StringLike.cpp File Reference

Functions to support the LIKE and ILIKE operator in SQL. Only single-byte character set is supported for now. More...

#include "StringLike.h"

Include dependency graph for StringLike.cpp:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Macros
#define	STR_LIKE_SIMPLE_NULLABLE(base_func)
#define	STR_LIKE_NULLABLE(base_func)
#define	STR_CMP_NULLABLE(base_func)

Enumerations
enum	LikeStatus { kLIKE_TRUE, kLIKE_FALSE, kLIKE_ABORT, kLIKE_ERROR }

Functions
static DEVICE int	lowercase (char c)
RUNTIME_EXPORT DEVICE bool	string_like_simple (const char *str, const int32_t str_len, const char *pattern, const int32_t pat_len, char escape_char)
RUNTIME_EXPORT DEVICE bool	string_ilike_simple (const char *str, const int32_t str_len, const char *pattern, const int32_t pat_len, char escape_char)
static DEVICE LikeStatus	string_like_match (const char *str, const int32_t str_len, const char *pattern, const int32_t pat_len, const char escape_char, const bool is_ilike)
RUNTIME_EXPORT DEVICE bool	string_like (const char *str, const int32_t str_len, const char *pattern, const int32_t pat_len, const char escape_char)
RUNTIME_EXPORT DEVICE bool	string_ilike (const char *str, const int32_t str_len, const char *pattern, const int32_t pat_len, const char escape_char)
RUNTIME_EXPORT DEVICE int32_t	StringCompare (const char *s1, const int32_t s1_len, const char *s2, const int32_t s2_len)
RUNTIME_EXPORT DEVICE bool	string_lt (const char *lhs, const int32_t lhs_len, const char *rhs, const int32_t rhs_len)
RUNTIME_EXPORT DEVICE bool	string_le (const char *lhs, const int32_t lhs_len, const char *rhs, const int32_t rhs_len)
RUNTIME_EXPORT DEVICE bool	string_gt (const char *lhs, const int32_t lhs_len, const char *rhs, const int32_t rhs_len)
RUNTIME_EXPORT DEVICE bool	string_ge (const char *lhs, const int32_t lhs_len, const char *rhs, const int32_t rhs_len)
RUNTIME_EXPORT DEVICE bool	string_eq (const char *lhs, const int32_t lhs_len, const char *rhs, const int32_t rhs_len)
RUNTIME_EXPORT DEVICE bool	string_ne (const char *lhs, const int32_t lhs_len, const char *rhs, const int32_t rhs_len)

Detailed Description

Functions to support the LIKE and ILIKE operator in SQL. Only single-byte character set is supported for now.

Definition in file StringLike.cpp.

Macro Definition Documentation

#define STR_CMP_NULLABLE

(

base_func

)

Value:

extern "C" RUNTIME_EXPORT DEVICE int8_t base_func##_nullable(const char* lhs,          \
                                                               const int32_t lhs_len,    \
                                                               const char* rhs,          \
                                                               const int32_t rhs_len,    \
                                                               const int8_t bool_null) { \
    if (!lhs || !rhs) {                                                                  \
      return bool_null;                                                                  \
    }                                                                                    \
    return base_func(lhs, lhs_len, rhs, rhs_len) ? 1 : 0;                                \
  }

Definition at line 350 of file StringLike.cpp.

#define STR_LIKE_NULLABLE

(

base_func

)

Value:

extern "C" RUNTIME_EXPORT DEVICE int8_t base_func##_nullable(const char* lhs,          \
                                                               const int32_t lhs_len,    \
                                                               const char* rhs,          \
                                                               const int32_t rhs_len,    \
                                                               const char escape_char,   \
                                                               const int8_t bool_null) { \
    if (!lhs || !rhs) {                                                                  \
      return bool_null;                                                                  \
    }                                                                                    \
    return base_func(lhs, lhs_len, rhs, rhs_len, escape_char) ? 1 : 0;                   \
  }

Definition at line 290 of file StringLike.cpp.

