X-Git-Url: http://git.chise.org/gitweb/?a=blobdiff_plain;f=chise.c;h=5d382c14ed37e8fab537556870485a95394a3642;hb=c334c5225035cef665c4e6c55d4167f065026d67;hp=6e8bb4650d0a09b15f87e907100d81ed45abcdb7;hpb=929758a18cec876b64c0d3d66f009ce7b5464965;p=chise%2Flibchise.git

diff --git a/chise.c b/chise.c
index 6e8bb46..5d382c1 100644
--- a/chise.c
+++ b/chise.c
@@ -2,10 +2,222 @@
 #  include "config.h"
 #endif
 
+#ifndef HAVE_STRNLEN
+/* original in mysql, strings/strnlen.c.
+uint strnlen(register const char *s, register uint maxlen)
+{
+  const char *end= (const char *)memchr(s, '\0', maxlen);
+  return end ? (uint) (end - s) : maxlen;
+}
+*/
+static inline int
+strnlen (register const char *s, register int maxlen)
+{
+  const char *end= (const char *)memchr(s, '\0', maxlen);
+  return end ? (int) (end - s) : maxlen;
+}
+#endif
+
 #include "chise.h"
 
 #define xzero(lvalue) ((void) memset (&(lvalue), '\0', sizeof (lvalue)))
 
+CHISE_Char_ID
+chise_char_id_parse_c_string (unsigned char *str, int len)
+{
+  int i = 0;
+
+  if ( (len >= 2) && (str[i++] == '?') )
+    {
+      unsigned char c = str[i++];
+      int counter;
+      CHISE_Char_ID cid;
+
+      if (c == '\\')
+	{
+	  if (len < 3)
+	    return -1;
+	  c = str[i++];
+	  if (c == '^')
+	    {
+	      if (len < 4)
+		return -1;
+	      c = str[i++];
+	      if (c == '?')
+		return 0x7F;
+	      else
+		return c & (0x80 | 0x1F);
+	    }
+	}
+      if ( c < 0xC0 )
+	{
+	  cid = c;
+	  counter = 0;
+	}
+      else if ( c < 0xE0 )
+	{
+	  cid = c & 0x1f;
+	  counter = 1;
+	}
+      else if ( c < 0xF0 )
+	{
+	  cid = c & 0x0f;
+	  counter = 2;
+	}
+      else if ( c < 0xF8 )
+	{
+	  cid = c & 0x07;
+	  counter = 3;
+	}
+      else if ( c < 0xFC )
+	{
+	  cid = c & 0x03;
+	  counter = 4;
+	}
+      else
+	{
+	  cid = c & 0x01;
+	  counter = 5;
+	}
+
+      if (counter + 2 <= len)
+	{
+	  int j;
+
+	  for (j = 0; j < counter; j++)
+	    cid = (cid << 6) | (str[j + i] & 0x3F);
+	  return cid;
+	}
+    }
+  return -1;
+}
+
+int
+chise_format_char_id (CHISE_Char_ID cid, unsigned char *dest, int len)
+{
+  int i = 0;
+
+  dest[i++] = '?';
+  if (cid == '\t')
+    {
+      dest[i++] = '\\';
+      dest[i++] = 't';
+      dest[i] = '\0';
+      return i;
+    }
+  else if (cid == '\n')
+    {
+      dest[i++] = '\\';
+      dest[i++] = 'n';
+      dest[i] = '\0';
+      return i;
+    }
+  else if (cid == '\r')
+    {
+      dest[i++] = '\\';
+      dest[i++] = 'r';
+      dest[i] = '\0';
+      return i;
+    }
+  else if (cid == 0x1C)
+    {
+      dest[i++] = '\\';
+      dest[i++] = '^';
+      dest[i++] = '\\';
+      dest[i++] = '\\';
+      dest[i] = '\0';
+      return i;
+    }
+  else if (cid <= 0x1F)
+    {
+      dest[i++] = '\\';
+      dest[i++] = '^';
+      dest[i++] = '@' + cid;
+      dest[i] = '\0';
+      return i;
+    }
+  else if ( (cid == ' ') || (cid == '"') ||
+	    (cid == '#') || (cid == '\'') ||
+	    (cid == '(') || (cid == ')') ||
+	    (cid == ',') || (cid == '.') ||
+	    (cid == ';') || (cid == '?') ||
+	    (cid == '[') || (cid == '\\') ||
+	    (cid == ']') || (cid == '`') )
+    {
+      dest[i++] = '\\';
+      dest[i++] = cid;
+      dest[i] = '\0';
+      return i;
+    }
+  else if (cid <= 0x7E)
+    {
+      dest[i++] = cid;
+      dest[i] = '\0';
+      return i;
+    }
+  else if (cid == 0x7F)
+    {
+      dest[i++] = '\\';
+      dest[i++] = '^';
+      dest[i++] = '?';
+      dest[i] = '\0';
+      return i;
+    }
+  else if (cid <= 0x9F)
+    {
+      dest[i++] = '\\';
+      dest[i++] = '^';
+      dest[i++] = ((cid + '@') >> 6) | 0xC0;
+      dest[i++] = ((cid + '@') & 0x3F) | 0x80;
+      dest[i] = '\0';
+      return i;
+    }
+  else if (cid <= 0x7FF)
+    {
+      dest[i++] = (cid >> 6) | 0xC0;
+      dest[i++] = (cid & 0x3F) | 0x80;
+      dest[i] = '\0';
+      return i;
+    }
+  else if (cid <= 0xFFFF)
+    {
+      dest[i++] = (cid >> 12) | 0xE0;
+      dest[i++]= ((cid >>  6) & 0x3F) | 0x80;
+      dest[i++]=  (cid        & 0x3F) | 0x80;
+      dest[i] = '\0';
+      return i;
+    }
+  else if (cid <= 0x1FFFFF)
+    {
+      dest[i++]=  (cid >> 18) | 0xF0;
+      dest[i++]= ((cid >> 12) & 0x3F) | 0x80;
+      dest[i++]= ((cid >>  6) & 0x3F) | 0x80;
+      dest[i++]=  (cid        & 0x3F) | 0x80;
+      dest[i] = '\0';
+      return i;
+    }
+  else if (cid <= 0x3FFFFFF)
+    {
+      dest[i++]=  (cid >> 24) | 0xF8;
+      dest[i++]= ((cid >> 18) & 0x3F) | 0x80;
+      dest[i++]= ((cid >> 12) & 0x3F) | 0x80;
+      dest[i++]= ((cid >>  6) & 0x3F) | 0x80;
+      dest[i++]=  (cid        & 0x3F) | 0x80;
+      dest[i] = '\0';
+      return i;
+    }
+  else
+    {
+      dest[i++]=  (cid >> 30) | 0xFC;
+      dest[i++]= ((cid >> 24) & 0x3F) | 0x80;
+      dest[i++]= ((cid >> 18) & 0x3F) | 0x80;
+      dest[i++]= ((cid >> 12) & 0x3F) | 0x80;
+      dest[i++]= ((cid >>  6) & 0x3F) | 0x80;
+      dest[i++]=  (cid        & 0x3F) | 0x80;
+      dest[i] = '\0';
+      return i;
+    }
+}
 
