teichmann@5829: package org.dive4elements.river.importer.parsers;
sascha@1211: 
sascha@1211: import org.w3c.dom.Document;
sascha@1211: import org.w3c.dom.NodeList;
sascha@1211: import org.w3c.dom.Element;
sascha@1211: 
sascha@1211: import javax.xml.xpath.XPathConstants;
sascha@1211: 
sascha@1211: import java.util.Map;
sascha@1211: import java.util.HashMap;
sascha@1211: import java.util.List;
sascha@1211: import java.util.ArrayList;
sascha@1211: 
sascha@1211: import java.util.regex.Pattern;
sascha@1211: import java.util.regex.Matcher;
sascha@1211: 
sascha@1211: import org.apache.log4j.Logger;
sascha@1211: 
teichmann@5829: import org.dive4elements.artifacts.common.utils.XMLUtils;
sascha@1211: 
teichmann@5829: import org.dive4elements.river.importer.ImportAnnotationType;
teichmann@5829: import org.dive4elements.river.importer.Importer;
sascha@1211: 
sascha@1211: public class AnnotationClassifier
sascha@1211: {
sascha@1211:     private static Logger log = Logger.getLogger(Importer.class);
sascha@1211: 
sascha@1211:     public static final String TYPES_XPATH =
sascha@1211:         "/annotation/types/type";
sascha@1211: 
sascha@1211:     public static final String FILE_PATTERNS_XPATH =
sascha@1211:         "/annotation/patterns/file";
sascha@1211: 
sascha@1211:     public static final String DESCRIPTION_PATTERNS_XPATH =
sascha@1211:         "/annotation/patterns/line";
sascha@1211: 
sascha@1211: 
sascha@1211:     public static class Pair {
sascha@1211: 
sascha@1211:         protected Pattern              pattern;
sascha@1211:         protected ImportAnnotationType annType;
sascha@1211: 
sascha@1211:         public Pair(Pattern pattern, ImportAnnotationType annType) {
sascha@1211:             this.pattern  = pattern;
sascha@1211:             this.annType = annType;
sascha@1211:         }
sascha@1211: 
sascha@1211:         public ImportAnnotationType match(String s) {
sascha@1211:             Matcher m = pattern.matcher(s);
sascha@1211:             return m.matches() ? annType : null;
sascha@1211:         }
sascha@1211:     } // class Pair
sascha@1211: 
sascha@1211: 
sascha@1211:     protected Map<String, ImportAnnotationType> types;
sascha@1211:     protected List<Pair>                        filePatterns;
sascha@1211:     protected List<Pair>                        descPatterns;
sascha@1211: 
sascha@1211:     protected ImportAnnotationType defaultType;
sascha@1211: 
sascha@1211:     public AnnotationClassifier() {
sascha@1211:     }
sascha@1211: 
sascha@1211:     public AnnotationClassifier(Document rules) {
sascha@1211:         types        = new HashMap<String, ImportAnnotationType>();
sascha@1211:         filePatterns = new ArrayList<Pair>();
sascha@1211:         descPatterns = new ArrayList<Pair>();
sascha@1211: 
sascha@1211:         buildRules(rules);
sascha@1211:     }
sascha@1211: 
sascha@1211:     protected void buildRules(Document rules) {
sascha@1211:         buildTypes(rules);
sascha@1211:         buildFilePatterns(rules);
sascha@1211:         buildDescriptionPatterns(rules);
sascha@1211:     }
sascha@1211: 
sascha@1211:     protected void buildTypes(Document rules) {
sascha@1211: 
sascha@1211:         NodeList typeList = (NodeList)XMLUtils.xpath(
sascha@1211:             rules,
sascha@1211:             TYPES_XPATH,
sascha@1211:             XPathConstants.NODESET,
sascha@1211:             null);
sascha@1211: 
sascha@1211:         if (typeList == null) {
sascha@1211:             log.info("no rules found.");
sascha@1211:             return;
sascha@1211:         }
sascha@1211: 
sascha@1211:         for (int i = 0, N = typeList.getLength(); i < N; ++i) {
sascha@1211:             Element typeElement = (Element)typeList.item(i);
sascha@1211:             String name = typeElement.getAttribute("name");
sascha@1211:             if (name.length() == 0) {
sascha@3662:                 log.warn("ANNCLASS: rule has no name");
sascha@1211:                 continue;
sascha@1211:             }
sascha@1211: 
sascha@1211:             ImportAnnotationType aic = new ImportAnnotationType(name);
sascha@1211: 
sascha@1211:             types.put(name, aic);
sascha@1211: 
sascha@1211:             if (typeElement.getAttribute("default").equals("true")) {
sascha@1211:                 defaultType = aic;
sascha@1211:             }
sascha@1211:         }
sascha@1211:     }
sascha@1211: 
sascha@1211:     protected void buildFilePatterns(Document rules) {
sascha@1211: 
sascha@1211:         NodeList patternList = (NodeList)XMLUtils.xpath(
sascha@1211:             rules,
sascha@1211:             FILE_PATTERNS_XPATH,
sascha@1211:             XPathConstants.NODESET,
sascha@1211:             null);
sascha@1211: 
sascha@1211:         if (patternList == null) {
