v2: modify, sisu markup version script (take from v3)

2 files changed, 35 insertions, 11 deletions

diff --git a/data/doc/sisu/CHANGELOG_v2 b/data/doc/sisu/CHANGELOG_v2
index 248fff73..7101d7af 100644
--- a/data/doc/sisu/CHANGELOG_v2
+++ b/data/doc/sisu/CHANGELOG_v2
@@ -16,6 +16,8 @@ Reverse Chronological:
 http://git.sisudoc.org/?p=code/sisu.git;a=log;h=refs/tags/sisu_3.1.11
 http://git.sisudoc.org/?p=code/sisu.git;a=log;h=refs/tags/debian/sisu_3.1.11-1
 
+  * modify sisu markup version script
+
 %% 2.9.10 (2011-10-15:41/6) [see 3.1.1.orig.tar.xz]
 http://git.sisudoc.org/?p=code/sisu.git;a=log;h=refs/tags/sisu_3.1.1
 http://git.sisudoc.org/?p=code/sisu.git;a=log;h=refs/tags/debian/sisu_3.1.1-1
diff --git a/data/sisu/v2/conf/convert/modify.rb b/data/sisu/v2/conf/convert/modify.rb
index 5f4f9ae0..c3fe47c9 100644
--- a/data/sisu/v2/conf/convert/modify.rb
+++ b/data/sisu/v2/conf/convert/modify.rb
@@ -1,4 +1,5 @@
 #!/usr/bin/env ruby
+# encoding: utf-8
 =begin
  * Name: modify.rb
  * Author: Ralph Amissah
@@ -36,13 +37,12 @@
     http://www.gnu.org/copyleft/gpl.html
     http://www.jus.uio.no/sisu/gpl3.fsf
 
-  © Ralph Amissah 1997, current 2006.
-  All Rights Reserved.
+  (C) Ralph Amissah 1997, current 2009.
+      All Rights Reserved.
 
  * Ralph Amissah: ralph@amissah.com
                   ralph.amissah@gmail.com
 
- * Notes: configure rc in ~/.sisu/sisurc.yaml
 =end
 module SiSU_Modify
   class Convert_markup
@@ -51,6 +51,7 @@ module SiSU_Modify
       @description="This is a script that contains canned text conversions for reuse"
     end
     def current_match_and_replace #Disable, edit manually
+      clean_non_regular_characters
       #convert_pre_37_to_38_experimental
     end
     def message(text)
@@ -93,6 +94,27 @@ WOK
       exit
     end
     #%% substitutions to be made
+    def clean_non_regular_characters
+      #/[^ a-zA-Z0-9,.!?;:/={}()%@#~\[\]"'´&$_+^-]
+      #/[^ a-zA-Z0-9,.!?;:/={}()%@#~\[\]"'´&$_+^•’‘`“”–ffifffflfifl-]
+      #message(%w{[^ a-zA-Z0-9,.!?;:/={}()%@#~\[\]"'´&$_+^-]})
+      message(%q{clean non regular characters})
+      [
+        [/’/,                     "'"],
+        [/‘/,                     "'"],
+        [/´/,                     "'"],
+        [/`/,                     "'"],
+        [/[“”]/,                  '"'],
+        [/–/,                     '-'],
+        [/ff/,                     'ff'],
+        [/ffi/,                     'ffi'],
+        [/ffl/,                     'ffl'],
+        [/fi/,                     'fi'],
+        [/fl/,                     'fl'],
+        [/^•/,                    '_* '],
+        [/•/,                     '*']
+      ]
+    end
     def convert_pre_37_to_38_experimental
       message('convert sisu markup from 0.37 to 0.38 experimental (rad)')
       [
@@ -149,13 +171,13 @@ WOK
         p @files
         mr=nil
         #%% changes to make m match, r replace      -------------------------->
-        if @cf =~/--help/: help
+        if @cf =~/--help/; help
         else
           message("WARNING, proceed at your own risk,\npermanent changes requested for the above named files\n  best that you check (manually) what this file is set to do\n  conversions set are at the top of the file")
           mr=case @cf
-          when /--convert|default/:     current_match_and_replace
-          when /--(?:(?:37)?to38|rad)/: convert_pre_37_to_38_experimental
-          when /--(?:(?:38)?to37)/:     convert_38_experimental_to_37
+          when /--convert|default/;     current_match_and_replace
+          when /--(?:(?:37)?to38|rad)/; convert_pre_37_to_38_experimental
+          when /--(?:(?:38)?to37)/;     convert_38_experimental_to_37
           else help
           end
         end
@@ -190,7 +212,7 @@ WOK
             puts "match in #{i}"
             @flag_start=true
             cont.each do |y|
-              if y =~end_processing: @flag_end=true
+              if y =~end_processing; @flag_end=true
               end
               if @flag_start and not @flag_end
                 match_and_replace.each do |m,r|
@@ -206,11 +228,11 @@ WOK
               #end
               #if y =~m1
               #end
-              if y=~/^\s*$/: @empty1=true
+              if y=~/^\s*$/; @empty1=true
               else           @empty1=false
               end
               @file.puts y unless (@empty1==true and @empty2==true)
-              if y=~/^\s*$/: @empty2=true
+              if y=~/^\s*$/; @empty2=true
               else           @empty2=false
               end
               #@file << y + "\n"
@@ -242,7 +264,7 @@ match_and_replace=Array.new
 #WOK
 #  exit
 #end
-unless f.length > 0: f=Dir.glob("[a-z]*.ss?")  #restricted to sisu type files, it need not be
+unless f.length > 0; f=Dir.glob("[a-z]*.ss?")  #restricted to sisu type files, it need not be
 end
 #f=Dir.glob("{bin,conf,data,lib}/**/*.rb")     #sisu development
 #f=Dir.glob("[^_]/**/*")                       #all files subdirectories beneath pwd except those starting with _