[5288] * tp/Texinfo/Convert/Line.pm, Texinfo/Convert/Paragraph.pm:

[Patrice Dumas]

Revision: 5288
          http://svn.sv.gnu.org/viewvc/?view=rev&root=texinfo&revision=5288
Author:   pertusus
Date:     2013-08-09 23:27:23 +0000 (Fri, 09 Aug 2013)
Log Message:
-----------
        * tp/Texinfo/Convert/Line.pm, Texinfo/Convert/Paragraph.pm:
        Do not break words at non-breaking spaces.

Modified Paths:
--------------
    trunk/ChangeLog
    trunk/tp/Texinfo/Convert/Line.pm
    trunk/tp/Texinfo/Convert/Paragraph.pm
    trunk/tp/t/results/plaintext_tests/all_spaces/res_plaintext/all_spaces.txt

Modified: trunk/ChangeLog
===================================================================
--- trunk/ChangeLog     2013-08-09 22:22:19 UTC (rev 5287)
+++ trunk/ChangeLog     2013-08-09 23:27:23 UTC (rev 5288)
@@ -2,6 +2,7 @@
 
        * tp/Texinfo/Convert/Line.pm, Texinfo/Convert/Paragraph.pm: preserve 
        spaces coming from the document.
+       Do not break words at non-breaking spaces.
 
 2013-08-09  Patrice Dumas  <address@hidden>
 

Modified: trunk/tp/Texinfo/Convert/Line.pm
===================================================================
--- trunk/tp/Texinfo/Convert/Line.pm    2013-08-09 22:22:19 UTC (rev 5287)
+++ trunk/tp/Texinfo/Convert/Line.pm    2013-08-09 23:27:23 UTC (rev 5288)
@@ -280,8 +280,9 @@
       $word = $line->{'word'} if (defined($line->{'word'}));
       print STDERR "s `$line->{'space'}', w `$word'\n";
     }
-    if ($text =~ s/^([^\S\n]+)//) {
-      $underlying_text =~ s/^([^\S\n]+)//;
+    # \x{202f}\x{00a0} are non breaking spaces
+    if ($text =~ s/^([^\S\x{202f}\x{00a0}\n]+)//) {
+      $underlying_text =~ s/^([^\S\x{202f}\x{00a0}\n]+)//;
       my $spaces = $1;
       print STDERR "SPACES.L\n" if ($line->{'DEBUG'});
       if ($line->{'protect_spaces'}) {
@@ -313,9 +314,9 @@
           }
         }
       }
-    } elsif ($text =~ s/^([^\s\p{InFullwidth}]+)//) {
+    } elsif ($text =~ s/^(([^\s\p{InFullwidth}]|[\x{202f}\x{00a0}])+)//) {
       my $added_word = $1;
-      $underlying_text =~ s/^([^\s\p{InFullwidth}]+)//;
+      $underlying_text =~ s/^(([^\s\p{InFullwidth}]|[\x{202f}\x{00a0}])+)//;
       my $underlying_added_word = $1;
 
       $result .= $line->_add_next($added_word, $underlying_added_word);

Modified: trunk/tp/Texinfo/Convert/Paragraph.pm
===================================================================
--- trunk/tp/Texinfo/Convert/Paragraph.pm       2013-08-09 22:22:19 UTC (rev 
5287)
+++ trunk/tp/Texinfo/Convert/Paragraph.pm       2013-08-09 23:27:23 UTC (rev 
5288)
@@ -328,9 +328,10 @@
       $word = $paragraph->{'word'} if (defined($paragraph->{'word'}));
       print STDERR "p ($paragraph->{'counter'}+$paragraph->{'word_counter'}) s 
`$paragraph->{'space'}', w `$word'\n";
     }
-    if ($text =~ s/^(\s+)//) {
+    # \x{202f}\x{00a0} are non breaking spaces
+    if ($text =~ s/^([^\S\x{202f}\x{00a0}]+)//) {
       my $spaces = $1;
-      $underlying_text =~ s/^(\s+)//;
+      $underlying_text =~ s/^([^\S\x{202f}\x{00a0}]+)//;
       print STDERR "SPACES($paragraph->{'counter'}) `$spaces'\n" if 
($paragraph->{'DEBUG'});
       #my $added_word = $paragraph->{'word'};
       if ($paragraph->{'protect_spaces'}) {
@@ -421,9 +422,9 @@
       $result .= $paragraph->_add_pending_word();
       delete $paragraph->{'end_sentence'};
       $paragraph->{'space'} = '';
-    } elsif ($text =~ s/^([^\s\p{InFullwidth}]+)//) {
+    } elsif ($text =~ s/^(([^\s\p{InFullwidth}]|[\x{202f}\x{00a0}])+)//) {
       my $added_word = $1;
-      $underlying_text =~ s/^([^\s\p{InFullwidth}]+)//;
+      $underlying_text =~ s/^(([^\s\p{InFullwidth}]|[\x{202f}\x{00a0}])+)//;
       my $underlying_added_word = $1;
 
       $result .= $paragraph->_add_next($added_word, $underlying_added_word);

Modified: 
trunk/tp/t/results/plaintext_tests/all_spaces/res_plaintext/all_spaces.txt
===================================================================
--- trunk/tp/t/results/plaintext_tests/all_spaces/res_plaintext/all_spaces.txt  
2013-08-09 22:22:19 UTC (rev 5287)
+++ trunk/tp/t/results/plaintext_tests/all_spaces/res_plaintext/all_spaces.txt  
2013-08-09 23:27:23 UTC (rev 5288)
@@ -10,7 +10,6 @@
    | PUNCTUATION SPACE: | |
 
    NO-BREAK SPACE:
-aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
-bbbbbbbbbbbbbbbbbbbbb NARROW NO-BREAK SPACE:
-ccccccccccccccccccccccccccccccccccccccccccccccccccccccccc
-dddddddddddddddddddd
+aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa bbbbbbbbbbbbbbbbbbbbb
+NARROW NO-BREAK SPACE:
+ccccccccccccccccccccccccccccccccccccccccccccccccccccccccc dddddddddddddddddddd