1、概述
经常使用isset
判断变量或数组中的键是否存在, 但是数组中可以使用array_key_exists
这个函数, 那么这两个谁最优呢?
官方文档对两者的定义
- | 分类 | 描述 | 文档 |
---|---|---|---|
isset | 语言构造器 | 检测变量是否已设置并且非 NULL | http://php.net/manual/zh/function.isset.php |
array_key_exists | 函数 | 检查数组里是否有指定的键名或索引 | http://php.net/manual/zh/function.array-key-exists.php |
isset()
对于数组中为 NULL 的值不会返回 TRUE,而array_key_exists()
会。array_key_exists()
仅仅搜索第一维的键。 多维数组里嵌套的键不会被搜索到。
要检查对象是否有某个属性,应该去用property_exists()
。
2、测试
2.1 测试环境
OS | PHP | PHPUnit |
---|---|---|
MacOS 10.13.6 | PHP 7.2.7 (cli) | PHPUnit 6.5.7 |
2.2 单元测试
class issetTest extends \PHPUnit\Framework\TestCase{ ???/** ????* @dataProvider dataArr ????*/ ???public function testName($arr) ???{ ???????$this->assertTrue(isset($arr[‘name‘])); ???????$this->assertFalse(isset($arr[‘age‘])); ???????$this->assertTrue(isset($arr[‘sex‘])); ???????$this->assertTrue(array_key_exists(‘name‘, $arr)); ???????$this->assertTrue(array_key_exists(‘age‘, $arr)); ???????$this->assertTrue(array_key_exists(‘sex‘, $arr)); ???????$this->assertFalse(empty($arr[‘name‘])); ???????$this->assertTrue(empty($arr[‘age‘])); ???????$this->assertTrue(empty($arr[‘sex‘])); ???} ???public function dataArr() ???{ ???????return [ ???????????[ ???????????????[‘name‘ => 123, ‘age‘ => null, ‘sex‘ => 0] ???????????] ???????]; ???}}/*PHPUnit 6.5.7 by Sebastian Bergmann and contributors.. ??????????????????????????????????????????????????????????????????1 / 1 (100%)Time: 113 ms, Memory: 8.00MBOK (1 test, 9 assertions)*/
2.3 性能-执行时间
如上, php cli
环境下, 执行10000000
次, 测试代码和执行时间如下:
<?php$arr = [ ???‘name‘ => 123, ???‘age‘ => null];$max = 10000000;testFunc($arr, ‘name‘, $max);testFunc($arr, ‘age‘, $max);function testFunc($arr, $key, $max = 1000){ ???echo ‘`$arr[\‘‘, $key, ‘\‘]` | - | -‘, PHP_EOL; ???$startTime = microtime(true); ???for ($i = 0; $i <= $max; $i++) { ???????isset($arr[$key]); ???} ???echo ‘^ | isset | ?‘, microtime(true) - $startTime, PHP_EOL; ???$startTime = microtime(true); ???for ($i = 0; $i <= $max; $i++) { ???????array_key_exists($key, $arr); ???} ???echo ‘^ | array_key_exists | ‘, microtime(true) - $startTime, PHP_EOL; ???$startTime = microtime(true); ???for ($i = 0; $i <= $max; $i++) { ???????isset($arr[$key]) || array_key_exists($key, $arr); ???} ???echo ‘^ | isset or array_key_exists | ‘, microtime(true) - $startTime, PHP_EOL;}
PHP 5.6
-|函数|执行时间(s)
$arr[‘name‘] | - | - |
^ | isset | 0.64719796180725 |
^ | array_key_exists | 2.5713651180267 |
^ | isset or array_key_exists | 1.1359150409698 |
