深入php内核之php in array

先给大家介绍php in array函数基本知识热热身。

定义和用法

in_array() 函数在数组中搜索给定的值。

语法
in_array(value,array,type)

 说明

如果给定的值 value 存在于数组 array 中则返回 true。如果第三个参数设置为 true，函数只有在元素存在于数组中且数据类型与给定值相同时才返回 true。

如果没有在数组中找到参数，函数返回 false。

注释：如果 value 参数是字符串，且 type 参数设置为 true，则搜索区分大小写。

无意中看到一段代码

<?php    
$y="1800";
$x = array();
for($j=0;$j<50000;$j++){
$x[]= "{$j}";
}
for($i=0;$i<30000;$i++){
if(in_array($y,$x)){
 continue;
}
}

测试了一下

[root@dev tmp]# time php b.php
real    0m9.517s
user    0m4.486s
sys     0m0.015s

竟然需要9s

in_array是这个样子的

bool in_array ( mixed $needle , array $haystack [, bool $strict = FALSE ] )

在 haystack 中搜索 needle，如果没有设置 strict 则使用宽松的比较。

needle

待搜索的值。如果 needle 是字符串，则比较是区分大小写的。

haystack

这个数组。

strict

如果第三个参数 strict 的值为 TRUE 则 in_array() 函数还会检查 needle 的类型是否和 haystack 中的相同。

那么我看一下源代码

第一步 在ext/standard/array.c 文件中

/* }}} */              
/* {{{ proto bool in_array(mixed needle, array haystack [, bool strict])
Checks if the given value exists in the array */  
PHP_FUNCTION(in_array)          
{                
php_search_array(INTERNAL_FUNCTION_PARAM_PASSTHRU, 0);
}                
/* }}} */            
/* {{{ proto mixed array_search(mixed needle, array haystack [, bool strict])
Searches the array for a given value and returns the corresponding key if successful */
PHP_FUNCTION(array_search)        
{            
php_search_array(INTERNAL_FUNCTION_PARAM_PASSTHRU, 1);
}                
/* }}} */

顺便看到了array_search，原来和in_array的内部实现基本一致

其中函数的参数 在./zend.h中

#define INTERNAL_FUNCTION_PARAM_PASSTHRU ht, return_value, return_value_ptr, this_ptr, return_value_used TSRMLS_CC

第二步 在ext/standard/array.c 文件中 查看php_search_array原型

/* void php_search_array(INTERNAL_FUNCTION_PARAMETERS, int behavior)
* 0 = return boolean
* 1 = return key
*/  
static void php_search_array(INTERNAL_FUNCTION_PARAMETERS, int behavior) /* {{{ */
{  
zval *value,   /* value to check for */
  *array,   /* array to check in */
  **entry,   /* pointer to array entry */
  res;    /* comparison result */
HashPosition pos;  /* hash iterator */
zend_bool strict = 0;  /* strict comparison or not */
ulong num_key;
uint str_key_len;
char *string_key;
int (*is_equal_func)(zval *, zval *, zval * TSRMLS_DC) = is_equal_function;
if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "za|b", &value, &array, &strict) == FAILURE) {
 return;
}
if (strict) {
 is_equal_func = is_identical_function;
}
zend_hash_internal_pointer_reset_ex(Z_ARRVAL_P(array), &pos);
while (zend_hash_get_current_data_ex(Z_ARRVAL_P(array), (void **)&entry, &pos) == SUCCESS) {
 is_equal_func(&res, value, *entry TSRMLS_CC);
 if (Z_LVAL(res)) {
  if (behavior == 0) {
   RETURN_TRUE;
  } else {
   /* Return current key */
   switch (zend_hash_get_current_key_ex(Z_ARRVAL_P(array), &string_key, &str_key_len, &num_key, 0, &pos)) {
    case HASH_KEY_IS_STRING:
     RETURN_STRINGL(string_key, str_key_len - 1, 1);
     break;
    case HASH_KEY_IS_LONG:
     RETURN_LONG(num_key);
     break;
   }
  }
 }
 zend_hash_move_forward_ex(Z_ARRVAL_P(array), &pos);
}
RETURN_FALSE;
}  
/* }}} */
/* {{{ proto bool in_array(mixed needle, array haystack [, bool strict])
Checks if the given value exists in the array */

