#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)
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
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 *dbp,
+ int (*func) (CHISE_Feature_Table *db,
+ CHISE_Char_ID cid, CHISE_Value *valdatum))
+{
+ DBT keydatum, valdatum;
+ DBC *dbcp;
+ int status;
+
+ xzero (keydatum);
+ xzero (valdatum);
+
+ status = dbp->cursor (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 (dbp, 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,
const char *db_dir,
status = db->get (db, NULL, &keydatum, valdatum, 0);
return status;
}
+
+int
+chise_put_attribute_table (CHISE_Attribute_Table *db,
+ 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 = db->put (db, NULL, &keydatum, &valdatum, 0);
+ return status;
+}
projects / chise / libchise.git / blobdiff