Add methods .hasprefix(), .nmatches(), .matched() and .which().

README.rst

@@ -26,7 +26,11 @@ CONTENTS
 * :ref:`obj_set`
 * :ref:`func_set.add`
 * :ref:`func_set.debug`
+* :ref:`func_set.hasprefix`
 * :ref:`func_set.match`
+* :ref:`func_set.matched`
+* :ref:`func_set.nmatches`
+* :ref:`func_set.which`
 * :ref:`func_version`
 
 
@@ -107,6 +111,68 @@ Example::
   }
 
 
+.. _func_set.hasprefix:
+
+BOOL xset.hasprefix(STRING)
+---------------------------
+
+Returns ``true`` if and only if ...
+
+Example::
+
+  if (myset.hasprefix(req.url)) {
+     call do_if_prefix_matched;
+  }
+
+
+.. _func_set.nmatches:
+
+INT xset.nmatches()
+-------------------
+
+Returns ...
+
+Example::
+
+  if (myset.hasprefix(req.url)) {
+     # ...
+  }
+
+
+.. _func_set.matched:
+
+BOOL xset.matched(INT)
+----------------------
+
+Returns ...
+
+Example::
+
+  if (myset.hasprefix(req.url)) {
+     # ...
+  }
+
+
+.. _func_set.which:
+
+INT xset.which(ENUM select)
+---------------------------
+
+::
+
+      INT xset.which(
+            ENUM {UNIQUE, FIRST, LAST, SHORTEST, LONGEST} select=UNIQUE
+      )
+
+Returns ...
+
+Example::
+
+  if (myset.hasprefix(req.url)) {
+     # ...
+  }
+
+
 .. _func_set.debug:
 
 STRING xset.debug()

src/patricia.c

@@ -256,6 +256,64 @@ PT_Lookup(const struct pt_y * const restrict root,
 	return (UINT_MAX);
 }
 
+static int
+pt_search(const struct pt_y * const restrict y,
+	  char * const restrict * const restrict strings,
+	  const unsigned char * restrict c,
+	  struct match_data * const restrict match)
+{
+	const unsigned char * restrict s;
+	unsigned short i;
+
+	if (y == NULL)
+		return (0);
+
+	CHECK_OBJ(y, PT_Y_MAGIC);
+	s = (unsigned char *)strings[y->idx] + y->off;
+	for (i = 0; *c != '\0' && i < y->len && s[i] == *c; i++)
+		c++;
+
+	if (s[i] == '\0') {
+		if (match->n == match->limit)
+			return (-1);
+		match->indices[match->n] = y->idx;
+		match->n++;
+		if (y->idx < match->min)
+			match->min = y->idx;
+		if (y->idx > match->max)
+			match->max = y->idx;
+	}
+
+	if (i < y->len)
+		return (0);
+
+	if (pt_search(y->leaf[0], strings, c, match) != 0)
+		return (-1);
+	if (pt_search(y->leaf[1], strings, c, match) != 0)
+		return (-1);
+
+	return (0);
+}
+
+int
+PT_Prefixes(const struct pt_y * const restrict root,
+	    char * const restrict * const restrict strings,
+	    const char * const restrict subject,
+	    struct match_data * const restrict match)
+{
+	CHECK_OBJ_NOTNULL(match, MATCH_DATA_MAGIC);
+	AN(match->indices);
+	AN(match->limit);
+	AN(strings);
+	AN(subject);
+
+	match->n = 0;
+	match->min = UINT_MAX;
+	match->max = 0;
+
+	return (pt_search(root, strings, (unsigned char *)subject, match));
+}
+
 void
 PT_Free(struct pt_y *y)
 {

src/patricia.h

@@ -34,16 +34,15 @@
 
 struct pt_y;
 
-#if 0
-struct t_stats {
+struct match_data {
 	unsigned	magic;
-#define T_STATS_MAGIC 0xf92e6603
-	double		d_mean;
-	unsigned	d_max;
-	unsigned	n_nodes;
-	unsigned	n_leaves;
-}
-#endif
+#define MATCH_DATA_MAGIC 0x0d9a845e
+	unsigned	*indices;
+	unsigned	limit;
+	unsigned	n;
+	unsigned	min;
+	unsigned	max;
+};
 
 void PT_Init(void);
 int PT_Inited(void);
@@ -52,5 +51,9 @@ int PT_Insert(struct pt_y * * restrict root, unsigned idx,
 unsigned PT_Lookup(const struct pt_y * const restrict root,
 		   char * const restrict * const restrict strings,
 		   const char * const restrict subject);
+int PT_Prefixes(const struct pt_y * const restrict root,
+		char * const restrict * const restrict strings,
+		const char * const restrict subject,
+		struct match_data * const restrict match);
 void PT_Free(struct pt_y *y);
 struct vsb * PT_Dump(struct pt_y *root, char **strings);

src/vmod_selector.c

@@ -39,6 +39,7 @@
 #include "vre.h"
 #include "cache/cache_director.h"
 