我们发现 strict  这个值的不同有两种比较方式，看一下两个函数的不同之处

is_identical_function 检查类型是否相同

ZEND_API int is_identical_function(zval *result, zval *op1, zval *op2 TSRMLS_DC) /* {{{ */
{  
Z_TYPE_P(result) = IS_BOOL;
if (Z_TYPE_P(op1) != Z_TYPE_P(op2)) {
 Z_LVAL_P(result) = 0;
 return SUCCESS;
}
switch (Z_TYPE_P(op1)) {
 case IS_NULL:
  Z_LVAL_P(result) = 1;
  break;
 case IS_BOOL:
 case IS_LONG:
 case IS_RESOURCE:
  Z_LVAL_P(result) = (Z_LVAL_P(op1) == Z_LVAL_P(op2));
  break;
 case IS_DOUBLE:
  Z_LVAL_P(result) = (Z_DVAL_P(op1) == Z_DVAL_P(op2));
  break;
 case IS_STRING:
  Z_LVAL_P(result) = ((Z_STRLEN_P(op1) == Z_STRLEN_P(op2))
  && (!memcmp(Z_STRVAL_P(op1), Z_STRVAL_P(op2), Z_STRLEN_P(op1))));
  break;
 case IS_ARRAY:
  Z_LVAL_P(result) = (Z_ARRVAL_P(op1) == Z_ARRVAL_P(op2)
  zend_hash_compare(Z_ARRVAL_P(op1), Z_ARRVAL_P(op2), (compare_func_t) hash_zval_identical_function, 1 TSRMLS_CC)==0);
  break;
 case IS_OBJECT:
  if (Z_OBJ_HT_P(op1) == Z_OBJ_HT_P(op2)) {
  Z_LVAL_P(result) = (Z_OBJ_HANDLE_P(op1) == Z_OBJ_HANDLE_P(op2));
  } else {
  Z_LVAL_P(result) = 0;
  }
  break;
 default:
  Z_LVAL_P(result) = 0;
  return FAILURE;
}
return SUCCESS;
}  
/* }}} */

is_equal_function 不检查类型是否相同，所以需要隐式转换

ZEND_API int is_equal_function(zval *result, zval *op1, zval *op2 TSRMLS_DC) /* {{{ */
{  
if (compare_function(result, op1, op2 TSRMLS_CC) == FAILURE) {
 return FAILURE;
}
ZVAL_BOOL(result, (Z_LVAL_P(result) == 0));
return SUCCESS;
}  
/* }}} */
==》compare_function

