the location search shows now not re-evaluated locations but only such locations that are attached as metadata to web pages

- added parser for in-text appearing geo-locations
- added geo-locations to rss search result
- added evaluation of metadata-attached geo-locations in yacysearch_location to show search results within a map


git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7631 6c8d7289-2bf4-0310-a012-ef5d649a1542

htroot/yacysearch.rss

@@ -6,6 +6,7 @@
     xmlns:media="http://search.yahoo.com/mrss/"
     xmlns:atom="http://www.w3.org/2005/Atom"
     xmlns:dc="http://purl.org/dc/elements/1.1/"
+    xmlns:geo="http://www.w3.org/2003/01/geo/wgs84_pos#"
 >
 	<!-- YaCy Search Engine; http://yacy.net -->
 	<channel>

htroot/yacysearch_location.html

@@ -9,11 +9,11 @@
     var map;
     var searchLayer_md = null;
     var searchLayer_co = null;
-    var path_mdsearch = 'yacysearch_location.rss?dom=title,publisher,creator,subject&query=';
+    var path_mdsearch = 'yacysearch_location.rss?dom=metatag&query=';
     var path_cosearch = 'yacysearch_location.rss?dom=query&query=';
     var marker_md = new OpenLayers.Icon("/env/grafics/marker_red.png", new OpenLayers.Size(11,16));
     var marker_co = new OpenLayers.Icon("/env/grafics/star_yellow.png", new OpenLayers.Size(25,25));
-    // possible values for dom: query,mdall,title,publisher,creator,subject
+    // possible values for dom: query,metatag,alltext,title,publisher,creator,subject
     
     function init() {
         map = new OpenLayers.Map('map', {

htroot/yacysearch_location.java

@@ -55,13 +55,13 @@ public class yacysearch_location {
             prop.put("kml", 1);
             if (post == null) return prop;
             String query = post.get("query", "");
-            boolean search_all = !post.containsKey("dom") || post.get("dom", "").equals("all");
-            boolean search_query = search_all || post.get("dom", "").indexOf("query") >= 0;
-            boolean search_mdall = search_all || post.get("dom", "").indexOf("mdall") >= 0;
-            boolean search_title = search_mdall || post.get("dom", "").indexOf("title") >= 0;
-            boolean search_publisher = search_mdall || post.get("dom", "").indexOf("publisher") >= 0;
-            boolean search_creator = search_mdall || post.get("dom", "").indexOf("creator") >= 0;
-            boolean search_subject = search_mdall || post.get("dom", "").indexOf("subject") >= 0;
+            boolean search_query = post.get("dom", "").indexOf("query") >= 0;
+            boolean metatag = post.get("dom", "").indexOf("metatag") >= 0;
+            boolean alltext = post.get("dom", "").indexOf("alltext") >= 0;
+            boolean search_title = alltext || post.get("dom", "").indexOf("title") >= 0;
+            boolean search_publisher = alltext || post.get("dom", "").indexOf("publisher") >= 0;
+            boolean search_creator = alltext || post.get("dom", "").indexOf("creator") >= 0;
+            boolean search_subject = alltext || post.get("dom", "").indexOf("subject") >= 0;
             long maximumTime = post.getLong("maximumTime", 3000);
             int maximumRecords = post.getInt("maximumRecords", 200);
             //i.e. http://localhost:8090/yacysearch_location.kml?query=berlin&maximumTime=2000&maximumRecords=100
@@ -89,7 +89,7 @@ public class yacysearch_location {
                 }
             }
             
