boxtec
/
yacy_search_server
mirror of https://github.com/yacy/yacy_search_server
4
1
Fork 0
Code Issues Projects Releases Wiki Activity

small change

Browse Source 
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@665 6c8d7289-2bf4-0310-a012-ef5d649a1542
pull/1/head
borg-0300 19 years ago
parent 578f36ae18
commit fbac053c03
1 changed files with 34 additions and 30 deletions
Show Stats Download Patch File Download Diff File

64
source/de/anomic/plasma/plasmaWordIndexEntry.java
Unescape View File

@ -51,6 +51,7 @@ import java.util.Properties;
import de.anomic.htmlFilter.htmlFilterContentScraper; import de.anomic.htmlFilter.htmlFilterContentScraper;
import de.anomic.server.serverCodings; import de.anomic.server.serverCodings;
import de.anomic.yacy.yacySeedDB; import de.anomic.yacy.yacySeedDB;
// import de.anomic.server.logging.serverLog;
public class plasmaWordIndexEntry { public class plasmaWordIndexEntry {
@ -105,43 +106,46 @@ public class plasmaWordIndexEntry {
// doctype calculation // doctype calculation
public static char docType(URL url) { public static char docType(URL url) {
String path = htmlFilterContentScraper.urlNormalform(url); String path = htmlFilterContentScraper.urlNormalform(url);
char doctype = DT_UNKNOWN; // serverLog.logFinest("PLASMA", "docType URL=" + path);
if (path.endsWith(".gif")) doctype = DT_IMAGE; char doctype = doctype = DT_UNKNOWN;
if (path.endsWith(".jpg")) doctype = DT_IMAGE; if (path.endsWith(".gif")) { doctype = DT_IMAGE; }
if (path.endsWith(".jpeg")) doctype = DT_IMAGE; else if (path.endsWith(".jpg")) { doctype = DT_IMAGE; }
if (path.endsWith(".png")) doctype = DT_IMAGE; else if (path.endsWith(".jpeg")) { doctype = DT_IMAGE; }
if (path.endsWith(".html")) doctype = DT_HTML; else if (path.endsWith(".png")) { doctype = DT_IMAGE; }
if (path.endsWith(".txt")) doctype = DT_TEXT; else if (path.endsWith(".html")) { doctype = DT_HTML; }
if (path.endsWith(".doc")) doctype = DT_DOC; else if (path.endsWith(".txt")) { doctype = DT_TEXT; }
if (path.endsWith(".rtf")) doctype = DT_DOC; else if (path.endsWith(".doc")) { doctype = DT_DOC; }
if (path.endsWith(".pdf")) doctype = DT_PDFPS; else if (path.endsWith(".rtf")) { doctype = DT_DOC; }
if (path.endsWith(".ps")) doctype = DT_PDFPS; else if (path.endsWith(".pdf")) { doctype = DT_PDFPS; }
if (path.endsWith(".avi")) doctype = DT_MOVIE; else if (path.endsWith(".ps")) { doctype = DT_PDFPS; }
if (path.endsWith(".mov")) doctype = DT_MOVIE; else if (path.endsWith(".avi")) { doctype = DT_MOVIE; }
if (path.endsWith(".qt")) doctype = DT_MOVIE; else if (path.endsWith(".mov")) { doctype = DT_MOVIE; }
if (path.endsWith(".mpg")) doctype = DT_MOVIE; else if (path.endsWith(".qt")) { doctype = DT_MOVIE; }
if (path.endsWith(".md5")) doctype = DT_SHARE; else if (path.endsWith(".mpg")) { doctype = DT_MOVIE; }
if (path.endsWith(".mpeg")) doctype = DT_MOVIE; else if (path.endsWith(".md5")) { doctype = DT_SHARE; }
if (path.endsWith(".asf")) doctype = DT_FLASH; else if (path.endsWith(".mpeg")) { doctype = DT_MOVIE; }
else if (path.endsWith(".asf")) { doctype = DT_FLASH; }
return doctype; return doctype;
} }
public static char docType(String mime) { public static char docType(String mime) {
// serverLog.logFinest("PLASMA", "docType mime=" + mime);
char doctype = DT_UNKNOWN; char doctype = DT_UNKNOWN;
if (mime == null) doctype = DT_UNKNOWN; if (mime == null) doctype = DT_UNKNOWN;
else if (mime.endsWith("/gif")) doctype = DT_IMAGE;
else if (mime.endsWith("/jpeg")) doctype = DT_IMAGE; else if (mime.endsWith("/jpeg")) doctype = DT_IMAGE;
else if (mime.endsWith("/html")) doctype = DT_HTML;
else if (mime.endsWith("/rtf")) doctype = DT_DOC; else if (mime.endsWith("/rtf")) doctype = DT_DOC;
else if (mime.endsWith("/msword")) doctype = DT_DOC;
else if (mime.endsWith("/mspowerpoint")) doctype = DT_DOC;
else if (mime.endsWith("/postscript")) doctype = DT_PDFPS;
else if (mime.endsWith("/pdf")) doctype = DT_PDFPS; else if (mime.endsWith("/pdf")) doctype = DT_PDFPS;
else if (mime.endsWith("/octet-stream")) doctype = DT_BINARY; else if (mime.endsWith("/octet-stream")) doctype = DT_BINARY;
else if (mime.endsWith("/x-shockwave-flash")) doctype = DT_FLASH; else if (mime.endsWith("/x-shockwave-flash")) doctype = DT_FLASH;
else if (mime.startsWith("audio/")) doctype = DT_AUDIO; else if (mime.endsWith("/msword")) doctype = DT_DOC;
else if (mime.startsWith("video/")) doctype = DT_MOVIE; else if (mime.endsWith("/mspowerpoint")) doctype = DT_DOC;
else if (mime.endsWith("/postscript")) doctype = DT_PDFPS;
else if (mime.startsWith("text/")) doctype = DT_TEXT; else if (mime.startsWith("text/")) doctype = DT_TEXT;
else if (mime.startsWith("image/")) doctype = DT_IMAGE; else if (mime.startsWith("image/")) doctype = DT_IMAGE;
else if (mime.endsWith("/html")) doctype = DT_HTML; else if (mime.startsWith("audio/")) doctype = DT_AUDIO;
else if (mime.startsWith("video/")) doctype = DT_MOVIE;
//bz2 = application/x-bzip2 //bz2 = application/x-bzip2
//dvi = application/x-dvi //dvi = application/x-dvi
//gz = application/gzip //gz = application/gzip
@ -165,14 +169,14 @@ public class plasmaWordIndexEntry {
//zip = application/zip //zip = application/zip
return doctype; return doctype;
} }
// language calculation // language calculation
public static String language(URL url) { public static String language(URL url) {
String host = url.getHost(); String language = "uk";
int pos = host.lastIndexOf("."); String host = url.getHost();
String language = "uk"; int pos = host.lastIndexOf(".");
if ((pos > 0) && (host.length() - pos == 3)) language = host.substring(pos + 1).toLowerCase(); if ((pos > 0) && (host.length() - pos == 3)) language = host.substring(pos + 1).toLowerCase();
return language; return language;
} }
// the class instantiation can only be done by a plasmaStore method // the class instantiation can only be done by a plasmaStore method

Write Preview
Loading…
Cancel
Save