tinybits 0.9.0 → 0.9.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d36aff126ea60de67bc4050896a329f39e53a155cf50f4b12f35602757dc4ab1
-  data.tar.gz: 610909057ad8ae5f8fa0dab0e1dd5ee03105b7fb0036b11e3ac22eea97522cee
+  metadata.gz: b196201c8c0379a74ae23e62d10022ed7deae323cba4d9281025941b1a8b5d77
+  data.tar.gz: 47acdb4977e7cbf79a48c74ea85a8917be38357f886a88ab33f483b8d396c34d
 SHA512:
-  metadata.gz: fb4ce2802d01a5a62a119edc8f8c7ea06aff12edf0eff1d9b0d3a758d096e36c40c49c3ca52c22368da199b65b75900db1a5296580e38b9d2c00da62b80d6535
-  data.tar.gz: 16355ff8ae048bf3ae8d50eb2067cc42e01e21da1a396713d3a74e831e163658f3631dc19c09fc52c867fcf4b24bd29399029f44fabe9ac92be99eaccaa66a9b
+  metadata.gz: ad078aa3bf66b321bf208d679393ba9856a553d321bacc8d3b62728feca300bf42fa5e497ecfd87c6b2bb80050dfbf61d70b8edd2378e1650095b4314303ccbb
+  data.tar.gz: 9b441821ee20ddaa0fbbecced7de678bf4dd1e0f7874c3174a73d843bee7b0ed720a37aa08a622117f24ab3a03acd15d448b90e325b1da9f662825ac85128c94

data/ext/tinybits/tinybits_ext.c

@@ -326,23 +326,59 @@ static VALUE rb_reset(VALUE self){
     return self;
 }
 
+#define KEY_CACHE_CAPA 63
+#define KEY_CACHE_MAX_MISSES 1024
+
 // Unpacker structure
 typedef struct {
     tiny_bits_unpacker* unpacker;
     int symbolize_keys; // map keys that are strings come back as symbols
     VALUE strings;      // Ruby strings by dedupe id, dictionary strings first
     VALUE buffer;       // the buffer set by #buffer=, kept alive while popping
+    int keys_count;
+    int keys_misses;    // misses while the cache is full
+    VALUE keys[KEY_CACHE_CAPA]; // interned map keys sorted by (length, bytes), kept across unpacks
 } UnpackerData;
 
 static void unpacker_mark(void* data) {
     UnpackerData* unpacker_data = (UnpackerData*)data;
     rb_gc_mark_movable(unpacker_data->strings);
     rb_gc_mark(unpacker_data->buffer); // pinned, the C unpacker points into it (embedded strings move)
+    for (int i = 0; i < unpacker_data->keys_count; i++) rb_gc_mark_movable(unpacker_data->keys[i]);
 }
 
 static void unpacker_compact(void* data) {
     UnpackerData* unpacker_data = (UnpackerData*)data;
     unpacker_data->strings = rb_gc_location(unpacker_data->strings);
+    for (int i = 0; i < unpacker_data->keys_count; i++) unpacker_data->keys[i] = rb_gc_location(unpacker_data->keys[i]);
+}
+
+// Map keys repeat across unpacks, a small sorted cache saves the global fstring table lookup
+// (same approach as the json gem's name cache, but persistent; when full, misses are not
+// inserted, and after KEY_CACHE_MAX_MISSES of them it restarts empty to follow new keys)
+static inline VALUE key_cache_fetch(UnpackerData* unpacker_data, const char* str, long length) {
+    if (length < 2 || length > 128) // same range as string dedupe
+        return rb_enc_interned_str(str, length, rb_utf8_encoding());
+    VALUE* keys = unpacker_data->keys;
+    int low = 0, high = unpacker_data->keys_count - 1;
+    while (low <= high) {
+        int mid = (low + high) >> 1;
+        long cmp = length - RSTRING_LEN(keys[mid]);
+        if (cmp == 0) cmp = memcmp(str, RSTRING_PTR(keys[mid]), length);
+        if (cmp == 0) return keys[mid];
+        if (cmp > 0) low = mid + 1; else high = mid - 1;
+    }
+    VALUE key = rb_enc_interned_str(str, length, rb_utf8_encoding());
+    if (unpacker_data->keys_count == KEY_CACHE_CAPA) {
+        if (++unpacker_data->keys_misses < KEY_CACHE_MAX_MISSES) return key;
+        unpacker_data->keys_misses = 0;
+        unpacker_data->keys_count = 0;
+        low = 0;
+    }
+    MEMMOVE(&keys[low + 1], &keys[low], VALUE, unpacker_data->keys_count - low);
+    keys[low] = key;
+    unpacker_data->keys_count++;
+    return key;
 }
 
 static void unpacker_free(void* data) {
@@ -369,6 +405,8 @@ static VALUE rb_unpacker_alloc(VALUE klass) {
     unpacker_data->symbolize_keys = 0;
     unpacker_data->strings = Qnil;
     unpacker_data->buffer = Qnil;
+    unpacker_data->keys_count = 0;
+    unpacker_data->keys_misses = 0;
     return TypedData_Wrap_Struct(klass, &unpacker_data_type, unpacker_data);
 }
 
@@ -421,7 +459,7 @@ static inline VALUE rb_unpack_str(VALUE strings, UnpackerData* unpacker_data, ti
     else if(id <= 0){
         VALUE str;
         if(interned > 0){
-            str = rb_enc_interned_str(value.str_blob_val.data, value.str_blob_val.length, rb_utf8_encoding());
+            str = key_cache_fetch(unpacker_data, value.str_blob_val.data, value.str_blob_val.length);
         } else {
             str = rb_utf8_str_new(value.str_blob_val.data, value.str_blob_val.length);
             rb_obj_freeze(str);
@@ -464,6 +502,14 @@ static VALUE unpack_ruby_object(VALUE strings, UnpackerData* unpacker_data, size
         case TINY_BITS_FALSE:
             return Qfalse;
         case TINY_BITS_ARRAY: {
+            if (value.length <= 16) { // small arrays are created in one call
+                VALUE items[16];
+                for (size_t i = 0; i < value.length; i++) {
+                    items[i] = unpack_ruby_object(strings, unpacker_data, 0);
+                    if (items[i] == Qundef) return Qundef; // Error
+                }
+                return rb_ary_new_from_values(value.length, items);
+            }
             VALUE arr = rb_ary_new_capa(value.length);
             for (size_t i = 0; i < value.length; i++) {
                 VALUE element = unpack_ruby_object(strings, unpacker_data, 0);

data/lib/tinybits/version.rb

@@ -1,3 +1,3 @@
 module TinyBits
-  VERSION = '0.9.0'.freeze
+  VERSION = '0.9.1'.freeze
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: tinybits
 version: !ruby/object:Gem::Version
-  version: 0.9.0
+  version: 0.9.1
 platform: ruby
 authors:
 - Mohamed Hassan