util.Map; |
---|
16 | import org.apache.commons.io.IOUtils; |
---|
17 | import org.slf4j.Logger; |
---|
18 | import org.slf4j.LoggerFactory; |
---|
19 | |
---|
20 | public class CMDIParserVTDXML implements CMDIDataProcessor { |
---|
21 | private final Map<String, PostProcessor> postProcessors; |
---|
22 | private final static Logger LOG = LoggerFactory.getLogger(CMDIParserVTDXML.class); |
---|
23 | |
---|
24 | public CMDIParserVTDXML(Map<String, PostProcessor> postProcessors) { |
---|
25 | this.postProcessors = postProcessors; |
---|
26 | } |
---|
27 | |
---|
28 | @Override |
---|
29 | public CMDIData process(File file) throws VTDException, IOException { |
---|
30 | CMDIData result = new CMDIData(); |
---|
31 | VTDGen vg = new VTDGen(); |
---|
32 | vg.setDoc(IOUtils.toByteArray(new FileInputStream(file))); |
---|
33 | vg.parse(true); |
---|
34 | VTDNav nav = vg.getNav(); |
---|
35 | setNameSpace(nav);//setting namespace once, all other instance of AutoPilot keep the setting (a bit tricky). |
---|
36 | FacetMapping facetMapping = getFacetMapping(nav.cloneNav(), file.getAbsolutePath()); |
---|
37 | /** New nice error log to find erroneous files */ |
---|
38 | if(facetMapping.getFacets().size() == 0){ |
---|
39 | LOG.error("Problems mapping facets for file: " + file.getAbsolutePath()); |
---|
40 | } |
---|
41 | |
---|
42 | nav.toElement(VTDNav.ROOT); |
---|
43 | processResources(result, nav); |
---|
44 | processFacets(result, nav, facetMapping); |
---|
45 | return result; |
---|
46 | } |
---|
47 | |
---|
48 | static void setNameSpace(VTDNav nav) { |
---|
49 | AutoPilot ap = new AutoPilot(nav); |
---|
50 | ap.declareXPathNameSpace("c", "http://www.clarin.eu/cmd/"); |
---|
51 | } |
---|
52 | |
---|
53 | private FacetMapping getFacetMapping(VTDNav nav, String tolog) throws VTDException { |
---|
54 | String xsd = extractXsd(nav); |
---|
55 | if (xsd == null) { |
---|
56 | throw new RuntimeException("Cannot get xsd schema so cannot get a proper mapping. Parse failed!"); |
---|
57 | } |
---|
58 | if (xsd.indexOf("http") != xsd.lastIndexOf("http")){ |
---|
59 | LOG.info("FILE WITH WEIRD HTTP THINGY! " + tolog); |
---|
60 | } |
---|
61 | return FacetMappingFactory.getFacetMapping(xsd); |
---|
62 | } |
---|
63 | |
---|
64 | String extractXsd(VTDNav nav) throws VTDException { |
---|
65 | String xsd = getXsdFromHeader(nav); |
---|
66 | if (xsd == null) { |
---|
67 | xsd = getXsdFromSchemaLocation(nav); |
---|
68 | } |
---|
69 | return xsd; |
---|
70 | } |
---|
71 | |
---|
72 | private String getXsdFromHeader(VTDNav nav) throws XPathParseException, XPathEvalException, NavException { |
---|
73 | String result = null; |
---|
74 | nav.toElement(VTDNav.ROOT); |
---|
75 | AutoPilot ap = new AutoPilot(nav); |
---|
76 | ap.selectXPath("/c:CMD/c:Header/c:MdProfile/text()"); |
---|
77 | int index = ap.evalXPath(); |
---|
78 | if (index != -1) { |
---|
79 | String profileId = nav.toString(index).trim(); |
---|
80 | result = VloConfig.getComponentRegistryProfileSchema(profileId); |
---|
81 | } |
---|
82 | return result; |
---|
83 | } |
---|
84 | |
---|
85 | private String getXsdFromSchemaLocation(VTDNav nav) throws NavException { |
---|
86 | String result = null; |
---|
87 | nav.toElement(VTDNav.ROOT); |
---|
88 | int index = nav.getAttrValNS("http://www.w3.org/2001/XMLSchema-instance", "schemaLocation"); |
---|
89 | if (index != -1) { |
---|
90 | String schemaLocation = nav.toNormalizedString(index); |
---|
91 | result = schemaLocation.split(" ")[1]; |
---|
92 | } else { |
---|