#define STR_LIKE_SIMPLE_NULLABLE

(

base_func

)

Value:

extern "C" RUNTIME_EXPORT DEVICE int8_t base_func##_nullable(const char* lhs,          \
                                                               const int32_t lhs_len,    \
                                                               const char* rhs,          \
                                                               const int32_t rhs_len,    \
                                                               char escape_char,         \
                                                               const int8_t bool_null) { \
    if (!lhs || !rhs) {                                                                  \
      return bool_null;                                                                  \
    }                                                                                    \
    return base_func(lhs, lhs_len, rhs, rhs_len, escape_char) ? 1 : 0;                   \
  }

Definition at line 78 of file StringLike.cpp.

Enumeration Type Documentation

enum LikeStatus

Enumerator
kLIKE_TRUE
kLIKE_FALSE
kLIKE_ABORT
kLIKE_ERROR

Definition at line 26 of file StringLike.cpp.

                {
  kLIKE_TRUE,
  kLIKE_FALSE,
  kLIKE_ABORT,  // means we run out of string characters to match against pattern, can
                // abort early
  kLIKE_ERROR   // error condition
};

Function Documentation

static DEVICE int lowercase ( char  c )

inlinestatic

Definition at line 34 of file StringLike.cpp.

Referenced by string_ilike_simple(), and string_like_match().

                                           {
  if ('A' <= c && c <= 'Z') {
    return 'a' + (c - 'A');
  }
  return c;
}

Here is the caller graph for this function:

RUNTIME_EXPORT DEVICE bool string_eq	(	const char *	lhs,
		const int32_t	lhs_len,
		const char *	rhs,
		const int32_t	rhs_len
	)

Definition at line 336 of file StringLike.cpp.

References StringCompare().

Referenced by StringDictionary::getCompare().

                                                                       {
  return StringCompare(lhs, lhs_len, rhs, rhs_len) == 0;
}

Here is the call graph for this function:

Here is the caller graph for this function:

RUNTIME_EXPORT DEVICE bool string_ge	(	const char *	lhs,
		const int32_t	lhs_len,
		const char *	rhs,
		const int32_t	rhs_len
	)

Definition at line 329 of file StringLike.cpp.

References StringCompare().

                                                                       {
  return StringCompare(lhs, lhs_len, rhs, rhs_len) >= 0;
}

Here is the call graph for this function:

RUNTIME_EXPORT DEVICE bool string_gt	(	const char *	lhs,
		const int32_t	lhs_len,
		const char *	rhs,
		const int32_t	rhs_len
	)

Definition at line 322 of file StringLike.cpp.

References StringCompare().

                                                                       {
  return StringCompare(lhs, lhs_len, rhs, rhs_len) > 0;
}

Here is the call graph for this function:

RUNTIME_EXPORT DEVICE bool string_ilike	(	const char *	str,
		const int32_t	str_len,
		const char *	pattern,
		const int32_t	pat_len,
		const char	escape_char
	)

Definition at line 261 of file StringLike.cpp.

References kLIKE_TRUE, and string_like_match().

Referenced by StringDictionaryProxy::getLike(), and StringDictionary::getLikeImpl().

                                                                           {
  // @TODO(wei/alex) add runtime error handling
  LikeStatus status =
      string_like_match(str, str_len, pattern, pat_len, escape_char, true);
  return status == kLIKE_TRUE;
}

Here is the call graph for this function:

Here is the caller graph for this function:

RUNTIME_EXPORT DEVICE bool string_ilike_simple	(	const char *	str,
		const int32_t	str_len,
		const char *	pattern,
		const int32_t	pat_len,
		char	escape_char
	)

Definition at line 61 of file StringLike.cpp.

References lowercase().

Referenced by StringDictionaryProxy::getLike(), and StringDictionary::getLikeImpl().

