| 1 | |
package org.paneris.bibliomania; |
| 2 | |
|
| 3 | |
import java.util.Enumeration; |
| 4 | |
|
| 5 | |
import javax.servlet.http.HttpServletRequest; |
| 6 | |
|
| 7 | |
import org.melati.Melati; |
| 8 | |
import org.melati.servlet.Form; |
| 9 | |
import org.melati.poem.AccessToken; |
| 10 | |
import org.melati.poem.Persistent; |
| 11 | |
import org.melati.poem.PoemTask; |
| 12 | |
import org.melati.poem.PoemThread; |
| 13 | |
import org.melati.poem.UnexpectedExceptionPoemException; |
| 14 | |
import org.melati.template.ServletTemplateContext; |
| 15 | |
import org.paneris.bibliomania.fti.ContextSearchResults; |
| 16 | |
import org.paneris.bibliomania.fti.HitText; |
| 17 | |
import org.paneris.bibliomania.fti.IndexOther; |
| 18 | |
import org.paneris.bibliomania.fti.Library; |
| 19 | |
import org.paneris.bibliomania.fti.Score; |
| 20 | |
import org.paneris.bibliomania.fti.ScoredHitTextsEnumeration; |
| 21 | |
import org.paneris.bibliomania.fti.SearchResults; |
| 22 | |
import org.webmacro.WebMacroException; |
| 23 | |
import org.webmacro.servlet.HandlerException; |
| 24 | |
import org.webmacro.servlet.WebContext; |
| 25 | |
|
| 26 | |
import com.sleepycat.db.DatabaseException; |
| 27 | |
|
| 28 | 0 | public class Search extends BibliomaniaServlet { |
| 29 | |
|
| 30 | |
private static final long serialVersionUID = 1L; |
| 31 | |
|
| 32 | |
protected String bibliomaniaHandle(Melati melati, WebContext context) throws WebMacroException { |
| 33 | 0 | ServletTemplateContext tc = melati.getServletTemplateContext(); |
| 34 | 0 | String queryText = Form.getFieldNulled(tc, "queryText"); |
| 35 | 0 | String queryHeadwords = Form.getFieldNulled(tc, "queryHeadwords"); |
| 36 | |
|
| 37 | 0 | if (queryText == null && queryHeadwords == null) |
| 38 | 0 | return bibliomaniaTemplate("search/Search.wm"); |
| 39 | |
|
| 40 | 0 | context.put("queryText", queryText == null ? "" : queryText); |
| 41 | 0 | context.put("queryHeadwords", queryHeadwords == null ? "" : queryHeadwords); |
| 42 | |
|
| 43 | 0 | int start = 0; |
| 44 | 0 | String startString = context.getForm("start"); |
| 45 | 0 | if (startString != null) { |
| 46 | |
try { |
| 47 | 0 | start = Math.max(0, Integer.parseInt(startString)); |
| 48 | 0 | } catch (NumberFormatException e) { |
| 49 | 0 | throw new HandlerException("`start' param must be a number"); |
| 50 | 0 | } |
| 51 | |
} |
| 52 | |
|
| 53 | 0 | BibliomaniaDatabase bib = (BibliomaniaDatabase)PoemThread.database(); |
| 54 | |
|
| 55 | 0 | if (!bib.openBerkelyDBs()) { |
| 56 | 0 | context.put("message", "Full text search has been temporarily switched off, sorry."); |
| 57 | |
} else { |
| 58 | |
|
| 59 | 0 | boolean infoSearch = "true".equals(context.getForm("info")); |
| 60 | |
|
| 61 | 0 | IndexOther fti = infoSearch ? bib.infoFTI() : bib.fti(); |
| 62 | 0 | Library lib = infoSearch ? bib.infoLibrary() : bib.getChapterTable(); |
