pg_query 5.1.0 → 6.1.0

data/ext/pg_query/include/postgres/common/hashfn_unstable.h

@@ -0,0 +1,407 @@
+/*
+ * hashfn_unstable.h
+ *
+ * Building blocks for creating fast inlineable hash functions. The
+ * functions in this file are not guaranteed to be stable between versions,
+ * and may differ by hardware platform. Hence they must not be used in
+ * indexes or other on-disk structures. See hashfn.h if you need stability.
+ *
+ *
+ * Portions Copyright (c) 2024, PostgreSQL Global Development Group
+ *
+ * src/include/common/hashfn_unstable.h
+ */
+#ifndef HASHFN_UNSTABLE_H
+#define HASHFN_UNSTABLE_H
+
+
+/*
+ * fasthash is a modification of code taken from
+ * https://code.google.com/archive/p/fast-hash/source/default/source
+ * under the terms of the MIT license. The original copyright
+ * notice follows:
+ */
+
+/* The MIT License
+
+   Copyright (C) 2012 Zilong Tan (eric.zltan@gmail.com)
+
+   Permission is hereby granted, free of charge, to any person
+   obtaining a copy of this software and associated documentation
+   files (the "Software"), to deal in the Software without
+   restriction, including without limitation the rights to use, copy,
+   modify, merge, publish, distribute, sublicense, and/or sell copies
+   of the Software, and to permit persons to whom the Software is
+   furnished to do so, subject to the following conditions:
+
+   The above copyright notice and this permission notice shall be
+   included in all copies or substantial portions of the Software.
+
+   THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+   EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+   MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+   NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+   BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+   ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+   CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+   SOFTWARE.
+*/
+
+/*
+ * fasthash as implemented here has two interfaces:
+ *
+ * 1) Standalone functions, e.g. fasthash32() for a single value with a
+ * known length. These return the same hash code as the original, at
+ * least on little-endian machines.
+ *
+ * 2) Incremental interface. This can used for incorporating multiple
+ * inputs. First, initialize the hash state (here with a zero seed):
+ *
+ * fasthash_state hs;
+ * fasthash_init(&hs, 0);
+ *
+ * If the inputs are of types that can be trivially cast to uint64, it's
+ * sufficient to do:
+ *
+ * hs.accum = value1;
+ * fasthash_combine(&hs);
+ * hs.accum = value2;
+ * fasthash_combine(&hs);
+ * ...
+ *
+ * For longer or variable-length input, fasthash_accum() is a more
+ * flexible, but more verbose method. The standalone functions use this
+ * internally, so see fasthash64() for an example of this.
+ *
+ * After all inputs have been mixed in, finalize the hash:
+ *
+ * hashcode = fasthash_final32(&hs, 0);
+ *
+ * The incremental interface allows an optimization for NUL-terminated
+ * C strings:
+ *
+ * len = fasthash_accum_cstring(&hs, str);
+ * hashcode = fasthash_final32(&hs, len);
+ *
+ * By handling the terminator on-the-fly, we can avoid needing a strlen()
+ * call to tell us how many bytes to hash. Experimentation has found that
+ * SMHasher fails unless we incorporate the length, so it is passed to
+ * the finalizer as a tweak.
+ */
+
+
+typedef struct fasthash_state
+{
+	/* staging area for chunks of input */
+	uint64		accum;
+
+	uint64		hash;
+} fasthash_state;
+
