/*
* Sherlock: Data structures used in indices
*
- * (c) 2001--2003 Martin Mares <mj@ucw.cz>
+ * (c) 2001--2004 Martin Mares <mj@ucw.cz>
*/
#ifndef _SHERLOCK_INDEX_H
#define _SHERLOCK_INDEX_H
-#include "lib/fastbuf.h"
-#include SHERLOCK_CUSTOM
-#include "charset/unistream.h"
+#include "custom/lib/custom.h"
-#define INDEX_VERSION (0x32240100+sizeof(struct card_attr)) /* Increase with each incompatible change in index format */
+#define INDEX_VERSION (0x32330100+sizeof(struct card_attr)) /* Increase with each incompatible change in index format */
/*
- * Words and word complexes
+ * Words
*
* 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
*
* 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 /* a multiple of 4 */
-#define MAX_COMPLEX_LEN 10
-/* Word and string types are defined in lib/custom.h */
+/* Word and string types are defined in custom/lib/custom.h */
+
+/* Types used for storing contexts */
+
+#ifdef CONFIG_CONTEXTS
+#if CONFIG_MAX_CONTEXTS == 32768
+typedef u16 context_t;
+#define bget_context bgetw
+#define bput_context bputw
+#define GET_CONTEXT GET_U16
+#define PUT_CONTEXT PUT_U16
+#elif CONFIG_MAX_CONTEXTS == 256
+typedef byte context_t;
+#define bget_context bgetc
+#define bput_context bputc
+#define GET_CONTEXT GET_U8
+#define PUT_CONTEXT PUT_U8
+#else
+#error CONFIG_MAX_CONTEXTS set to an invalid value.
+#endif
+#else
+struct fastbuf;
+typedef struct { } context_t;
+static inline uns bget_context(struct fastbuf *b UNUSED) { return 0; }
+static inline void bput_context(struct fastbuf *b UNUSED, uns context UNUSED) { }
+#define GET_CONTEXT(p) 0
+#define PUT_CONTEXT(p,x) do {} while(0)
+#endif
/* Index card attributes */
#ifdef CONFIG_SITES
u32 site_id;
#endif
+ area_t area;
CUSTOM_CARD_ATTRS /* Include all custom attributes */
byte weight;
byte flags;
CARD_FLAG_MERGED = 8, /* Destination of a merge [merger] */
CARD_FLAG_IMAGE = 16, /* Is an image object [scanner] */
CARD_FLAG_FRAMESET = 32, /* Contains a frameset to be ignored [scanner] */
- CARD_FLAG_GIANT_CLASS = 64, /* Belongs to a very large class, subject to penalties [merger] */
+ CARD_FLAG_OVERRIDEN = 64, /* Overriden by another index [sherlockd] */
};
#define CARD_POS_SHIFT 5 /* Card positions are shifted this # of bits to the right */
#define CA_GET_FILE_TYPE(a) ((a)->type_flags >> 5)
#define CA_GET_FILE_INFO(a) ((a)->type_flags & 0x1f)
#define CA_GET_FILE_LANG(a) ((a)->type_flags & 0x80 ? 0 : CA_GET_FILE_INFO(a))
-#define FILETYPE_ATTRS SMALL_SET_ATTR(ftype, FILETYPE, CA_GET_FILE_TYPE, ext_ft_parse)
#define MAX_FILE_TYPES 8
+#define FILETYPE_IS_TEXT(f) ((f) < 4)
byte *ext_ft_parse(u32 *dest, byte *value, uns intval);
extern byte *custom_file_type_names[MAX_FILE_TYPES];
+#define FILETYPE_STAT_VARS uns matching_per_type[MAX_FILE_TYPES];
+#define FILETYPE_SHOW_STATS(q,f) ext_ft_show(q,f)
+#define FILETYPE_INIT_STATS(q) bzero(q->matching_per_type, sizeof(q->matching_per_type))
+#ifdef CONFIG_COUNT_ALL_FILETYPES
+#define FILETYPE_ATTRS LATE_SMALL_SET_ATTR(ftype, FILETYPE, CA_GET_FILE_TYPE, ext_ft_parse)
+#define FILETYPE_EARLY_STATS(q,a) q->matching_per_type[CA_GET_FILE_TYPE(a)]++
+#define FILETYPE_LATE_STATS(q,a)
+#else
+#define FILETYPE_ATTRS SMALL_SET_ATTR(ftype, FILETYPE, CA_GET_FILE_TYPE, ext_ft_parse)
+#define FILETYPE_EARLY_STATS(q,a)
+#define FILETYPE_LATE_STATS(q,a) q->matching_per_type[CA_GET_FILE_TYPE(a)]++
+#endif
#else
#define FILETYPE_ATTRS
+#define FILETYPE_STAT_VARS
+#define FILETYPE_INIT_STATS(q)
+#define FILETYPE_EARLY_STATS(q,a)
+#define FILETYPE_LATE_STATS(q,a)
+#define FILETYPE_SHOW_STATS(q,f)
#endif
#ifdef CONFIG_LANG
#define LANG_ATTRS
#endif
-#define EXTENDED_ATTRS CUSTOM_ATTRS LANG_ATTRS /* Beware, FILETYPE_ATTRS are handled separately */
+#ifdef CONFIG_AREAS
+#define CA_GET_AREA(a) ((a)->area)
+#define SPLIT_ATTRS INT_ATTR(area, AREA, CA_GET_AREA, ext_area_parse)
+byte *ext_area_parse(u32 *dest, byte *value, uns intval);
+#else
+#define SPLIT_ATTRS
+#endif
+
+/*
+ * A list of all extended attributes: custom attributes and also some
+ * built-in attributes treated in the same way.