| 63 | 0 | context.put("info", infoSearch ? Boolean.TRUE : null); |
| 64 | |
|
| 65 | |
SearchResults raw; |
| 66 | |
try { |
| 67 | 0 | raw = fti.querySearchResults(queryText, queryHeadwords); |
| 68 | 0 | PoemThread.toTidy().add(raw); |
| 69 | 0 | } catch (DatabaseException e) { |
| 70 | 0 | throw new HandlerException( |
| 71 | |
"Problem with Full Text index:" + e.toString()); |
| 72 | 0 | } |
| 73 | |
|
| 74 | |
ContextSearchResults results; |
| 75 | 0 | results = new ContextSearchResults(lib, raw, fti); |
| 76 | 0 | PoemThread.toTidy().add(results); |
| 77 | |
|
| 78 | 0 | String hitsPerTextString = context.getForm("hitsPerText"); |
| 79 | |
int hitsPerText; |
| 80 | 0 | if (hitsPerTextString != null && !hitsPerTextString.trim().equals("")) |
| 81 | |
try { |
| 82 | 0 | hitsPerText = Math.max(0, Integer.parseInt(hitsPerTextString)); |
| 83 | 0 | if (hitsPerText == 0) |
| 84 | 0 | hitsPerText = |
| 85 | |
((BibliomaniaDatabase)melati.getDatabase()) |
| 86 | |
.getDefaultSearchHitsPerText(); |
| 87 | 0 | } catch (NumberFormatException e) { |
| 88 | 0 | throw new HandlerException("`hitsPerText' param must be a number"); |
| 89 | 0 | } else |
| 90 | 0 | hitsPerText = |
| 91 | |
((BibliomaniaDatabase)melati.getDatabase()) |
| 92 | |
.getDefaultSearchHitsPerText(); |
| 93 | |
|
| 94 | 0 | long startTextID = 0; |
| 95 | 0 | long limitTextID = Long.MAX_VALUE; |
| 96 | |
|
| 97 | 0 | boolean restrict = "true".equals(context.getForm("restrict")); |
| 98 | |
|
| 99 | 0 | if (restrict) { |
| 100 | 0 | Persistent object = melati.getObject(); |
| 101 | 0 | if (object != null && object instanceof Unit) { |
| 102 | 0 | startTextID = ((Unit)object).ftiTextID_start(); |
| 103 | 0 | limitTextID = ((Unit)object).ftiTextID_limit(); |
| 104 | |
} |
| 105 | |
} |
| 106 | 0 | context.put("restrict", restrict ? "true" : "false"); |
| 107 | |
|
| 108 | |
Score.Hit[] order; |
| 109 | |
try { |
| 110 | 0 | order = Score.scoredHits(raw, startTextID, limitTextID, 50 |
| 111 | |
); |
| 112 | 0 | } catch (DatabaseException e) { |
| 113 | 0 | throw new HandlerException( |
| 114 | |
"Problem with Full Text index:" + e.toString()); |
| 115 | 0 | } |
| 116 | |
|
| 117 | 0 | ScoredHitTextsEnumeration hits = |
| 118 | |
new ScoredHitTextsEnumeration( |
| 119 | |
results, |
| 120 | |
order, |
| 121 | |
start, |
| 122 | |
bib.getSearchHitsPerPage(), |
| 123 | |
hitsPerText); |
| 124 | 0 | context.put("hitTexts", hits); |
| 125 | 0 | context.put("count", order.length); |
| 126 | |
|
| 127 | 0 | context.put("hitsPerText", new Integer(hitsPerText)); |
| 128 | |
|
| 129 | 0 | context.put("start", new Integer(hits.getPageStart())); |
| 130 | 0 | context.put("end", new Integer(hits.getPageEnd())); |
| 131 | |
|
| 132 | 0 | HttpServletRequest req = context.getRequest(); |
| 133 | |
|
| 134 | 0 | Integer prevStart = hits.getPrevPageStart(); |
| 135 | 0 | if (prevStart != null) |
| 136 | 0 | context.put( |
