#define _SHERLOCK_INDEX_H
#include "lib/fastbuf.h"
+#include SHERLOCK_CUSTOM
#include "charset/unistream.h"
/* Words */
#define MAX_WORD_LEN 64
+#define MAX_COMPLEX_LEN 10
/* Word and string types are defined in lib/custom.h */
struct card_attr {
u32 card; /* Reference to card description (either oid or filepos) */
+#ifdef CONFIG_SITES
u32 site_id;
-#define INT_ATTR(t,i,o,k,g,p) t i;
- CUSTOM_ATTRS /* Include all custom attributes */
-#undef INT_ATTR
+#endif
+ CUSTOM_CARD_ATTRS /* Include all custom attributes */
byte weight;
byte flags;
byte age; /* Document age in pseudo-logarithmic units wrt. reference time */
CARD_FLAG_ACCENTED = 2, /* Document contains accented characters [scanner] */
CARD_FLAG_DUP = 4, /* Removed as a duplicate [merger] */
CARD_FLAG_MERGED = 8, /* Destination of a merge [merger] */
+ CARD_FLAG_IMAGE = 16, /* Is an image object [scanner] */
};
#define CARD_POS_SHIFT 5 /* Card positions are shifted this # of bytes to the right */
p++; \
} while (0)
+#define SKIP_TAGGED_CHAR(p) do { \
+ if (*p >= 0x80 && *p < 0xc0) \
+ { \
+ uns u = *p++; \
+ if (u >= 0xa0 && u < 0xb0 && *p >= 0x80 && *p < 0xc0) \
+ p++; \
+ } \
+ else \
+ UTF8_SKIP(p); \
+} while (0)
+
static inline uns
bget_tagged_char(struct fastbuf *f)
{