Produce table for the ucstoname hash

5 files changed, 177 insertions, 16 deletions

diff --git a/Makefile b/Makefile
index 47bdf29..625545c 100644
--- a/Makefile
+++ b/Makefile
@@ -1,14 +1,14 @@
 CC	     = cc
-CFLAGS	     = -g -O -I.
+CFLAGS	     = -g -O -I. -W -Wall
 PICFLAGS     = -fPIC
 LDFLAGS      =
 
 HOST_CC      = cc
-HOST_CFLAGS  = -g -O
+HOST_CFLAGS  = -g -O -I. -W -Wall
 HOST_LDFLAGS =
 HOST_LIBS    =
 
-PERL         = perl
+PERL         = time perl
 
 #
 # These are the files produced by convert_ucd.pl
@@ -18,29 +18,40 @@ CVT_FILES = gen/jamo.c gen/nameslist.tab gen/nametoucs.keys gen/nametoucs.tab \
 
 # -----------------------------------------------------------------------
 
-.SUFFIXES: .c .o .lo .s .ls .i .li .cc .h
+.SUFFIXES: .c .o .lo .s .ls .i .ho .hs .hi .cc .h
 .c.o:
 	$(CC) $(CFLAGS) -c -o $@ $<
+.c.lo:
+	$(CC) $(CFLAGS) $(PICFLAGS) -c -o $@ $<
 .c.s:
 	$(CC) $(CFLAGS) -S -o $@ $<
+.c.ls:
+	$(CC) $(CFLAGS) $(PICFLAGS) -S -o $@ $<
 .c.i:
 	$(CC) $(CFLAGS) -E -o $@ $<
-.c.lo:
+.c.ho:
 	$(HOST_CC) $(HOST_CFLAGS) -c -o $@ $<
-.c.ls:
+.c.hs:
 	$(HOST_CC) $(HOST_CFLAGS) -S -o $@ $<
-.c.li:
+.c.hi:
 	$(HOST_CC) $(HOST_CFLAGS) -E -o $@ $<
 
 # -----------------------------------------------------------------------
 
-all : perfect/perfect gen/nametoucs_hash.o gen/ucstoname_hash.o \
-	proparray.o
+LIBSRC = proparray.c gen/nametoucs_hash.c gen/ucstoname_hash.c \
+	 gen/jamo.c gen/nameslist.c gen/nameslist_dict.c \
+	 gen/ucstoname_tab.c
+
+LIBOBJ = $(patsubst %.c,%.o,$(LIBSRC))
+SO_OBJ = $(patsubst %.c,%.lo,$(LIBSRC))
+
+# -----------------------------------------------------------------------
 
+all : $(LIBOBJ) $(SO_OBJ)
 
 clean:
 	rm -rf gen
-	rm -f *.o *.a *.so *.so.*
+	rm -f *.o *.i *.*.a *.so *.so.*
 	$(MAKE) -C perfect clean
 
 # -----------------------------------------------------------------------
@@ -56,14 +67,14 @@ perfect/perfect: $(wildcard perfect/*.c perfect/*.h)
 	$(MAKE) -C perfect
 
 gen/nametoucs_hash.c: gen/nametoucs.keys perfect/perfect
-	perfect/perfect -im libucd_nametoucs_hash gen/nametoucs_hash.c \
+	perfect/perfect -im _libucd_nametoucs_hash gen/nametoucs_hash.c \
 		gen/nametoucs_hash.h < gen/nametoucs.keys 
 
 gen/nametoucs_hash.h: gen/nametoucs_hash.c
 	: Generated by side effect
 
 gen/ucstoname_hash.c: gen/ucstoname.keys perfect/perfect
-	perfect/perfect -hm libucd_ucstoname_hash gen/ucstoname_hash.c \
+	perfect/perfect -hm _libucd_ucstoname_hash gen/ucstoname_hash.c \
 		gen/ucstoname_hash.h < gen/ucstoname.keys 
 
 gen/ucstoname_hash.h: gen/ucstoname_hash.c
@@ -76,8 +87,24 @@ gen/nameslist_tab.c gen/nameslist.offset: gen/nameslist.compr
 	: Generated by side effect
 
 gen/nameslist.c: gen/nameslist.compr bin2c.pl
-	$(PERL) bin2c.pl < $< > $@ || rm -f $@
+	$(PERL) bin2c.pl _libucd_names_list < $< > $@ || rm -f $@
+
+gen/mk_nametoucs_tab: mk_nametoucs_tab.ho gen/ucstoname_hash.ho
+	$(HOST_CC) $(HOST_LDFLAGS) -o $@ $^ $(HOST_LIBS)
+
+gen/ucstoname_tab.c: gen/mk_nametoucs_tab \
+		     gen/proparrayindex gen/nameslist.offset
+	gen/mk_nametoucs_tab
 
 # -----------------------------------------------------------------------
 
 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_nametoucs_tab.ho: gen/ucstoname_hash.h
+
+gen/ucstoname_tab.o: gen/ucstoname_tab.c libucd_int.h
+gen/ucstoname_tab.lo: gen/ucstoname_tab.c libucd_int.h
+
+gen/nameslist_dict.o: gen/nameslist_dict.c
+gen/nameslist_dict.lo: gen/nameslist_dict.c
diff --git a/bin2c.pl b/bin2c.pl
index 8c56ca6..93907b9 100644
--- a/bin2c.pl
+++ b/bin2c.pl
@@ -26,7 +26,7 @@ if ( $#ARGV != 0 ) {
 
 ($table_name) = @ARGV;
 
-printf "unsigned char %s[] = {\n", $table_name;
+printf "const unsigned char %s[] = {\n", $table_name;
 
