Update to 3.02. From the changelog:

3.02:
* Use optparse for parsing and checking arguments
* Encode all output as UTF-8
* Accept optional encoding for local file or URL
* Use chardet for guessing local file character sets
* Fix double-newlines inside code blocks

3.01:
* Fix bug with unknown entities

3.0:
* Add forward compatibility with Python3

2.40:
* Update from sgmllib to HTMLParser so that valid XHTML doesn't fail

3 files changed, 20 insertions, 22 deletions

diff --git a/textproc/py-html2text/Makefile b/textproc/py-html2text/Makefile
index 0e20daabf44..62eb2ab962c 100644
--- a/textproc/py-html2text/Makefile
+++ b/textproc/py-html2text/Makefile
@@ -1,11 +1,13 @@
-# $NetBSD: Makefile,v 1.19 2011/05/10 16:45:35 schnoebe Exp $
+# $NetBSD: Makefile,v 1.20 2011/06/09 06:21:38 schmonz Exp $
 #
 
-DISTNAME=		html2text-2.39
-PKGNAME=		${PYPKGPREFIX}-${DISTNAME}
+DISTNAME=		aaronsw-${GHNAME}-${GHVERSION}-0-g${GHCOMMIT}
+GHNAME=			html2text
+GHVERSION=		3.02
+GHCOMMIT=		d9bf7d6
+PKGNAME=		${PYPKGPREFIX}-${GHNAME}-${GHVERSION}
 CATEGORIES=		textproc python
-MASTER_SITES=		${HOMEPAGE}
-EXTRACT_SUFX=		.py
+MASTER_SITES=		http://download.github.com/
 
 MAINTAINER=		schmonz@NetBSD.org
 HOMEPAGE=		http://www.aaronsw.com/2002/html2text/
@@ -14,20 +16,14 @@ LICENSE=		gnu-gpl-v2
 
 PKG_DESTDIR_SUPPORT=	user-destdir
 
-WRKSRC=			${WRKDIR}
+WRKSRC=			${WRKDIR}/aaronsw-${GHNAME}-${GHCOMMIT}
 
-USE_TOOLS+=		tr
 NO_BUILD=		yes
 
 REPLACE_PYTHON=		html2text.py
 
 INSTALLATION_DIRS=	share/html2text
 
-do-extract:
-	${RUN}								\
-	${TR} -d '\r' < ${DISTDIR}/${DISTNAME}${EXTRACT_SUFX}		\
-		> ${WRKSRC}/html2text.py
-
 do-install:
 	${INSTALL_SCRIPT} ${WRKSRC}/html2text.py ${DESTDIR}${PREFIX}/share/html2text
 
diff --git a/textproc/py-html2text/distinfo b/textproc/py-html2text/distinfo
index 4a0d03272ae..c751683bdb3 100644
--- a/textproc/py-html2text/distinfo
+++ b/textproc/py-html2text/distinfo
@@ -1,6 +1,6 @@
-$NetBSD: distinfo,v 1.14 2010/09/19 14:05:35 wiz Exp $
+$NetBSD: distinfo,v 1.15 2011/06/09 06:21:38 schmonz Exp $
 
-SHA1 (html2text-2.39.py) = eb9800b30929b5ed2a522930eb9e8b28e117342f
-RMD160 (html2text-2.39.py) = 97f9bdd09ece52b7ea9c31563ac09993efccae56
-Size (html2text-2.39.py) = 14325 bytes
-SHA1 (patch-aa) = 2395e54a2c2f72231f6766375b8406928dfa7f3b
+SHA1 (aaronsw-html2text-3.02-0-gd9bf7d6.tar.gz) = 021be508ba1835f7543ee9daf1e23ade85011418
+RMD160 (aaronsw-html2text-3.02-0-gd9bf7d6.tar.gz) = 68170480d7a37857a1722bae1b462f0de3c443cf
+Size (aaronsw-html2text-3.02-0-gd9bf7d6.tar.gz) = 5067 bytes
+SHA1 (patch-aa) = e26d95e8b7ad7a44d0951fc2dab73ec1e87593de
diff --git a/textproc/py-html2text/patches/patch-aa b/textproc/py-html2text/patches/patch-aa
index 367e5258c33..28ee1208bd6 100644
--- a/textproc/py-html2text/patches/patch-aa
+++ b/textproc/py-html2text/patches/patch-aa
@@ -1,16 +1,18 @@
-$NetBSD: patch-aa,v 1.6 2008/10/19 18:39:56 schmonz Exp $
+$NetBSD: patch-aa,v 1.7 2011/06/09 06:21:38 schmonz Exp $
 
---- html2text.py.orig	2008-10-19 14:36:07.000000000 -0400
+Small cleanup patch from Debian.
+
+--- html2text.py.orig	2011-03-08 23:35:25.000000000 +0000
 +++ html2text.py
-@@ -282,6 +282,7 @@ class _html2text(sgmllib.SGMLParser):
-             if attrs.has_key('src'):
+@@ -305,6 +305,7 @@ class _html2text(HTMLParser.HTMLParser):
+             if has_key(attrs, 'src'):
                  attrs['href'] = attrs['src']
                  alt = attrs.get('alt', '')
 +                alt = re.sub('\n', ' ', alt)
                  i = self.previousIndex(attrs)
                  if i is not None:
                      attrs = self.a[i]
-@@ -302,7 +303,7 @@ class _html2text(sgmllib.SGMLParser):
+@@ -325,7 +326,7 @@ class _html2text(HTMLParser.HTMLParser):
          if tag in ["ol", "ul"]:
              if start:
                  self.list.append({'name':tag, 'num':0})