-            if (search_title || search_publisher || search_creator || search_subject) try {
+            if (metatag || search_title || search_publisher || search_creator || search_subject) try {
                 // get a queue of search results
                 String rssSearchServiceURL = "http://127.0.0.1:" + sb.getConfig("port", "8090") + "/yacysearch.rss";
                 BlockingQueue<RSSMessage> results = new LinkedBlockingQueue<RSSMessage>();
@@ -98,6 +98,7 @@ public class yacysearch_location {
                 // take the results and compute some locations
                 RSSMessage message;
                 loop: while ((message = results.poll(maximumTime, TimeUnit.MILLISECONDS)) != RSSMessage.POISON) {
+
                     // find all associated locations
                     Set<Location> locations = new HashSet<Location>();
                     StringBuilder words = new StringBuilder(120);
@@ -112,6 +113,13 @@ public class yacysearch_location {
                     for (int i = 0; i < wordlist.length - 1; i++) locations.addAll(LibraryProvider.geoLoc.find(wordlist[i] + space + wordlist[i + 1], true));
                     for (int i = 0; i < wordlist.length - 2; i++) locations.addAll(LibraryProvider.geoLoc.find(wordlist[i] + space + wordlist[i + 1] + space + wordlist[i + 2], true));
                     
+                    // add locations from metatag
+                    if (metatag) {
+                        if (message.getLat() != 0.0f && message.getLon() != 0.0f) {
+                            locations.add(new Location(message.getLon(), message.getLat(), message.getTitle().trim()));
+                        }
+                    }
+                    
                     for (Location location: locations) {
                         // write for all locations a point to this message
                         prop.put("kml_placemark_" + placemarkCounter + "_location", location.getName());

htroot/yacysearchitem.java

@@ -193,6 +193,13 @@ public class yacysearchitem {
             } else {
                 prop.put("content_code", "");
             }
+            if (result.lat() == 0.0f || result.lon() == 0.0f) {
+                prop.put("content_loc", 0);
+            } else {
+                prop.put("content_loc", 1);
+                prop.put("content_loc_lat", result.lat());
+                prop.put("content_loc_lon", result.lon());
+            }
             theQuery.transmitcount = item + 1;
             return prop;
         }

htroot/yacysearchitem.xml

@@ -12,6 +12,7 @@
 <yacy:path>#[path]#</yacy:path>
 <yacy:file>#[file]#</yacy:file>
 <guid isPermaLink="false">#[urlhash]#</guid>
+#(loc)#::<geo:lat>#[lat]#</geo:lat><geo:long>#[lon]#</geo:long>#(/loc)#
 </item>::
 #(item)#::<item>
 <title>#[name]#</title>

source/de/anomic/search/ResultEntry.java

@@ -168,6 +168,12 @@ public class ResultEntry implements Comparable<ResultEntry>, Comparator<ResultEn
     public int lapp() {
         return urlentry.lapp();
     }
+    public float lat() {
+        return urlentry.metadata().lat();
+    }
+    public float lon() {
+        return urlentry.metadata().lon();
+    }
     public WordReferenceVars word() {
         final Reference word = urlentry.word();
         assert word instanceof WordReferenceVars;

source/net/yacy/document/parser/html/ContentScraper.java

@@ -54,6 +54,9 @@ import net.yacy.kelondro.util.ISO639;
 
 public class ContentScraper extends AbstractScraper implements Scraper {
 
+    private final char degree = '\u00B0';
+    private final char[] minuteCharsHTML = "'".toCharArray();
+    
     // statics: for initialization of the HTMLFilterAbstractScraper
     private static final Set<String> linkTags0 = new HashSet<String>(9,0.99f);
     private static final Set<String> linkTags1 = new HashSet<String>(7,0.99f);
@@ -121,6 +124,55 @@ public class ContentScraper extends AbstractScraper implements Scraper {
     