                                                                            {
  int i, j;
  int search_len = str_len - pat_len + 1;
  for (i = 0; i < search_len; ++i) {
    for (j = 0; j < pat_len && pattern[j] == lowercase(str[j + i]); ++j) {
    }
    if (j >= pat_len) {
      return true;
    }
  }
  return false;
}

Here is the call graph for this function:

Here is the caller graph for this function:

RUNTIME_EXPORT DEVICE bool string_le	(	const char *	lhs,
		const int32_t	lhs_len,
		const char *	rhs,
		const int32_t	rhs_len
	)

Definition at line 315 of file StringLike.cpp.

References StringCompare().

                                                                       {
  return StringCompare(lhs, lhs_len, rhs, rhs_len) <= 0;
}

Here is the call graph for this function:

RUNTIME_EXPORT DEVICE bool string_like	(	const char *	str,
		const int32_t	str_len,
		const char *	pattern,
		const int32_t	pat_len,
		const char	escape_char
	)

Definition at line 250 of file StringLike.cpp.

References kLIKE_TRUE, and string_like_match().

Referenced by StringDictionaryProxy::getLike(), and StringDictionary::getLikeImpl().

                                                                          {
  // @TODO(wei/alex) add runtime error handling
  LikeStatus status =
      string_like_match(str, str_len, pattern, pat_len, escape_char, false);
  return status == kLIKE_TRUE;
}

Here is the call graph for this function:

Here is the caller graph for this function:

static DEVICE LikeStatus string_like_match ( const char *  str, const int32_t  str_len, const char *  pattern, const int32_t  pat_len, const char  escape_char, const bool  is_ilike  )

static

Definition at line 98 of file StringLike.cpp.

References kLIKE_ABORT, kLIKE_ERROR, kLIKE_FALSE, kLIKE_TRUE, and lowercase().

Referenced by string_ilike(), and string_like().

                                                                {
  const char* s = str;
  int slen = str_len;
  const char* p = pattern;
  int plen = pat_len;

  while (slen > 0 && plen > 0) {
    if (*p == escape_char) {
      // next pattern char must match literally, whatever it is
      p++;
      plen--;
      if (plen <= 0) {
        return kLIKE_ERROR;
      }
      if ((!is_ilike && *s != *p) || (is_ilike && lowercase(*s) != *p)) {
        return kLIKE_FALSE;
      }
    } else if (*p == '%') {
      char firstpat;
      p++;
      plen--;
      while (plen > 0) {
        if (*p == '%') {
          p++;
          plen--;
        } else if (*p == '_') {
          if (slen <= 0) {
            return kLIKE_ABORT;
          }
          s++;
          slen--;
          p++;
          plen--;
        } else {
          break;
        }
      }
      if (plen <= 0) {
        return kLIKE_TRUE;
      }
      if (*p == escape_char) {
        if (plen < 2) {
          return kLIKE_ERROR;
        }
        firstpat = p[1];
      } else {
        firstpat = *p;
      }

