Defined a GET_TAGGED_CHAR macro to read our internal representation

[libucw.git] / lib / index.h

diff --git a/lib/index.h b/lib/index.h
index 33e7ef733cfea07af3bf8afe975ca39d50fddf23..70a8186d7e2eaf6078811ba45cd4ff4264acba34 100644 (file)
--- a/lib/index.h
+++ b/lib/index.h
@@ -4,6 +4,8 @@
  *     (c) 2001 Martin Mares <mj@ucw.cz>
  */
 
  *     (c) 2001 Martin Mares <mj@ucw.cz>
  */
 

+#define CLAMP(x,min,max) ({ int _t=x; (_t < min) ? min : (_t > max) ? max : _t; })
+

 /* Words */
 
 #define MAX_WORD_LEN           64
 /* Words */
 
 #define MAX_WORD_LEN           64


@@ -11,6 +13,7 @@
 /* Word types */
 
 enum word_type {
 /* Word types */
 
 enum word_type {

+  WT_RESERVED,                         /* Reserved word type */

   WT_TEXT,                             /* Ordinary text */
   WT_EMPH,                             /* Emphasized text */
   WT_SMALL,                            /* Small font */
   WT_TEXT,                             /* Ordinary text */
   WT_EMPH,                             /* Emphasized text */
   WT_SMALL,                            /* Small font */


@@ -21,3 +24,61 @@ enum word_type {
   WT_META,                             /* Various meta-information */
   WT_ALT                               /* Alternate texts for graphical elements */
 };
   WT_META,                             /* Various meta-information */
   WT_ALT                               /* Alternate texts for graphical elements */
 };

+
+/* String types */
+
+enum string_type {
+  ST_RESERVED,                         /* Reserved string type */
+  ST_URL,                              /* URL of the document */
+  ST_HOST,                             /* Host name */
+  ST_DOMAIN,                           /* Domain name */
+  ST_REF,                              /* URL reference */
+  ST_BACKREF,                          /* Back-reference (frame or redirect source) */
+};
+
+/* Index card attributes */
+
+struct card_attr {
+  u32 card;                            /* Reference to card description (either oid or filepos) */
+  u32 site_id;
+  byte weight;
+  byte flags;
+  byte rfu[2];
+};
+
+enum card_flag {
+  CARD_FLAG_EMPTY = 1,                 /* Empty document (redirect, robot file etc.) [scanner] */
+  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] */
+};
+
+#define CARD_POS_SHIFT 5               /* Card positions are shifted this # of bytes to the right */
+
+/* String fingerprints */
+
+struct fingerprint {
+  byte hash[12];
+};
+
+void fingerprint(byte *string, 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(p,u);                                             \
+  else if (u >= 0x80)                                          \
+    {                                                          \
+      p++;                                                     \
+      if (u >= 0xb0)                                           \
+       u += 0x80020000;                                        \
+      else if (u >= 0xa0)                                      \
+       u = 0x80010000 + ((u & 0x0f) << 6) + (*p++ & 0x3f);     \
+      else                                                     \
+       u += 0x80000000;                                        \
+    }                                                          \
+  else                                                         \
+    p++;                                                       \
+} while (0)