 int
 chise_open_data_source (CHISE_DS *ds, CHISE_DS_Type type, char *location)
@@ -25,7 +237,7 @@ chise_close_data_source (CHISE_DS *ds)
 
 
 int
-chise_open_decoding_table (CHISE_Decoding_Table **db,
+chise_open_decoding_table (CHISE_Decoding_Table *db,
 			   CHISE_DS *ds, const char *ccs,
 			   DBTYPE real_subtype,
 			   u_int32_t accessmask, int modemask)
@@ -58,78 +270,28 @@ chise_dt_get_char (CHISE_Decoding_Table *db, int code_point)
       unsigned char *str
 	= (unsigned char *)chise_value_data (&valdatum);
       int len = strnlen (str, chise_value_size (&valdatum));
-      int i = 0;
-
-      if ( (len >= 2) && (str[i++] == '?') )
-	{
-	  unsigned char c = str[i++];
-	  int counter;
-	  CHISE_Char_ID cid;
-
-	  if (c == '\\')
-	    {
-	      if (len < 3)
-		return -1;
-	      c = str[i++];
-	      if (c == '^')
-		{
-		  if (len < 4)
-		    return -1;
-		  c = str[i++];
-		  if (c == '?')
-		    return 0x7F;
-		  else
-		    return c & (0x80 | 0x1F);
-		}
-	    }
-	  if ( c < 0xC0 )
-	    {
-	      cid = c;
-	      counter = 0;
-	    }
-	  else if ( c < 0xE0 )
-	    {
-	      cid = c & 0x1f;
-	      counter = 1;
-	    }
-	  else if ( c < 0xF0 )
-	    {
-	      cid = c & 0x0f;
-	      counter = 2;
-	    }
-	  else if ( c < 0xF8 )
-	    {
-	      cid = c & 0x07;
-	      counter = 3;
-	    }
-	  else if ( c < 0xFC )
-	    {
-	      cid = c & 0x03;
-	      counter = 4;
-	    }
-	  else
-	    {
-	      cid = c & 0x01;
-	      counter = 5;
-	    }
 
