Hyphenation moved to dw (soon to be used by dw::Textlock)

5 files changed, 222 insertions, 174 deletions

diff --git a/dw/Makefile.am b/dw/Makefile.am
index 4cd53099..bab63682 100644
--- a/dw/Makefile.am
+++ b/dw/Makefile.am
@@ -57,6 +57,8 @@ libDw_widgets_a_SOURCES = \
 	alignedtextblock.hh \
 	bullet.cc \
 	bullet.hh \
+	hyphenator.cc \
+	hyphenator.hh \
 	image.cc \
 	image.hh \
 	listitem.cc \
diff --git a/dw/hyphenator.cc b/dw/hyphenator.cc
new file mode 100644
index 00000000..d06ac94e
--- /dev/null
+++ b/dw/hyphenator.cc
@@ -0,0 +1,162 @@
+#include "hyphenator.hh"
+
+#include "../lout/misc.hh"
+#include <stdio.h>
+#include <string.h>
+
+#define LEN 1000
+
+/*
+ * This is a direct translation of the Python implementation by Ned
+ * Batchelder.
+ */
+
+using namespace lout::object;
+using namespace lout::container::typed;
+
+namespace dw {
+
+Hyphenator::Hyphenator (core::Platform *platform, const char *filename)
+{
+   this->platform = platform;
+   tree = new HashTable <Integer, Collection <Integer> > (true, true);
+
+   FILE *file = fopen (filename, "r");
+   while (!feof (file)) {
+      char buf[LEN + 1];
+      char *s = fgets (buf, LEN, file);
+      if (s) {
+         int l = strlen (s);
+         if (s[l - 1] == '\n')
+            s[l - 1] = 0;
+         insertPattern (s);
+      }
+   }
+   fclose (file);
+}
+
+void Hyphenator::insertPattern (char *s)
+{
+   // Convert the a pattern like 'a1bc3d4' into a string of chars 'abcd'
+   // and a list of points [ 0, 1, 0, 3, 4 ].
+   int l = strlen (s);
+   char chars [l + 1];
+   Vector <Integer> *points = new Vector <Integer> (1, true);
+
+   // TODO numbers consisting of multiple digits?
+   // TODO Encoding: This implementation works exactly like the Python
+   // implementation, based on UTF-8. Does this always work?
+   int numChars = 0;
+   for (int i = 0; s[i]; i++)
+      if (s[i] >= '0' && s[i] <= '9')
+         points->put (new Integer (s[i] - '0'), numChars);
+      else
+         chars[numChars++] = s[i];
+   chars[numChars] = 0;
+
+   for (int i = 0; i < numChars + 1; i++) {
+      Integer *val = points->get (i);
+      if (val == NULL)
+         points->put (new Integer (0), i);
+   }
+
+   // Insert the pattern into the tree.  Each character finds a dict
+   // another level down in the tree, and leaf nodes have the list of
+   // points.
+
+   HashTable <Integer, Collection <Integer> > *t = tree;
+   for (int i = 0; chars[i]; i++) {
+      Integer c (chars[i]);
+      if (!t->contains(&c))
+         t->put (new Integer (chars[i]),
+                 new HashTable <Integer, Collection <Integer> > (true, true));
+      t = (HashTable <Integer, Collection <Integer> >*) t->get (&c);
+   }
+
+   t->put (new Integer (0), points);
+}
+
+/**
+ * Given a word, returns a list of pieces, broken at the possible
+ * hyphenation points.
+ */
+Vector <String> *Hyphenator::hyphenateWord(const char *word)
+{
+   Vector <String> *pieces = new Vector <String> (1, true);
+
+   // Short words aren't hyphenated.
+   if (strlen (word) <= 4) { // TODO UTF-8?
+      pieces->put (new String (word));
+      return pieces;
+   }
+
+   // If the word is an exception, get the stored points.
+   // TODO
+
+   char work[strlen (word) + 3];
+   strcpy (work, ".");
+   char *wordLc = platform->textToLower (word, strlen (word));
+   strcat (work, wordLc);
+   delete wordLc;
+   strcat (work, ".");
+   
+   int l = strlen (work);
+   Vector <Integer> points (l + 1, true);
+   for (int i = 0; i < l + 1; i++)
+      points.put (new Integer (0), i);
+
+   Integer null (0);
+
+   for (int i = 0; i < l; i++) {
+      HashTable <Integer, Collection <Integer> > *t = tree;
+      for (int j = i; j < l; j++) {
+         Integer c (work[j]);
+         if (t->contains (&c)) {
+            t = (HashTable <Integer, Collection <Integer> >*) t->get (&c);
+            if (t->contains (&null)) {
+               Vector <Integer> *p = (Vector <Integer>*) t->get (&null);
+
+               for (int k = 0; k < p->size (); k++) {
+                  Integer *v1 = points.get (i + k);
+                  Integer *v2 = p->get (k);
+                  // TODO Not very efficient, especially here: too much
+                  // calls of "new"
+                  points.put(new Integer (lout::misc::max (v1->getValue (),
+                                                           v2->getValue ())),
+                             i + k);
+               }
+            }
+         } else
+            break;
+      }
+   }  
+
+   // No hyphens in the first two chars or the last two.
+   points.put (new Integer (0), 1);
+   points.put (new Integer (0), 2);
+   points.put (new Integer (0), points.size () - 2);
+   points.put (new Integer (0), points.size () - 3);
+ 
+   // Examine the points to build the pieces list.
+   char temp[strlen (word) + 1], *ptemp = temp;
+
+   int n = lout::misc::min ((int)strlen (word), points.size () - 2);
+   for (int i = 0; i < n; i++) {
+      char c = word[i];
+      int p = points.get(i + 2)->getValue ();
+      
+      *(ptemp++) = c;
+      if (p % 2) {
+         *ptemp = 0;
+         pieces->put (new String (temp));
+         ptemp = temp;
+      }
+   }
+
+   *ptemp = 0;
+   pieces->put (new String (temp));
+         
+   return pieces;
+}
+
+} // namespace dw
diff --git a/dw/hyphenator.hh b/dw/hyphenator.hh
new file mode 100644
index 00000000..089680bc
--- /dev/null
+++ b/dw/hyphenator.hh
@@ -0,0 +1,33 @@
+#ifndef __DW_HYPHENATOR_HH__
+#define __DW_HYPHENATOR_HH__
+
+#include "../lout/object.hh"
+#include "../lout/container.hh"
+#include "../dw/core.hh"
+
+namespace dw {
+
+class Hyphenator
+{
+private:
+   /*
+    * Actually, only one method in Platform is needed:
+    * textToLower(). And, IMO, this method is actually not platform
+    * independant, but based on UTF-8. Clarify? Change?
+    */
+   core::Platform *platform;
+   lout::container::typed::HashTable <lout::object::Integer,
+                                      lout::container::typed::Collection
+                                      <lout::object::Integer> > *tree;
+   void insertPattern (char *s);
+
+public:
+   Hyphenator (core::Platform *platform, const char *filename);
+   
+   lout::container::typed::Vector <lout::object::String>
+   *hyphenateWord(const char *word);
+};
+
+} // namespace dw
+
+#endif // __DW_HYPHENATOR_HH__
diff --git a/test/Makefile.am b/test/Makefile.am
index daa2cef5..425a11a8 100644
--- a/test/Makefile.am
+++ b/test/Makefile.am
@@ -163,4 +163,9 @@ cookies_LDADD = \
 