+#include "vcc_if.h"
 #include "patricia.h"
 
 #define VFAIL(ctx, fmt, ...) \
@@ -48,6 +49,9 @@
 	VSLb((ctx)->vsl, SLT_VCL_Error, "vmod selector error: " fmt,	\
 	     __VA_ARGS__)
 
+#define VERRNOMEM(ctx, fmt, ...) \
+	VERR((ctx), "out of space: " fmt,  __VA_ARGS__)
+
 struct entry {
 	unsigned	magic;
 #define VMOD_SELECTOR_ENTRY_MAGIC 0x733dbe63
@@ -205,11 +209,42 @@ vmod_set_add(VRT_CTX, struct vmod_selector_set *set, VCL_STRING member,
 	set->table[n - 1] = entry;
 }
 
+static struct match_data *
+get_match_data(VRT_CTX, struct vmod_selector_set *set, const char *method)
+{
+	struct vmod_priv *task;
+	struct match_data *match;
+
+	task = VRT_priv_task(ctx, set);
+	AN(task);
+	if (task->priv == NULL) {
+		if ((task->priv = WS_Alloc(ctx->ws, sizeof(*match))) == NULL) {
+			VERRNOMEM(ctx, "Allocating match data in %s.%s()",
+				  set->vcl_name, method);
+			return (NULL);
+		}
+		task->len = sizeof(*match);
+		task->free = NULL;
+		match = (struct match_data *)task->priv;
+		match->magic = MATCH_DATA_MAGIC;
+	}
+	else {
+		WS_Assert_Allocated(ctx->ws, task->priv, sizeof(*match));
+		CAST_OBJ(match, task->priv, MATCH_DATA_MAGIC);
+	}
+	return match;
+}
+
 VCL_BOOL
 vmod_set_match(VRT_CTX, struct vmod_selector_set *set, VCL_STRING subject)
 {
+	unsigned idx;
+	struct match_data *match;
+
 	CHECK_OBJ_NOTNULL(ctx, VRT_CTX_MAGIC);
 	CHECK_OBJ_NOTNULL(set, VMOD_SELECTOR_SET_MAGIC);
+	AN(set->origo);
+	AN(set->members);
 
 	if (subject == NULL) {
 		VERR(ctx, "%s.match(): subject string is NULL", set->vcl_name);
@@ -220,9 +255,162 @@ vmod_set_match(VRT_CTX, struct vmod_selector_set *set, VCL_STRING subject)
 		return (0);
 	}
 
