name-to-ucs mapping function, first attempt

4 files changed, 70 insertions, 5 deletions

diff --git a/Makefile b/Makefile
index 605e788..0d61fe5 100644
--- a/Makefile
+++ b/Makefile
@@ -47,7 +47,7 @@ CVT_FILES = gen/jamo.c gen/nameslist.tab gen/nametoucs.keys gen/nametoucs.tab \
 
 LIBSRCS = proparray.c gen/nametoucs_hash.c gen/ucstoname_hash.c \
 	  gen/jamo.c gen/nameslist.c gen/nameslist_dict.c \
-	  gen/ucstoname_tab.c gen/nametoucs_tab.c
+	  gen/ucstoname_tab.c gen/nametoucs_tab.c nametoucs.c
 
 LIBOBJS = $(patsubst %.c,%.o,$(LIBSRCS))
 SO_OBJS = $(patsubst %.c,%.lo,$(LIBSRCS))
@@ -128,8 +128,8 @@ endif
 proparray.o: proparray.c ucd.h libucd_int.h gen/proparray.c
 proparray.lo: proparray.c ucd.h libucd_int.h gen/proparray.c
 
-mk_ucstoname_tab.ho: gen/ucstoname_hash.h
-mk_nametoucs_tab.ho: gen/nametoucs_hash.h
+mk_ucstoname_tab.ho: mk_ucstoname_tab.c gen/ucstoname_hash.h
+mk_nametoucs_tab.ho: mk_nametoucs_tab.c gen/nametoucs_hash.h
 
 gen/ucstoname_tab.o: gen/ucstoname_tab.c libucd_int.h
 gen/ucstoname_tab.lo: gen/ucstoname_tab.c libucd_int.h
@@ -139,3 +139,6 @@ gen/nametoucs_tab.lo: gen/nametoucs_tab.c libucd_int.h
 
 gen/nameslist_dict.o: gen/nameslist_dict.c
 gen/nameslist_dict.lo: gen/nameslist_dict.c
+
+nametoucs.o: nametoucs.c libucd_int.h gen/nametoucs_hash.h
+nametoucs.lo: nametoucs.c libucd_int.h gen/nametoucs_hash.h
diff --git a/compiler.h b/compiler.h
new file mode 100644
index 0000000..a975a83
--- /dev/null
+++ b/compiler.h
@@ -0,0 +1,16 @@
+#ifndef LIBUCD_COMPILER_H
+#define LIBUCD_COMPILER_H
+
+#if defined(__GNUC__)
+
+#define ALIGNED(x) __attribute__((aligned(x)))
+#define noreturn void __attribute__((noreturn))
+
+#else
+
+#define ALIGNED(x)
+#define noreturn void
+
+#endif
+
+#endif /* LIBUCD_COMPILER_H */
diff --git a/libucd_int.h b/libucd_int.h
index 58df27a..675ab79 100644
--- a/libucd_int.h
+++ b/libucd_int.h
@@ -10,6 +10,7 @@
 #include <stdlib.h>
 #include "ucd.h"
 #include "int24.h"
+#include "compiler.h"
 
 extern const char _libucd_hangul_jamo_l[][4];
 extern const char _libucd_hangul_jamo_v[][4];
@@ -39,16 +40,19 @@ struct _libucd_property_array {
   unsigned numeric_type         :2;
   unsigned bidi_class           :5;
   unsigned /* unused */         :1;
-};
+} ALIGNED(32);
+extern const struct _libucd_property_array _libucd_property_array[];
 
 struct _libucd_ucstoname_tab {
   int24    ucs;
   uint24   names_offset;
   uint16_t proparray_offset;
-};
+} ALIGNED(8);
+extern const struct _libucd_ucstoname_tab _libucd_ucstoname_tab[];
 
 struct _libucd_nametoucs_tab {
   int24	   ucs;
 };
+extern const struct _libucd_nametoucs_tab _libucd_nametoucs_tab[];
 
 #endif
diff --git a/nametoucs.c b/nametoucs.c
new file mode 100644
index 0000000..7a54600
--- /dev/null
+++ b/nametoucs.c
@@ -0,0 +1,42 @@
+#include <stdlib.h>
+#include <string.h>
+#include <stdlib.h>
+#include <inttypes.h>
+#include "libucd_int.h"
+#include "gen/nametoucs_hash.h"
+
+static uint32_t prehash(const char *str)
+{
+  uint32_t hash = PHASHSALT;
+  const uint8_t *p = (const uint8_t *)str;
+
+  while ( *p )
+    hash = (hash ^ *p++) + ((hash << 27)+(hash >> 5));
+
+  return hash;
+}
+
+/* This returns a candidate UCS for a given name. */
+int32_t _libucd_ucs_to_name(const char *name)
+{
+  int32_t ucs;
+  uint32_t hash;
+
+  if ( !strncmp(name, "CJK UNIFIED IDEOGRAPH-", 22) ) {
+    char *q;
+
+    ucs = strtol(name+22, &q, 16);
+    if ( *q || q < name+26 || q > name+27 )
+      return -1;
+  } else {
+    hash = _libucd_nametoucs_hash(prehash(name));
+    if ( hash > PHASHNKEYS )
+      return -1;
+    ucs = getint24(_libucd_nametoucs_tab[hash].ucs);
+  }
+
+  return ucs;
+}
+
+    
+