-	  if (counter + 2 <= len)
-	    {
-	      int j;
-
-	      for (j = 0; j < counter; j++)
-		cid = (cid << 6) | (str[j + i] & 0x3F);
-	      return cid;
-	    }
-	}
+      return chise_char_id_parse_c_string (str, len);
     }
   return -1;
 }
 
+int
+chise_dt_put_char (CHISE_Decoding_Table *db,
+		   int code_point, CHISE_Char_ID cid)
+{
+  CHISE_Value valdatum;
+  char key_buf[16], val_buf[8];
+
+  sprintf(key_buf, "%d", code_point);
+  chise_format_char_id (cid, val_buf, 8);
+  return chise_put_attribute_table (db, key_buf, val_buf);
+}
+
 
 
 int
-chise_open_feature_table (CHISE_Feature_Table **db,
+chise_open_feature_table (CHISE_Feature_Table *db,
 			  CHISE_DS *ds, const char *feature,
 			  DBTYPE real_subtype,
 			  u_int32_t accessmask, int modemask)
@@ -148,64 +310,46 @@ chise_close_feature_table (CHISE_Feature_Table *db)
   return -1;
 }
 
-int chise_ft_get_value (CHISE_Feature_Table *db,
-			CHISE_Char_ID cid, CHISE_Value *valdatum)
+int
+chise_ft_get_value (CHISE_Feature_Table *db,
+		    CHISE_Char_ID cid, CHISE_Value *valdatum)
 {
   unsigned char key_buf[8];
 
-  key_buf[0] = '?';
-  if (cid <= 0x7f)
-    {
-      key_buf[1] = cid;
-      key_buf[2] = '\0';
-    }
-  else if (cid <= 0x7ff)
-    {
-      key_buf[1] = (cid >> 6) | 0xc0;
-      key_buf[2] = (cid & 0x3f) | 0x80;
-      key_buf[3] = '\0';
-    }
-  else if (cid <= 0xffff)
-    {
-      key_buf[1] = (cid >> 12) | 0xe0;
-      key_buf[2]= ((cid >>  6) & 0x3f) | 0x80;
-      key_buf[3]=  (cid        & 0x3f) | 0x80;
-      key_buf[4] = '\0';
-    }
-  else if (cid <= 0x1fffff)
-    {
-      key_buf[1]=  (cid >> 18) | 0xf0;
-      key_buf[2]= ((cid >> 12) & 0x3f) | 0x80;
-      key_buf[3]= ((cid >>  6) & 0x3f) | 0x80;
-      key_buf[4]=  (cid        & 0x3f) | 0x80;
-      key_buf[5] = '\0';
-    }
-  else if (cid <= 0x3ffffff)
-    {
-      key_buf[1]=  (cid >> 24) | 0xf8;
-      key_buf[2]= ((cid >> 18) & 0x3f) | 0x80;
-      key_buf[3]= ((cid >> 12) & 0x3f) | 0x80;
-      key_buf[4]= ((cid >>  6) & 0x3f) | 0x80;
-      key_buf[5]=  (cid        & 0x3f) | 0x80;
-      key_buf[6] = '\0';
-    }
-  else
+  chise_format_char_id (cid, key_buf, 8);
+  return chise_get_attribute_table (db, key_buf, valdatum);
+}
+
+void
+chise_ft_iterate (CHISE_Feature_Table *ft,
+		  int (*func) (CHISE_Feature_Table *ft,
+			       CHISE_Char_ID cid, CHISE_Value *valdatum))
+{
+  DBT keydatum, valdatum;
+  DBC *dbcp;
+  int status;
+
+  xzero (keydatum);
+  xzero (valdatum);
+
+  status = ft->dbp->cursor (ft->dbp, NULL, &dbcp, 0);
+  for (status = dbcp->c_get (dbcp, &keydatum, &valdatum, DB_FIRST);
+       status == 0;
+       status = dbcp->c_get (dbcp, &keydatum, &valdatum, DB_NEXT))
     {
-      key_buf[1]=  (cid >> 30) | 0xfc;
-      key_buf[2]= ((cid >> 24) & 0x3f) | 0x80;
-      key_buf[3]= ((cid >> 18) & 0x3f) | 0x80;
-      key_buf[4]= ((cid >> 12) & 0x3f) | 0x80;
-      key_buf[5]= ((cid >>  6) & 0x3f) | 0x80;
-      key_buf[6]=  (cid        & 0x3f) | 0x80;
-      key_buf[7] = '\0';
+      unsigned char *key_str = (unsigned char *)keydatum.data;
+      int key_len = strnlen (key_str, keydatum.size);
+      CHISE_Char_ID key = chise_char_id_parse_c_string (key_str, key_len);
+      int ret;
+
+      if (ret = func (ft, key, &valdatum))
+	break;
     }
-  return
-    chise_get_attribute_table (db, key_buf, valdatum);
+  dbcp->c_close (dbcp);
 }
 
