Add port textproc/py-html2text:

html2text is a Python script that convers a page of HTML into clean,
easy-to-read plain ASCII text. Better yet, that ASCII also happens to
be valid Markdown (a text-to-HTML format).

WWW: http://www.aaronsw.com/2002/html2text/
Author: Aaron Swartz <me@aaronsw.com>

Inspired by:	pkgsrc package

textproc/Makefile

@@ -754,6 +754,7 @@
     SUBDIR += py-expat
     SUBDIR += py-feedparser
     SUBDIR += py-genshi
+    SUBDIR += py-html2text
     SUBDIR += py-hyperestraier
     SUBDIR += py-hyperestraier-python
     SUBDIR += py-jaxml

textproc/py-html2text/Makefile

@@ -0,0 +1,24 @@
+# New ports collection makefile for:	html2text
+# Date created:		23 April 2007
+# Whom:			Andrew Pantyukhin <infofarmer@FreeBSD.org>
+#
+# $FreeBSD$
+#
+
+PORTNAME=	html2text
+PORTVERSION=	2.2.8
+CATEGORIES=	textproc python
+MASTER_SITES=	CENKES
+PKGNAMEPREFIX=	${PYTHON_PKGNAMEPREFIX}
+
+MAINTAINER=	infofarmer@FreeBSD.org
+COMMENT=	Convert HTML into clean plain ASCII text
+
+USE_PYTHON=	yes
+USE_PYDISTUTILS=yes
+PLIST_FILES=	bin/${PORTNAME}.py
+
+post-patch:
+	@${REINPLACE_CMD} -e '1s|.*|#!${PYTHON_CMD}|' ${WRKSRC}/${PORTNAME}.py
+
+.include <bsd.port.mk>

textproc/py-html2text/distinfo

@@ -0,0 +1,3 @@
+MD5 (html2text-2.2.8.tar.gz) = 8f84bd9456972ee1cccd2489f8b2535a
+SHA256 (html2text-2.2.8.tar.gz) = 81a2304eeb7006f351343e8be59214eac8352ece6ff020fd11028b5a6e10d890
+SIZE (html2text-2.2.8.tar.gz) = 3998

textproc/py-html2text/files/patch-html2text.py

@@ -0,0 +1,28 @@
+--- html2text.py.orig	2007-01-18 19:06:49.000000000 -0500
++++ html2text.py
+@@ -150,7 +150,7 @@ class _html2text(sgmllib.SGMLParser):
+ 		self.lastWasNL = 0
+ 	
+ 	def outtextf(self, s): 
+-		if type(s) is type(''): s = codecs.utf_8_decode(s)[0]
++		if type(s) is type(''): s = codecs.utf_8_decode(s, "replace")[0]
+ 		self.outtext += s
+ 	
+ 	def close(self):
+@@ -259,6 +259,7 @@ class _html2text(sgmllib.SGMLParser):
+ 			if attrs.has_key('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]
+@@ -279,7 +280,7 @@ class _html2text(sgmllib.SGMLParser):
+ 		if tag in ["ol", "ul"]:
+ 			if start:
+ 				self.list.append({'name':tag, 'num':0})
+-			else:
++			elif self.list:
+ 				if self.list: self.list.pop()
+ 			
+ 			self.p()

textproc/py-html2text/pkg-descr

@@ -0,0 +1,6 @@
+html2text is a Python script that convers a page of HTML into clean,
+easy-to-read plain ASCII text. Better yet, that ASCII also happens to
+be valid Markdown (a text-to-HTML format).
+
+WWW: http://www.aaronsw.com/2002/html2text/
+Author: Aaron Swartz <me@aaronsw.com>