[CBRD-25043] refactor functions related to isspace, tolower, and toup…

…per (#4932)

http://jira.cubrid.org/browse/CBRD-25043

* Refactor the functions in the following list to inline.
  - char_islower, char_isupper, char_isalpha, char_isdigit, char_isalnum,
  - char_isspace, char_iseol, char_isxdigit, char_tolower, char_toupper,
  - char_isupper_iso8859, char_islower_iso8859, char_tolower_iso8859, char_toupper_iso8859
* Each function can immediately check and obtain values from a pre-prepared array.
Conflicts:
	src/base/hide_password.cpp
	src/base/system_parameter.h
	win/cubridcs/cubridcs.def

src/base/chartype.h

@@ -29,33 +29,119 @@
 
 #ident "$Id$"
 
+#include <assert.h>
+
 #ifdef __cplusplus
 extern "C"
 {
 #endif
 
-  extern int char_islower (int c);
-  extern int char_isupper (int c);
-  extern int char_isalpha (int c);
-  extern int char_isdigit (int c);
-  extern int char_isalnum (int c);
-  extern int char_isspace (int c);
-  extern int char_iseol (int c);
-  extern int char_isxdigit (int c);
-#if defined (ENABLE_UNUSED_FUNCTION)
-  extern int char_isascii (int c);
+#define CHAR_PROP_NONE        (0x00)
+#define CHAR_PROP_UPPER       (0x01)	/* uppercase.  */
+#define CHAR_PROP_LOWER       (0x02)	/* lowercase.  */
+#define CHAR_PROP_DIGIT       (0x04)	/* Numeric.    */
+#define CHAR_PROP_SPACE       (0x08)	/* space, \t \n \r \f \v */
+#define CHAR_PROP_HEXNUM      (0x10)	/* 0~9, a~f, A~F  */
+#define CHAR_PROP_EOL         (0x20)	/* \r \n  */
+#define CHAR_PROP_ISO8859_UPPER (0x40)
+#define CHAR_PROP_ISO8859_LOWER (0x80)
+
+#define CHAR_PROP_ALPHA       (CHAR_PROP_UPPER | CHAR_PROP_LOWER)	/* Alphabetic.  */
+#define CHAR_PROP_ALPHA_NUM   (CHAR_PROP_ALPHA | CHAR_PROP_DIGIT)	/* Alpha-Numeric */
+
+  typedef unsigned char char_type_prop;
+  extern const char_type_prop *char_properties_ptr;
+  extern const unsigned char *char_lower_mapper_ptr;
+  extern const unsigned char *char_upper_mapper_ptr;
+  extern const unsigned char *iso8859_lower_mapper_ptr;
+  extern const unsigned char *iso8859_upper_mapper_ptr;
+
+#ifndef NDEBUG
+#define CHECK_OVER_CODE_VALUE(c)   assert((unsigned char)(c) < 256)
+#else
+#define CHECK_OVER_CODE_VALUE(c)
 #endif
 
-  extern int char_tolower (int c);
-  extern int char_toupper (int c);
+  inline int char_isspace (int c)
+  {
+    CHECK_OVER_CODE_VALUE (c);
+    return (char_properties_ptr[(unsigned char) c] & CHAR_PROP_SPACE);
+  }
+  inline int char_isupper (int c)
+  {
+    CHECK_OVER_CODE_VALUE (c);
+    return (char_properties_ptr[(unsigned char) c] & CHAR_PROP_UPPER);
+  }
+  inline int char_tolower (int c)
+  {
+    CHECK_OVER_CODE_VALUE (c);
+    return ((int) char_lower_mapper_ptr[(unsigned char) c]);
+  }
+  inline int char_islower (int c)
+  {
+    CHECK_OVER_CODE_VALUE (c);
+    return (char_properties_ptr[(unsigned char) c] & CHAR_PROP_LOWER);
+  }
+  inline int char_isalpha (int c)
+  {
+    CHECK_OVER_CODE_VALUE (c);
+    return (char_properties_ptr[(unsigned char) c] & CHAR_PROP_ALPHA);
+  }
+  inline int char_isdigit (int c)
+  {
+    CHECK_OVER_CODE_VALUE (c);
+    return (char_properties_ptr[(unsigned char) c] & CHAR_PROP_DIGIT);
+  }
+  inline int char_isalnum (int c)
+  {
+    CHECK_OVER_CODE_VALUE (c);
+    return (char_properties_ptr[(unsigned char) c] & CHAR_PROP_ALPHA_NUM);
+  }
+  inline int char_iseol (int c)
+  {
+    CHECK_OVER_CODE_VALUE (c);
+    return (char_properties_ptr[(unsigned char) c] & CHAR_PROP_EOL);
+  }
+  inline int char_isxdigit (int c)
+  {
+    CHECK_OVER_CODE_VALUE (c);
+    return (char_properties_ptr[(unsigned char) c] & CHAR_PROP_HEXNUM);
+  }
+  inline int char_toupper (int c)
+  {
+    CHECK_OVER_CODE_VALUE (c);
+    return ((int) char_upper_mapper_ptr[(unsigned char) c]);
+  }
+
+  inline int char_tolower_iso8859 (int c)
+  {
+    CHECK_OVER_CODE_VALUE (c);
+    return ((int) iso8859_lower_mapper_ptr[(unsigned char) c]);
+  }
+  inline int char_toupper_iso8859 (int c)
+  {
+    CHECK_OVER_CODE_VALUE (c);
+    return ((int) iso8859_upper_mapper_ptr[(unsigned char) c]);
+  }
+  inline int char_islower_iso8859 (int c)
+  {
+    CHECK_OVER_CODE_VALUE (c);
+    return (char_properties_ptr[(unsigned char) c] & (CHAR_PROP_LOWER | CHAR_PROP_ISO8859_LOWER));
+  }
+  inline int char_isupper_iso8859 (int c)
+  {
+    CHECK_OVER_CODE_VALUE (c);
+    return (char_properties_ptr[(unsigned char) c] & (CHAR_PROP_UPPER | CHAR_PROP_ISO8859_UPPER));
+  }
+
+/* In some codes, only four items (' ', '\t', '\n', '\r') were checked.
+ * In order to maintain its original form, it was given a different name. */
+#define char_isspace2  char_isspace
 