-
 int
-chise_open_attribute_table (CHISE_Attribute_Table **db,
+chise_open_attribute_table (CHISE_Attribute_Table *ft,
 			    const char *db_dir,
 			    const char *encoding, const char *feature,
 			    DBTYPE real_subtype,
@@ -213,16 +357,17 @@ chise_open_attribute_table (CHISE_Attribute_Table **db,
 {
   DB* dbase;
   int status;
-  int len;
+  int len, flen, i;
   int size;
-  char *db_file_name;
+  char *db_file_name, *sp;
 
   status = db_create (&dbase, NULL, 0);
   if (status)
     return -1;
 
   len = strlen (db_dir);
-  size = len + strlen (encoding) + strlen (feature) + 4;
+  flen = strlen (feature);
+  size = len + strlen (encoding) + flen * 3 + 4;
   db_file_name = alloca (size);
   strcpy (db_file_name, db_dir);
   if (db_file_name[len - 1] != '/')
@@ -232,7 +377,21 @@ chise_open_attribute_table (CHISE_Attribute_Table **db,
     }
   strcat (db_file_name, encoding);
   strcat (db_file_name, "/");
-  strcat (db_file_name, feature);
+  /* strcat (db_file_name, feature); */
+  sp = &db_file_name[strlen (db_file_name)];
+  for (i = 0; i < flen; i++)
+    {
+      int c = feature[i];
+
+      if ( (c == '/') || (c == '%') )
+	{
+	  sprintf (sp, "%%%02X", c);
+	  sp += 3;
+	}
+      else
+	*sp++ = c;
+    }
+  *sp = '\0';
   status = dbase->open (dbase, db_file_name, NULL,
 			real_subtype, accessmask, modemask);
   if (status)
@@ -240,23 +399,23 @@ chise_open_attribute_table (CHISE_Attribute_Table **db,
       dbase->close (dbase, 0);
       return -1;
     }
-  *db = dbase;
+  ft->dbp = dbase;
   return 0;
 }
 
 int
-chise_close_attribute_table (CHISE_Attribute_Table *db)
+chise_close_attribute_table (CHISE_Attribute_Table *ft)
 {
-  if (db)
+  if (ft->dbp)
     {
-      db->sync  (db, 0);
-      db->close (db, 0);
+      ft->dbp->sync  (ft->dbp, 0);
+      ft->dbp->close (ft->dbp, 0);
     }
   return 0;
 }
 
 int
-chise_get_attribute_table (CHISE_Attribute_Table *db,
+chise_get_attribute_table (CHISE_Attribute_Table *ft,
 			   char *key, CHISE_Value *valdatum)
 {
   DBT keydatum;
@@ -269,6 +428,27 @@ chise_get_attribute_table (CHISE_Attribute_Table *db,
   keydatum.data = key;
   keydatum.size = strlen (key);
 
-  status = db->get (db, NULL, &keydatum, valdatum, 0);
+  status = ft->dbp->get (ft->dbp, NULL, &keydatum, valdatum, 0);
+  return status;
+}
+
+int
+chise_put_attribute_table (CHISE_Attribute_Table *ft,
+			   char *key, char *value)
+{
+  DBT keydatum, valdatum;
+  int status = 0;
+
+  /* DB Version 2 requires DBT's to be zeroed before use. */
+  xzero (keydatum);
+  xzero (valdatum);
+
+  keydatum.data = key;
+  keydatum.size = strlen (key);
+
+  valdatum.data = value;
+  valdatum.size = strlen (value);
+
+  status = ft->dbp->put (ft->dbp, NULL, &keydatum, &valdatum, 0);
   return status;
 }