$arr[‘age‘] | - | - |
^ | isset | 0.53988218307495 |
^ | array_key_exists | 2.7240340709686 |
^ | isset or array_key_exists | 2.9613540172577 |
PHP 7.2.4
-|函数|执行时间(s)
$arr[‘name‘] | - | - |
^ | isset | 0.24308800697327 |
^ | array_key_exists | 0.3645191192627 |
^ | isset or array_key_exists | 0.28933310508728 |
$arr[‘age‘] | - | - |
^ | isset | 0.23279714584351 |
^ | array_key_exists | 0.33850502967834 |
^ | isset or array_key_exists | 0.54935812950134 |
2.4 性能-使用VLD查看opcode
/usr/local/Cellar/php/7.2.7/bin/php -d vld.active=1 -dvld.verbosity=3 vld.php
描述 | isset | array_key_exists |
---|---|---|
code | $arr = [‘name‘ => ‘li‘]; isset($arr[‘name‘]); | $arr = [‘name‘ => ‘li‘]; array_key_exists(‘name‘, $arr); |
-dvld.active=1 | ||
-dvld.verbosity=3 |
3、源码
3.1 isset 源码分析
Zend/zend_language_scanner.l (Scanning阶段)
Scanning阶段,程序会扫描zend_language_scanner.l文件将代码文件转换成语言片段。
<ST_IN_SCRIPTING>"isset" { ???RETURN_TOKEN(T_ISSET);}
可见 isset 生成对应的token为 T_ISSET
3.1.2 Zend/zend_language_parser.y (Parsing阶段)
当执行PHP源码,会先进行语法分析,isset的yacc如下:
接下来就到了Parsing阶段,这个阶段,程序将 T_ISSET
等Tokens转换成有意义的表达式,此时会做语法分析,Tokens的yacc保存在zend_language_parser.y
文件中。isset
的yacc如下(T_ISSET
):
internal_functions_in_yacc: ???????T_ISSET ‘(‘ isset_variables ‘)‘ { $$ = $3; } ???| ???T_EMPTY ‘(‘ expr ‘)‘ { $$ = zend_ast_create(ZEND_AST_EMPTY, $3); } ???| ???T_INCLUDE expr ???????????{ $$ = zend_ast_create_ex(ZEND_AST_INCLUDE_OR_EVAL, ZEND_INCLUDE, $2); } ???| ???T_INCLUDE_ONCE expr ???????????{ $$ = zend_ast_create_ex(ZEND_AST_INCLUDE_OR_EVAL, ZEND_INCLUDE_ONCE, $2); } ???| ???T_EVAL ‘(‘ expr ‘)‘ ???????????{ $$ = zend_ast_create_ex(ZEND_AST_INCLUDE_OR_EVAL, ZEND_EVAL, $3); } ???| ???T_REQUIRE expr ???????????{ $$ = zend_ast_create_ex(ZEND_AST_INCLUDE_OR_EVAL, ZEND_REQUIRE, $2); } ???| ???T_REQUIRE_ONCE expr ???????????{ $$ = zend_ast_create_ex(ZEND_AST_INCLUDE_OR_EVAL, ZEND_REQUIRE_ONCE, $2); };isset_variables: ???????isset_variable { $$ = $1; } ???| ???isset_variables ‘,‘ isset_variable ???????????{ $$ = zend_ast_create(ZEND_AST_AND, $1, $3); };isset_variable: ???????expr { $$ = zend_ast_create(ZEND_AST_ISSET, $1); };%%/* Zend/zend_ast.c */# zend_ast_export_excase ZEND_AST_EMPTY: ???FUNC_OP("empty");case ZEND_AST_ISSET: ???FUNC_OP("isset");
最终执行了zend_ast_create(ZEND_AST_ISSET, $1);
我们知道, PHP7开始, 语法解析过程的产物保存于CG(AST),接着zend引擎会把AST进一步编译为 zend_op_array ,它是编译阶段最终的产物,也是执行阶段的输入
3.1.3 Zend/zend_compile.c(将表达式编译成opcodes)
将表达式编译成opcodes,可见isset
对应的opcodes为ZEND_AST_ISSET