sascha@1211:             log.info("no file patterns found.");
sascha@1211:             return;
sascha@1211:         }
sascha@1211: 
sascha@1211:         for (int i = 0, N = patternList.getLength(); i < N; ++i) {
sascha@1211:             Element element = (Element)patternList.item(i);
sascha@1211:             Pair pair = buildPair(element);
sascha@1211:             if (pair != null) {
sascha@1211:                 filePatterns.add(pair);
sascha@1211:             }
sascha@1211:         }
sascha@1211:     }
sascha@1211: 
sascha@1211:     protected void buildDescriptionPatterns(Document rules) {
sascha@1211: 
sascha@1211:         NodeList patternList = (NodeList)XMLUtils.xpath(
sascha@1211:             rules,
sascha@1211:             DESCRIPTION_PATTERNS_XPATH,
sascha@1211:             XPathConstants.NODESET,
sascha@1211:             null);
sascha@1211: 
sascha@1211:         if (patternList == null) {
sascha@1211:             log.info("no line patterns found.");
sascha@1211:             return;
sascha@1211:         }
sascha@1211: 
sascha@1211:         for (int i = 0, N = patternList.getLength(); i < N; ++i) {
sascha@1211:             Element element = (Element)patternList.item(i);
sascha@1211:             Pair pair = buildPair(element);
sascha@1211:             if (pair != null) {
sascha@1211:                 descPatterns.add(pair);
sascha@1211:             }
sascha@1211:         }
sascha@1211:     }
sascha@1211: 
sascha@1211:     protected Pair buildPair(Element element) {
sascha@1211:         String pattern = element.getAttribute("pattern");
sascha@1211:         String type    = element.getAttribute("type");
sascha@1211: 
sascha@1211:         if (pattern.length() == 0) {
sascha@3662:             log.warn("ANNCLASS: pattern has no 'pattern' attribute.");
sascha@1211:             return null;
sascha@1211:         }
sascha@1211: 
sascha@1211:         if (type.length() == 0) {
sascha@3662:             log.warn("ANNCLASS: pattern has no 'type' attribute.");
sascha@1211:             return null;
sascha@1211:         }
sascha@1211: 
sascha@1211:         ImportAnnotationType annType = types.get(type);
sascha@1211: 
sascha@1211:         if (annType == null) {
sascha@3662:             log.warn("ANNCLASS: pattern has unknown type '" + type + "'");
sascha@1211:             return null;
sascha@1211:         }
sascha@1211: 
sascha@1211:         Pattern p;
sascha@1211: 
sascha@1211:         try {
sascha@1211:             p = Pattern.compile(pattern,
sascha@1211:                     Pattern.CASE_INSENSITIVE|Pattern.UNICODE_CASE);
sascha@1211:         }
sascha@1211:         catch (IllegalArgumentException iae) {
sascha@3662:             log.warn("ANNCLASS: pattern '" + pattern + "' is invalid.", iae);
sascha@1211:             return null;
sascha@1211:         }
sascha@1211: 
sascha@1211:         return new Pair(p, annType);
sascha@1211:     }
sascha@1211: 
sascha@1211:     public ImportAnnotationType getDefaultType() {
sascha@1211:         return defaultType;
sascha@1211:     }
sascha@1211: 
sascha@1211:     public ImportAnnotationType classifyFile(String filename) {
sascha@1211:         return classifyFile(filename, null);
sascha@1211:     }
sascha@1211: 
sascha@1211:     public ImportAnnotationType classifyFile(
sascha@1211:         String                filename,
sascha@1211:         ImportAnnotationType def
sascha@1211:     ) {
sascha@1211:         if (filename.toLowerCase().endsWith(".km")) {
sascha@1211:             filename = filename.substring(0, filename.length()-3);
sascha@1211:         }
sascha@1211: 
sascha@1211:         for (Pair pair: filePatterns) {
sascha@1211:             ImportAnnotationType annType = pair.match(filename);
sascha@1211:             if (annType != null) {
sascha@1211:                 return annType;
sascha@1211:             }
sascha@1211:         }
sascha@1211: 
sascha@1211:         return def;
sascha@1211:     }
sascha@1211: 
sascha@1211:     public ImportAnnotationType classifyDescription(String description) {
sascha@1211:         return classifyDescription(description, null);
sascha@1211:     }
sascha@1211: 
sascha@1211:     public ImportAnnotationType classifyDescription(
sascha@1211:         String                description,
sascha@1211:         ImportAnnotationType def
sascha@1211:     ) {
sascha@1211:         for (Pair pair: descPatterns) {
sascha@1211:             ImportAnnotationType annType = pair.match(description);
sascha@1211:             if (annType != null) {
sascha@1211:                 return annType;
sascha@1211:             }
sascha@1211:         }
sascha@1211: 
sascha@1211:         return def;
sascha@1211:     }
sascha@1211: }
sascha@1211: // vim:set ts=4 sw=4 si et sta sts=4 fenc=utf8 :