dict_view for (Base)Trie

.gitignore

@@ -5,3 +5,6 @@ src/*.html
 
 *.so
 build/
+
+.idea/
+.cache/

Makefile

@@ -0,0 +1,6 @@
+.PHONY: build
+
+build:
+	./update_c.sh
+	python setup.py build
+	python setup.py build_ext --inplace

libdatrie/datrie/libdatrie.def

@@ -10,6 +10,7 @@ trie_fread
 trie_free
 trie_save
 trie_fwrite
+trie_size
 trie_is_dirty
 trie_retrieve
 trie_store

libdatrie/datrie/trie.c

@@ -44,6 +44,7 @@ struct _Trie {
     DArray     *da;
     Tail       *tail;
 
+    uint32      size;
     Bool        is_dirty;
 };
 
@@ -133,6 +134,7 @@ trie_new (const AlphaMap *alpha_map)
     if (UNLIKELY (!trie->tail))
         goto exit_da_created;
 
+    trie->size = 0;
     trie->is_dirty = TRUE;
     return trie;
 
@@ -203,6 +205,11 @@ trie_fread (FILE *file)
     if (NULL == (trie->tail = tail_fread (file)))
         goto exit_da_created;
 
+    uint32 counter = 0;
+    if (!trie_enumerate (trie, len_enumerator, &counter)) {
+        goto exit_trie_created;
+    }
+    trie->size = counter;
     trie->is_dirty = FALSE;
     return trie;
 
@@ -290,6 +297,19 @@ trie_fwrite (Trie *trie, FILE *file)
     return 0;
 }
 
+/**
+ * @brief Check pending changes
+ *
+ * @param trie  : the trie object
+ *
+ * @return total count of trie keys
+ */
+uint32
+trie_size (const Trie *trie)
+{
+    return trie->size;
+}
+
 /**
  * @brief Check pending changes
  *
@@ -431,6 +451,9 @@ trie_store_conditionally (Trie            *trie,
             res = trie_branch_in_branch (trie, s, key_str, data);
             free (key_str);
 
+            if (res) {
+                trie->size++;
+            }
             return res;
         }
         if (0 == *p)
@@ -455,6 +478,9 @@ trie_store_conditionally (Trie            *trie,
             res = trie_branch_in_tail (trie, s, tail_str, data);
             free (tail_str);
 
+            if (res) {
+                trie->size++;
+            }
             return res;
         }
         if (0 == *p)
@@ -580,6 +606,7 @@ trie_delete (Trie *trie, const AlphaChar *key)
     da_set_base (trie->da, s, TRIE_INDEX_ERROR);
     da_prune (trie->da, s);
 
+    trie->size--;
     trie->is_dirty = TRUE;
     return TRUE;
 }
@@ -630,6 +657,14 @@ trie_enumerate (const Trie *trie, TrieEnumFunc enum_func, void *user_data)
 }
 
 
+Bool
+len_enumerator (const AlphaChar *key, TrieData key_data, uint32 *counter_ptr)
+{
+    (*counter_ptr)++;
+    return TRUE;
+}
+
+
 /*-------------------------------*
  *   STEPWISE QUERY OPERATIONS   *
  *-------------------------------*/

libdatrie/datrie/trie.h

@@ -129,6 +129,8 @@ int     trie_save (Trie *trie, const char *path);
 
 int     trie_fwrite (Trie *trie, FILE *file);
 
+uint32    trie_size (const Trie *trie);
+
 Bool    trie_is_dirty (const Trie *trie);
 
 
@@ -150,6 +152,10 @@ Bool    trie_enumerate (const Trie     *trie,
                         TrieEnumFunc    enum_func,
                         void           *user_data);
 
+Bool    len_enumerator (const AlphaChar *key,
+                        TrieData key_data,
+                        uint32 *counter_ptr);
+
 
