feat(lsm): write str tests; start trie tests

lsm/include/lsm/str.h

@@ -1,6 +1,8 @@
 #ifndef LSM_STR
 #define LSM_STR
 
+#include <stdbool.h>
+
 #include "lsm.h"
 
 /**
@@ -10,22 +12,6 @@
  */
 typedef struct lsm_str lsm_str;
 
-/**
- * Allocate a new string struct of length 0.
- *
- * @param ptr pointer to store newly allocated pointer in
- */
-lsm_error lsm_str_init_zero(lsm_str **ptr);
-
-/**
- * Update an existing lsm_str so it now represents the new provided string. The
- * string pointer of the original object is free'd if needed.
- *
- * @param str lsm_str object to modify
- * @param s string to convert into lsm string; ownership is taken over
- */
-void lsm_str_init_prealloc(lsm_str *str, char *s);
-
 /**
  * Allocate and initialize a new lsm_str object
  *
@@ -35,17 +21,45 @@ void lsm_str_init_prealloc(lsm_str *str, char *s);
 lsm_error lsm_str_init(lsm_str **ptr, char *s);
 
 /**
- * Same as lsm_str_init, except it copies the original string instead of taking
- * over ownership, leaving the original string untouched.
+ * Allocate a new string struct of length 0.
+ *
+ * @param ptr pointer to store newly allocated pointer in
+ */
+lsm_error lsm_str_init_zero(lsm_str **ptr);
+
+/**
+ * Allocate and initialize a new lsm_str object, but copy the original string
+ * instead of taking over ownership, leaving the original string untouched.
  *
  * @param ptr pointer to store newly allocated pointer
  * @param s string to copy into lsm string
  */
 lsm_error lsm_str_init_copy(lsm_str **ptr, char *s);
 
+/**
+ * Overwrite an existing lsm_str so it now represents the new provided string.
+ * The string pointer of the original object is free'd if needed. Ownership of
+ * the pointer is taken over.
+ *
+ * @param str lsm_str object to modify
+ * @param s string to convert into lsm string; ownership is taken over
+ */
+void lsm_str_overwrite(lsm_str *str, char *s);
+
+/**
+ * Overwrite an existing lsm_str so it now represents the new provided string.
+ * The string pointer of the original object is free'd if needed. The provided
+ * string is copied, leaving the original untouched.
+ *
+ * @param str lsm_str object to modify
+ * @param s string to convert into lsm string; ownership is taken over
+ */
+lsm_error lsm_str_overwrite_copy(lsm_str *str, char *s);
+
 /**
  * Deallocate the existing internal string if needed and replace the lsm_str
- * with a string of length 0, wiping its contents.
+ * with a string of length 0, wiping its contents. This function can be used as
+ * a substitute for lsm_str_free for stack-allocated structs.
  *
  * @param str string to wipe
  */
@@ -85,7 +99,8 @@ char lsm_str_char(lsm_str *str, uint64_t index);
  * Take a substring and copy it to a provided string object.
  *
  * @param out string to store new substring in. The contents of this string will
- * be replaced.
+ * be replaced. This string is assumed to be unitialized, so zero this string
+ * manually if you're overwriting an existing string.
  * @param str string to take substring from
  * @param start inclusive start index for the substring. If this is greater than
  * or equal to the string's length, out will be a zero-length string.
@@ -109,7 +124,16 @@ uint64_t lsm_str_cmp(lsm_str *s1, uint64_t s1_offset, lsm_str *s2,
                      uint64_t s2_offset);
 
 /**
- * Truncate a string in-place.
+ * Checks whether the two strings are identical.
+ *
+ * @param s1 first string to compare
+ * @param s2 second string to compare
+ * @return true if their values are equal, false otherwise
+ */
+bool lsm_str_eq(lsm_str *s1, lsm_str *s2);
+
+/**
+ * Truncate an already initialized string in-place.
  *
  * @param s string to truncate
  * @param new_len new length of the string. If new_len is >= the original

lsm/src/str/lsm_str.c

@@ -7,6 +7,20 @@
 
 #define MIN(x, y) (((x) < (y)) ? (x) : (y))
 
