From fc826abef5f83543f2cf00938b6d656e3b2c2821 Mon Sep 17 00:00:00 2001
From: Ralph Amissah <ralph@amissah.com>
Date: Wed, 3 Sep 2008 22:40:23 -0400
Subject: dal, left and right quotes (single & double) kept; previously
 converted to ordinary quotes (watch)

---
 lib/sisu/v0/dal.rb        |  4 ----
 lib/sisu/v0/shared_xml.rb | 12 +++++++++---
 2 files changed, 9 insertions(+), 7 deletions(-)

diff --git a/lib/sisu/v0/dal.rb b/lib/sisu/v0/dal.rb
index a8bf1cd0..e75309f9 100644
--- a/lib/sisu/v0/dal.rb
+++ b/lib/sisu/v0/dal.rb
@@ -240,13 +240,9 @@ module SiSU_DAL
         para.gsub!(/(#{Mx[:en_a_o]})\s*/,'\1 '); para.gsub!(/(~\{\s*)\s+/,'\1 ')
         para.gsub!(/ \/\//,"#{Mx[:br_line]}")                                           #added 2004w29
         para.gsub!(/<br>/,"#{Mx[:br_line]}")                                            #needed by xml, xhtml etc.
-        #para.gsub!(/<p>/,'<p />')                                             #consider
-        para.gsub!(/`/,"'")
         para.gsub!(/\t/,' ')
         para.gsub!(/\342\200\231/u,"'") #if para =~/’/  #Avoid #&lsquo; &rsquo; #&ldquo; &rdquo;
         para.gsub!(/�/u,' ') #watch, replace with char code
-        para.gsub!(/[“”]/u,'""')
-        para.gsub!(/[­–—]/u,'-')   #— – chk
         para.gsub!(/·/u,'*')
         para.gsub!(/\\copy(?:right)?\b/,'&#169;')
         para.gsub!(/\\trademark\b|\\tm\b/,'&#174;')
diff --git a/lib/sisu/v0/shared_xml.rb b/lib/sisu/v0/shared_xml.rb
index 228a5c14..7ecc52bb 100644
--- a/lib/sisu/v0/shared_xml.rb
+++ b/lib/sisu/v0/shared_xml.rb
@@ -254,6 +254,8 @@ module SiSU_XML_munge
           para.gsub!(/ÿ/u, '&#255;') # '&yuml;'   # &#255;
           para.gsub!(/‘/u, '&#8216;') # '&lsquo;' # &#8216;
           para.gsub!(/’/u, '&#8217;') # '&rsquo;' # &#8217;
+          para.gsub!(/“/u, '&#8220;') # &ldquo;   # &#8220;
+          para.gsub!(/”/u, '&#8221;') # &rdquo;   # &#8221;
           para.gsub!(/–/u, '&#8211;') # &ndash;   # &#8211;
           para.gsub!(/—/u, '&#8212;') # &mdash;   # &#8212;
           para.gsub!(/∝/u, '&#8733;') # &prop;    # &#8733;
@@ -355,6 +357,8 @@ module SiSU_XML_munge
           para.gsub!(/ÿ/u, '&yuml;')      # &#255;
           para.gsub!(/‘/u, '&#lsquo;')    # &lsquo;  # &#8216;
           para.gsub!(/’/u, '&#rsquo;')    # &rsquo;  # &#8217;
+          para.gsub!(/“/u, '&ldquo;')     # &ldquo;   # &#8220;
+          para.gsub!(/”/u, '&rdquo;')     # &rdquo;   # &#8221;
           para.gsub!(/–/u, '&ndash;')     # &ndash;  # &#8211;
           para.gsub!(/—/u, '&mdash;')     # &mdash;  # &#8212;
           para.gsub!(/∝/u, '&prop;')      # &prop;   # &#8733;
@@ -369,11 +373,13 @@ module SiSU_XML_munge
     end
     def tidywords(wordlist)
       wordlist.each do |x|
+        #imperfect solution will not catch all possible cases
         x.gsub!(/&/,'&amp;') unless x =~/&\S+;/
+        x.gsub!(/&([A-Z])/,'&amp;\1')
       end
     end
     def markup(para='')
-      wordlist=para.scan(/\S+|\n/) #\n needed for tables, check though added 2005w17
+      wordlist=para.scan(/&[#0-9a-z]+;|\S+|\n/) #\n needed for tables, check though added 2005w17
       para=tidywords(wordlist).join(' ').strip
       para.gsub!(/#{Mx[:br_line]}|#{Mx[:br_nl]}/,'<br />')
       para.gsub!(/#{Mx[:mk_o]}:name#\S+?#{Mx[:mk_c]}/,'')
@@ -439,7 +445,7 @@ module SiSU_XML_munge
         "<image.path>#{@dir.url.images_local}\/\\1</image.path>")
       para.gsub!(/&nbsp;/,'&#160;')
       #para.gsub!(/&nbsp;/,' ') #clean
-      wordlist=para.scan(/\S+|\n/) #\n needed for tables, check though added 2005w17
+      wordlist=para.scan(/&[#0-9a-z]+;|\S+|\n/) #\n needed for tables, check though added 2005w17
       para=tidywords(wordlist).join(' ').strip
       para
     end
@@ -462,7 +468,7 @@ module SiSU_XML_munge
         "<image.path>#{@dir.url.images_local}\/\\1</image.path>")
       para.gsub!(/&nbsp;/,'&#160;')
       #para.gsub!(/&nbsp;/,' ') #clean
-      wordlist=para.scan(/\S+|\n/) #\n needed for tables, check though added 2005w17
+      wordlist=para.scan(/&[#0-9a-z]+;|\S+|\n/) #\n needed for tables, check though added 2005w17
       para=tidywords(wordlist).join(' ').strip
       para
     end
-- 
cgit v1.2.3