 liang_SOURCES = liang.cc
 
-liang_LDADD = $(top_builddir)/lout/liblout.a
+liang_LDADD = \
+	$(top_builddir)/dw/libDw-widgets.a \
+	$(top_builddir)/dw/libDw-fltk.a \
+	$(top_builddir)/dw/libDw-core.a \
+	$(top_builddir)/lout/liblout.a \
+	@LIBFLTK_LIBS@
diff --git a/test/liang.cc b/test/liang.cc
index c8fb5f45..c303466f 100644
--- a/test/liang.cc
+++ b/test/liang.cc
@@ -1,186 +1,32 @@
-#include "../lout/misc.hh"
-#include "../lout/object.hh"
-#include "../lout/container.hh"
-#include <stdio.h>
-#include <string.h>
-
-#define LEN 1000
-
-/*
- * This is a direct translation of the Python implementation by Ned
- * Batchelder.
- */
-
-class Hyphenator
-{
-private:
-   lout::container::typed::HashTable <lout::object::Integer,
-                                      lout::container::typed::Collection
-                                      <lout::object::Integer> > *tree;
-   void insertPattern (char *s);
-
-public:
-   Hyphenator (const char *filename);
-   
-   lout::container::typed::Vector <lout::object::String>
-   *hyphenateWord(const char *word);
-};
+#include "../dw/fltkcore.hh"
+#include "../dw/hyphenator.hh"
 
 using namespace lout::object;
 using namespace lout::container::typed;
 
-Hyphenator::Hyphenator (const char *filename)
+void hyphenateWord (dw::Hyphenator *h, const char *word)
 {
-   tree = new HashTable <Integer, Collection <Integer> > (true, true);
-
-   FILE *file = fopen (filename, "r");
-   while (!feof (file)) {
-      char buf[LEN + 1];
-      char *s = fgets (buf, LEN, file);
-      if (s) {
-         int l = strlen (s);
-         if (s[l - 1] == '\n')
-            s[l - 1] = 0;
-         insertPattern (s);
-      }
-   }
-   fclose (file);
-}
-
-void Hyphenator::insertPattern (char *s)
-{
-   // Convert the a pattern like 'a1bc3d4' into a string of chars 'abcd'
-   // and a list of points [ 0, 1, 0, 3, 4 ].
-   int l = strlen (s);
-   char chars [l + 1];
-   Vector <Integer> *points = new Vector <Integer> (1, true);
-
-   // TODO numbers consisting of multiple digits?
-   // TODO Encoding: This implementation works exactly like the Python
-   // implementation, based on UTF-8. Does this always work?
-   int numChars = 0;
-   for (int i = 0; s[i]; i++)
-      if (s[i] >= '0' && s[i] <= '9')
-         points->put (new Integer (s[i] - '0'), numChars);
-      else
-         chars[numChars++] = s[i];
-   chars[numChars] = 0;
-
-   for (int i = 0; i < numChars + 1; i++) {
-      Integer *val = points->get (i);
-      if (val == NULL)
-         points->put (new Integer (0), i);
-   }
-
-   // Insert the pattern into the tree.  Each character finds a dict
-   // another level down in the tree, and leaf nodes have the list of
-   // points.
-
-   HashTable <Integer, Collection <Integer> > *t = tree;
-   for (int i = 0; chars[i]; i++) {
-      Integer c (chars[i]);
-      if (!t->contains(&c))
-         t->put (new Integer (chars[i]),
-                 new HashTable <Integer, Collection <Integer> > (true, true));
-      t = (HashTable <Integer, Collection <Integer> >*) t->get (&c);
+   Vector <String> *pieces = h->hyphenateWord (word);
+   for (int i = 0; i < pieces->size (); i++) {
+      if (i != 0)
+         putchar ('-');
+      printf ("%s", pieces->get(i)->chars ());
    }
-
-   t->put (new Integer (0), points);
-}
-
-/**
- * Given a word, returns a list of pieces, broken at the possible
- * hyphenation points.
- */
-Vector <String> *Hyphenator::hyphenateWord(const char *word)
-{
-   // Short words aren't hyphenated.
-   if (strlen (word) <= 4) // TODO UTF-8
-      return NULL; // TODO
-
-   // If the word is an exception, get the stored points.
-   // TODO
-
-   char work[strlen (word) + 3];
-   strcpy (work, ".");
-   strcat (work, word); // TODO tolower
-   strcat (work, ".");
-   
-   int l = strlen (work);
-   Vector <Integer> points (l + 1, true);
-   for (int i = 0; i < l + 1; i++)
-      points.put (new Integer (0), i);
-
-   Integer null (0);
-
-   for (int i = 0; i < l; i++) {
-      HashTable <Integer, Collection <Integer> > *t = tree;
-      for (int j = i; j < l; j++) {
-         Integer c (work[j]);
-         if (t->contains (&c)) {
-            t = (HashTable <Integer, Collection <Integer> >*) t->get (&c);
-            if (t->contains (&null)) {
-               Vector <Integer> *p = (Vector <Integer>*) t->get (&null);
-
-               for (int k = 0; k < p->size (); k++) {
-                  Integer *v1 = points.get (i + k);
-                  Integer *v2 = p->get (k);
-                  // TODO Not very efficient, especially here: too much
-                  // calls of "new"
-                  points.put(new Integer (lout::misc::max (v1->getValue (),
-                                                           v2->getValue ())),
-                             i + k);
-               }
-            }
-         } else
-            break;
-      }
-   }  
-
-   // No hyphens in the first two chars or the last two.
-   points.put (new Integer (0), 1);
-   points.put (new Integer (0), 2);
-   points.put (new Integer (0), points.size () - 2);
-   points.put (new Integer (0), points.size () - 3);
- 
-   // Examine the points to build the pieces list.
-   Vector <String> *pieces = new Vector <String> (1, true);
-   char temp[strlen (word) + 1], *ptemp = temp;
-
-   int n = lout::misc::min ((int)strlen (word), points.size () - 2);
-   for (int i = 0; i < n; i++) {
-      char c = word[i];
-      int p = points.get(i + 2)->getValue ();
-      
-      *(ptemp++) = c;
-      if (p % 2) {
-         *ptemp = 0;
-         printf ("'%s'\n", temp);
-         ptemp = temp;
-      }
-   }
-
-   *ptemp = 0;
-   printf ("'%s'\n", temp);
-
-   return pieces;
+   putchar ('\n');
 }
 
 int main (int argc, char *argv[])
 {
-   Hyphenator h ("test/hyph-de-1996.pat");
-   h.hyphenateWord ("jahrhundertroman");
-   puts ("---");
-   h.hyphenateWord ("währenddessen");
-   puts ("---");
-   h.hyphenateWord ("ückendorf");
-   puts ("---");
-   h.hyphenateWord ("über");
-   puts ("---");
-   h.hyphenateWord ("aber");
-   puts ("---");
-   h.hyphenateWord ("ackermann");
-   puts ("---");
+   dw::fltk::FltkPlatform p;
+   dw::Hyphenator h (&p, "test/hyph-de-1996.pat");
+
+   hyphenateWord (&h, "Jahrhundertroman");
+   hyphenateWord (&h, "JAHRHUNDERTROMAN");
+   hyphenateWord (&h, "währenddessen");
+   hyphenateWord (&h, "Ückendorf");
+   hyphenateWord (&h, "über");
+   hyphenateWord (&h, "aber");
+   hyphenateWord (&h, "Ackermann");
 
    return 0;
 }