Xomw: Add basic namespace parsing to title

gplx.xowa.mediawiki/src/gplx/xowa/mediawiki/includes/XomwNamespace.java

@@ -232,19 +232,21 @@ public class XomwNamespace {
 	public static int getCanonicalIndex(byte[] name) {
 		if (xNamespaces == null) {
 			xNamespaces = Hash_adp_bry.cs();
-			int len = namespaces.Len();
+			XomwNamespacesById namespacesHash = getCanonicalNamespaces();
+			int len = namespacesHash.Len();
 			for (int i = 0; i < len; i++) {
-				XomwNamespaceItem item = (XomwNamespaceItem)namespaces.GetAtOrNull(i);
+				XomwNamespaceItem item = (XomwNamespaceItem)namespacesHash.GetAtOrNull(i);
 				xNamespaces.Add(Bry_.Lcase__all(item.name), item);	// NOTE: MW does "strtolower($text)"; canonical namespaces are always ascii
 			}
 		}
-//			if (array_key_exists($name, $xNamespaces)) {
+		XomwNamespaceItem xNs = (XomwNamespaceItem)xNamespaces.Get_by(name);
-//				return $xNamespaces[$name];
+		if (xNs != null) {
-//			} else {
+			return xNs.id;
-//				return null;
+		}
-//			}
+		else {
 			return XomwNamespace.NULL_NS_ID;
 		}
+	}
 
 //		/**
 //		* Returns an array of the namespaces (by integer id) that exist on the

gplx.xowa.mediawiki/src/gplx/xowa/mediawiki/includes/title/XomwMediaWikiTitleCodec.java

@@ -233,7 +233,7 @@ public class XomwMediaWikiTitleCodec implements XomwTitleFormatter {
 	* @return array A map with the fields 'interwiki', 'fragment', 'namespace',
 	*         'user_case_dbkey', and 'dbkey'.
 	*/
-	// private final    byte[][] tmpPrefixRegex = new byte[2][];
+	private final    byte[][] tmpPrefixRegex = new byte[2][];
 	public XomwMediaWikiTitleCodecParts splitTitleString(byte[] text, int defaultNamespace) {
 		byte[] dbkey = XophpString.str_replace(Byte_ascii.Space, Byte_ascii.Underline, text);
 
@@ -276,74 +276,70 @@ public class XomwMediaWikiTitleCodec implements XomwTitleFormatter {
 		}
 