     public void scrapeText(final char[] newtext, final String insideTag) {
         // System.out.println("SCRAPE: " + UTF8.String(newtext));
+        int p, q, s = 0;
+
+        // try to find location information in text
+        location: while (s < newtext.length) {
+            p = CharBuffer.indexOf(newtext, s, degree);
+            if (p < 0) break location;
+            // try to find a coordinate
+            // <nobr>N 50o 05.453&#039;</nobr><nobr>E 008o 30.191&#039;</nobr>
+            // N 52o 28.025 E 013o 20.299
+            q = CharBuffer.indexOf(newtext, p, minuteCharsHTML);
+            if (q < 0) q = CharBuffer.indexOf(newtext, p, " E".toCharArray());
+            if (q < 0) q = CharBuffer.indexOf(newtext, p, " W".toCharArray());
+            if (q < 0 && newtext.length - p == 8) q = newtext.length; 
+            if (q < 0) break location;
+            int r = p;
+            while (r-- > 1) {
+                if (newtext[r] == ' ') {
+                    r--;
+                    if (newtext[r] == 'N') {
+                        this.lat =  Float.parseFloat(new String(newtext, r + 2, p - r - 2)) +
+                                    Float.parseFloat(new String(newtext, p + 2, q - p - 2)) / 60.0f;
+                        s = q + 6;
+                        continue location;
+                    }
+                    if (newtext[r] == 'S') {
+                        this.lat = -Float.parseFloat(new String(newtext, r + 2, p - r - 2)) -
+                                    Float.parseFloat(new String(newtext, p + 2, q - p - 2)) / 60.0f;
+                        s = q + 6;
+                        continue location;
+                    }
+                    if (newtext[r] == 'E') {
+                        this.lon =  Float.parseFloat(new String(newtext, r + 2, p - r - 2)) +
+                                    Float.parseFloat(new String(newtext, p + 2, q - p - 2)) / 60.0f;
+                        s = q + 6;
+                        continue location;
+                    }
+                    if (newtext[r] == 'W') {
+                        this.lon = -Float.parseFloat(new String(newtext, r + 2, p - r - 2)) -
+                                    Float.parseFloat(new String(newtext, p + 2, q - p - 2)) / 60.0f;
+                        s = q + 6;
+                        continue location;
+                    }
+                    break location;
+                }
+            }
+            break location;
+        }
+        
+        // find tags inside text
         String b = cleanLine(super.stripAll(newtext));
         if ((insideTag != null) && (!(insideTag.equals("a")))) {
             // texts inside tags sometimes have no punctuation at the line end
@@ -132,7 +184,7 @@ public class ContentScraper extends AbstractScraper implements Scraper {
             //System.out.println("*** Appended dot: " + b.toString());
         }
         // find http links inside text
-        int p, q, s = 0;
+        s = 0;
         String u;
         MultiProtocolURI url;
         while (s < b.length()) {

source/net/yacy/kelondro/data/meta/URIMetadataRow.java

@@ -191,8 +191,8 @@ public class URIMetadataRow implements URIMetadata {
         final CharBuffer s = new CharBuffer(360);
         s.append(url.toNormalform(false, true)).append(10);
         s.append(dc_title).append(10);
-        s.append(dc_creator).append(10);
-        s.append(dc_subject).append(10);
+        s.append(dc_creator.length() > 80 ? dc_creator.substring(0, 80) : dc_creator).append(10);
+        s.append(dc_subject.length() > 120 ? dc_subject.substring(0, 120) : dc_subject).append(10);
         s.append(dc_publisher).append(10);
         if (lon == 0.0f && lat == 0.0f) s.append(10); else s.append(Float.toString(lat)).append(',').append(Float.toString(lon)).append(10);
 		return UTF8.getBytes(s.toString());

source/net/yacy/kelondro/io/CharBuffer.java

@@ -247,6 +247,36 @@ public final class CharBuffer extends Writer {
         return -1;
     }
     
+    public static int indexOf(final char[] b, final char c) {
+        return indexOf(b, 0, c);
+    }
+
+    public static int indexOf(final char[] b, final int offset, final char c) {
+        for (int i = offset; i < b.length; i++) if (b[i] == c) return i;
+        return -1;
+    }
+    
+    public static int indexOf(final char[] b, final char[] s) {
+        return indexOf(b, 0, s);
+    }
+
+    public static int indexOf(final char[] b, final int start, final char[] bs) {
+        if (start + bs.length > b.length) return -1;
+        loop: for (int i = start; i <= b.length - bs.length; i++) {
+            // first test only first char
+            if (b[i] != bs[0]) continue loop;
+            
+            // then test all remaining char
+            for (int j = 1; j < bs.length; j++) {
+                if (b[i + j] != bs[j]) continue loop;
+            }
+            
+            // found hit
+            return i;
+        }
+        return -1;
+    }
+
     public int lastIndexOf(final char b) {
         for (int i = length - 1; i >= 0; i--) if (buffer[offset + i] == b) return i;
         return -1;