+lsm_error lsm_str_init(lsm_str **ptr, char *s) {
+  lsm_str *str = calloc(1, sizeof(lsm_str));
+
+  if (str == NULL) {
+    return lsm_error_failed_alloc;
+  }
+
+  lsm_str_overwrite(str, s);
+
+  *ptr = str;
+
+  return lsm_error_ok;
+}
+
 lsm_error lsm_str_init_zero(lsm_str **ptr) {
   lsm_str *str = calloc(1, sizeof(lsm_str));
 
@@ -19,7 +33,21 @@ lsm_error lsm_str_init_zero(lsm_str **ptr) {
   return lsm_error_ok;
 }
 
-void lsm_str_init_prealloc(lsm_str *str, char *s) {
+lsm_error lsm_str_init_copy(lsm_str **ptr, char *s) {
+  lsm_str *str = calloc(1, sizeof(lsm_str));
+
+  if (str == NULL) {
+    return lsm_error_failed_alloc;
+  }
+
+  lsm_str_overwrite_copy(str, s);
+
+  *ptr = str;
+
+  return lsm_error_ok;
+}
+
+void lsm_str_overwrite(lsm_str *str, char *s) {
   str->len = strlen(s);
 
   if (str->len <= 8) {
@@ -30,27 +58,7 @@ void lsm_str_init_prealloc(lsm_str *str, char *s) {
   }
 }
 
-lsm_error lsm_str_init(lsm_str **ptr, char *s) {
-  lsm_str *str = calloc(1, sizeof(lsm_str));
-
-  if (str == NULL) {
-    return lsm_error_failed_alloc;
-  }
-
-  lsm_str_init_prealloc(str, s);
-
-  *ptr = str;
-
-  return lsm_error_ok;
-}
-
-lsm_error lsm_str_init_copy(lsm_str **ptr, char *s) {
-  lsm_str *str = calloc(1, sizeof(lsm_str));
-
-  if (str == NULL) {
-    return lsm_error_failed_alloc;
-  }
-
+lsm_error lsm_str_overwrite_copy(lsm_str *str, char *s) {
   str->len = strlen(s);
 
   if (str->len <= 8) {
@@ -66,8 +74,6 @@ lsm_error lsm_str_init_copy(lsm_str **ptr, char *s) {
     str->data.ptr = buf;
   }
 
-  *ptr = str;
-
   return lsm_error_ok;
 }
 
@@ -80,10 +86,7 @@ void lsm_str_zero(lsm_str *str) {
 }
 
 void lsm_str_free(lsm_str *str) {
-  if (str->len > 8) {
-    free(str->data.ptr);
-  }
-
+  lsm_str_zero(str);
   free(str);
 }
 
@@ -112,7 +115,7 @@ lsm_error lsm_str_substr(lsm_str *out, lsm_str *str, uint64_t start,
   const char *str_ptr = lsm_str_ptr(str);
 
   if (len <= 8) {
-    lsm_str_zero(out);
+    /* lsm_str_zero(out); */
     memcpy(out->data.val, &str_ptr[start], len);
   } else {
     char *buf = malloc(len * sizeof(char));
@@ -123,7 +126,7 @@ lsm_error lsm_str_substr(lsm_str *out, lsm_str *str, uint64_t start,
 
     memcpy(buf, &str_ptr[start], len);
 
-    lsm_str_zero(out);
+    /* lsm_str_zero(out); */
     out->data.ptr = buf;
   }
 
@@ -185,3 +188,11 @@ lsm_error lsm_str_split(lsm_str *s, lsm_str *s2, uint64_t index) {
 
   return lsm_str_truncate(s, index);
 }
+
+bool lsm_str_eq(lsm_str *s1, lsm_str *s2) {
+  if (s1->len != s2->len) {
+    return false;
+  }
+
+  return memcmp(lsm_str_ptr(s1), lsm_str_ptr(s2), s1->len) == 0;
+}

lsm/src/trie/lsm_trie.c

@@ -35,6 +35,8 @@ lsm_error lsm_trie_init(lsm_trie **ptr) {
   return lsm_error_ok;
 }
 
