refactor(lsm): switch to uint8_t attr types; refactor disk code

include/lander.h

@@ -18,10 +18,10 @@ typedef struct lander_ctx {
   uint64_t remaining_data;
 } lander_ctx;
 
-typedef enum lander_attr_type : uint64_t {
-  lander_attr_type_entry_type = 1 << 0,
-  lander_attr_type_content_type = 1 << 1,
-  lander_attr_type_url = 1 << 2,
+typedef enum lander_attr_type : uint8_t {
+  lander_attr_type_entry_type = 0,
+  lander_attr_type_content_type = 1,
+  lander_attr_type_url = 2,
 } lander_attr_type;
 
 typedef enum lander_entry_type {

lsm/include/lsm/store.h

@@ -21,7 +21,7 @@ typedef struct lsm_entry_handle lsm_entry_handle;
  * @param entry entry to check
  * @param type type of attribute to check for
  */
-bool lsm_entry_attr_present(lsm_entry_handle *handle, uint64_t type);
+bool lsm_entry_attr_present(lsm_entry_handle *handle, uint8_t type);
 
 /**
  * Retrieve the contents of an attribute from an entry, if present
@@ -31,7 +31,7 @@ bool lsm_entry_attr_present(lsm_entry_handle *handle, uint64_t type);
  * @param type type of attribute to return
  */
 lsm_error lsm_entry_attr_get(lsm_str **out, lsm_entry_handle *handle,
-                             uint64_t type);
+                             uint8_t type);
 
 /**
  * Convenience wrapper around `lsm_entry_attr_get` that can be used if we know
@@ -42,7 +42,7 @@ lsm_error lsm_entry_attr_get(lsm_str **out, lsm_entry_handle *handle,
  * @param type type of attribute to return
  */
 lsm_error lsm_entry_attr_get_num(uint64_t *out, lsm_entry_handle *handle,
-                                 uint64_t type);
+                                 uint8_t type);
 
 /**
  * Add a new attribute to the entry.
@@ -51,7 +51,7 @@ lsm_error lsm_entry_attr_get_num(uint64_t *out, lsm_entry_handle *handle,
  * @param type type of attribute to add
  * @param data data of attribute; ownership of pointer is taken over
  */
-lsm_error lsm_entry_attr_insert(lsm_entry_handle *handle, uint64_t type,
+lsm_error lsm_entry_attr_insert(lsm_entry_handle *handle, uint8_t type,
                                 lsm_str *data);
 
 /**
@@ -62,7 +62,7 @@ lsm_error lsm_entry_attr_insert(lsm_entry_handle *handle, uint64_t type,
  * @param type type of attribute to add
  * @param data data of attribute
  */
-lsm_error lsm_entry_attr_insert_num(lsm_entry_handle *handle, uint64_t type,
+lsm_error lsm_entry_attr_insert_num(lsm_entry_handle *handle, uint8_t type,
                                     uint64_t data);
 
 /**
@@ -74,7 +74,7 @@ lsm_error lsm_entry_attr_insert_num(lsm_entry_handle *handle, uint64_t type,
  * @param type type of attribute to remove
  */
 lsm_error lsm_entry_attr_remove(lsm_str **out, lsm_entry_handle *handle,
-                                uint64_t type);
+                                uint8_t type);
 
 /**
  * A store consisting of LSM entries.

lsm/src/_include/lsm/store_internal.h

@@ -12,7 +12,7 @@
 #define LSM_IDX_FILE_NAME "lsm.idx"
 
 typedef struct lsm_attr {
-  uint64_t type;
+  uint8_t type;
   lsm_str *str;
 } lsm_attr;
 