 		// Namespace or interwiki prefix
-		if (Bry_.Has_at_bgn(dbkey, Bry_.new_a7("File:"))) {
+		do {
-			parts.ns = XomwDefines.NS_FILE;
+			byte[][] m = tmpPrefixRegex;
-			dbkey = Bry_.Mid(dbkey, 5);
+			if (XomwRegexTitlePrefix.preg_match(m, dbkey)) {
+				byte[] p = m[0];
+				int ns = this.language.getNsIndex(p);
+				if (ns != XophpUtility.NULL_INT) {
+					// Ordinary namespace
+					dbkey = m[1];
+					parts.ns = ns;
+					// For Talk:X pages, check if X has a "namespace" prefix
+					if (ns == XomwDefines.NS_TALK && XomwRegexTitlePrefix.preg_match(m, dbkey)) {
+						if (this.language.getNsIndex(m[0]) != XophpUtility.NULL_INT) {
+							// Disallow Talk:File:x type titles...
+							throw new XomwMalformedTitleException("title-invalid-talk-namespace", text);
 						}
-//			do {
+//							else if (Interwiki::isValidInterwiki($x[1])) {
-//				byte[][] m = tmpPrefixRegex;
+//								// TODO: get rid of global state!
-//				if (XomwRegexTitlePrefix.preg_match(m, dbkey)) {
+//								// Disallow Talk:Interwiki:x type titles...
-//					byte[] p = m[0];
+//								throw new XomwMalformedTitleException('title-invalid-talk-namespace', text);
-//					int ns = this.language.getNsIndex(p);
-//					if (ns != XophpUtility.NULL_INT) {
-//						// Ordinary namespace
-//						dbkey = m[1];
-//						parts.ns = ns;
-//						// For Talk:X pages, check if X has a "namespace" prefix
-//						if (ns == XomwDefines.NS_TALK && XomwRegexTitlePrefix.preg_match(m, dbkey)) {
-//							if (this.language.getNsIndex(m[0]) != XophpUtility.NULL_INT) {
-//								// Disallow Talk:File:x type titles...
-//								throw new XomwMalformedTitleException("title-invalid-talk-namespace", text);
 //							}
-////							else if (Interwiki::isValidInterwiki($x[1])) {
+					}
-////								// TODO: get rid of global state!
+				}
-////								// Disallow Talk:Interwiki:x type titles...
+//					else if (Interwiki::isValidInterwiki($p)) {
-////								throw new XomwMalformedTitleException('title-invalid-talk-namespace', text);
+//						// Interwiki link
-////							}
+//						dbkey = $m[2];
+//						parts['interwiki'] = this.language->lc($p);
+//
+//						// Redundant interwiki prefix to the local wiki
+//						foreach (this.localInterwikis as $localIW) {
+//							if (0 == strcasecmp(parts['interwiki'], $localIW)) {
+//								if (dbkey == '') {
+//									// Empty self-links should point to the Main Page, to ensure
+//									// compatibility with cross-wiki transclusions and the like.
+//									$mainPage = Title::newMainPage();
+//									return [
+//										'interwiki' => $mainPage->getInterwiki(),
+//										'local_interwiki' => true,
+//										'fragment' => $mainPage->getFragment(),
+//										'namespace' => $mainPage->getNamespace(),
+//										'dbkey' => $mainPage->getDBkey(),
+//										'user_case_dbkey' => $mainPage->getUserCaseDBKey()
+//									];
+//								}
+//								parts['interwiki'] = '';
+//								// local interwikis should behave like initial-colon links
+//								parts['local_interwiki'] = true;
+//
+//								// Do another namespace split...
+//								continue 2;
 //							}
 //						}
-////					else if (Interwiki::isValidInterwiki($p)) {
+//
-////						// Interwiki link
+//						// If there's an initial colon after the interwiki, that also
-////						dbkey = $m[2];
+//						// resets the default namespace
-////						parts['interwiki'] = this.language->lc($p);
+//						if (dbkey !== '' && dbkey[0] == ':') {
-////
+//							parts['namespace'] = NS_MAIN;
-////						// Redundant interwiki prefix to the local wiki
+//							dbkey = substr(dbkey, 1);
-////						foreach (this.localInterwikis as $localIW) {
-////							if (0 == strcasecmp(parts['interwiki'], $localIW)) {
-////								if (dbkey == '') {
-////									// Empty self-links should point to the Main Page, to ensure
-////									// compatibility with cross-wiki transclusions and the like.
-////									$mainPage = Title::newMainPage();
-////									return [
-////										'interwiki' => $mainPage->getInterwiki(),
-////										'local_interwiki' => true,
-////										'fragment' => $mainPage->getFragment(),
-////										'namespace' => $mainPage->getNamespace(),
-////										'dbkey' => $mainPage->getDBkey(),
-////										'user_case_dbkey' => $mainPage->getUserCaseDBKey()
-////									];
-////								}
-////								parts['interwiki'] = '';
-////								// local interwikis should behave like initial-colon links
-////								parts['local_interwiki'] = true;
-////
-////								// Do another namespace split...
-////								continue 2;
-////							}
-////						}
-////
-////						// If there's an initial colon after the interwiki, that also
-////						// resets the default namespace
-////						if (dbkey !== '' && dbkey[0] == ':') {
-////							parts['namespace'] = NS_MAIN;
-////							dbkey = substr(dbkey, 1);
-////						}
-////					}
-//					// If there's no recognized interwiki or namespace,
-//					// then let the colon expression be part of the title.
 //						}
-//				break;
+//					}
-//			} while (true);
+				// If there's no recognized interwiki or namespace,
+				// then let the colon expression be part of the title.
+			}
+			break;
+		} while (true);
 