。打开zend_compile.c
文件
# void zend_compile_expr(znode *result, zend_ast *ast) /* {{{ */ ???case ZEND_AST_ISSET:case ZEND_AST_EMPTY: ???zend_compile_isset_or_empty(result, ast); ???return;
最终执行了zend_compile_isset_or_empty
函数,在源码目录中查找, 可以发现,此函数也在 zend_compile.c 文件中定义。
void zend_compile_isset_or_empty(znode *result, zend_ast *ast) /* {{{ */{ ???zend_ast *var_ast = ast->child[0]; ???znode var_node; ???zend_op *opline = NULL; ???ZEND_ASSERT(ast->kind == ZEND_AST_ISSET || ast->kind == ZEND_AST_EMPTY); ???if (!zend_is_variable(var_ast) || zend_is_call(var_ast)) { ???????if (ast->kind == ZEND_AST_EMPTY) { ???????????/* empty(expr) can be transformed to !expr */ ???????????zend_ast *not_ast = zend_ast_create_ex(ZEND_AST_UNARY_OP, ZEND_BOOL_NOT, var_ast); ???????????zend_compile_expr(result, not_ast); ???????????return; ???????} else { ???????????zend_error_noreturn(E_COMPILE_ERROR, ???????????????"Cannot use isset() on the result of an expression " ???????????????"(you can use \"null !== expression\" instead)"); ???????} ???} ???switch (var_ast->kind) { ???????case ZEND_AST_VAR: ???????????if (is_this_fetch(var_ast)) { ???????????????opline = zend_emit_op(result, ZEND_ISSET_ISEMPTY_THIS, NULL, NULL); ???????????} else if (zend_try_compile_cv(&var_node, var_ast) == SUCCESS) { ???????????????opline = zend_emit_op(result, ZEND_ISSET_ISEMPTY_VAR, &var_node, NULL); ???????????????opline->extended_value = ZEND_FETCH_LOCAL | ZEND_QUICK_SET; ???????????} else { ???????????????opline = zend_compile_simple_var_no_cv(result, var_ast, BP_VAR_IS, 0); ???????????????opline->opcode = ZEND_ISSET_ISEMPTY_VAR; ???????????} ???????????break; ???????case ZEND_AST_DIM: ???????????opline = zend_compile_dim_common(result, var_ast, BP_VAR_IS); ???????????opline->opcode = ZEND_ISSET_ISEMPTY_DIM_OBJ; ???????????break; ???????case ZEND_AST_PROP: ???????????opline = zend_compile_prop_common(result, var_ast, BP_VAR_IS); ???????????opline->opcode = ZEND_ISSET_ISEMPTY_PROP_OBJ; ???????????break; ???????case ZEND_AST_STATIC_PROP: ???????????opline = zend_compile_static_prop_common(result, var_ast, BP_VAR_IS, 0); ???????????opline->opcode = ZEND_ISSET_ISEMPTY_STATIC_PROP; ???????????break; ???????EMPTY_SWITCH_DEFAULT_CASE() ???} ???result->op_type = opline->result_type = IS_TMP_VAR; ???opline->extended_value |= ast->kind == ZEND_AST_ISSET ? ZEND_ISSET : ZEND_ISEMPTY;}/* }}} */
从这个函数最后一行可以看出,最终执行的还是ZEND_ISSET
, 网站被黑,根据不同的用法会使用不同的opcode处理, 此处以ZEND_ISSET_ISEMPTY_DIM_OBJ
为例。