93 | index = nav.getAttrValNS("http://www.w3.org/2001/XMLSchema-instance", "noNamespaceSchemaLocation"); |
---|
94 | if (index != -1) { |
---|
95 | result = nav.toNormalizedString(index); |
---|
96 | } |
---|
97 | } |
---|
98 | return result; |
---|
99 | } |
---|
100 | |
---|
101 | /* |
---|
102 | * kj: describe |
---|
103 | */ |
---|
104 | private void processResources(CMDIData result, VTDNav nav) throws VTDException { |
---|
105 | |
---|
106 | AutoPilot resourceProxy = new AutoPilot(nav); |
---|
107 | resourceProxy.selectXPath("/c:CMD/c:Resources/c:ResourceProxyList/c:ResourceProxy"); |
---|
108 | |
---|
109 | AutoPilot resourceRef = new AutoPilot(nav); |
---|
110 | resourceRef.selectXPath("c:ResourceRef"); |
---|
111 | AutoPilot resourceType = new AutoPilot(nav); |
---|
112 | resourceType.selectXPath("c:ResourceType"); |
---|
113 | AutoPilot resourceMimeType = new AutoPilot(nav); |
---|
114 | resourceMimeType.selectXPath("c:ResourceType/@mimetype"); |
---|
115 | |
---|
116 | while (resourceProxy.evalXPath() != -1) { |
---|
117 | String ref = resourceRef.evalXPathToString(); |
---|
118 | String type = resourceType.evalXPathToString(); |
---|
119 | String mimeType = resourceMimeType.evalXPathToString(); |
---|
120 | |
---|
121 | // if (!ref.equals("") && !type.equals("")){ |
---|
122 | if (!ref.equals("")){ |
---|
123 | // mime type is allowed to be an empty string, kj: check if this is allowed in general |
---|
124 | result.addResource(ref, type, mimeType); |
---|
125 | } |
---|
126 | } |
---|
127 | } |
---|
128 | |
---|
129 | private void processFacets(CMDIData result, VTDNav nav, FacetMapping facetMapping) throws VTDException { |
---|
130 | List<FacetConfiguration> facetList = facetMapping.getFacets(); |
---|
131 | for (FacetConfiguration config : facetList) { |
---|
132 | List<String> patterns = config.getPatterns(); |
---|
133 | for (String pattern : patterns) { |
---|
134 | boolean matchedPattern = matchPattern(result, nav, config, pattern, config.getAllowMultipleValues()); |
---|
135 | if (matchedPattern && !config.getAllowMultipleValues()) { |
---|
136 | break; |
---|
137 | } |
---|
138 | } |
---|
139 | } |
---|
140 | } |
---|
141 | |
---|
142 | private boolean matchPattern(CMDIData result, VTDNav nav, FacetConfiguration config, String pattern, Boolean allowMultipleValues) throws VTDException { |
---|
143 | boolean matchedPattern = false; |
---|
144 | AutoPilot ap = new AutoPilot(nav); |
---|
145 | ap.selectXPath(pattern); |
---|
146 | int index = ap.evalXPath(); |
---|
147 | while (index != -1) { |
---|
148 | matchedPattern = true; |
---|
149 | if (nav.getTokenType(index) == VTDNav.TOKEN_ATTR_NAME) { |
---|
150 | //if it is an attribute you need to add 1 to the index to get the right value |
---|
151 | index++; |
---|
152 | } |
---|
153 | String value = nav.toString(index); |
---|
154 | value = postProcess(config.getName(), value); |
---|
155 | result.addDocField(config.getName(), value, config.isCaseInsensitive()); |
---|
156 | index = ap.evalXPath(); |
---|
157 | |
---|
158 | if(!allowMultipleValues) |
---|
159 | break; |
---|
160 | } |
---|
161 | return matchedPattern; |
---|
162 | } |
---|
163 | |
---|
164 | private String postProcess(String name, String value) { |
---|
165 | String result = value; |
---|
166 | if (postProcessors.containsKey(name)) { |
---|
167 | PostProcessor processor = postProcessors.get(name); |
---|
168 | result = processor.process(value); |
---|
169 | } |
---|
170 | return result.trim(); |
---|
171 | } |
---|
172 | |
---|
173 | } |
---|