@@ -26,8 +26,8 @@ typedef struct lsm_attr {
 typedef struct lsm_entry {
   lsm_str *key;
   struct {
-    uint64_t count;
-    uint64_t bitmap;
+    uint64_t bitmap[4];
+    uint8_t count;
     lsm_attr *items;
   } attrs;
   struct {

lsm/src/store/lsm_store_disk_read.c

@@ -1,6 +1,7 @@
 #include <stdlib.h>
 #include <string.h>
 
+#include "lsm.h"
 #include "lsm/store_internal.h"
 
 lsm_error lsm_store_load(lsm_store **ptr, lsm_str *data_path) {
@@ -79,48 +80,59 @@ lsm_error lsm_store_load(lsm_store **ptr, lsm_str *data_path) {
   return lsm_error_ok;
 }
 
-static lsm_error lsm_entry_read_attrs(lsm_entry_handle *handle, FILE *db_file) {
-  uint64_t attr_count;
-  size_t res = fread(&attr_count, sizeof(uint64_t), 1, db_file);
+static lsm_error lsm_fread(void *out, uint64_t *sum, FILE *f, uint64_t size,
+                           uint64_t count) {
+  size_t res = fread(out, size, count, f);
 
-  if (res == 0) {
+  if (res < count) {
     return lsm_error_failed_io;
   }
 
+  *sum += size * count;
+
+  return lsm_error_ok;
+}
+
+static lsm_error lsm_entry_read_str(lsm_str **out, uint64_t *sum, FILE *f) {
+  uint64_t len;
+  LSM_RES(lsm_fread(&len, sum, f, sizeof(uint64_t), 1));
+
+  char *buf = malloc(len + 1);
+
+  if (buf == NULL) {
+    return lsm_error_failed_alloc;
+  }
+
+  uint64_t read = 0;
+
+  while (read < len) {
+    read += fread(&buf[read], 1, len - read, f);
+  }
+
+  *sum += len;
+
+  return lsm_str_init(out, buf);
+}
+
+static lsm_error lsm_entry_read_attrs(uint64_t *sum, lsm_entry_handle *handle,
+                                      FILE *db_file) {
+  uint8_t attr_count;
+  LSM_RES(lsm_fread(&attr_count, sum, db_file, sizeof(uint8_t), 1));
+
   // attr_type, val_len
-  uint64_t nums[2];
+  uint8_t attr_type;
   lsm_str *val;
 
   for (uint64_t i = 0; i < attr_count; i++) {
-    res = fread(nums, sizeof(uint64_t), 2, db_file);
-
-    if (res < 2) {
-      return lsm_error_failed_io;
-    }
-
-    char *val_s = malloc(nums[1] + 1);
-    val_s[nums[1]] = '\0';
-
-    if (val_s == NULL) {
-      return lsm_error_failed_alloc;
-    }
-
-    uint64_t read = 0;
-
-    while (read < nums[1]) {
-      read += fread(&val_s[read], 1, nums[1] - read, db_file);
-    }
-
-    LSM_RES(lsm_str_init(&val, val_s));
-    ;
-    lsm_entry_attr_insert(handle, nums[0], val);
+    LSM_RES(lsm_fread(&attr_type, sum, db_file, sizeof(uint8_t), 1));
+    LSM_RES(lsm_entry_read_str(&val, sum, db_file));
+    lsm_entry_attr_insert(handle, attr_type, val);
   }
 
   return lsm_error_ok;
 }
 
 lsm_error lsm_store_load_db(lsm_store *store) {
-  uint64_t key_len;
   uint64_t db_dim[2];
   lsm_str *key;
   lsm_entry_handle *handle;
@@ -128,43 +140,13 @@ lsm_error lsm_store_load_db(lsm_store *store) {
   rewind(store->idx_file);
 