+#define FH_SIZEOF_ACCUM sizeof(uint64)
+
+
+/*
+ * Initialize the hash state.
+ *
+ * 'seed' can be zero.
+ */
+static inline void
+fasthash_init(fasthash_state *hs, uint64 seed)
+{
+	memset(hs, 0, sizeof(fasthash_state));
+	hs->hash = seed ^ 0x880355f21e6d1965;
+}
+
+/* both the finalizer and part of the combining step */
+static inline uint64
+fasthash_mix(uint64 h, uint64 tweak)
+{
+	h ^= (h >> 23) + tweak;
+	h *= 0x2127599bf4325c37;
+	h ^= h >> 47;
+	return h;
+}
+
+/* combine one chunk of input into the hash */
+static inline void
+fasthash_combine(fasthash_state *hs)
+{
+	hs->hash ^= fasthash_mix(hs->accum, 0);
+	hs->hash *= 0x880355f21e6d1965;
+}
+
+/* accumulate up to 8 bytes of input and combine it into the hash */
+static inline void
+fasthash_accum(fasthash_state *hs, const char *k, size_t len)
+{
+	uint32		lower_four;
+
+	Assert(len <= FH_SIZEOF_ACCUM);
+	hs->accum = 0;
+
+	/*
+	 * For consistency, bytewise loads must match the platform's endianness.
+	 */
+#ifdef WORDS_BIGENDIAN
+	switch (len)
+	{
+		case 8:
+			memcpy(&hs->accum, k, 8);
+			break;
+		case 7:
+			hs->accum |= (uint64) k[6] << 8;
+			/* FALLTHROUGH */
+		case 6:
+			hs->accum |= (uint64) k[5] << 16;
+			/* FALLTHROUGH */
+		case 5:
+			hs->accum |= (uint64) k[4] << 24;
+			/* FALLTHROUGH */
+		case 4:
+			memcpy(&lower_four, k, sizeof(lower_four));
+			hs->accum |= (uint64) lower_four << 32;
+			break;
+		case 3:
+			hs->accum |= (uint64) k[2] << 40;
+			/* FALLTHROUGH */
+		case 2:
+			hs->accum |= (uint64) k[1] << 48;
+			/* FALLTHROUGH */
+		case 1:
+			hs->accum |= (uint64) k[0] << 56;
+			break;
+		case 0:
+			return;
+	}
+#else
+	switch (len)
+	{
+		case 8:
+			memcpy(&hs->accum, k, 8);
+			break;
+		case 7:
+			hs->accum |= (uint64) k[6] << 48;
+			/* FALLTHROUGH */
+		case 6:
+			hs->accum |= (uint64) k[5] << 40;
+			/* FALLTHROUGH */
+		case 5:
+			hs->accum |= (uint64) k[4] << 32;
+			/* FALLTHROUGH */
+		case 4:
+			memcpy(&lower_four, k, sizeof(lower_four));
+			hs->accum |= lower_four;
+			break;
+		case 3:
+			hs->accum |= (uint64) k[2] << 16;
+			/* FALLTHROUGH */
+		case 2:
+			hs->accum |= (uint64) k[1] << 8;
+			/* FALLTHROUGH */
+		case 1:
+			hs->accum |= (uint64) k[0];
+			break;
+		case 0:
+			return;
+	}
+#endif
+
+	fasthash_combine(hs);
+}
+
+/*
+ * Set high bit in lowest byte where the input is zero, from:
+ * https://graphics.stanford.edu/~seander/bithacks.html#ZeroInWord
+ */
+#define haszero64(v) \
+	(((v) - 0x0101010101010101) & ~(v) & 0x8080808080808080)
+
+/*
+ * all-purpose workhorse for fasthash_accum_cstring
+ */
+static inline size_t
+fasthash_accum_cstring_unaligned(fasthash_state *hs, const char *str)
+{
+	const char *const start = str;
+
+	while (*str)
+	{
+		size_t		chunk_len = 0;
+
+		while (chunk_len < FH_SIZEOF_ACCUM && str[chunk_len] != '\0')
+			chunk_len++;
+
+		fasthash_accum(hs, str, chunk_len);
+		str += chunk_len;
+	}
+
+	return str - start;
+}
+
+/*
+ * specialized workhorse for fasthash_accum_cstring
+ *
+ * With an aligned pointer, we consume the string a word at a time.
+ * Loading the word containing the NUL terminator cannot segfault since
+ * allocation boundaries are suitably aligned. To keep from setting
+ * off alarms with address sanitizers, exclude this function from
+ * such testing.
+ */
+pg_attribute_no_sanitize_address()
+static inline size_t
+fasthash_accum_cstring_aligned(fasthash_state *hs, const char *str)
+{
+	const char *const start = str;
+	size_t		remainder;
+	uint64		zero_byte_low;
+
