X-Git-Url: http://git.madism.org/?p=apps%2Fpfixtools.git;a=blobdiff_plain;f=common%2Ftrie.c;h=f16f0a64d7f24255c9d116b31728e39254303375;hp=3cc093a3036c2f0df88c99a04f63a4da87d78c3b;hb=ab82546953101224d09dcf2e44ff170e454e7282;hpb=fb519b0c3a78183d3fad3778d7d57bc2693598fc

diff --git a/common/trie.c b/common/trie.c
index 3cc093a..f16f0a6 100644
--- a/common/trie.c
+++ b/common/trie.c
@@ -33,30 +33,27 @@
  * Copyright © 2008 Florent Bruneau
  */
 
+#include "array.h"
 #include "str.h"
 #include "trie.h"
 
 typedef struct trie_entry_t trie_entry_t;
 
 struct trie_entry_t {
-    int  c_offset;
-    int  c_len;
+    uint32_t c_offset;
+    uint32_t children_offset;
 
-    int* children;
-    int children_len;
-    int children_size;
-
-    bool locked;
+    uint16_t c_len;
+    uint16_t children_len;
 };
+#define TRIE_ENTRY_INIT { 0, 0, 0, 0 }
+ARRAY(trie_entry_t)
 
 struct trie_t {
-    trie_entry_t *entries;
-    int entries_len;
-    int entries_size;
-
-    char *c;
-    int  c_len;
-    int  c_size;
+    A(trie_entry_t) entries;
+    A(char)         c;
+    A(char)         keys;
+    A(int)          keys_offset;
 
     bool locked;
 };
@@ -66,15 +63,19 @@ trie_t *trie_new(void)
     return p_new(trie_t, 1);
 }
 
+static inline void trie_cleanup_build_data(trie_t *trie)
+{
+    array_wipe(trie->keys);
+    array_wipe(trie->keys_offset);
+}
+
 void trie_delete(trie_t **trie)
 {
     if (*trie) {
-        for (int i = 0 ; i < (*trie)->entries_len ; ++i) {
-            trie_entry_t *entry = &(*trie)->entries[i];
-            p_delete(&(entry->children));
-        }
-        p_delete(&(*trie)->entries);
-        p_delete(&(*trie)->c);
+        trie_cleanup_build_data(*trie);
+        trie_unlock(*trie);
+        array_wipe((*trie)->entries);
+        array_wipe((*trie)->c);
         p_delete(trie);
     }
 }
@@ -83,28 +84,32 @@ void trie_delete(trie_t **trie)
  */
 static inline bool trie_entry_c_match(const trie_t *trie,
                                       const trie_entry_t *entry,
-                                      const char *key, int *pos)
+                                      const char *key)
 {
-    const char *c = trie->c + entry->c_offset;
+    const char *c = array_ptr(trie->c, entry->c_offset);
     int i = 0;
     for (i = 0 ; i < entry->c_len ; ++i) {
         if (key[i] != c[i]) {
-            if (pos) {
-                *pos = i;
-            }
             return false;
         }
     }
-    if (pos) {
-        *pos = i;
-    }
     return true;
 }
 
 static inline bool trie_entry_match(const trie_t *trie,
                                     const trie_entry_t *entry, const char *key)
 {
-    return !!(strcmp(trie->c + entry->c_offset, key) == 0);
+    return !!(strcmp(array_ptr(trie->c, entry->c_offset), key) == 0);
+}
+
+static inline bool trie_entry_prefix(const trie_t *trie,
+                                     const trie_entry_t *entry, const char *key)
+{
+    int len = entry->c_len;
+    if (len > 0 && array_elt(trie->c, entry->c_offset + len - 1) == '\0') {
+        --len;
+    }
+    return !!(strncmp(array_ptr(trie->c, entry->c_offset), key, len) == 0);
 }
 
 static inline bool trie_entry_is_leaf(const trie_entry_t *entry)
@@ -116,109 +121,69 @@ static inline bool trie_entry_is_leaf(const trie_entry_t *entry)
  * Only the first character of the children is taken into account in the
  * lookup. The current entry is assumed to match the key.
  */
-static inline trie_entry_t* trie_entry_child(const trie_t *trie,
-                                             const trie_entry_t* entry,
-                                             const char *key)
+static inline const trie_entry_t* trie_entry_child(const trie_t *trie,
+                                                   const trie_entry_t* entry,
+                                                   const char *key)
 {
-    int start = 0;
-    int end   = entry->children_len;
+    uint32_t start = entry->children_offset;
+    uint32_t end   = start + entry->children_len;
     const char c = *key;
 
     while (start < end) {
-        int mid = (start + end) / 2;
-        trie_entry_t* child = &trie->entries[entry->children[mid]];
-        const char c2 = trie->c[child->c_offset];
+        uint32_t mid = (start + end) >> 1;
+        const trie_entry_t* child = array_ptr(trie->entries, mid);
+        const char c2 = array_elt(trie->c, child->c_offset);
 