3.1.4 Zend/zend_vm_execute.h (执行opcodes)
opcode 对应处理函数的命名规律:
ZEND_[opcode]_SPEC_(变量类型1)_(变量类型2)_HANDLER
变量类型1和变量类型2是可选的,如果同时存在,那就是左值和右值,动漫资源,归纳有下几类: VAR TMP CV UNUSED CONST 这样可以根据相关的执行场景来判定。
zend_vm_execute.h: ZEND_ISSET_ISEMPTY_DIM_OBJ_SPEC_CONST_CONST_HANDLER,zend_vm_execute.h: ZEND_ISSET_ISEMPTY_DIM_OBJ_SPEC_CONST_TMPVAR_HANDLER,zend_vm_execute.h: ZEND_ISSET_ISEMPTY_DIM_OBJ_SPEC_CONST_CV_HANDLER,zend_vm_execute.h: ZEND_ISSET_ISEMPTY_DIM_OBJ_SPEC_TMPVAR_CONST_HANDLER,zend_vm_execute.h: ZEND_ISSET_ISEMPTY_DIM_OBJ_SPEC_TMPVAR_TMPVAR_HANDLER,zend_vm_execute.h: ZEND_ISSET_ISEMPTY_DIM_OBJ_SPEC_TMPVAR_CV_HANDLER,zend_vm_execute.h: ZEND_ISSET_ISEMPTY_DIM_OBJ_SPEC_TMPVAR_CONST_HANDLER,zend_vm_execute.h: ZEND_ISSET_ISEMPTY_DIM_OBJ_SPEC_TMPVAR_TMPVAR_HANDLER,zend_vm_execute.h: ZEND_ISSET_ISEMPTY_DIM_OBJ_SPEC_TMPVAR_CV_HANDLER,zend_vm_execute.h: ZEND_ISSET_ISEMPTY_DIM_OBJ_SPEC_CV_CONST_HANDLER,zend_vm_execute.h: ZEND_ISSET_ISEMPTY_DIM_OBJ_SPEC_CV_TMPVAR_HANDLER,zend_vm_execute.h: ZEND_ISSET_ISEMPTY_DIM_OBJ_SPEC_CV_CV_HANDLER,
我们看下 ZEND_ISSET_ISEMPTY_DIM_OBJ_SPEC_CV_CV_HANDLER
这个处理函数
if (opline->extended_value & ZEND_ISSET) { ???/* > IS_NULL means not IS_UNDEF and not IS_NULL */ ???result = value != NULL && Z_TYPE_P(value) > IS_NULL && ???????(!Z_ISREF_P(value) || Z_TYPE_P(Z_REFVAL_P(value)) != IS_NULL);} else /* if (opline->extended_value & ZEND_ISEMPTY) */ { ???result = (value == NULL || !i_zend_is_true(value));}
上面的 if ... else 就是判断是isset,还是empty,然后做不同处理,Z_TYPE_P, i_zend_is_true 不同判断。
可见,isset
的最终实现是通过 Z_TYPE_P 获取变量类型,然后再进行判断的。
函数的完整定义请查看Zend/zend_vm_execute.h
,以下是 i_zend_is_true
和 Z_TYPE_P
的定义:
- Zend/zend_operators.h 中的i_zend_is_true()
- Zend/zend_types.h 中的 Z_TYPE_P()
3.2 array_key_exists 源码分析
3.2.1 ext/standard/array.c (数组扩展中实现)
array_key_exists
是php内置函数,通过扩展方式实现的。打开php源码,ext/standard/目录下
// ? ?standard git:(master) ? grep -r ‘PHP_FUNCTION(array_key_exists)‘ *array.c: PHP_FUNCTION(array_key_exists)php_array.h: PHP_FUNCTION(array_key_exists);
具体实现如下:
/* {{{ proto bool array_key_exists(mixed key, array search) ??Checks if the given key or index exists in the array */PHP_FUNCTION(array_key_exists){ ???zval *key; ???????????????????/* key to check for */ ???HashTable *array; ???????????/* array to check in */#ifndef FAST_ZPP ???if (zend_parse_parameters(ZEND_NUM_ARGS(), "zH", &key, &array) == FAILURE) { ???????return; ???