Added non-breakable space to division characters, so that hyphenation works correctly.

4 files changed, 14 insertions, 4 deletions

diff --git a/dw/textblock.cc b/dw/textblock.cc
index 20ac7fa5..226b5ab6 100644
--- a/dw/textblock.cc
+++ b/dw/textblock.cc
@@ -48,7 +48,10 @@ Textblock::DivChar Textblock::divChars[NUM_DIV_CHARS] = {
    { "\xe2\x80\x90", false, true, true, -1, PENALTY_HYPHEN },
    // em dash (U+2014): breaks on both sides are allowed (but see below).
    { "\xe2\x80\x94", false, true, false,
-     PENALTY_EM_DASH_LEFT, PENALTY_EM_DASH_RIGHT }
+     PENALTY_EM_DASH_LEFT, PENALTY_EM_DASH_RIGHT },
+   // non-breakable space
+   { "\xc2\xa0", false, true, true,
+     PENALTY_PROHIBIT_BREAK_I, PENALTY_PROHIBIT_BREAK_I }
 };
 
 // Standard values are defined here. The values are already multiplied
@@ -74,7 +77,9 @@ int Textblock::penalties[PENALTY_NUM][2] = {
    // so that a break on the *right* side is preferred.
    { 800, 800 },
    // Penalties for a break point *right* of an em-dash: like hyphens.
-   { 100, 800 }
+   { 100, 800 },
+   // Prohibit: added here to have a defined index in divChars[].
+   { INT_MAX, INT_MAX }
 };
 
 /**
diff --git a/dw/textblock.hh b/dw/textblock.hh
index b1ad3ee8..f448cda3 100644
--- a/dw/textblock.hh
+++ b/dw/textblock.hh
@@ -217,8 +217,8 @@ private:
    };
 
    enum { PENALTY_HYPHEN, PENALTY_EM_DASH_LEFT, PENALTY_EM_DASH_RIGHT,
-          PENALTY_NUM };
-   enum { NUM_DIV_CHARS = 4 };
+          PENALTY_PROHIBIT_BREAK_I, PENALTY_NUM };
+   enum { NUM_DIV_CHARS = 5 };
 
    typedef struct
    {
diff --git a/test/hyphenate-nbsp.html b/test/hyphenate-nbsp.html
new file mode 100644
index 00000000..1e9bf379
--- /dev/null
+++ b/test/hyphenate-nbsp.html
@@ -0,0 +1,4 @@
+<div style="font-size:10em" lang=de>
+Weiß
+kurz&nbsp;und&nbsp;knapp
+</div>
diff --git a/test/liang.cc b/test/liang.cc
index 71ed6210..4ed717ee 100644
--- a/test/liang.cc
+++ b/test/liang.cc
@@ -49,6 +49,7 @@ int main (int argc, char *argv[])
                   "carbonsäure-7²-(Z)-(O-methyloxim)");
    hyphenateWord (&p, "Abtei-Stadt");
    hyphenateWord (&p, "Nordrhein-Westfalen");
+   hyphenateWord (&p, "kurz\xc2\xa0und\xc2\xa0knapp");
 
    return 0;
 }