-        if (child->c_len) {
-            if (c2 == c) {
-                return child;
-            }
-            if (c < c2) {
-                end = mid;
-            } else {
-                start = mid + 1;
-            }
+        if (c2 == c) {
+          return child;
+        }
+        if (c < c2) {
+          end = mid;
         } else {
-            abort();
+          start = mid + 1;
         }
     }
     return NULL;
 }
 
-#define GROW(Buffer, Added, Len, Size)                  \
-  do {                                                  \
-      const int required_size = (Len) + (Added);        \
-      int next_size = (Size);                           \
-      if (next_size >= required_size) {                 \
-          break;                                        \
-      }                                                 \
-      do {                                              \
-          next_size = p_alloc_nr(next_size);            \
-      } while (next_size < required_size);              \
-      p_allocgrow(&(Buffer), next_size, &(Size));       \
-  } while(0)
-
-static inline void trie_grow(trie_t *trie, int delta)
-{
-    GROW(trie->entries, delta, trie->entries_len, trie->entries_size);
-}
-
-static inline int trie_entry_new(trie_t *trie)
+static inline uint32_t trie_entry_new(trie_t *trie)
 {
-    memset(trie->entries + trie->entries_len, 0, sizeof(trie_entry_t));
-    return trie->entries_len++;
+    const trie_entry_t e = TRIE_ENTRY_INIT;
+    array_add(trie->entries, e);
+    return trie->entries.len - 1;
 }
 
-static inline int trie_add_leaf(trie_t *trie, const char *key)
+static inline uint32_t trie_add_leaf(trie_t *trie, const char *key)
 {
     trie_entry_t *entry;
     int len = m_strlen(key) + 1;
-    entry = &trie->entries[trie_entry_new(trie)];
-    GROW(trie->c, len, trie->c_len, trie->c_size);
-    memcpy(trie->c + trie->c_len, key, len);
-    entry->c_offset = trie->c_len;
+    entry = array_ptr(trie->entries, trie_entry_new(trie));
+    entry->c_offset = trie->c.len;
     entry->c_len    = len;
-    trie->c_len    += len;
-    return trie->entries_len - 1;
+    array_append(trie->c, key, len);
+    return trie->entries.len - 1;
 }
 
 static inline void trie_entry_insert_child(trie_t *trie, trie_entry_t *entry,
-                                           int pchild)
+                                           uint32_t pchild)
 {
-    const char c = trie->c[trie->entries[pchild].c_offset];
-    int start = 0;
-    int end   = entry->children_len;
-
-    p_allocgrow(&entry->children, entry->children_len + 1, &entry->children_size);
-    while (start < end) {
-        int mid = (start + end) / 2;
-        const trie_entry_t* child = &trie->entries[entry->children[mid]];
-        const char c2 = trie->c[child->c_offset];
-
-        if (child->c_len) {
-            if (c2 == c) {
-                abort();
-            }
-            if (c < c2) {
-                end = mid;
-            } else {
-                start = mid + 1;
-            }
-        } else {
+    if (entry->children_len == 0) {
+        entry->children_offset = pchild;
+        entry->children_len    = 1;
+    } else {
+        if (entry->children_offset + entry->children_len != pchild) {
+            printf("Inserting child %d while offset is %d[%d]\n",
+                   pchild, entry->children_offset, entry->children_len);
             abort();
         }
+        ++entry->children_len;
     }
-    memmove(entry->children + start + 1,
-            entry->children + start,
-            sizeof(int) * (entry->children_len - start));
-    entry->children[start] = pchild;
-    ++entry->children_len;
 }
 
-static inline void trie_entry_split(trie_t *trie, trie_entry_t *entry, int pos)
+static inline void trie_entry_split(trie_t *trie, trie_entry_t *entry, uint16_t pos)
 {
     trie_entry_t *child;
-    child    = &trie->entries[trie_entry_new(trie)];
+    child    = array_ptr(trie->entries, trie_entry_new(trie));
     if (pos == 0) {
         child->c_offset = entry->c_offset;
         child->c_len    = entry->c_len;
@@ -229,60 +194,120 @@ static inline void trie_entry_split(trie_t *trie, trie_entry_t *entry, int pos)
         child->c_len    = entry->c_len - pos;
         entry->c_len    = pos;
     }
-    child->children      = entry->children;
-    child->children_len  = entry->children_len;
-    child->children_size = entry->children_size;
-    entry->children      = NULL;
-    entry->children_len  = 0;
-    entry->children_size = 0;
-    trie_entry_insert_child(trie, entry, trie->entries_len - 1);
+    child->children_offset = entry->children_offset;
+    child->children_len    = entry->children_len;
+    entry->children_offset = trie->entries.len - 1;
+    entry->children_len    = 1;
 }
 