+	Assert(PointerIsAligned(start, uint64));
+
+	/*
+	 * For every chunk of input, check for zero bytes before mixing into the
+	 * hash. The chunk with zeros must contain the NUL terminator.
+	 */
+	for (;;)
+	{
+		uint64		chunk = *(uint64 *) str;
+
+		zero_byte_low = haszero64(chunk);
+		if (zero_byte_low)
+			break;
+
+		hs->accum = chunk;
+		fasthash_combine(hs);
+		str += FH_SIZEOF_ACCUM;
+	}
+
+	/* mix in remaining bytes */
+	remainder = fasthash_accum_cstring_unaligned(hs, str);
+	str += remainder;
+
+	return str - start;
+}
+
+/*
+ * Mix 'str' into the hash state and return the length of the string.
+ */
+static inline size_t
+fasthash_accum_cstring(fasthash_state *hs, const char *str)
+{
+#if SIZEOF_VOID_P >= 8
+
+	size_t		len;
+#ifdef USE_ASSERT_CHECKING
+	size_t		len_check;
+	fasthash_state hs_check;
+
+	memcpy(&hs_check, hs, sizeof(fasthash_state));
+	len_check = fasthash_accum_cstring_unaligned(&hs_check, str);
+#endif
+	if (PointerIsAligned(str, uint64))
+	{
+		len = fasthash_accum_cstring_aligned(hs, str);
+		Assert(len_check == len);
+		Assert(hs_check.hash == hs->hash);
+		return len;
+	}
+#endif							/* SIZEOF_VOID_P */
+
+	/*
+	 * It's not worth it to try to make the word-at-a-time optimization work
+	 * on 32-bit platforms.
+	 */
+	return fasthash_accum_cstring_unaligned(hs, str);
+}
+
+/*
+ * The finalizer
+ *
+ * 'tweak' is intended to be the input length when the caller doesn't know
+ * the length ahead of time, such as for NUL-terminated strings, otherwise
+ * zero.
+ */
+static inline uint64
+fasthash_final64(fasthash_state *hs, uint64 tweak)
+{
+	return fasthash_mix(hs->hash, tweak);
+}
+
+/*
+ * Reduce a 64-bit hash to a 32-bit hash.
+ *
+ * This optional step provides a bit more additional mixing compared to
+ * just taking the lower 32-bits.
+ */
+static inline uint32
+fasthash_reduce32(uint64 h)
+{
+	/*
+	 * Convert the 64-bit hashcode to Fermat residue, which shall retain
+	 * information from both the higher and lower parts of hashcode.
+	 */
+	return h - (h >> 32);
+}
+
+/* finalize and reduce */
+static inline uint32
+fasthash_final32(fasthash_state *hs, uint64 tweak)
+{
+	return fasthash_reduce32(fasthash_final64(hs, tweak));
+}
+
+/*
+ * The original fasthash64 function, re-implemented using the incremental
+ * interface. Returns a 64-bit hashcode. 'len' controls not only how
+ * many bytes to hash, but also modifies the internal seed.
+ * 'seed' can be zero.
+ */
+static inline uint64
+fasthash64(const char *k, size_t len, uint64 seed)
+{
+	fasthash_state hs;
+
+	fasthash_init(&hs, 0);
+
+	/* re-initialize the seed according to input length */
+	hs.hash = seed ^ (len * 0x880355f21e6d1965);
+
+	while (len >= FH_SIZEOF_ACCUM)
+	{
+		fasthash_accum(&hs, k, FH_SIZEOF_ACCUM);
+		k += FH_SIZEOF_ACCUM;
+		len -= FH_SIZEOF_ACCUM;
+	}
+
+	fasthash_accum(&hs, k, len);
+	return fasthash_final64(&hs, 0);
+}
+
+/* like fasthash64, but returns a 32-bit hashcode */
+static inline uint32
+fasthash32(const char *k, size_t len, uint64 seed)
+{
+	return fasthash_reduce32(fasthash64(k, len, seed));
+}
+
+/*
+ * Convenience function for hashing NUL-terminated strings
+ */
+static inline uint32
+hash_string(const char *s)
+{
+	fasthash_state hs;
+	size_t		s_len;
+
+	fasthash_init(&hs, 0);
+
+	/*
+	 * Combine string into the hash and save the length for tweaking the final
+	 * mix.
+	 */
+	s_len = fasthash_accum_cstring(&hs, s);
+
+	return fasthash_final32(&hs, s_len);
+}
+
+#endif							/* HASHFN_UNSTABLE_H */