      while (slen > 0) {
        bool match = false;
        if (firstpat == '[' && *p != escape_char) {
          const char* pp = p + 1;
          int pplen = plen - 1;
          while (pplen > 0 && *pp != ']') {
            if ((!is_ilike && *s == *pp) || (is_ilike && lowercase(*s) == *pp)) {
              match = true;
              break;
            }
            pp++;
            pplen--;
          }
          if (pplen <= 0) {
            return kLIKE_ERROR;  // malformed
          }
        } else if ((!is_ilike && *s == firstpat) ||
                   (is_ilike && lowercase(*s) == firstpat)) {
          match = true;
        }
        if (match) {
          LikeStatus status = string_like_match(s, slen, p, plen, escape_char, is_ilike);
          if (status != kLIKE_FALSE) {
            return status;
          }
        }
        s++;
        slen--;
      }
      return kLIKE_ABORT;
    } else if (*p == '_') {
      s++;
      slen--;
      p++;
      plen--;
      continue;
    } else if (*p == '[') {
      const char* pp = p + 1;
      int pplen = plen - 1;
      bool match = false;
      while (pplen > 0 && *pp != ']') {
        if ((!is_ilike && *s == *pp) || (is_ilike && lowercase(*s) == *pp)) {
          match = true;
          break;
        }
        pp++;
        pplen--;
      }
      if (match) {
        s++;
        slen--;
        pplen--;
        const char* x;
        for (x = pp + 1; *x != ']' && pplen > 0; x++, pplen--) {
          ;
        }
        if (pplen <= 0) {
          return kLIKE_ERROR;  // malformed
        }
        plen -= (x - p + 1);
        p = x + 1;
        continue;
      } else {
        return kLIKE_FALSE;
      }
    } else if ((!is_ilike && *s != *p) || (is_ilike && lowercase(*s) != *p)) {
      return kLIKE_FALSE;
    }
    s++;
    slen--;
    p++;
    plen--;
  }
  if (slen > 0) {
    return kLIKE_FALSE;
  }
  while (plen > 0 && *p == '%') {
    p++;
    plen--;
  }
  if (plen <= 0) {
    return kLIKE_TRUE;
  }
  return kLIKE_ABORT;
}

Here is the call graph for this function:

Here is the caller graph for this function:

RUNTIME_EXPORT DEVICE bool string_like_simple	(	const char *	str,
		const int32_t	str_len,
		const char *	pattern,
		const int32_t	pat_len,
		char	escape_char
	)

Definition at line 43 of file StringLike.cpp.

Referenced by StringDictionaryProxy::getLike(), and StringDictionary::getLikeImpl().

                                                                           {
  int i, j;
  int search_len = str_len - pat_len + 1;
  for (i = 0; i < search_len; ++i) {
    for (j = 0; j < pat_len && pattern[j] == str[j + i]; ++j) {
    }
    if (j >= pat_len) {
      return true;
    }
  }
  return false;
}

Here is the caller graph for this function:

RUNTIME_EXPORT DEVICE bool string_lt	(	const char *	lhs,
		const int32_t	lhs_len,
		const char *	rhs,
		const int32_t	rhs_len
	)

Definition at line 308 of file StringLike.cpp.

References StringCompare().

Referenced by StringDictionary::getCompare(), StringDictionary::mergeSortedCache(), and StringDictionary::sortCache().

                                                                       {
  return StringCompare(lhs, lhs_len, rhs, rhs_len) < 0;
}

Here is the call graph for this function:

Here is the caller graph for this function:

RUNTIME_EXPORT DEVICE bool string_ne	(	const char *	lhs,
		const int32_t	lhs_len,
		const char *	rhs,
		const int32_t	rhs_len
	)

Definition at line 343 of file StringLike.cpp.

References StringCompare().

                                                                       {
  return StringCompare(lhs, lhs_len, rhs, rhs_len) != 0;
}

Here is the call graph for this function:

RUNTIME_EXPORT DEVICE int32_t StringCompare	(	const char *	s1,
		const int32_t	s1_len,
		const char *	s2,
		const int32_t	s2_len
	)

Definition at line 272 of file StringLike.cpp.

Referenced by string_eq(), string_ge(), string_gt(), string_le(), string_lt(), and string_ne().

                                                                             {
  const char* s1_ = s1;
  const char* s2_ = s2;

  while (s1_ < s1 + s1_len && s2_ < s2 + s2_len && *s1_ == *s2_) {
    s1_++;
    s2_++;
  }

  unsigned char c1 = (s1_ < s1 + s1_len) ? (*(unsigned char*)s1_) : 0;
  unsigned char c2 = (s2_ < s2 + s2_len) ? (*(unsigned char*)s2_) : 0;

  return c1 - c2;
}

Here is the caller graph for this function: