feat: use custom binary file format for data

trie/src/trie.c

@@ -43,42 +43,45 @@ TrieExitCode trie_init(Trie **trie_ptr, const char *file_path) {
     return FileError;
   }
 
-  // We read in lines of at most 8192 characters (sounds like enough)
-  char buffer[8192];
-  EntryType type;
-  Entry *entry;
-  char *string;
-  int i, j;
+  uint64_t key_size, data_size;
+  char *key;
+  void *data;
   TrieExitCode status;
+  size_t items_read;
 
-  while (fgets(buffer, 8192, fp)) {
-    i = 0;
-
-    // Move index in buffer until we encounter first space character
-    while (buffer[i] != ' ') {
-      i++;
+  while (!feof(fp)) {
+    items_read = fread(&key_size, sizeof(uint64_t), 1, fp);
+    if (items_read < 1) {
+      break;
     }
 
-    // Split the buffer into two strings, the key and the payload
-    buffer[i] = '\0';
+    key = malloc(key_size + 1);
 
-    type = entry_type_from_char(buffer[i + 1]);
-
-    // Skip type character & its surrounding spaces
-    j = i + 3;
-
-    // Now remove the newline character
-    while (buffer[j] != '\n') {
-      j++;
+    items_read = fread(key, 1, key_size, fp);
+    if (items_read < key_size) {
+      break;
     }
 
-    buffer[j] = '\0';
+    key[key_size] = '\0';
 
-    entry = entry_new(type, buffer + i + 3);
-    status = trie_add_no_lock(trie, buffer, entry);
+    items_read = fread(&data_size, sizeof(uint64_t), 1, fp);
+    if (items_read < 1) {
+      break;
+    }
+
+    data = malloc(data_size);
+    items_read = fread(data, 1, data_size, fp);
+    if (items_read < data_size) {
+      break;
+    }
+
+    status = trie_add_no_lock(trie, key, data);
 
     if (status != Ok) {
       trie_free(trie);
+      free(key);
+      free(data);
+
       return status;
     }
   }
@@ -275,7 +278,8 @@ TrieExitCode trie_add_no_lock(Trie *trie, const char *string, void *data) {
   return Ok;
 }
 
-TrieExitCode trie_add(Trie *trie, const char *key, void *entry) {
+TrieExitCode trie_add(Trie *trie, const char *key, void *data,
+                      uint64_t data_len) {
   if (trie->file_path != NULL) {
     // Easiest way to make sure we don't add duplicate entries
     // We use an internal function that doesn't require a read lock, as we're
@@ -290,23 +294,24 @@ TrieExitCode trie_add(Trie *trie, const char *key, void *entry) {
       return FileError;
     }
 
-    /* fputs(key, fp); */
-    /* fputs(" ", fp); */
-    /* fputc(entry_type_to_char(entry->type), fp); */
-    /* fputs(" ", fp); */
-    /* fputs(entry->string, fp); */
-    /* fputs("\n", fp); */
+    // First we write the key, then the actual data
+    uint64_t key_len = (uint64_t)strlen(key);
+    fwrite(&key_len, 1, sizeof(uint64_t), fp);
+    fwrite(key, 1, key_len, fp);
+
+    fwrite(&data_len, 1, sizeof(uint64_t), fp);
+    fwrite(data, 1, data_len, fp);
 
     fclose(fp);
   }
 
   // This function *should* always return Ok. Otherwise, the function would've
   // exited because the string was found in the trie.
-  return trie_add_no_lock(trie, key, entry);
+  return trie_add_no_lock(trie, key, data);
 }
 
 TrieExitCode trie_add_random(Trie *trie, char **key_ptr, void *data,
-                             bool secure) {
+                             uint64_t data_len, bool secure) {
   // Generate random key
   bool ok = false;
   int key_length = secure ? RANDOM_KEY_LENGTH_LONG : RANDOM_KEY_LENGTH_SHORT;
@@ -324,7 +329,7 @@ TrieExitCode trie_add_random(Trie *trie, char **key_ptr, void *data,
     ok = trie_search_node(trie, key).child == NULL;
   }
 
-  TrieExitCode return_value = trie_add(trie, key, data);
+  TrieExitCode return_value = trie_add(trie, key, data, data_len);
 
   if (return_value == Ok) {
     *key_ptr = key;

trie/src/trie_entry.c

@@ -23,23 +23,20 @@ char entry_type_to_char(EntryType et) {
   }
 }
 
-Entry *entry_new(EntryType type, const char *string) {
-  Entry *entry = malloc(sizeof(Entry));
+uint64_t entry_new(Entry **entry_ptr, EntryType type, const char *string) {
+  size_t str_len = strlen(string);
+  uint64_t entry_size = sizeof(EntryType) + str_len + 1;
+  Entry *entry = malloc(entry_size);
   entry->type = type;
 
   if (string != NULL) {
-    entry->string = strdup(string);
+    memcpy(entry->string, string, str_len + 1);
   } else {
-    entry->string = NULL;
+    entry->string[0] = '\0';
   }
 
-  return entry;
+  *entry_ptr = entry;
+  return entry_size;
 }
 
-void entry_free(Entry *entry) {
-  if (entry->string != NULL) {
-    free(entry->string);
-  }
-
-  free(entry);
-}
+void entry_free(Entry *entry) { free(entry); }