unictype/joininggroup-byname: Allow hyphens, omitted word separators.

* lib/unictype/joininggroup_byname.c (uc_joining_group_byname): Convert
also hyphens to space.
* lib/unictype/joininggroup_byname.gperf: Recognize the names also
without spaces.
* tests/unictype/test-joininggroup_byname.c (main): Add more tests.

diff --git a/ChangeLog b/ChangeLog
index 5742cd3f3c1cff5453d3b94a2b3c0f4dea694b64..afcd124e5da2348bf26a34d296ac334a29cb5452 100644 (file)
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,12 @@
+2011-03-26  Bruno Haible  <bruno@clisp.org>
+
+       unictype/joininggroup-byname: Allow hyphens, omitted word separators.
+       * lib/unictype/joininggroup_byname.c (uc_joining_group_byname): Convert
+       also hyphens to space.
+       * lib/unictype/joininggroup_byname.gperf: Recognize the names also
+       without spaces.
+       * tests/unictype/test-joininggroup_byname.c (main): Add more tests.
+
 2011-03-26  Bruno Haible  <bruno@clisp.org>
 
        unictype/joiningtype-byname: Recognize long names as well.

diff --git a/lib/unictype/joininggroup_byname.c b/lib/unictype/joininggroup_byname.c
index ca9f1a6594ed8879a36300c62af0923cdcb512d9..063ec2df5bad495b5216d769cbf53404d1288ed6 100644 (file)
--- a/lib/unictype/joininggroup_byname.c
+++ b/lib/unictype/joininggroup_byname.c
@@ -35,7 +35,7 @@ uc_joining_group_byname (const char *joining_group_name)
       char buf[MAX_WORD_LENGTH + 1];
       const struct named_joining_group *found;
 
-      /* Copy joining_group_name into buf, converting '_' to ' '.  */
+      /* Copy joining_group_name into buf, converting '_' and '-' to ' '.  */
       {
         const char *p = joining_group_name;
         char *q = buf;
@@ -44,7 +44,7 @@ uc_joining_group_byname (const char *joining_group_name)
           {
             char c = *p;
 
-            if (c == '_')
+            if (c == '_' || c == '-')
               c = ' ';
             *q = c;
             if (c == '\0')

diff --git a/lib/unictype/joininggroup_byname.gperf b/lib/unictype/joininggroup_byname.gperf
index 518d1c9cd0b1a7713a4539ae8a85298bb333081b..bc2fbc8d9b3fa05e717f23bd6ee4ca6df6db6e79 100644 (file)
--- a/lib/unictype/joininggroup_byname.gperf
+++ b/lib/unictype/joininggroup_byname.gperf
@@ -12,30 +12,37 @@ struct named_joining_group { int name; int joining_group; };
 %define string-pool-name joining_group_stringpool
 %%
 No Joining Group, UC_JOINING_GROUP_NONE
+NoJoiningGroup, UC_JOINING_GROUP_NONE
 Ain, UC_JOINING_GROUP_AIN
 Alaph, UC_JOINING_GROUP_ALAPH