 void trie_insert(trie_t *trie, const char* key)
 {
-    trie_grow(trie, 2);
-    if (trie->entries_len == 0) {
-        (void)trie_add_leaf(trie, key);
+    assert(trie->entries.len == 0 && "Trie already compiled");
+
+    int len = m_strlen(key) + 1;
+    array_add(trie->keys_offset, trie->keys.len);
+    array_append(trie->keys, key, len);
+}
+
+
+static inline void trie_compile_aux(trie_t *trie, uint32_t id,
+                                    uint32_t first_key, uint32_t last_key,
+                                    int offset, int initial_diff)
+{
+    uint32_t forks[256];
+    uint32_t fork_pos = 0;
+    char current = '\0';
+
+    for (int off_diff = initial_diff ; fork_pos == 0 ; ++off_diff, ++offset) {
+        current = array_elt(trie->keys, array_elt(trie->keys_offset, first_key) + offset);
+        for (uint32_t i = first_key + 1 ; i < last_key ; ++i) {
+            const char *str = array_ptr(trie->keys, array_elt(trie->keys_offset, i));
+            const char c = str[offset];
+            if (c != current) {
+                array_ensure_capacity_delta(trie->entries, 2);
+                if (fork_pos == 0) {
+                    trie_entry_split(trie, array_ptr(trie->entries, id), off_diff);
+                }
+                trie_entry_insert_child(trie, array_ptr(trie->entries, id),
+                                        trie_add_leaf(trie, str + offset));
+                forks[fork_pos++] = i;
+                current = c;
+            }
+        }
+        if (fork_pos == 0 && current == '\0') {
+            return;
+        }
+    }
+    forks[fork_pos] = last_key;
+
+    const uint8_t children_len = array_elt(trie->entries, id).children_len;
+    for (uint16_t i = 0 ; i < children_len ; ++i) {
+        int child = array_elt(trie->entries, id).children_offset + i;
+        if (forks[i] - 1 > first_key) {
+            trie_compile_aux(trie, child, first_key, forks[i], offset, 1);
+        }
+        first_key = forks[i];
+    }
+}
+
+void trie_compile(trie_t *trie, bool memlock)
+{
+    assert(trie->entries.len == 0 && "Trie already compiled");
+    assert(trie->keys.len != 0 && "Trying to compile an empty trie");
+    {
+#       define QSORT_TYPE int
+#       define QSORT_BASE trie->keys_offset.data
+#       define QSORT_NELT trie->keys_offset.len
+#       define QSORT_LT(a,b) strcmp(trie->keys.data + *a, trie->keys.data + *b) < 0
+#       include "qsort.c"
+    }
+
+    array_ensure_capacity(trie->entries, trie->keys.len);
+    trie_compile_aux(trie, trie_add_leaf(trie, trie->keys.data),
+                     0, trie->keys_offset.len, 0, 0);
+    trie_cleanup_build_data(trie);
+    array_adjust(trie->entries);
+    array_adjust(trie->c);
+    if (memlock) {
+        trie_lock(trie);
+    }
+}
+
+bool trie_lookup(const trie_t *trie, const char *key)
+{
+    assert(trie->keys.len == 0L && "Can't lookup: trie not compiled");
+    if (trie->entries.len == 0) {
+        return false;
     } else {
-        trie_entry_t *current = trie->entries;
+        const trie_entry_t *current = array_ptr(trie->entries, 0);
         while (true) {
-            int pos = 0;
-            if (trie_entry_c_match(trie, current, key, &pos)) {
-                if (trie_entry_is_leaf(current)) {
-                    return;
-                }
-                trie_entry_t *next = NULL;
-                key += pos;
-                next = trie_entry_child(trie, current, key);
-                if (next == NULL) {
-                    trie_entry_insert_child(trie, current,
-                                            trie_add_leaf(trie, key));
-                    return;
-                } else {
-                    current = next;
+            if (trie_entry_is_leaf(current)) {
+                return trie_entry_match(trie, current, key);
+            } else if (trie_entry_c_match(trie, current, key)) {
+                key += current->c_len;
+                current = trie_entry_child(trie, current, key);
+                if (current == NULL) {
+                    return false;
                 }
             } else {
-                trie_entry_split(trie, current, pos);
-                trie_entry_insert_child(trie, current,
-                                        trie_add_leaf(trie, key + pos));
-                return;
+                return false;
             }
         }
     }
 }
 
