X-Git-Url: http://mj.ucw.cz/gitweb/?a=blobdiff_plain;f=lib%2Findex.h;h=7a462942349cd93b1f351c36008dddc64c46d49b;hb=36fac2a8ffbf1275176e5eae0a9397288417bf0d;hp=8bfbe88d4d6862f58118de370d52f41d3d483553;hpb=7a84ff60630fb4211e60320843955ff2cf013ac2;p=libucw.git

diff --git a/lib/index.h b/lib/index.h
index 8bfbe88d..7a462942 100644
--- a/lib/index.h
+++ b/lib/index.h
@@ -1,7 +1,7 @@
 /*
  *	Sherlock: Data structures used in indices
  *
- *	(c) 2001--2002 Martin Mares <mj@ucw.cz>
+ *	(c) 2001--2003 Martin Mares <mj@ucw.cz>
  */
 
 #ifndef _SHERLOCK_INDEX_H
@@ -11,9 +11,21 @@
 #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 */
@@ -34,8 +46,12 @@ struct card_attr {
   CUSTOM_CARD_ATTRS			/* Include all custom attributes */
   byte weight;
   byte flags;
+#ifdef CONFIG_LASTMOD
   byte age;				/* Document age in pseudo-logarithmic units wrt. reference time */
-  // byte rfu[1];			/* If no custom attributes are defined */
+#endif
+#ifdef CONFIG_FILETYPE
+  byte type_flags;			/* File type flags (see below) */
+#endif
 };
 
 enum card_flag {
@@ -44,9 +60,23 @@ enum card_flag {
   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] */
+  CARD_FLAG_FRAMESET = 32,		/* Contains a frameset [scanner] */
 };
 
-#define CARD_POS_SHIFT 5		/* Card positions are shifted this # of bytes to the right */
+#define CARD_POS_SHIFT 5		/* Card positions are shifted this # of bits to the right */
+
+/*
+ *  We store document type and several other properties in card_attr->type_flags.
+ *  Here we define only the basic structure, the details and also how to match the
+ *  types are defined in custom.h.
+ *
+ *  bits 7--5	file type: (0-3: text types, 4-7: other types, defined by custom.h)
+ *  bits 4--0	type-dependent information, for text types it's document language code
+ */
+
+#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))
 
 /* String fingerprints */