 Alef, UC_JOINING_GROUP_ALEF
 Beh, UC_JOINING_GROUP_BEH
 Beth, UC_JOINING_GROUP_BETH
 Burushaski Yeh Barree, UC_JOINING_GROUP_BURUSHASKI_YEH_BARREE
+BurushaskiYehBarree, UC_JOINING_GROUP_BURUSHASKI_YEH_BARREE
 Dal, UC_JOINING_GROUP_DAL
 Dalath Rish, UC_JOINING_GROUP_DALATH_RISH
+DalathRish, UC_JOINING_GROUP_DALATH_RISH
 E, UC_JOINING_GROUP_E
 Farsi Yeh, UC_JOINING_GROUP_FARSI_YEH
+FarsiYeh, UC_JOINING_GROUP_FARSI_YEH
 Fe, UC_JOINING_GROUP_FE
 Feh, UC_JOINING_GROUP_FEH
 Final Semkath, UC_JOINING_GROUP_FINAL_SEMKATH
+FinalSemkath, UC_JOINING_GROUP_FINAL_SEMKATH
 Gaf, UC_JOINING_GROUP_GAF
 Gamal, UC_JOINING_GROUP_GAMAL
 Hah, UC_JOINING_GROUP_HAH
 He, UC_JOINING_GROUP_HE
 Heh, UC_JOINING_GROUP_HEH
 Heh Goal, UC_JOINING_GROUP_HEH_GOAL
+HehGoal, UC_JOINING_GROUP_HEH_GOAL
 Heth, UC_JOINING_GROUP_HETH
 Kaf, UC_JOINING_GROUP_KAF
 Kaph, UC_JOINING_GROUP_KAPH
 Khaph, UC_JOINING_GROUP_KHAPH
 Knotted Heh, UC_JOINING_GROUP_KNOTTED_HEH
+KnottedHeh, UC_JOINING_GROUP_KNOTTED_HEH
 Lam, UC_JOINING_GROUP_LAM
 Lamadh, UC_JOINING_GROUP_LAMADH
 Meem, UC_JOINING_GROUP_MEEM
@@ -48,23 +55,31 @@ Qaf, UC_JOINING_GROUP_QAF
 Qaph, UC_JOINING_GROUP_QAPH
 Reh, UC_JOINING_GROUP_REH
 Reversed Pe, UC_JOINING_GROUP_REVERSED_PE
+ReversedPe, UC_JOINING_GROUP_REVERSED_PE
 Sad, UC_JOINING_GROUP_SAD
 Sadhe, UC_JOINING_GROUP_SADHE
 Seen, UC_JOINING_GROUP_SEEN
 Semkath, UC_JOINING_GROUP_SEMKATH
 Shin, UC_JOINING_GROUP_SHIN
 Swash Kaf, UC_JOINING_GROUP_SWASH_KAF
+SwashKaf, UC_JOINING_GROUP_SWASH_KAF
 Syriac Waw, UC_JOINING_GROUP_SYRIAC_WAW
+SyriacWaw, UC_JOINING_GROUP_SYRIAC_WAW
 Tah, UC_JOINING_GROUP_TAH
 Taw, UC_JOINING_GROUP_TAW
 Teh Marbuta, UC_JOINING_GROUP_TEH_MARBUTA
+TehMarbuta, UC_JOINING_GROUP_TEH_MARBUTA
 Teh Marbuta Goal, UC_JOINING_GROUP_TEH_MARBUTA_GOAL
+TehMarbutaGoal, UC_JOINING_GROUP_TEH_MARBUTA_GOAL
 Teth, UC_JOINING_GROUP_TETH
 Waw, UC_JOINING_GROUP_WAW
 Yeh, UC_JOINING_GROUP_YEH
 Yeh Barree, UC_JOINING_GROUP_YEH_BARREE
+YehBarree, UC_JOINING_GROUP_YEH_BARREE
 Yeh with tail, UC_JOINING_GROUP_YEH_WITH_TAIL
+YehWithTail, UC_JOINING_GROUP_YEH_WITH_TAIL
 Yudh, UC_JOINING_GROUP_YUDH
 Yudh He, UC_JOINING_GROUP_YUDH_HE
+YudhHe, UC_JOINING_GROUP_YUDH_HE
 Zain, UC_JOINING_GROUP_ZAIN
 Zhain, UC_JOINING_GROUP_ZHAIN

diff --git a/tests/unictype/test-joininggroup_byname.c b/tests/unictype/test-joininggroup_byname.c
index 31b073bedeb5f0142696d6a549f83fcf319612ec..9f2119af662cfa44394403a925dff5a5d4b7f4ae 100644 (file)
--- a/tests/unictype/test-joininggroup_byname.c
+++ b/tests/unictype/test-joininggroup_byname.c
@@ -30,6 +30,7 @@ main ()