 $pos = 0;
 $linelen = 8;
diff --git a/convert_ucd.pl b/convert_ucd.pl
index 34470d4..9d85cc2 100755
--- a/convert_ucd.pl
+++ b/convert_ucd.pl
@@ -502,9 +502,9 @@ sub make_properties_array()
 	    print $fh $mine;
 	    $cnt++;
 	    $prev = $mine;
+	    printf $fhi "0x%05x $cnt\n", $c, $cnt;
 	}
 	$prop_array_position{$c} = $cnt;
-	printf $fhi "0x%05x $cnt\n", $c, $cnt;
     }
     print $fh "\t/* Total: $cnt ranges */\n";
 
diff --git a/libucd_int.h b/libucd_int.h
index c23b8ed..3da1857 100644
--- a/libucd_int.h
+++ b/libucd_int.h
@@ -17,7 +17,7 @@ extern const char _libucd_hangul_jamo_t[][4];
 
 /* This structure is exactly 32 bytes long, nice and alignable. */
 struct _libucd_property_array {
-  int32_t  ucd;			/* Wasteful but fast (used in search) */
+  int32_t  ucs;			/* Wasteful but fast (used in search) */
   uint8_t  general_category;
   uint8_t  script;
   uint8_t  numeric_value_num;
@@ -41,4 +41,10 @@ struct _libucd_property_array {
   unsigned /* unused */         :1;
 };
 
+struct _libucd_ucstoname_tab {
+  int24    ucs;
+  uint24   names_offset;
+  uint16_t proparray_offset;
+};
+
 #endif
diff --git a/mk_nametoucs_tab.c b/mk_nametoucs_tab.c
new file mode 100644
index 0000000..1205d49
--- /dev/null
+++ b/mk_nametoucs_tab.c
@@ -0,0 +1,128 @@
+#include <stdio.h>
+#include <string.h>
+#include <stdlib.h>
+#include <inttypes.h>
+#include "gen/ucstoname_hash.h"
+
+#define UCS_CNT 0x110000
+
+const char *program;
+
+void die(const char *msg)
+{
+  fprintf(stderr, "%s: %s\n", program, msg);
+  exit(1);
+}
+
+static const char *int24str(int32_t n)
+{
+  static char str[17];
+
+  sprintf(str, "{0x%02x,0x%02x,0x%02x}",
+	  (n & 0xff),
+	  ((n >> 8) & 0xff),
+	  ((n >> 16) & 0xff));
+  return str;
+}
+
+static int proparrayindex[UCS_CNT];
+static void read_proparrayindex(void)
+{
+  FILE *f = fopen("gen/proparrayindex", "rt");
+  if ( !f )
+    die("could not open gen/proparrayindex");
+  int last = 0, prev_ix = -1;
+  int curr, i, ix;
+
+  while ( fscanf(f, "%x %d\n", &curr, &ix) == 2 ) {
+    for ( i = last ; i < curr ; i++ )
+      proparrayindex[i] = prev_ix;
+    last = curr;
+    prev_ix = ix;
+  }
+
+  for ( i = last ; i < UCS_CNT ; i++ )
+    proparrayindex[i] = prev_ix;
+
+  fclose(f);
+}
+
+
+static int nameslistoffset[UCS_CNT];
+static void read_nameslistoffset(void)
+{
+  FILE *f = fopen("gen/nameslist.offset", "rt");
+  if ( !f )
+    die("could not open gen/nameslist.offset");
+  int curr, offset;
+  int i;
+
+  for ( i = 0 ; i < UCS_CNT ; i++ )
+    nameslistoffset[i] = -1;
+
+  while ( fscanf(f, "%x %d\n", &curr, &offset) == 2 )
+    nameslistoffset[curr] = offset;
+    
+  fclose(f);
+}
+  
+
+int32_t hash_to_ucs[PHASHNKEYS];
+static void compute_hash_to_ucs(void)
+{
+  uint32_t hash;
+  int i;
+
+  for ( i = 0 ; i < PHASHNKEYS ; i++ )
+    hash_to_ucs[i] = -1;
+
+  for ( i = 0 ; i < UCS_CNT ; i++ ) {
+    if ( nameslistoffset[i] != -1 ) {
+      hash = _libucd_ucstoname_hash(i);
+
+      if ( hash >= PHASHNKEYS )
+	die("hash not minimal");
+
+      if ( hash_to_ucs[hash] != -1 )
+	die("hash collision");
+
+      hash_to_ucs[hash] = i;
+    }
+  }
+}
+
+
+static void make_ucstoname_tab(void)
+{
+  FILE *f = fopen("gen/ucstoname_tab.c", "wt");
+  if ( !f )
+    die("could not create gen/ucstoname_tab.c");
+  int i;
+  int32_t ucs;
+
+  fprintf(f,
+	  "#include \"libucd_int.h\"\n"
+	  "const struct _libucd_ucstoname_tab _libucd_ucstoname_tab[] =\n"
+	  "{\n");
+
+  for ( i = 0 ; i < PHASHNKEYS ; i++ ) {
+    ucs = hash_to_ucs[i];
+    fprintf(f, "\t{ %s, ", int24str(ucs));
+    fprintf(f, "%s, ", int24str(nameslistoffset[ucs]));
+    fprintf(f, "%d },\n", proparrayindex[ucs]);
+  }
+  fprintf(f, "};\n");
+  fclose(f);
+}
+
+
+int main(void)
+{
+  read_proparrayindex();
+  read_nameslistoffset();
+  compute_hash_to_ucs();
+  make_ucstoname_tab();
+
+  return 0;
+}
+