}#else ???ZEND_PARSE_PARAMETERS_START(2, 2) ???????Z_PARAM_ZVAL(key) ???????Z_PARAM_ARRAY_OR_OBJECT_HT(array) ???ZEND_PARSE_PARAMETERS_END();#endif ???switch (Z_TYPE_P(key)) { ???????case IS_STRING: ???????????if (zend_symtable_exists_ind(array, Z_STR_P(key))) { ???????????????RETURN_TRUE; ???????????} ???????????RETURN_FALSE; ???????case IS_LONG: ???????????if (zend_hash_index_exists(array, Z_LVAL_P(key))) { ???????????????RETURN_TRUE; ???????????} ???????????RETURN_FALSE; ???????case IS_NULL: ???????????if (zend_hash_exists_ind(array, ZSTR_EMPTY_ALLOC())) { ???????????????RETURN_TRUE; ???????????} ???????????RETURN_FALSE; ???????default: ???????????php_error_docref(NULL, E_WARNING, "The first argument should be either a string or an integer"); ???????????RETURN_FALSE; ???}}/* }}} */
可以看到, 是通过 Z_TYPE_P
宏获取变量类型, 通过 zend_hash
相关函数判断 key
是否存在。以key
为字符串为例,在Zend/zend_hash.h
追踪具体实现:
3.2.2 Zend/zend_hash.h
ZEND_API zval* ZEND_FASTCALL zend_hash_find(const HashTable *ht, zend_string *key);...static zend_always_inline int zend_symtable_exists_ind(HashTable *ht, zend_string *key){ ???zend_ulong idx; ???if (ZEND_HANDLE_NUMERIC(key, idx)) { ???????return zend_hash_index_exists(ht, idx); ???} else { ???????return zend_hash_exists_ind(ht, key); ???}}static zend_always_inline int zend_hash_exists_ind(const HashTable *ht, zend_string *key){ ???zval *zv; ???zv = zend_hash_find(ht, key); ???return zv && (Z_TYPE_P(zv) != IS_INDIRECT || ???????????Z_TYPE_P(Z_INDIRECT_P(zv)) != IS_UNDEF);}
再次先通过函数ZEND_HANDLE_NUMERIC
对key做判断,看这个字符串是不是数字类型的, 当key
为数字时执行 zend_hash_index_exists
, 实现如下:
3.2.3 Zend/zend_hash.c
3.2.3.1 zend_hash_index_exists()
/** * 这里有一个宏HASH_FLAG_PACKED,为真就代表当前数组的key都是系统生成的,也就是说是按从0到1,2,3等等按序排列的,所以判读键为key的是否存在,直接检查arData数组中第idx个元素是否有定义就行了,这里不涉及什么hash查找,冲突解决等一系列问题。 * ??* 但如果HASH_FLAG_PACKED为假,那么肯定就需要先计算idx的hash值,找到key为idx的数据应该在arData的第几位才行。这就要通过函数zend_hash_index_find_bucket了。 */ZEND_API zend_bool ZEND_FASTCALL zend_hash_index_exists(const HashTable *ht, zend_ulong h){ ???Bucket *p; ???IS_CONSISTENT(ht); ???if (ht->u.flags & HASH_FLAG_PACKED) { ???????if (h < ht->nNumUsed) { ???????????if (Z_TYPE(ht->arData[h].val) != IS_UNDEF) { ???????????????return 1; ???????????} ???????} ???????return 0; ???} ???p = zend_hash_index_find_bucket(ht, h); ???return p ? 1 : 0;}
3.2.3.2 zend_hash_find()
在Zend/zend_hash.c
中有zend_hash_find()
的实现, code如下:
/*++-- zend_hash_find --++*//* Returns the hash table data if found and NULL if not. */ZEND_API zval* ZEND_FASTCALL zend_hash_find(const HashTable *ht, zend_string *key){ ???Bucket *p; ???IS_CONSISTENT(ht); ???p = zend_hash_find_bucket(ht, key); ???return p ? &p->val : NULL;}