   ASSERT (uc_joining_group_byname ("NO JOINING GROUP") == UC_JOINING_GROUP_NONE);
   ASSERT (uc_joining_group_byname ("No Joining Group") == UC_JOINING_GROUP_NONE);
   ASSERT (uc_joining_group_byname ("No_Joining_Group") == UC_JOINING_GROUP_NONE);
+  ASSERT (uc_joining_group_byname ("NoJoiningGroup") == UC_JOINING_GROUP_NONE);
   ASSERT (uc_joining_group_byname ("AIN") == UC_JOINING_GROUP_AIN);
   ASSERT (uc_joining_group_byname ("Ain") == UC_JOINING_GROUP_AIN);
   ASSERT (uc_joining_group_byname ("ALAPH") == UC_JOINING_GROUP_ALAPH);
@@ -43,15 +44,18 @@ main ()
   ASSERT (uc_joining_group_byname ("BURUSHASKI YEH BARREE") == UC_JOINING_GROUP_BURUSHASKI_YEH_BARREE);
   ASSERT (uc_joining_group_byname ("Burushaski Yeh Barree") == UC_JOINING_GROUP_BURUSHASKI_YEH_BARREE);
   ASSERT (uc_joining_group_byname ("Burushaski_Yeh_Barree") == UC_JOINING_GROUP_BURUSHASKI_YEH_BARREE);
+  ASSERT (uc_joining_group_byname ("BurushaskiYehBarree") == UC_JOINING_GROUP_BURUSHASKI_YEH_BARREE);
   ASSERT (uc_joining_group_byname ("DAL") == UC_JOINING_GROUP_DAL);
   ASSERT (uc_joining_group_byname ("Dal") == UC_JOINING_GROUP_DAL);
   ASSERT (uc_joining_group_byname ("DALATH RISH") == UC_JOINING_GROUP_DALATH_RISH);
   ASSERT (uc_joining_group_byname ("Dalath Rish") == UC_JOINING_GROUP_DALATH_RISH);
   ASSERT (uc_joining_group_byname ("Dalath_Rish") == UC_JOINING_GROUP_DALATH_RISH);
+  ASSERT (uc_joining_group_byname ("DalathRish") == UC_JOINING_GROUP_DALATH_RISH);
   ASSERT (uc_joining_group_byname ("E") == UC_JOINING_GROUP_E);
   ASSERT (uc_joining_group_byname ("FARSI YEH") == UC_JOINING_GROUP_FARSI_YEH);
   ASSERT (uc_joining_group_byname ("Farsi Yeh") == UC_JOINING_GROUP_FARSI_YEH);
   ASSERT (uc_joining_group_byname ("Farsi_Yeh") == UC_JOINING_GROUP_FARSI_YEH);
+  ASSERT (uc_joining_group_byname ("FarsiYeh") == UC_JOINING_GROUP_FARSI_YEH);
   ASSERT (uc_joining_group_byname ("FE") == UC_JOINING_GROUP_FE);
   ASSERT (uc_joining_group_byname ("Fe") == UC_JOINING_GROUP_FE);
   ASSERT (uc_joining_group_byname ("FEH") == UC_JOINING_GROUP_FEH);
@@ -59,6 +63,7 @@ main ()
   ASSERT (uc_joining_group_byname ("FINAL SEMKATH") == UC_JOINING_GROUP_FINAL_SEMKATH);
   ASSERT (uc_joining_group_byname ("Final Semkath") == UC_JOINING_GROUP_FINAL_SEMKATH);
   ASSERT (uc_joining_group_byname ("Final_Semkath") == UC_JOINING_GROUP_FINAL_SEMKATH);
+  ASSERT (uc_joining_group_byname ("FinalSemkath") == UC_JOINING_GROUP_FINAL_SEMKATH);
   ASSERT (uc_joining_group_byname ("GAF") == UC_JOINING_GROUP_GAF);