+	match = get_match_data(ctx, set, "match");
+	if ((idx = PT_Lookup(set->origo, set->members, subject)) == UINT_MAX) {
+		match->n = 0;
+		return (0);
+	}
+
+	if ((match->indices = WS_Alloc(ctx->ws, sizeof(unsigned))) == NULL) {
+		VERRNOMEM(ctx, "Reserving space for index in "
+			  "%s.match(\"%.40s\")", set->vcl_name, subject);
+		return (0);
+	}
+	*match->indices = idx;
+	match->n = 1;
+	return (1);
+}
+
+VCL_BOOL
+vmod_set_hasprefix(VRT_CTX, struct vmod_selector_set *set, VCL_STRING subject)
+{
+	struct match_data *match;
+
+	CHECK_OBJ_NOTNULL(ctx, VRT_CTX_MAGIC);
+	CHECK_OBJ_NOTNULL(set, VMOD_SELECTOR_SET_MAGIC);
 	AN(set->origo);
 	AN(set->members);
-	return (PT_Lookup(set->origo, set->members, subject) != UINT_MAX);
+
+	if (subject == NULL) {
+		VERR(ctx, "%s.hasprefix(): subject string is NULL",
+		     set->vcl_name);
+		return (0);
+	}
+	if (set->nmembers == 0) {
+		VERR(ctx, "%s.hasprefix(): no entries were added",
+		     set->vcl_name);
+		return (0);
+	}
+
+	match = get_match_data(ctx, set, "hasprefix");
+	if ((match->limit = WS_ReserveLumps(ctx->ws, sizeof(unsigned))) == 0) {
+		VERRNOMEM(ctx, "Reserving index array in "
+			  "%s.hasprefix(\"%.40s\")", set->vcl_name, subject);
+		return (0);
+	}
+	match->indices = (unsigned *)WS_Front(ctx->ws);
+
+	if (PT_Prefixes(set->origo, set->members, subject, match) != 0) {
+		VERRNOMEM(ctx, "Adding indices in %s.hasprefix(\"%.40s\")",
+			  set->vcl_name, subject);
+		WS_Release(ctx->ws, 0);
+		return (0);
+	}
+
+	WS_Release(ctx->ws, match->n * sizeof(unsigned));
+	return (match->n > 0);
+}
+
+static struct match_data *
+get_existing_match_data(VRT_CTX,
+			const struct vmod_selector_set * const restrict set,
+			const char * const restrict method)
+{
+	struct vmod_priv *task;
+	struct match_data *match;
+
+	task = VRT_priv_task(ctx, set);
+	AN(task);
+	if (task->priv == NULL) {
+		VERR(ctx, "%s.%s() called without prior match", set->vcl_name,
+		     method);
+		return (NULL);
+	}
+	WS_Assert_Allocated(ctx->ws, task->priv, sizeof(*match));
+	CAST_OBJ(match, task->priv, MATCH_DATA_MAGIC);
+	return match;
+}
+
+VCL_INT
+vmod_set_nmatches(VRT_CTX, struct vmod_selector_set *set)
+{
+	struct match_data *match;
+
+	CHECK_OBJ_NOTNULL(ctx, VRT_CTX_MAGIC);
+	CHECK_OBJ_NOTNULL(set, VMOD_SELECTOR_SET_MAGIC);
+
+	match = get_existing_match_data(ctx, set, "nmatches");
+	if (match == NULL)
+		return (0);
+	return (match->n);
+}
+
+VCL_BOOL
+vmod_set_matched(VRT_CTX, struct vmod_selector_set *set, VCL_INT idx)
+{
+	struct match_data *match;
+
+	CHECK_OBJ_NOTNULL(ctx, VRT_CTX_MAGIC);
+	CHECK_OBJ_NOTNULL(set, VMOD_SELECTOR_SET_MAGIC);
+
+	if (idx < 1 || idx > set->nmembers) {
+		VERR(ctx, "%s.matched(%ld) out of range (%d members)",
+		     set->vcl_name, idx, set->nmembers);
+		return (0);
+	}
+
+	match = get_existing_match_data(ctx, set, "matched");
+	if (match == NULL || match->n == 0)
+		return (0);
+
+	AN(match->indices);
+	WS_Assert_Allocated(ctx->ws, match->indices,
+			    match->n * sizeof(unsigned));
+	/* XXX search algorithm? */
+	idx--;
+	for (unsigned i = 0; i < match->n; i++)
+		if (match->indices[i] == idx)
+			return (1);
+	return (0);
+}
+
+static unsigned
+select(VRT_CTX, const struct match_data * const restrict match,
+       const char * const restrict obj, VCL_ENUM const restrict selects,
+       const char * const restrict method)
+{
+	if (selects == vmod_enum_UNIQUE) {
+		if (match->n != 1) {
+			VERR(ctx, "%s.%s(select=UNIQUE): "
+			     "%d elements were matched", obj, method, match->n);
+			return (UINT_MAX);
+		}
+		return match->indices[0];
+	}
+	if (selects == vmod_enum_FIRST)
+		return match->min;
+	if (selects == vmod_enum_LAST)
+		return match->max;
+	if (selects == vmod_enum_SHORTEST)
+		return match->indices[0];
+	if (selects == vmod_enum_LONGEST)
+		return match->indices[match->n - 1];
+	WRONG("illegal select enum");
+	return (UINT_MAX);
+}
+
+VCL_INT
+vmod_set_which(VRT_CTX, struct vmod_selector_set *set, VCL_ENUM selects)
+{
+	struct match_data *match;
+
+	CHECK_OBJ_NOTNULL(ctx, VRT_CTX_MAGIC);
+	CHECK_OBJ_NOTNULL(set, VMOD_SELECTOR_SET_MAGIC);
+
+	match = get_existing_match_data(ctx, set, "which");
+	if (match == NULL || match->n == 0)
+		return (0);
+	return (select(ctx, match, set->vcl_name, selects, "which") + 1);
 }
 
 VCL_STRING

src/vmod_selector.vcc

@@ -64,6 +64,46 @@ Example::
      call do_on_match;
   }
 
+$Method BOOL .hasprefix(STRING)
+
+Returns ``true`` if and only if ...
+
+Example::
+
+  if (myset.hasprefix(req.url)) {
+     call do_if_prefix_matched;
+  }
+
+$Method INT .nmatches()
+
+Returns ...
+
+Example::
+
+  if (myset.hasprefix(req.url)) {
+     # ...
+  }
+
+$Method BOOL .matched(INT)
+
+Returns ...
+
+Example::
+
+  if (myset.hasprefix(req.url)) {
+     # ...
+  }
+
+$Method INT .which(ENUM {UNIQUE, FIRST, LAST, SHORTEST, LONGEST} select=UNIQUE)
+
+Returns ...
+
+Example::
+
+  if (myset.hasprefix(req.url)) {
+     # ...
+  }
+
 $Method STRING .debug()
 
 Intentionally not documented.