| 137 | |
"prevPageURI", |
| 138 | |
req.getRequestURI() |
| 139 | |
+ "?" |
| 140 | |
+ Form.sameQueryWith( |
| 141 | |
req.getQueryString(), |
| 142 | |
"start", |
| 143 | |
prevStart.toString())); |
| 144 | |
|
| 145 | 0 | Integer nextStart = hits.getNextPageStart(); |
| 146 | 0 | if (nextStart != null) |
| 147 | 0 | context.put( |
| 148 | |
"nextPageURI", |
| 149 | |
req.getRequestURI() |
| 150 | |
+ "?" |
| 151 | |
+ Form.sameQueryWith( |
| 152 | |
req.getQueryString(), |
| 153 | |
"start", |
| 154 | |
nextStart.toString())); |
| 155 | |
} |
| 156 | 0 | return bibliomaniaTemplate("search/Results.wm"); |
| 157 | |
} |
| 158 | |
|
| 159 | |
public static void main(final String[] args) throws Exception { |
| 160 | 0 | IndexOther.debug = true; |
| 161 | |
|
| 162 | 0 | final BibliomaniaDatabase bib = new BibliomaniaDatabase(false); |
| 163 | 0 | bib.connect("bibliomania", |
| 164 | |
"org.melati.poem.dbms.Postgresql", |
| 165 | |
"jdbc:postgresql:bibliomania", |
| 166 | |
"postgres", |
| 167 | |
"*", |
| 168 | |
8); |
| 169 | |
|
| 170 | 0 | bib.inSession(AccessToken.root, new PoemTask() { |
| 171 | |
public void run() { |
| 172 | |
try { |
| 173 | 0 | SearchResults raw = bib.fti().querySearchResults(args[0]); |
| 174 | |
|
| 175 | 0 | int hitsPerText = 10; |
| 176 | |
|
| 177 | |
long startTextID; |
| 178 | 0 | if (args.length < 3) |
| 179 | 0 | startTextID = 0L; |
| 180 | 0 | else if (args[1].equals("tid")) |
| 181 | 0 | startTextID = Long.parseLong(args[2]); |
| 182 | |
else |
| 183 | 0 | startTextID = |
| 184 | |
((Unit)bib |
| 185 | |
.getTable(args[1]) |
| 186 | |
.getObject(Integer.parseInt(args[2]))) |
| 187 | |
.ftiTextID_start(); |
| 188 | |
|
| 189 | |
long limitTextID; |
| 190 | 0 | if (args.length < 5) |
| 191 | 0 | limitTextID = Long.MAX_VALUE; |
| 192 | 0 | else if (args[3].equals("tid")) |
| 193 | 0 | limitTextID = Long.parseLong(args[4]); |
| 194 | |
else |
| 195 | 0 | limitTextID = |
| 196 | |
((Unit)bib |
| 197 | |
.getTable(args[3]) |
| 198 | |
.getObject(Integer.parseInt(args[4]))) |
| 199 | |
.ftiTextID_start(); |
| 200 | |
|
| 201 | 0 | System.err.println("[ " + startTextID + ", " + limitTextID + ")"); |
| 202 | |
|
| 203 | 0 | Score.Hit[] order = |
| 204 | |
Score.scoredHits(raw, startTextID, limitTextID, 50); |
| 205 | |
|
| 206 | 0 | ContextSearchResults results = |
| 207 | |
new ContextSearchResults(bib.getChapterTable(), raw, bib.fti()); |
| 208 | |
|
| 209 | 0 | Enumeration hits = |
| 210 | |
new ScoredHitTextsEnumeration( |
| 211 | |
results, |
| 212 | |
order, |
| 213 | |
0, |
| 214 | |
bib.getSearchHitsPerPage(), |
| 215 | |
hitsPerText); |
| 216 | |
|
| 217 | 0 | while (hits.hasMoreElements()) |
| 218 | 0 | ((HitText)hits.nextElement()).dump(); |
| 219 | 0 | } catch (Exception e) { |
| 220 | 0 | throw new UnexpectedExceptionPoemException(e); |
| 221 | 0 | } |
| 222 | 0 | } |
| 223 | |
}); |
| 224 | 0 | } |
| 225 | |
} |