   ASSERT (uc_joining_group_byname ("Gaf") == UC_JOINING_GROUP_GAF);
   ASSERT (uc_joining_group_byname ("GAMAL") == UC_JOINING_GROUP_GAMAL);
@@ -72,6 +77,7 @@ main ()
   ASSERT (uc_joining_group_byname ("HEH GOAL") == UC_JOINING_GROUP_HEH_GOAL);
   ASSERT (uc_joining_group_byname ("Heh Goal") == UC_JOINING_GROUP_HEH_GOAL);
   ASSERT (uc_joining_group_byname ("Heh_Goal") == UC_JOINING_GROUP_HEH_GOAL);
+  ASSERT (uc_joining_group_byname ("HehGoal") == UC_JOINING_GROUP_HEH_GOAL);
   ASSERT (uc_joining_group_byname ("HETH") == UC_JOINING_GROUP_HETH);
   ASSERT (uc_joining_group_byname ("Heth") == UC_JOINING_GROUP_HETH);
   ASSERT (uc_joining_group_byname ("KAF") == UC_JOINING_GROUP_KAF);
@@ -83,6 +89,7 @@ main ()
   ASSERT (uc_joining_group_byname ("KNOTTED HEH") == UC_JOINING_GROUP_KNOTTED_HEH);
   ASSERT (uc_joining_group_byname ("Knotted Heh") == UC_JOINING_GROUP_KNOTTED_HEH);
   ASSERT (uc_joining_group_byname ("Knotted_Heh") == UC_JOINING_GROUP_KNOTTED_HEH);
+  ASSERT (uc_joining_group_byname ("KnottedHeh") == UC_JOINING_GROUP_KNOTTED_HEH);
   ASSERT (uc_joining_group_byname ("LAM") == UC_JOINING_GROUP_LAM);
   ASSERT (uc_joining_group_byname ("Lam") == UC_JOINING_GROUP_LAM);
   ASSERT (uc_joining_group_byname ("LAMADH") == UC_JOINING_GROUP_LAMADH);
@@ -108,6 +115,7 @@ main ()
   ASSERT (uc_joining_group_byname ("REVERSED PE") == UC_JOINING_GROUP_REVERSED_PE);
   ASSERT (uc_joining_group_byname ("Reversed Pe") == UC_JOINING_GROUP_REVERSED_PE);
   ASSERT (uc_joining_group_byname ("Reversed_Pe") == UC_JOINING_GROUP_REVERSED_PE);
+  ASSERT (uc_joining_group_byname ("ReversedPe") == UC_JOINING_GROUP_REVERSED_PE);
   ASSERT (uc_joining_group_byname ("SAD") == UC_JOINING_GROUP_SAD);
   ASSERT (uc_joining_group_byname ("Sad") == UC_JOINING_GROUP_SAD);
   ASSERT (uc_joining_group_byname ("SADHE") == UC_JOINING_GROUP_SADHE);
@@ -121,8 +129,11 @@ main ()
   ASSERT (uc_joining_group_byname ("SWASH KAF") == UC_JOINING_GROUP_SWASH_KAF);
   ASSERT (uc_joining_group_byname ("Swash Kaf") == UC_JOINING_GROUP_SWASH_KAF);
   ASSERT (uc_joining_group_byname ("Swash_Kaf") == UC_JOINING_GROUP_SWASH_KAF);
+  ASSERT (uc_joining_group_byname ("SwashKaf") == UC_JOINING_GROUP_SWASH_KAF);
   ASSERT (uc_joining_group_byname ("SYRIAC WAW") == UC_JOINING_GROUP_SYRIAC_WAW);
   ASSERT (uc_joining_group_byname ("Syriac Waw") == UC_JOINING_GROUP_SYRIAC_WAW);
+  ASSERT (uc_joining_group_byname ("Syriac_Waw") == UC_JOINING_GROUP_SYRIAC_WAW);
+  ASSERT (uc_joining_group_byname ("SyriacWaw") == UC_JOINING_GROUP_SYRIAC_WAW);
