Attachment 121758 Details for Bug 164813 – jericho-html-3.2.patch

[patch] jericho-html-3.2.patch

jericho-html-3.2.patch (text/plain), 4.82 KB, created by Greg Lewis on 2012-02-06 04:30:12 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Greg Lewis

Created: 2012-02-06 04:30:12 UTC

Size: 4.82 KB

patch

obsolete

>Index: Makefile
>===================================================================
>RCS file: /var/fcvs/ports/www/jericho-html/Makefile,v
>retrieving revision 1.5
>diff -u -r1.5 Makefile
>--- Makefile	31 Dec 2009 10:03:33 -0000	1.5
>+++ Makefile	5 Feb 2012 21:23:37 -0000
>@@ -6,7 +6,7 @@
> #
> 
> PORTNAME=	jericho-html
>-PORTVERSION=	2.6
>+PORTVERSION=	3.2
> CATEGORIES=	www java
> MASTER_SITES=	SF/jerichohtml/${PORTNAME}/${PORTVERSION}
> 
>@@ -14,38 +14,29 @@
> COMMENT=	A java library to analyse and manipulate HTML
> 
> USE_ZIP=	yes
>-USE_JAVA=	1.3+
>+USE_JAVA=	1.5+
> 
>-INTERFACES:=	"compile-time-dependencies/slf4j-api-1.5.2.jar:\
>+INTERFACES:=	"compile-time-dependencies/slf4j-api-1.6.1.jar:\
> 		compile-time-dependencies/commons-logging-api-1.1.1.jar:\
>-		compile-time-dependencies/log4j-api-1.2.15.jar"
>+		compile-time-dependencies/log4j-1.2.16.jar"
> 
>-PORTDOCS=	api
>+PORTDOCS=	*
> PLIST_FILES+=	%%JAVAJARDIR%%/${PORTNAME}.jar
> 
> do-build:
> 	(cd ${WRKSRC} &&  ${RM} -rf classes/* && ${JAVAC}         \
> 		-classpath ${INTERFACES:S, ,,g}                   \
>-		-d classes src/java/au/id/jericho/lib/html/*.java \
>-		src/java/au/id/jericho/lib/html/nodoc/*.java)
>+		-d classes src/java/net/htmlparser/jericho/*.java \
>+		src/java/net/htmlparser/jericho/nodoc/*.java)
>+	${MKDIR} ${WRKSRC}/lib
> 	${JAR} -cf ${WRKSRC}/lib/${PORTNAME}.jar                  \
> 		 -C ${WRKSRC}/classes .
>-.if !defined(NOPORTDOCS)
>-	(cd ${WRKSRC} && ${RM} -rf doc/* && ${JAVADOC} -quiet     \
>-		-windowtitle "Jericho HTML Parser ${PORTVERSION}" \
>-		-classpath ${INTERFACES:S, ,,g}:src/java:classes  \
>-		-use -d ${WRKSRC}/doc/api                         \
>-		-subpackages au.id.jericho.lib.html               \
>-		-exclude au.id.jericho.lib.html.nodoc             \
>-		-noqualifier au.id.jericho.lib.html               \
>-		-group "Core package" au.id.jericho.lib.html)
>-.endif
> 
> do-install:
> 	${INSTALL_DATA} ${WRKSRC}/lib/${PORTNAME}.jar ${JAVAJARDIR}
> .if !defined(NOPORTDOCS)
> 	${MKDIR} ${DOCSDIR}
>-	(cd ${WRKSRC}/doc && ${FIND} api | ${CPIO} -pdmu ${DOCSDIR})
>+	(cd ${WRKSRC}/docs && ${COPYTREE_SHARE} . ${DOCSDIR})
> .endif
> 
> .include <bsd.port.mk>
>Index: distinfo
>===================================================================
>RCS file: /var/fcvs/ports/www/jericho-html/distinfo,v
>retrieving revision 1.3
>diff -u -r1.3 distinfo
>--- distinfo	19 Mar 2011 12:34:23 -0000	1.3
>+++ distinfo	5 Feb 2012 20:11:23 -0000
>@@ -1,2 +1,2 @@
>-SHA256 (jericho-html-2.6.zip) = 6968459488579d17c88d2cbd7f46d6e07ea1bd086caac1015a1845975a550f68
>-SIZE (jericho-html-2.6.zip) = 1522427
>+SHA256 (jericho-html-3.2.zip) = 35787b825bd2fbf78d7c521e27a1ca164caaa01426ac38ae8f3c9697efc0dc13
>+SIZE (jericho-html-3.2.zip) = 2396280
>Index: files/patch-encoding
>===================================================================
>RCS file: files/patch-encoding
>diff -N files/patch-encoding
>--- files/patch-encoding	19 Jul 2008 17:02:19 -0000	1.1
>+++ /dev/null	1 Jan 1970 00:00:00 -0000
>@@ -1,15 +0,0 @@
>---- src/java/au/id/jericho/lib/html/StreamEncodingDetector.java.orig	2008-06-17 21:01:53.890292905 +0200
>-+++ src/java/au/id/jericho/lib/html/StreamEncodingDetector.java	2008-06-17 21:02:43.940300330 +0200
>-@@ -203,9 +203,9 @@
>- 		// Assume the more likely case of four 8-bit characters <= U+00FF.
>- 		// Check whether it fits some common EBCDIC strings that might be found at the start of a document:
>- 		if (b1==0x4C) { // first character is EBCDIC '<' (ASCII 'L'), check a couple more characters before assuming EBCDIC encoding:
>--			if (b2==0x6F && b3==0xA7 && b4==0x94) return setEncoding(EBCDIC,"default EBCDIC encoding (<?xml...> detected)"); // first four bytes are "<?xm" in EBCDIC ("Lo§" in Windows-1252)
>--			if (b2==0x5A && b3==0xC4 && b4==0xD6) return setEncoding(EBCDIC,"default EBCDIC encoding (<!DOCTYPE...> detected)"); // first four bytes are "<!DO" in EBCDIC ("LZÄÖ" in Windows-1252)
>--			if ((b2&b3&b4&0x80)!=0) return setEncoding(EBCDIC,"default EBCDIC-compatible encoding (HTML element detected)"); // all of the 3 bytes after the '<' have the high-order bit set, indicating EBCDIC letters such as "<HTM" ("LÈãÔ" in Windows-1252), or "<htm" ("L£" in Windows-1252)
>-+			if (b2==0x6F && b3==0xA7 && b4==0x94) return setEncoding(EBCDIC,"default EBCDIC encoding (<?xml...> detected)"); // first four bytes are "<?xm" in EBCDIC 
>-+			if (b2==0x5A && b3==0xC4 && b4==0xD6) return setEncoding(EBCDIC,"default EBCDIC encoding (<!DOCTYPE...> detected)"); // first four bytes are "<!DO" in EBCDIC 
>-+			if ((b2&b3&b4&0x80)!=0) return setEncoding(EBCDIC,"default EBCDIC-compatible encoding (HTML element detected)"); // all of the 3 bytes after the '<' have the high-order bit set, indicating EBCDIC letters such as "<HTM" or "<htm" 
>- 			// although this is not an exhaustive check for EBCDIC, it is safer to assume a more common preliminary encoding if none of these conditions are met.
>- 		}
>- 		// Now confident that it is not EBCDIC, but some other 8-bit encoding.

Actions: View | Diff

Attachments on bug 164813: 121758