+uint64_t lsm_trie_size(lsm_trie *trie) { return trie->size; }
+
 lsm_error lsm_trie_insert(lsm_trie *trie, lsm_str *key, void *data) {
   // NULL is not allowed as a data value, as it's used to indicate a lack of
   // data
@@ -48,6 +50,7 @@ lsm_error lsm_trie_insert(lsm_trie *trie, lsm_str *key, void *data) {
   if (key_len == 0) {
     if (trie->root->data == NULL) {
       trie->root->data = data;
+      trie->size++;
 
       return lsm_error_ok;
     } else {
@@ -75,6 +78,8 @@ lsm_error lsm_trie_insert(lsm_trie *trie, lsm_str *key, void *data) {
       }
 
       new_node->data = data;
+      trie->size++;
+
       lsm_str_substr(&new_node->skip, key, index + 1, key_len);
 
       return lsm_bt_insert(&node->bt, c, new_node);
@@ -125,6 +130,7 @@ lsm_error lsm_trie_insert(lsm_trie *trie, lsm_str *key, void *data) {
   }
 
   node->data = data;
+  trie->size++;
 
   return lsm_error_ok;
 }

lsm/test/str/str.c

@@ -0,0 +1,91 @@
+#include "test.h"
+#include "lsm/str_internal.h"
+
+void test_cmp() {
+  lsm_str s1, s2, s3;
+  lsm_str_overwrite_copy(&s1, "some_string");
+  lsm_str_overwrite_copy(&s2, "some");
+  lsm_str_overwrite_copy(&s3, "some_string_extra");
+
+  TEST_CHECK(lsm_str_cmp(&s1, 0, &s2, 0) == 4);
+  TEST_CHECK(lsm_str_cmp(&s1, 0, &s2, 1) == 0);
+  TEST_CHECK(lsm_str_cmp(&s1, 1, &s2, 1) == 3);
+  TEST_CHECK(lsm_str_cmp(&s1, 1, &s2, 0) == 0);
+
+  TEST_CHECK(lsm_str_cmp(&s1, 0, &s3, 0) == lsm_str_len(&s1));
+}
+
+void test_eq() {
+  lsm_str s1, s2;
+  lsm_str_overwrite_copy(&s1, "longerthan8");
+  lsm_str_overwrite_copy(&s2, "longerthan8");
+
+  TEST_CHECK(lsm_str_eq(&s1, &s2));
+
+  lsm_str_overwrite_copy(&s1, "longerthan8");
+  lsm_str_overwrite_copy(&s2, "lmaolongerthan8");
+
+  TEST_CHECK(!lsm_str_eq(&s1, &s2));
+
+  lsm_str_overwrite_copy(&s1, "short");
+  lsm_str_overwrite_copy(&s2, "short");
+
+  TEST_CHECK(lsm_str_eq(&s1, &s2));
+
+  lsm_str_overwrite_copy(&s1, "short");
+  lsm_str_overwrite_copy(&s1, "shorte");
+
+  TEST_CHECK(!lsm_str_eq(&s1, &s2));
+
+  lsm_str_overwrite_copy(&s1, "longerthan8");
+  lsm_str_overwrite_copy(&s2, "short");
+
+  TEST_CHECK(!lsm_str_eq(&s1, &s2));
+}
+
+void test_substr() {
+  lsm_str s1, s2, s3;
+  lsm_str_overwrite_copy(&s1, "some_string");
+  lsm_str_overwrite_copy(&s3, "string");
+  lsm_str_substr(&s2, &s1, 5, lsm_str_len(&s1));
+
+  TEST_CHECK(lsm_str_eq(&s2, &s3));
+
+  lsm_str_zero(&s2);
+  lsm_str_substr(&s2, &s1, 25, lsm_str_len(&s1));
+
+  TEST_CHECK(lsm_str_len(&s2) == 0);
+}
+
+void test_truncate() {
+  lsm_str s1, s2, s3;
+  lsm_str_overwrite_copy(&s1, "some_longer_string_thing");
+  lsm_str_overwrite_copy(&s2, "some_longer_string");
+  lsm_str_overwrite_copy(&s3, "some");
+
+  lsm_str_truncate(&s1, 18);
+  TEST_CHECK(lsm_str_eq(&s1, &s2));
+
+  lsm_str_truncate(&s1, 4);
+  TEST_CHECK(lsm_str_eq(&s1, &s3));
+}
+
+void test_init_copy() {
+  char orig[] = "some_string";
+  lsm_str *s;
+  lsm_str_init_copy(&s, orig);
+
+  TEST_CHECK(s->data.ptr != orig);
+  TEST_CHECK(strcmp(s->data.ptr, orig) == 0);
+
+  lsm_str_free(s);
+}
+
+TEST_LIST = {
+  { "str init_copy", test_init_copy },
+  { "str cmp", test_cmp },
+  { "str eq", test_eq },
+  { "str substr", test_substr },
+  { "str truncate", test_truncate },
+  { NULL, NULL }
+};