   // idx file starts with block count
-  size_t res =
-      fread(&store->idx_file_block_count, sizeof(uint64_t), 1, store->idx_file);
-
-  if (res == 0) {
-    return lsm_error_failed_io;
-  }
-
-  store->idx_file_size += sizeof(uint64_t);
+  LSM_RES(lsm_fread(&store->idx_file_block_count, &store->idx_file_size,
+                    store->idx_file, sizeof(uint64_t), 1));
 
   for (uint64_t i = 0; i < store->idx_file_block_count; i++) {
-    // Read in idx metadata
-    res = fread(&key_len, sizeof(uint64_t), 1, store->idx_file);
-
-    if (res == 0) {
-      return lsm_error_failed_io;
-    }
-
-    char *key_s = malloc(key_len + 1);
-    key_s[key_len] = '\0';
-
-    if (key_s == NULL) {
-      return lsm_error_failed_alloc;
-    }
-
-    res = fread(key_s, 1, key_len, store->idx_file);
-
-    if (res < key_len) {
-      return lsm_error_failed_io;
-    }
-
-    res = fread(db_dim, sizeof(uint64_t), 2, store->idx_file);
-
-    if (res < 2) {
-      return lsm_error_failed_io;
-    }
-
-    LSM_RES(lsm_str_init(&key, key_s));
+    LSM_RES(lsm_entry_read_str(&key, &store->idx_file_size, store->idx_file));
+    LSM_RES(lsm_fread(&db_dim, &store->idx_file_size, store->idx_file,
+                      sizeof(uint64_t), 2));
     LSM_RES(lsm_store_insert(&handle, store, key));
 
     // Read attributes from database file
@@ -172,10 +154,10 @@ lsm_error lsm_store_load_db(lsm_store *store) {
       return lsm_error_failed_io;
     }
 
-    LSM_RES(lsm_entry_read_attrs(handle, store->db_file));
+    LSM_RES(
+        lsm_entry_read_attrs(&store->idx_file_size, handle, store->db_file));
     lsm_entry_close(handle);
 
-    store->idx_file_size += 3 * sizeof(uint64_t) + key_len;
     store->db_file_size += db_dim[1];
   }
 

lsm/src/store/lsm_store_disk_write.c

@@ -1,9 +1,8 @@
 #include "lsm/store_internal.h"
 
-static lsm_error lsm_entry_write_uint64_t(FILE *f, uint64_t num) {
-  size_t res = fwrite(&num, sizeof(uint64_t), 1, f);
+static lsm_error lsm_entry_write_single(FILE *f, uint64_t size, void *val) {
+  size_t res = fwrite(val, size, 1, f);
 
-  // Such a small write should succeed in one go
   if (res == 0) {
     return lsm_error_failed_io;
   }
@@ -11,6 +10,10 @@ static lsm_error lsm_entry_write_uint64_t(FILE *f, uint64_t num) {
   return lsm_error_ok;
 }
 
+static lsm_error lsm_entry_write_uint64_t(FILE *f, uint64_t num) {
+  return lsm_entry_write_single(f, sizeof(uint64_t), &num);
+}
+
 static lsm_error lsm_entry_write_str(FILE *f, lsm_str *s) {
   uint64_t to_write = lsm_str_len(s);
   uint64_t written = 0;
@@ -35,17 +38,19 @@ lsm_error lsm_entry_write_db(uint64_t *size, FILE *db_file, lsm_entry *entry,
   LSM_RES(lsm_seek(db_file, pos));
 
   // First we write how many attributes follow
-  LSM_RES(lsm_entry_write_uint64_t(db_file, entry->attrs.count));
-  *size = sizeof(uint64_t);
+  LSM_RES(
+      lsm_entry_write_single(db_file, sizeof(uint8_t), &entry->attrs.count));
+  *size = sizeof(uint8_t);
 
-  for (uint64_t i = 0; i < entry->attrs.count; i++) {
+  for (uint8_t i = 0; i < entry->attrs.count; i++) {
     // Write attribute type, length & value
-    LSM_RES(lsm_entry_write_uint64_t(db_file, entry->attrs.items[i].type));
+    LSM_RES(lsm_entry_write_single(db_file, sizeof(uint8_t),
+                                   &entry->attrs.items[i].type));
     LSM_RES(lsm_entry_write_uint64_t(db_file,
                                      lsm_str_len(entry->attrs.items[i].str)));
     LSM_RES(lsm_entry_write_str(db_file, entry->attrs.items[i].str));
 
-    *size += 2 * sizeof(uint64_t) +
+    *size += sizeof(uint8_t) + sizeof(uint64_t) +
              lsm_str_len(entry->attrs.items[i].str) * sizeof(char);
   }
 

lsm/src/store/lsm_store_entry.c

@@ -56,12 +56,13 @@ void lsm_entry_close(lsm_entry_handle *handle) {
   free(handle);
 }
 
-bool lsm_entry_attr_present(lsm_entry_handle *handle, uint64_t type) {
-  return (handle->wrapper->entry->attrs.bitmap & type) != 0;
+bool lsm_entry_attr_present(lsm_entry_handle *handle, uint8_t type) {
+  return (handle->wrapper->entry->attrs.bitmap[type / 64] &
+          (((uint64_t)1) << (type % 64))) != 0;
 }
 
 lsm_error lsm_entry_attr_get(lsm_str **out, lsm_entry_handle *handle,
-                             uint64_t type) {
+                             uint8_t type) {
   if (!lsm_entry_attr_present(handle, type)) {
     return lsm_error_not_found;
   }
@@ -79,7 +80,7 @@ lsm_error lsm_entry_attr_get(lsm_str **out, lsm_entry_handle *handle,
 }
 
 lsm_error lsm_entry_attr_get_num(uint64_t *out, lsm_entry_handle *handle,
-                                 uint64_t type) {
+                                 uint8_t type) {
   lsm_str *s;
 
   LSM_RES(lsm_entry_attr_get(&s, handle, type));
@@ -96,7 +97,7 @@ lsm_error lsm_entry_attr_get_num(uint64_t *out, lsm_entry_handle *handle,
 }
 
 lsm_error lsm_entry_attr_remove(lsm_str **out, lsm_entry_handle *handle,
-                                uint64_t type) {
+                                uint8_t type) {
   if (!lsm_entry_attr_present(handle, type)) {
     return lsm_error_not_found;
   }
@@ -109,7 +110,7 @@ lsm_error lsm_entry_attr_remove(lsm_str **out, lsm_entry_handle *handle,
     free(entry->attrs.items);
     entry->attrs.items = NULL;
     entry->attrs.count = 0;
-    entry->attrs.bitmap = 0;
+    entry->attrs.bitmap[type / 64] &= ~(((uint64_t)1) << (type % 64));
 
     return lsm_error_ok;
   }
@@ -138,12 +139,12 @@ lsm_error lsm_entry_attr_remove(lsm_str **out, lsm_entry_handle *handle,
 
   entry->attrs.items = new_attrs;
   entry->attrs.count--;
-  entry->attrs.bitmap &= ~type;
+  entry->attrs.bitmap[type / 64] &= ~(((uint64_t)1) << (type % 64));
 
   return lsm_error_ok;
 }
 
-lsm_error lsm_entry_attr_insert(lsm_entry_handle *handle, uint64_t type,
+lsm_error lsm_entry_attr_insert(lsm_entry_handle *handle, uint8_t type,
                                 lsm_str *data) {
   if (lsm_entry_attr_present(handle, type)) {
     return lsm_error_already_present;
@@ -163,12 +164,12 @@ lsm_error lsm_entry_attr_insert(lsm_entry_handle *handle, uint64_t type,
 
   entry->attrs.items = new_attrs;
   entry->attrs.count++;
-  entry->attrs.bitmap |= type;
+  entry->attrs.bitmap[type / 64] |= ((uint64_t)1) << (type % 64);
 
   return lsm_error_ok;
 }
 
-lsm_error lsm_entry_attr_insert_num(lsm_entry_handle *handle, uint64_t type,
+lsm_error lsm_entry_attr_insert_num(lsm_entry_handle *handle, uint8_t type,
                                     uint64_t data) {
   lsm_str *s;
   LSM_RES(