 //			$fragment = strstr(dbkey, '#');
 //			if (false !== $fragment) {

gplx.xowa.mediawiki/src/gplx/xowa/mediawiki/includes/title/XomwMediaWikiTitleCodecParts.java

@@ -29,4 +29,10 @@ public class XomwMediaWikiTitleCodecParts {
 		this.dbkey = dbkey;
 		this.user_case_dbkey = dbkey;
 	}
+	public String ToStr() {
+		Bry_bfr bfr = Bry_bfr_.New();
+		bfr.Add_int_variable(ns).Add_byte_pipe();
+		bfr.Add(dbkey).Add_byte_pipe();
+		return bfr.To_str_and_clear();
+	}
 }

gplx.xowa.mediawiki/src/gplx/xowa/mediawiki/includes/title/XomwMediaWikiTitleCodecTest.java

@@ -14,7 +14,10 @@ GPLv3 License: https://github.com/gnosygnu/xowa/blob/master/LICENSE-GPLv3.txt
 Apache License: https://github.com/gnosygnu/xowa/blob/master/LICENSE-APACHE2.txt
 */
 package gplx.xowa.mediawiki.includes.title; import gplx.*; import gplx.xowa.*; import gplx.xowa.mediawiki.*; import gplx.xowa.mediawiki.includes.*;
-import org.junit.*; import gplx.core.tests.*; import gplx.core.btries.*; import gplx.xowa.mediawiki.includes.parsers.*;
+import org.junit.*; import gplx.core.tests.*;
+import gplx.core.btries.*;
+import gplx.xowa.mediawiki.includes.parsers.*;
+import gplx.xowa.mediawiki.languages.*; import gplx.xowa.langs.*;
 public class XomwMediaWikiTitleCodecTest {
 	private final    XomwMediaWikiTitleCodecFxt fxt = new XomwMediaWikiTitleCodecFxt();
 	@Test  public void regexTitlePrefix() {
@@ -25,6 +28,11 @@ public class XomwMediaWikiTitleCodecTest {
 		// underscores
 		fxt.Test_regexTitlePrefix("a__:___b"  , "a", "b");
 	}
+	@Test  public void splitTitleString() {
+		XomwMediaWikiTitleCodec codec = fxt.Make_codec(fxt.Make_lang());
+		// ns
+		fxt.Test_splitTitleString(codec, "File:A"    , fxt.Make_parts(XomwDefines.NS_FILE, "A"));
+	}
 }
 class XomwMediaWikiTitleCodecFxt {
 	private byte[][] regexTitlePrefixResult = new byte[2][];
@@ -33,4 +41,19 @@ class XomwMediaWikiTitleCodecFxt {
 		Gftest.Eq__str(expd_ns,  String_.new_u8(regexTitlePrefixResult[0]));
 		Gftest.Eq__str(expd_ttl, String_.new_u8(regexTitlePrefixResult[1]));
 	}
+	public XomwMediaWikiTitleCodecParts Make_parts(int ns, String dbkey) {
+		return new XomwMediaWikiTitleCodecParts(Bry_.new_u8(dbkey), ns);
+	}
+	public XomwMediaWikiTitleCodec Make_codec(XomwLanguage lang) {
+		return new XomwMediaWikiTitleCodec(lang);
+	}
+	public XomwLanguage Make_lang() {
+		Xoae_app app = Xoa_app_fxt.Make__app__edit();
+		Xol_lang_itm lang = new Xol_lang_itm(app.Lang_mgr(), Xol_lang_itm_.Key_en);
+		return new XomwLanguage(lang);
+	}
+	public void Test_splitTitleString(XomwMediaWikiTitleCodec codec, String src, XomwMediaWikiTitleCodecParts expd) {
+		XomwMediaWikiTitleCodecParts actl = codec.splitTitleString(Bry_.new_u8(src), XomwDefines.NS_MAIN);
+		Gftest.Eq__str(expd.ToStr(), actl.ToStr());
+	}
 }