lsm/test/trie/fuzzy.h

@@ -0,0 +1,222 @@
+#ifndef LSM_TRIE_FUZZY_TEST
+#define LSM_TRIE_FUZZY_TEST
+
+#include <stdlib.h>
+#include <stdbool.h>
+#include <string.h>
+#include <stdio.h>
+
+#include "lsm/trie.h"
+#include "lsm/str_internal.h"
+
+typedef struct fuzzyconfig {
+    int seed;
+    int word_length;
+    int word_count;
+} FuzzyConfig;
+
+void random_clean_string(char* s, int len) {
+    char charset[] = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789,?";
+    int charset_len = strlen(charset);
+
+    // len - 1 ensures that we can still set the null byte for the final byte
+    int actual_len = rand() % (len - 1);
+    int key;
+    int i;
+
+    for (i = 0; i < actual_len; i++) {
+        key = rand() % charset_len;
+        s[i] = charset[key];
+    }
+
+    s[i] = '\0';
+}
+
+void random_string(char* s, int len) {
+    int val = rand();
+
+    // String can't be an empty string as they aren't supported
+    s[0] = (char)(val % 255 + 1);
+
+    for (int i = 1; i < len - 1; i++) {
+        val = rand();
+        s[i] = (char)(val % 255 + 1);
+    }
+
+    // Just in case no null characters were created
+    s[len - 1] = '\0';
+}
+
+void random_string_matrix(char** s, int count, int len) {
+    for (int i = 0; i < count; i++) {
+        random_string(s[i], len);
+    }
+}
+
+char** init_string_matrix(int count, int len) {
+    char** matrix = malloc(count * sizeof(char*));
+
+    for (int i = 0; i < count; i++) {
+        matrix[i] = calloc(len, sizeof(char));
+    }
+
+    return matrix;
+}
+
+lsm_str *lsm_random_string_matrix(int count, int max_len) {
+  lsm_str *matrix = calloc(count, sizeof(lsm_str));
+
+  for (int i = 0; i < count; i++) {
+    int len = rand() % max_len;
+    char *buf = malloc(len * sizeof(char));
+
+    for (int i = 0; i < len; i++) {
+      buf[i] = (char)(rand() % 255 + 1);
+    }
+
+    lsm_str_overwrite(&matrix[i], buf);
+  }
+
+  return matrix;
+}
+
+/**
+ * Test a given trie implementation using randomly generated strings generated
+ * using a given seed.
+ *
+ * @param seed seed to use for generating random strings
+ * @param count how many strings to test with
+ * @param len maximum length of each string
+ * @param init_func function to creat a new trie of the wanted type
+ * @param free_func function to free the given trie
+ * @param add_func function to add a string to the given trie
+ * @param remove_func function to remove a string from the given trie
+ * @param size_func function to get the size of the given trie
+ * @return exit code describing failures, if any
+ */
+int fuzzy_test_trie_seed(FuzzyConfig conf) {
+    srand(conf.seed);
+
+    lsm_str *matrix = lsm_random_string_matrix(conf.word_count, conf.word_length);
+    bool* contains = calloc(conf.word_count, sizeof(bool));
+
+    // It's possible that the string matrix contains duplicate strings
+    bool** contains_dedupped = calloc(conf.word_count, sizeof(bool*));
+
+    for (int i = 0; i < conf.word_count; i++) {
+        if (contains_dedupped[i] == NULL) {
+            contains_dedupped[i] = &contains[i];
+
+            for (int j = i + 1; j < conf.word_count; j++) {
+                if (lsm_str_eq(&matrix[i], &matrix[j])) {
+                    contains_dedupped[j] = &contains[i];
+                }
+            }
+        }
+    }
+    
+    // We keep track of the size as well so that we can check whether this is
+    // also correct
+    size_t size = 0;
+
+    lsm_trie *trie;
+    lsm_trie_init(&trie);
+
+    bool changed;
+    lsm_error status;
+
+    // 0: success
+    // 1: invalid add
+    // 2: invalid remove
+    // 3: bad size after adds
+    // 4: bad size after removes
+    int exit_code = 0;
+
+    // Add all strings to trie, checking for duplicates
+    for (int i = 0; i < conf.word_count; i++) {
+      status = lsm_trie_insert(trie, &matrix[i], (void **)1);
+
+      // if changed is false, *contains_dedupped[i] should be true, as changed
+      // can only be false if the string is already contained in the trie. if
+      // changed is true, *contains_dedupped[i] should be false, as the string
+      // cannot be in the trie yet.
+      if (status == lsm_error_ok && *contains_dedupped[i]) {
+        exit_code = 1;
+        goto END;
+      }
+
+      if (!*contains_dedupped[i]) {
+        *contains_dedupped[i] = true;
+        size++;
+      }
+    }
+
+    // Ensure size is correct
+    if (lsm_trie_size(trie) != size) {
+      printf("%lu %lu\n", lsm_trie_size(trie), size);
+      exit_code = 3;
+      goto END;
+    }
+
+    // Remove all strings again, again taking duplicates into consideration
+    /* for (int i = 0; i < conf.word_count; i++) { */
+    /*   changed = remove_func(ct, matrix[i]); */
+
+    /*   // The string shouldn't be in the trie, yet another add operation */
+    /*   // says it added it as well */
+    /*   if (changed != *contains_dedupped[i]) { */
+    /*     exit_code = 2; */
+    /*     goto END; */
+    /*   } */
+
+    /*   if (*contains_dedupped[i]) { */
+    /*     *contains_dedupped[i] = false; */
+    /*     size--; */
+    /*   } */
+    /* } */
+
+    // Finally, check that the trie is completely empty
+    /* if (size_func(ct) != 0) { */
+    /*   exit_code = 4; */
+    /* } */
+
+END:
+    /* trie_free(ct); */
+
+    // Even testing functions should properly free memory
+    free(contains);
+    free(contains_dedupped);
+
+    for (int i = 0; i < conf.word_count; i++) {
+        lsm_str_zero(&matrix[i]);
+    }
+
+    free(matrix);
+
+    return exit_code;
+}
+
+/**
+ * Same as fuzzy_test_trie_seed, except that the seed is randomly generated.
+ *
+ * @param count how many strings to test with
+ * @param len maximum length of each string
+ * @param init_func function to creat a new trie of the wanted type
+ * @param free_func function to free the given trie
+ * @param add_func function to add a string to the given trie
+ * @param remove_func function to remove a string from the given trie
+ * @param size_func function to get the size of the given trie
+ * @return the generated seed if the test wasn't successful, -1 otherwise.
+ */
+/* int fuzzy_test_trie(int count, int len, void* (*init_func) (), void (*free_func) (void*), bool (*add_func) (void*, char*), bool (*remove_func) (void*, char*), int (*size_func) (void*)) { */
+/*     int seed = rand(); */
+/*     bool succeeded = fuzzy_test_trie_seed(seed, count, len, init_func, free_func, add_func, remove_func, size_func); */
+
+/*     if (!succeeded) { */
+/*         return seed; */
+/*     } */
+
+/*     return -1; */
+/* } */
+
+#endif