3.2.3.3 zend_hash_index_find_bucket()
static zend_always_inline Bucket *zend_hash_index_find_bucket(const HashTable *ht, zend_ulong h){ ???uint32_t nIndex; ???uint32_t idx; ???Bucket *p, *arData; ???arData = ht->arData; ???nIndex = h | ht->nTableMask; ???idx = HT_HASH_EX(arData, nIndex); ???while (idx != HT_INVALID_IDX) { ???????ZEND_ASSERT(idx < HT_IDX_TO_HASH(ht->nTableSize)); ???????p = HT_HASH_TO_BUCKET_EX(arData, idx); ???????if (p->h == h && !p->key) { ???????????return p; ???????} ???????idx = Z_NEXT(p->val); ???} ???return NULL;}
3.2.3.4 zend_hash_find_bucket()
static zend_always_inline Bucket *zend_hash_find_bucket(const HashTable *ht, zend_string *key){ ???zend_ulong h; ???uint32_t nIndex; ???uint32_t idx; ???Bucket *p, *arData; ???h = zend_string_hash_val(key); ???arData = ht->arData; ???nIndex = h | ht->nTableMask; ???idx = HT_HASH_EX(arData, nIndex); ???while (EXPECTED(idx != HT_INVALID_IDX)) { ???????p = HT_HASH_TO_BUCKET_EX(arData, idx); ???????if (EXPECTED(p->key == key)) { /* check for the same interned string */ ???????????return p; ???????} else if (EXPECTED(p->h == h) && ????????????EXPECTED(p->key) && ????????????EXPECTED(ZSTR_LEN(p->key) == ZSTR_LEN(key)) && ????????????EXPECTED(memcmp(ZSTR_VAL(p->key), ZSTR_VAL(key), ZSTR_LEN(key)) == 0)) { ???????????return p; ???????} ???????idx = Z_NEXT(p->val); ???} ???return NULL;}
这里需要明白一点,数字的哈希值就等于他本身,所以才有不计算h的哈希值,就执行h | ht->nTableMask。
然后处理一下冲突,最后得出key为idx的数据是否存在于数组中。
如果idx确确实实是字符串,那么思路更简单一点,最后通过zen_hash_find_bucket来判断是否存在,与上面zend_hash_index_find_bucket不同的是,函数中要先计算字符串key的哈希值,然后再执行h | ht->nTableMask。
如下,
zend_symtable_exists_ind --ZEND_HANDLE_NUMERIC{ZEND_HANDLE_NUMERIC}ZEND_HANDLE_NUMERIC --> zend_hash_index_existsZEND_HANDLE_NUMERIC --> zend_hash_exists_indzend_hash_index_exists-->zend_hash_index_find_bucketzend_hash_exists_ind-->zend_hash_findzend_hash_find-->zend_hash_find_bucket
4、总结
isset
效率高于array_key_exists
, PHP7之后有30%左右的提升, php5.6有将近70%的提升。isset
是语法结构,array_key_exists
是函数, 调用开销要小。isset
通过Z_TYPE_P
获取变量类型,然后再进行判断实现的;array_key_exists
则是通过hash查找来实现的。- 对于数组,
isset
的性能要高于array_key_exists
所以,如果数组比较大,我们应该用如下方法保证性能和准确性isset or array_key_exists
。
5、扩展阅读
- PHP源码
- PHP7内核剖析
- 升级 php7 后 isset 不太对了
- [[PHP源码阅读]empty和isset函数](http://www.hoohack.me/2016/05...
- 深入理解PHP内核
- PHP安装与使用VLD查看opcode代码
- array_key_exists官方文档
- isset官方文档
深入理解PHP之isset和array_key_exists对比
原文地址:https://www.cnblogs.com/asd667/p/9798781.html