+ */
+
+#define EXTENDED_ATTRS CUSTOM_ATTRS FILETYPE_ATTRS LANG_ATTRS SPLIT_ATTRS
+
+/*
+ * A list of all statistics collectors, also composed of custom parts
+ * and built-in parts.
+ */
+
+#ifndef CUSTOM_STAT_VARS
+#define CUSTOM_STAT_VARS
+#define CUSTOM_INIT_STATS(q)
+#define CUSTOM_EARLY_STATS(q,a)
+#define CUSTOM_LATE_STATS(q,a)
+#define CUSTOM_SHOW_STATS(q,f)
+#endif
+
+#define EXTENDED_STAT_VARS CUSTOM_STAT_VARS FILETYPE_STAT_VARS
+#define EXTENDED_INIT_STATS(q) CUSTOM_INIT_STATS(q) FILETYPE_INIT_STATS(q)
+#define EXTENDED_EARLY_STATS(q,a) CUSTOM_EARLY_STATS(q,a) FILETYPE_EARLY_STATS(q,a)
+#define EXTENDED_LATE_STATS(q,a) CUSTOM_LATE_STATS(q,a) FILETYPE_LATE_STATS(q,a)
+#define EXTENDED_SHOW_STATS(q,f) CUSTOM_SHOW_STATS(q,f) FILETYPE_SHOW_STATS(q,f)
/* String fingerprints */
static inline u32
fp_hash(struct fingerprint *fp)
{
- return fp->hash[0] ^ fp->hash[1] ^ fp->hash[2] ^ fp->hash[3];
+ return (fp->hash[0] << 24) | (fp->hash[1] << 16) | (fp->hash[2] << 8) | fp->hash[3];
}
+/* The card fingerprints */
+
+struct card_print {
+ struct fingerprint fp;
+ u32 cardid;
+};
+
/* URL keys */
+#define URL_KEY_BUF_SIZE (3*MAX_URL_SIZE)
byte *url_key(byte *url, byte *buf);
void url_fingerprint(byte *url, struct fingerprint *fp);
-
-/* Reading of tagged text (Unicode values, tags mapped to 0x80000000 and higher) */
-
-#define GET_TAGGED_CHAR(p,u) do { \
- u = *p; \
- if (u >= 0xc0) \
- GET_UTF8_CHAR(p,u); \
- else if (u >= 0x80) \
- { \
- p++; \
- if (u >= 0xb0) \
- { \
- ASSERT(u == 0xb0); \
- u += 0x80020000; \
- } \
- else if (u >= 0xa0) \
- { \
- ASSERT(*p >= 0x80 && *p <= 0xbf); \
- u = 0x80010000 + ((u & 0x0f) << 6) + (*p++ & 0x3f); \
- } \
- else \
- u += 0x80000000; \
- } \
- else \
- 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)
-{
- uns u = bgetc(f);
- if ((int)u < 0x80)
- ;
- else if (u < 0xc0)
- {
- if (u >= 0xb0)
- {
- ASSERT(u == 0xb0);
- u += 0x80020000;
- }
- else if (u >= 0xa0)
- {
- uns v = bgetc(f);
- ASSERT(v >= 0x80 && v <= 0xbf);
- u = 0x80010000 + ((u & 0x0f) << 6) + (v & 0x3f);
- }
- else
- u += 0x80000000;
- }
- else
- {
- bungetc(f);
- u = bget_utf8(f);
- }
- return u;
-}
+void url_key_init(void);
/* Conversion of document age from seconds to our internal units */