   ASSERT (uc_joining_group_byname ("TAH") == UC_JOINING_GROUP_TAH);
   ASSERT (uc_joining_group_byname ("Tah") == UC_JOINING_GROUP_TAH);
   ASSERT (uc_joining_group_byname ("TAW") == UC_JOINING_GROUP_TAW);
@@ -130,9 +141,11 @@ main ()
   ASSERT (uc_joining_group_byname ("TEH MARBUTA") == UC_JOINING_GROUP_TEH_MARBUTA);
   ASSERT (uc_joining_group_byname ("Teh Marbuta") == UC_JOINING_GROUP_TEH_MARBUTA);
   ASSERT (uc_joining_group_byname ("Teh_Marbuta") == UC_JOINING_GROUP_TEH_MARBUTA);
+  ASSERT (uc_joining_group_byname ("TehMarbuta") == UC_JOINING_GROUP_TEH_MARBUTA);
   ASSERT (uc_joining_group_byname ("TEH MARBUTA GOAL") == UC_JOINING_GROUP_TEH_MARBUTA_GOAL);
   ASSERT (uc_joining_group_byname ("Teh Marbuta Goal") == UC_JOINING_GROUP_TEH_MARBUTA_GOAL);
   ASSERT (uc_joining_group_byname ("Teh_Marbuta_Goal") == UC_JOINING_GROUP_TEH_MARBUTA_GOAL);
+  ASSERT (uc_joining_group_byname ("TehMarbutaGoal") == UC_JOINING_GROUP_TEH_MARBUTA_GOAL);
   ASSERT (uc_joining_group_byname ("TETH") == UC_JOINING_GROUP_TETH);
   ASSERT (uc_joining_group_byname ("Teth") == UC_JOINING_GROUP_TETH);
   ASSERT (uc_joining_group_byname ("WAW") == UC_JOINING_GROUP_WAW);
@@ -142,13 +155,17 @@ main ()
   ASSERT (uc_joining_group_byname ("YEH BARREE") == UC_JOINING_GROUP_YEH_BARREE);
   ASSERT (uc_joining_group_byname ("Yeh Barree") == UC_JOINING_GROUP_YEH_BARREE);
   ASSERT (uc_joining_group_byname ("Yeh_Barree") == UC_JOINING_GROUP_YEH_BARREE);
+  ASSERT (uc_joining_group_byname ("YehBarree") == UC_JOINING_GROUP_YEH_BARREE);
   ASSERT (uc_joining_group_byname ("YEH WITH TAIL") == UC_JOINING_GROUP_YEH_WITH_TAIL);
   ASSERT (uc_joining_group_byname ("Yeh with tail") == UC_JOINING_GROUP_YEH_WITH_TAIL);
   ASSERT (uc_joining_group_byname ("Yeh_with_tail") == UC_JOINING_GROUP_YEH_WITH_TAIL);
+  ASSERT (uc_joining_group_byname ("YehWithTail") == UC_JOINING_GROUP_YEH_WITH_TAIL);
   ASSERT (uc_joining_group_byname ("YUDH") == UC_JOINING_GROUP_YUDH);
   ASSERT (uc_joining_group_byname ("Yudh") == UC_JOINING_GROUP_YUDH);
   ASSERT (uc_joining_group_byname ("YUDH HE") == UC_JOINING_GROUP_YUDH_HE);
+  ASSERT (uc_joining_group_byname ("Yudh He") == UC_JOINING_GROUP_YUDH_HE);
   ASSERT (uc_joining_group_byname ("Yudh_He") == UC_JOINING_GROUP_YUDH_HE);
+  ASSERT (uc_joining_group_byname ("YudhHe") == UC_JOINING_GROUP_YUDH_HE);
   ASSERT (uc_joining_group_byname ("ZAIN") == UC_JOINING_GROUP_ZAIN);
   ASSERT (uc_joining_group_byname ("Zain") == UC_JOINING_GROUP_ZAIN);
   ASSERT (uc_joining_group_byname ("ZHAIN") == UC_JOINING_GROUP_ZHAIN);