#define _SHERLOCK_INDEX_H
#include "lib/fastbuf.h"
+#include SHERLOCK_CUSTOM
#include "charset/unistream.h"
-/* Words */
+/*
+ * Words and word complexes
+ *
+ * MAX_WORD_LEN is the maximum length (measured in UTF-8 characters, excluding
+ * the terminating zero byte if there's any) of any word which may appear in the
+ * indices or in the bucket file. Naturally, the same constant also bounds
+ * the number of UCS-2 characters in a word.
+ *
+ * Caveat: If you are upcasing/downcasing the word, the UTF-8 encoding can
+ * expand, although at most twice, so you need to reserve 2*MAX_WORD_LEN bytes.
+ *
+ * MAX_COMPLEX_LEN is the upper bound on number of words in any word complex.
+ */
-#define MAX_WORD_LEN 64
+#define MAX_WORD_LEN 64 /* a multiple of 4 */
#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 */
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)
{