-bool trie_lookup(const trie_t *trie, const char *key)
+bool trie_prefix(const trie_t *trie, const char *key)
 {
-    if (trie->entries_len == 0) {
+    assert(trie->keys.len == 0L && "Can't lookup: trie not compiled");
+    if (trie->entries.len == 0) {
         return false;
     } else {
-        trie_entry_t *current = trie->entries;
+        const trie_entry_t *current = array_ptr(trie->entries, 0);
         while (true) {
-            int pos = 0;
             if (trie_entry_is_leaf(current)) {
-                return trie_entry_match(trie, current, key);
-            } else if (trie_entry_c_match(trie, current, key, &pos)) {
-                key += pos;
+                return trie_entry_prefix(trie, current, key);
+            } else if (trie_entry_c_match(trie, current, key)) {
+                key += current->c_len;
                 current = trie_entry_child(trie, current, key);
                 if (current == NULL) {
                     return false;
@@ -294,43 +319,92 @@ bool trie_lookup(const trie_t *trie, const char *key)
     }
 }
 
+void trie_lock(trie_t *trie)
+{
+    if (trie->locked) {
+        return;
+    }
+    if (!array_lock(trie->entries)) {
+        UNIXERR("mlock");
+    }
+    if (!array_lock(trie->c)) {
+        UNIXERR("mlock");
+    }
+    if (mlock(trie, sizeof(trie_t)) != 0) {
+        UNIXERR("mlock");
+        return;
+    }
+    trie->locked = true;
+}
+
+void trie_unlock(trie_t *trie)
+{
+    if (!trie->locked) {
+        return;
+    }
+    array_unlock(trie->entries);
+    array_unlock(trie->c);
+    munlock(trie, sizeof(trie_t));
+    trie->locked = false;
+}
 
 /* Debug {{{1
  */
 
-static inline void trie_entry_inspect(const trie_t *trie,
+static inline void trie_entry_inspect(const trie_t *trie, bool show_content,
                                       const trie_entry_t *entry, int level)
 {
-    static int c_sum = 0;
-    static int nodes = 0;
-
-    ++nodes;
-    c_sum += entry->c_len;
-    for (int i = 0 ; i < level ; ++i) {
-        fputs("  ", stdout);
+    static int max_depth = 0;
+    static int leaves    = 0;
+    static int depth_sum = 0;
+
+    if (entry == array_ptr(trie->entries, 0)) {
+      max_depth = 0;
+      leaves    = 0;
+      depth_sum = 0;
     }
-    if (entry->c_len == 0) {
-        fputs("(nil)", stdout);
-    } else {
-        const char *c = trie->c + entry->c_offset;
-        for (int i = 0 ; i < entry->c_len ; ++i) {
-            if (c[i]) {
-                printf("%c ", c[i]);
-            } else {
-                fputs("\\0 ", stdout);
+    if (trie_entry_is_leaf(entry)) {
+        if (level > max_depth) {
+            max_depth = level;
+        }
+        ++leaves;
+        depth_sum += level;
+    }
+    if (show_content) {
+        for (int i = 0 ; i < level ; ++i) {
+            fputs("  ", stdout);
+        }
+        if (entry->c_len == 0) {
+            fputs("(nil)", stdout);
+        } else {
+            const char *c = array_ptr(trie->c, entry->c_offset);
+            printf("(%d) ", entry->c_len);
+            for (int i = 0 ; i < entry->c_len ; ++i) {
+                if (c[i]) {
+                    printf("%c ", c[i]);
+                } else {
+                    fputs("\\0 ", stdout);
+                }
             }
         }
+        fputs("\n", stdout);
     }
-    fputs("\n", stdout);
-    for (int i = 0 ; i < entry->children_len ; ++i) {
-        trie_entry_inspect(trie, &trie->entries[entry->children[i]], level + 1);
+    for (uint32_t i = entry->children_offset ;
+          i < entry->children_offset + entry->children_len ; ++i) {
+        trie_entry_inspect(trie, show_content, array_ptr(trie->entries, i), level + 1);
     }
     if (level == 0) {
-        printf("Mean char per node: %d\n", c_sum / nodes);
+        printf("Average char per node: %d\n", trie->c.len / trie->entries.len);
+        printf("Number of nodes: %d\n", trie->entries.len);
+        printf("Number of leaves: %d\n", leaves);
+        printf("Max depth: %d\n", max_depth);
+        printf("Average leaf depth: %d\n", depth_sum / leaves);
+        printf("Memory used: %d\n", (trie->entries.size * sizeof(trie_entry_t))
+                                  + (trie->c.size) + sizeof(trie_t));
     }
 }
 
-void trie_inspect(const trie_t *trie)
+void trie_inspect(const trie_t *trie, bool show_content)
 {
-    trie_entry_inspect(trie, trie->entries, 0);
+    trie_entry_inspect(trie, show_content, trie->entries.data, 0);
 }