X-Git-Url: http://mj.ucw.cz/gitweb/?a=blobdiff_plain;f=lib%2Findex.h;h=6579fc4d09aec556531726687a2ef1a3715272f0;hb=89556b90f1afb9b7ebe247c05721e0a04a7232c0;hp=0bd9a63583a1b26df843c02394ced78b28efbf5c;hpb=13bc5e1c30abb77c6e1120b2e0bf50cd0ab94a6a;p=libucw.git

diff --git a/lib/index.h b/lib/index.h
index 0bd9a635..6579fc4d 100644
--- a/lib/index.h
+++ b/lib/index.h
@@ -1,23 +1,39 @@
 /*
  *	Sherlock: Data structures used in indices
  *
- *	(c) 2001 Martin Mares <mj@ucw.cz>
+ *	(c) 2001--2002 Martin Mares <mj@ucw.cz>
  */
 
+#ifndef _SHERLOCK_INDEX_H
+#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 */
 
+/* Global index parameters */
+
+struct index_params {
+  sh_time_t ref_time;			/* Reference time (for document ages etc.) */
+};
+
 /* Index card attributes */
 
 struct card_attr {
   u32 card;				/* Reference to card description (either oid or filepos) */
   u32 site_id;
+  CUSTOM_CARD_ATTRS			/* Include all custom attributes */
   byte weight;
   byte flags;
-  byte rfu[2];
+  byte age;				/* Document age in pseudo-logarithmic units wrt. reference time */
+  // byte rfu[1];			/* If no custom attributes are defined */
 };
 
 enum card_flag {
@@ -25,6 +41,7 @@ enum card_flag {
   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 */
@@ -54,8 +71,7 @@ fp_hash(struct fingerprint *fp)
       p++;							\
       if (u >= 0xb0)						\
         {							\
-	  if (u != 0xb0)					\
-            ASSERT(0);						\
+	  ASSERT(u == 0xb0);					\
 	  u += 0x80020000;					\
         }							\
       else if (u >= 0xa0)					\
@@ -69,3 +85,58 @@ fp_hash(struct fingerprint *fp)
   else								\
     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;
+}
+
+/* Conversion of document age from seconds to our internal units */
+
+static inline int
+convert_age(sh_time_t lastmod, sh_time_t reftime)
+{
+  sh_time_t age;
+  if (reftime < lastmod)		/* past times */
+    return -1;
+  age = (reftime - lastmod) / 3600;
+  if (age < 48)				/* last 2 days: 1 hour resolution */
+    return age;
+  age = (age-48) / 24;
+  if (age < 64)				/* next 64 days: 1 day resolution */
+    return 48 + age;
+  age = (age-64) / 7;
+  if (age < 135)			/* next 135 weeks: 1 week resolution */
+    return 112 + age;
+  age = (age-135) / 52;
+  if (age < 8)				/* next 8 years: 1 year resolution */
+    return 247 + age;
+  return 255;				/* then just "infinite future" */
+}
+
+#endif