ZEND_API int compare_function(zval *result, zval *op1, zval *op2 TSRMLS_DC) /* {{{ */
{  
int ret;
int converted = 0;
zval op1_copy, op2_copy;
zval *op_free;
while (1) {
 switch (TYPE_PAIR(Z_TYPE_P(op1), Z_TYPE_P(op2))) {
  case TYPE_PAIR(IS_LONG, IS_LONG):
  ZVAL_LONG(result, Z_LVAL_P(op1)>Z_LVAL_P(op2)?1:(Z_LVAL_P(op1)<Z_LVAL_P(op2)?-1:0));
  return SUCCESS;
  case TYPE_PAIR(IS_DOUBLE, IS_LONG):
  Z_DVAL_P(result) = Z_DVAL_P(op1) - (double)Z_LVAL_P(op2);
  ZVAL_LONG(result, ZEND_NORMALIZE_BOOL(Z_DVAL_P(result)));
  return SUCCESS;
  case TYPE_PAIR(IS_LONG, IS_DOUBLE):
  Z_DVAL_P(result) = (double)Z_LVAL_P(op1) - Z_DVAL_P(op2);
  ZVAL_LONG(result, ZEND_NORMALIZE_BOOL(Z_DVAL_P(result)));
  return SUCCESS;
  case TYPE_PAIR(IS_DOUBLE, IS_DOUBLE):
  if (Z_DVAL_P(op1) == Z_DVAL_P(op2)) {
   ZVAL_LONG(result, 0);
  } else {
   Z_DVAL_P(result) = Z_DVAL_P(op1) - Z_DVAL_P(op2);
   ZVAL_LONG(result, ZEND_NORMALIZE_BOOL(Z_DVAL_P(result)));
  }
  return SUCCESS;
  case TYPE_PAIR(IS_ARRAY, IS_ARRAY):
  zend_compare_arrays(result, op1, op2 TSRMLS_CC);
  return SUCCESS;
  case TYPE_PAIR(IS_NULL, IS_NULL):
  ZVAL_LONG(result, 0);
  return SUCCESS;
  case TYPE_PAIR(IS_NULL, IS_BOOL):
  ZVAL_LONG(result, Z_LVAL_P(op2) ? -1 : 0);
  return SUCCESS;
  case TYPE_PAIR(IS_BOOL, IS_NULL):
  ZVAL_LONG(result, Z_LVAL_P(op1) ? 1 : 0);
  return SUCCESS;
  case TYPE_PAIR(IS_BOOL, IS_BOOL):
  ZVAL_LONG(result, ZEND_NORMALIZE_BOOL(Z_LVAL_P(op1) - Z_LVAL_P(op2)));
  return SUCCESS;
  case TYPE_PAIR(IS_STRING, IS_STRING):
  zendi_smart_strcmp(result, op1, op2);
  return SUCCESS;
  case TYPE_PAIR(IS_NULL, IS_STRING):
  ZVAL_LONG(result, zend_binary_strcmp("", 0, Z_STRVAL_P(op2), Z_STRLEN_P(op2)));
  return SUCCESS;
  case TYPE_PAIR(IS_STRING, IS_NULL):
  ZVAL_LONG(result, zend_binary_strcmp(Z_STRVAL_P(op1), Z_STRLEN_P(op1), "", 0));
  return SUCCESS;
  case TYPE_PAIR(IS_OBJECT, IS_NULL):
  ZVAL_LONG(result, 1);
  return SUCCESS;
  case TYPE_PAIR(IS_NULL, IS_OBJECT):
  ZVAL_LONG(result, -1);
  return SUCCESS;
  case TYPE_PAIR(IS_OBJECT, IS_OBJECT):
  /* If both are objects sharing the same comparision handler then use is */
  if (Z_OBJ_HANDLER_P(op1,compare_objects) == Z_OBJ_HANDLER_P(op2,compare_objects)) {
   if (Z_OBJ_HANDLE_P(op1) == Z_OBJ_HANDLE_P(op2)) {
    /* object handles are identical, apparently this is the same object */
    ZVAL_LONG(result, 0);
    return SUCCESS;
   }
   ZVAL_LONG(result, Z_OBJ_HT_P(op1)->compare_objects(op1, op2 TSRMLS_CC));
   return SUCCESS;
  }
  /* break missing intentionally */
  default:
  if (Z_TYPE_P(op1) == IS_OBJECT) {
   if (Z_OBJ_HT_P(op1)->get) {
    op_free = Z_OBJ_HT_P(op1)->get(op1 TSRMLS_CC);
    ret = compare_function(result, op_free, op2 TSRMLS_CC);
    zend_free_obj_get_result(op_free TSRMLS_CC);
    return ret;
   } else if (Z_TYPE_P(op2) != IS_OBJECT && Z_OBJ_HT_P(op1)->cast_object) {
    ALLOC_INIT_ZVAL(op_free);
    if (Z_OBJ_HT_P(op1)->cast_object(op1, op_free, Z_TYPE_P(op2) TSRMLS_CC) == FAILURE) {
     ZVAL_LONG(result, 1);
     zend_free_obj_get_result(op_free TSRMLS_CC);
     return SUCCESS;
    }
    ret = compare_function(result, op_free, op2 TSRMLS_CC);
    zend_free_obj_get_result(op_free TSRMLS_CC);
    return ret;
   }
  }
  if (Z_TYPE_P(op2) == IS_OBJECT) {
   if (Z_OBJ_HT_P(op2)->get) {
    op_free = Z_OBJ_HT_P(op2)->get(op2 TSRMLS_CC);
    ret = compare_function(result, op1, op_free TSRMLS_CC);
    zend_free_obj_get_result(op_free TSRMLS_CC);
    return ret;
   } else if (Z_TYPE_P(op1) != IS_OBJECT && Z_OBJ_HT_P(op2)->cast_object) {
    ALLOC_INIT_ZVAL(op_free);
    if (Z_OBJ_HT_P(op2)->cast_object(op2, op_free, Z_TYPE_P(op1) TSRMLS_CC) == FAILURE) {
     ZVAL_LONG(result, -1);
     zend_free_obj_get_result(op_free TSRMLS_CC);
     return SUCCESS;
    }
    ret = compare_function(result, op1, op_free TSRMLS_CC);
    zend_free_obj_get_result(op_free TSRMLS_CC);
    return ret;
   } else if (Z_TYPE_P(op1) == IS_OBJECT) {
    ZVAL_LONG(result, 1);
    return SUCCESS;
   }
  }
  if (!converted) {
   if (Z_TYPE_P(op1) == IS_NULL) {
    zendi_convert_to_boolean(op2, op2_copy, result);
    ZVAL_LONG(result, Z_LVAL_P(op2) ? -1 : 0);
    return SUCCESS;
   } else if (Z_TYPE_P(op2) == IS_NULL) {
    zendi_convert_to_boolean(op1, op1_copy, result);
    ZVAL_LONG(result, Z_LVAL_P(op1) ? 1 : 0);
    return SUCCESS;
   } else if (Z_TYPE_P(op1) == IS_BOOL) {
    zendi_convert_to_boolean(op2, op2_copy, result);
    ZVAL_LONG(result, ZEND_NORMALIZE_BOOL(Z_LVAL_P(op1) - Z_LVAL_P(op2)));
    return SUCCESS;
   } else if (Z_TYPE_P(op2) == IS_BOOL) {
    zendi_convert_to_boolean(op1, op1_copy, result);
    ZVAL_LONG(result, ZEND_NORMALIZE_BOOL(Z_LVAL_P(op1) - Z_LVAL_P(op2)));
    return SUCCESS;
   } else {
    zendi_convert_scalar_to_number(op1, op1_copy, result);
    zendi_convert_scalar_to_number(op2, op2_copy, result);
    converted = 1;
   }
  } else if (Z_TYPE_P(op1)==IS_ARRAY) {
   ZVAL_LONG(result, 1);
   return SUCCESS;
  } else if (Z_TYPE_P(op2)==IS_ARRAY) {
   ZVAL_LONG(result, -1);
   return SUCCESS;
  } else if (Z_TYPE_P(op1)==IS_OBJECT) {
   ZVAL_LONG(result, 1);
   return SUCCESS;
  } else if (Z_TYPE_P(op2)==IS_OBJECT) {
   ZVAL_LONG(result, -1);
   return SUCCESS;
  } else {
   ZVAL_LONG(result, 0);
   return FAILURE;
  }
 }
}  
}  
/* }}} */