-  extern int char_isupper_iso8859 (int c);
-  extern int char_islower_iso8859 (int c);
-  extern int char_tolower_iso8859 (int c);
-  extern int char_toupper_iso8859 (int c);
+  extern char *trim (char *str);
 
 #ifdef __cplusplus
 }
 #endif
 
-#endif				/* _CHARTYPE_H_ */
+#endif /* _CHARTYPE_H_ */

src/base/intl_support.c

@@ -66,10 +66,6 @@
               && ((unsigned char) ch <= (unsigned char) 0xfe) )
 #endif
 
-#define CHAR_BYTE_TO_LOWER(c) ((c) + ('a' - 'A'))
-
-#define CHAR_BYTE_TO_UPPER(c) ((c) - ('a' - 'A'))
-
 /* conversion from turkish ISO 8859-9 to UTF-8 */
 #define ISO_8859_9_FIRST_CP 0x11e
 #define ISO_8859_9_LAST_CP 0x15f
@@ -727,10 +723,7 @@ intl_tolower_iso8859 (unsigned char *s, int length)
 
   for (end = s + length; s < end; s++)
     {
-      if (char_isupper_iso8859 (*s))
-	{
-	  *s = CHAR_BYTE_TO_LOWER (*s);
-	}
+      *s = char_tolower_iso8859 (*s);
     }
 
   return char_count;
@@ -753,10 +746,7 @@ intl_toupper_iso8859 (unsigned char *s, int length)
 
   for (end = s + length; s < end; s++)
     {
-      if (char_islower_iso8859 (*s))
-	{
-	  *s = CHAR_BYTE_TO_UPPER (*s);
-	}
+      *s = char_toupper_iso8859 (*s);
     }
 
   return char_count;
@@ -3257,14 +3247,7 @@ intl_identifier_mht_1strlowerhash (const void *key, const unsigned int ht_size)
     case INTL_CODESET_ISO88591:
       for (hash = 0; *byte_p; byte_p++)
 	{
-	  if (char_isupper_iso8859 (*byte_p))
-	    {
-	      ch = char_tolower_iso8859 (*byte_p);
-	    }
-	  else
-	    {
-	      ch = char_tolower (*byte_p);
-	    }
+	  ch = char_tolower_iso8859 (*byte_p);
 	  hash = (hash << 5) - hash + ch;
 	}
       break;

src/base/porting.c

@@ -2257,36 +2257,6 @@ port_close_memstream (FILE * fp, char **ptr, size_t * sizeloc)
     }
 }
 
-char *
-trim (char *str)
-{
-  char *p;
-  char *s;
-
-  if (str == NULL)
-    return (str);
-
-  for (s = str; *s != '\0' && (*s == ' ' || *s == '\t' || *s == '\n' || *s == '\r'); s++)
-    ;
-  if (*s == '\0')
-    {
-      *str = '\0';
-      return (str);
-    }
-
-  /* *s must be a non-white char */
-  for (p = s; *p != '\0'; p++)
-    ;
-  for (p--; *p == ' ' || *p == '\t' || *p == '\n' || *p == '\r'; p--)
-    ;
-  *++p = '\0';
-
-  if (s != str)
-    memmove (str, s, strlen (s) + 1);
-
-  return (str);
-}
-
 int
 parse_int (int *ret_p, const char *str_p, int base)
 {

src/base/porting.h

@@ -986,8 +986,6 @@ extern FILE *port_open_memstream (char **ptr, size_t * sizeloc);
 
 extern void port_close_memstream (FILE * fp, char **ptr, size_t * sizeloc);
 