 /*-------------------------------*
  *   STEPWISE QUERY OPERATIONS   *

libdatrie/tests/test_store-retrieve.c

@@ -48,6 +48,34 @@ main ()
         goto err_trie_not_created;
     }
 
+    msg_step ("Check initial trie size");
+    if (trie_size(test_trie) != 0) {
+        printf ("Wrong trie size; expected 0, got %d.\n", trie_size(test_trie));
+        goto err_trie_size;
+    }
+
+    msg_step ("Delete non-existent key from trie and check size");
+    trie_delete (test_trie, (AlphaChar *)L"a");
+    if (trie_size(test_trie) != 0) {
+        printf ("Wrong trie size; expected 0, got %d.\n", trie_size(test_trie));
+        goto err_trie_size;
+    }
+    msg_step ("Add non-existent key with trie_store_if_absent and check size");
+    if (!trie_store_if_absent (test_trie, (AlphaChar *)L"a", TRIE_DATA_UNREAD)) {
+        printf ("Failed to add non-existing key '%ls'.\n", (AlphaChar *)L"a");
+        goto err_trie_created;
+    }
+    if (trie_size(test_trie) != 1) {
+        printf ("Wrong trie size; expected 1, got %d.\n", trie_size(test_trie));
+        goto err_trie_size;
+    }
+    msg_step ("Delete existing key from trie and check size");
+    trie_delete (test_trie, (AlphaChar *)L"a");
+    if (trie_size(test_trie) != 0) {
+        printf ("Wrong trie size; expected 0, got %d.\n", trie_size(test_trie));
+        goto err_trie_size;
+    }
+
     /* store */
     msg_step ("Adding data to trie");
     for (dict_p = dict_src; dict_p->key; dict_p++) {
@@ -58,6 +86,51 @@ main ()
         }
     }
 
+    msg_step ("Check trie size");
+    if (trie_size(test_trie) != dict_src_n_entries()) {
+        printf ("Wrong trie size; expected %d, got %d.\n",
+                dict_src_n_entries(), trie_size(test_trie));
+        goto err_trie_size;
+    }
+
+    msg_step ("Update existing trie element and check trie size");
+    if (!trie_store (test_trie, dict_src[1].key, dict_src[1].data)) {
+            printf ("Failed to add key '%ls', data %d.\n",
+                    dict_src[1].key, dict_src[1].data);
+            goto err_trie_created;
+        }
+    if (trie_size(test_trie) != dict_src_n_entries()) {
+        printf ("Wrong trie size; expected %d, got %d.\n",
+                dict_src_n_entries(), trie_size(test_trie));
+        goto err_trie_size;
+    }
+
+    msg_step ("Update existing trie element with trie_store_if_absent and check trie size");
+    if (trie_store_if_absent (test_trie, dict_src[1].key, dict_src[1].data)) {
+            printf ("Value for existing key '%ls' was updated with trie_store_if_absent.\n",
+                    dict_src[1].key);
+            goto err_trie_created;
+        }
+    if (trie_size(test_trie) != dict_src_n_entries()) {
+        printf ("Wrong trie size; expected %d, got %d.\n",
+                dict_src_n_entries(), trie_size(test_trie));
+        goto err_trie_size;
+    }
+
+    msg_step ("Add trie element with wrong alphabet and check trie size");
+    if (trie_store (test_trie, (AlphaChar *)L"я", TRIE_DATA_UNREAD)) {
+            printf ("Key '%ls' with wrong alphabet was added.\n",
+                    (AlphaChar *)L"я");
+            goto err_trie_created;
+        }
+    if (trie_size(test_trie) != dict_src_n_entries()) {
+        printf ("Wrong trie size; expected %d, got %d.\n",
+                dict_src_n_entries(), trie_size(test_trie));
+        goto err_trie_size;
+    }
+
+    // TODO: add key with wrong alphabet and check size?
+
     /* retrieve */
     msg_step ("Retrieving data from trie");
     is_failed = FALSE;
@@ -99,6 +172,14 @@ main ()
         goto err_trie_created;
     }
 
+    msg_step ("Check trie size after deleting some entries.");
+    if (trie_size(test_trie) != (n_entries - (n_entries/3 + 1))) {
+        printf ("Wrong trie size; expected %d, got %d.\n",
+                (n_entries - (n_entries/3 + 1)), trie_size(test_trie));
+        goto err_trie_size;
+    }
+
+
     /* retrieve */
     msg_step ("Retrieving data from trie again after deletions");
     for (dict_p = dict_src; dict_p->key; dict_p++) {
@@ -192,6 +273,8 @@ main ()
     trie_state_free (trie_root_state);
 err_trie_created:
     trie_free (test_trie);
+err_trie_size:
+    trie_free (test_trie);
 err_trie_not_created:
     return 1;
 }