lsm/test/trie/trie.c

@@ -1,5 +1,5 @@
-#include "lsm.h"
 #include "test.h"
+#include "lsm.h"
 #include "lsm/trie_internal.h"
 
 #define TRIE_INIT() \

lsm/test/trie/trie_fuzzy.c

@@ -0,0 +1,35 @@
+#include "test.h"
+#include "lsm.h"
+#include "lsm/trie_internal.h"
+#include "fuzzy.h"
+
+void test_fuzzy() {
+  // Randomize seed
+  srand(time(NULL));
+
+  FuzzyConfig config;
+  int counter = 0;
+  int res;
+
+  for (int len = 1; len < 25; len += 5) {
+    for (int count = 10; count <= 500; count += 10) {
+      for (int i = 0; i < 1; i++) {
+        counter++;
+
+        config.seed = rand();
+        config.word_length = len;
+        config.word_count = count;
+
+        res = fuzzy_test_trie_seed(config);
+        TEST_CHECK_(res == 0,
+            "Failed config, seed = %i, len = %i, count = %i, code = %i", config.seed, config.word_length, config.word_count, res);
+      }
+    }
+  }
+  TEST_MSG("fuzzy tests done = %i", counter);
+}
+
+TEST_LIST = {
+  /* { "trie fuzzy", test_fuzzy }, */
+  { NULL, NULL}
+};