-extern char *trim (char *str);
-
 extern int parse_bigint (INT64 * ret_p, const char *str_p, int base);
 
 extern int str_to_int32 (int *ret_p, char **end_p, const char *str_p, int base);

src/base/system_parameter.h

@@ -33,6 +33,8 @@
 #include "error_manager.h"
 #include "error_code.h"
 #include "porting.h"
+#include "porting_inline.hpp"
+#include "chartype.h"
 
 typedef enum
 {

src/broker/broker_log_sql_list.c

@@ -241,7 +241,7 @@ sql_change_comp_form (char *src, char *dest)
 
   for (p = src, q = dest; *p; p++)
     {
-      if (*p == '\r' || *p == '\n' || *p == '\t' || *p == ' ')
+      if (char_isspace2 (*p))
 	{
 	  if (write_space_flag)
 	    {

src/broker/broker_log_util.c

@@ -39,36 +39,6 @@
 
 static bool is_bind_with_size (char *buf, int *tot_val_size, int *info_size);
 
-char *
-ut_trim (char *str)
-{
-  char *p;
-  char *s;
-
-  if (str == NULL)
-    return (str);
-
-  for (s = str; *s != '\0' && (*s == ' ' || *s == '\t' || *s == '\n' || *s == '\r'); s++)
-    ;
-  if (*s == '\0')
-    {
-      *str = '\0';
-      return (str);
-    }
-
-  /* *s must be a non-white char */
-  for (p = s; *p != '\0'; p++)
-    ;
-  for (p--; *p == ' ' || *p == '\t' || *p == '\n' || *p == '\r'; p--)
-    ;
-  *++p = '\0';
-
-  if (s != str)
-    memmove (str, s, strlen (s) + 1);
-
-  return (str);
-}
-
 void
 ut_tolower (char *str)
 {
@@ -79,8 +49,7 @@ ut_tolower (char *str)
 
   for (p = str; *p; p++)
     {
-      if (*p >= 'A' && *p <= 'Z')
-	*p = *p - 'A' + 'a';
+      *p = char_tolower (*p);
     }
 }
 

src/broker/broker_log_util.h

@@ -49,7 +49,7 @@
           BUF[DATE_STR_LEN] = '\0';             \
         } while (0)
 
-extern char *ut_trim (char *);
+#define ut_trim  trim
 extern void ut_tolower (char *str);
 extern int ut_get_line (FILE * fp, T_STRING * t_str, char **out_str, int *lineno);
 extern int is_cas_log (char *str);

src/broker/broker_util.c

@@ -102,7 +102,7 @@ ut_access_log (int as_index, struct timeval *start_time, char error_flag, int er
     clt_appl = (char *) "-";
   for (p = clt_appl; *p; p++)
     {
-      if (*p == ' ' || *p == '\t' || *p == '\n' || *p == '\r')
+      if (char_isspace2 (*p))
 	*p = '_';
     }
 

src/broker/cas_common.h

@@ -32,6 +32,7 @@
 #endif
 
 #include "porting.h"
+#include "chartype.h"
 #define makestring1(x) #x
 #define makestring(x) makestring1(x)
 

src/broker/cas_util.c

@@ -51,37 +51,6 @@ ut_uchar2ipstr (unsigned char *ip_addr)
   return (ip_str);
 }
 
-char *
-ut_trim (char *str)
-{
-  char *p;
-  char *s;
-
-  if (str == NULL)
-    return (str);
-
-  for (s = str; *s != '\0' && (*s == ' ' || *s == '\t' || *s == '\n' || *s == '\r'); s++)
-    ;
-  if (*s == '\0')
-    {
-      *str = '\0';
-      return (str);
-    }
-
-  /* *s must be a non-white char */
-  for (p = s; *p != '\0'; p++)
-    ;
-  for (p--; *p == ' ' || *p == '\t' || *p == '\n' || *p == '\r'; p--)
-    ;
-  *++p = '\0';
-
-  if (s != str)
-    memmove (str, s, strlen (s) + 1);
-
-  return (str);
-}
-
-
 void
 ut_tolower (char *str)
 {
@@ -92,8 +61,7 @@ ut_tolower (char *str)
 
   for (p = str; *p; p++)
     {
-      if (*p >= 'A' && *p <= 'Z')
-	*p = *p - 'A' + 'a';
+      *p = char_tolower (*p);
     }
 }
 

src/broker/cas_util.h

@@ -31,8 +31,9 @@
 #include <windows.h>
 #endif
 
+#define ut_trim  trim
+
 extern char *ut_uchar2ipstr (unsigned char *ip_addr);
-extern char *ut_trim (char *);
 extern void ut_tolower (char *str);
 extern void ut_timeval_diff (struct timeval *start, struct timeval *end, int *res_sec, int *res_msec);
 extern int ut_check_timeout (struct timeval *start_time, struct timeval *end_time, int timeout_msec, int *res_sec,

src/broker/shard_proxy_log.c

@@ -330,7 +330,7 @@ proxy_access_log (struct timeval *start_time, int client_ip_addr, const char *db
 
   for (p = clt_appl; *p; p++)
     {
-      if (*p == ' ' || *p == '\t' || *p == '\n' || *p == '\r')
+      if (char_isspace2 (*p))
 	*p = '_';
     }