data/ext/pg_query/include/postgres/common/int.h

@@ -1,7 +1,7 @@
 /*-------------------------------------------------------------------------
  *
  * int.h
- *	  Routines to perform integer math, while checking for overflows.
+ *	  Overflow-aware integer math and integer comparison routines.
  *
  * The routines in this file are intended to be well defined C, without
  * relying on compiler flags like -fwrapv.
@@ -11,7 +11,7 @@
  * the 64 bit cases can be considerably faster with intrinsics. In case no
  * intrinsics are available 128 bit math is used where available.
  *
- * Copyright (c) 2017-2023, PostgreSQL Global Development Group
+ * Copyright (c) 2017-2024, PostgreSQL Global Development Group
  *
  * src/include/common/int.h
  *
@@ -22,7 +22,7 @@
 
 
 /*---------
- * The following guidelines apply to all the routines:
+ * The following guidelines apply to all the overflow routines:
  * - If a + b overflows, return true, otherwise store the result of a + b
  * into *result. The content of *result is implementation defined in case of
  * overflow.
@@ -200,8 +200,12 @@ pg_sub_s64_overflow(int64 a, int64 b, int64 *result)
 	*result = (int64) res;
 	return false;
 #else
+	/*
+	 * Note: overflow is also possible when a == 0 and b < 0 (specifically,
+	 * when b == PG_INT64_MIN).
+	 */
 	if ((a < 0 && b > 0 && a < PG_INT64_MIN + b) ||
-		(a > 0 && b < 0 && a > PG_INT64_MAX + b))
+		(a >= 0 && b < 0 && a > PG_INT64_MAX + b))
 	{
 		*result = 0x5EED;		/* to avoid spurious warnings */
 		return true;
@@ -434,4 +438,75 @@ pg_mul_u64_overflow(uint64 a, uint64 b, uint64 *result)
 #endif
 }
 
+/*------------------------------------------------------------------------
+ *
+ * Comparison routines for integer types.
+ *
+ * These routines are primarily intended for use in qsort() comparator
+ * functions and therefore return a positive integer, 0, or a negative
+ * integer depending on whether "a" is greater than, equal to, or less
+ * than "b", respectively.  These functions are written to be as efficient
+ * as possible without introducing overflow risks, thereby helping ensure
+ * the comparators that use them are transitive.
+ *
+ * Types with fewer than 32 bits are cast to signed integers and
+ * subtracted.  Other types are compared using > and <, and the results of
+ * those comparisons (which are either (int) 0 or (int) 1 per the C
+ * standard) are subtracted.
+ *
+ * NB: If the comparator function is inlined, some compilers may produce
+ * worse code with these helper functions than with code with the
+ * following form:
+ *
+ *     if (a < b)
+ *         return -1;
+ *     if (a > b)
+ *         return 1;
+ *     return 0;
+ *
+ *------------------------------------------------------------------------
+ */
+
+static inline int
+pg_cmp_s16(int16 a, int16 b)
+{
+	return (int32) a - (int32) b;
+}
+
+static inline int
+pg_cmp_u16(uint16 a, uint16 b)
+{
+	return (int32) a - (int32) b;
+}
+
+static inline int
+pg_cmp_s32(int32 a, int32 b)
+{
+	return (a > b) - (a < b);
+}
+
+static inline int
+pg_cmp_u32(uint32 a, uint32 b)
+{
+	return (a > b) - (a < b);
+}
+
+static inline int
+pg_cmp_s64(int64 a, int64 b)
+{
+	return (a > b) - (a < b);
+}
+
+static inline int
+pg_cmp_u64(uint64 a, uint64 b)
+{
+	return (a > b) - (a < b);
+}
+
+static inline int
+pg_cmp_size(size_t a, size_t b)
+{
+	return (a > b) - (a < b);
+}
+
 #endif							/* COMMON_INT_H */

data/ext/pg_query/include/postgres/common/keywords.h

@@ -4,7 +4,7 @@
  *	  PostgreSQL's list of SQL keywords
  *
  *
- * Portions Copyright (c) 1996-2023, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1996-2024, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
  * src/include/common/keywords.h

data/ext/pg_query/include/postgres/common/kwlookup.h

@@ -4,7 +4,7 @@
  *	  Key word lookup for PostgreSQL
  *
  *
- * Portions Copyright (c) 1996-2023, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1996-2024, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
  * src/include/common/kwlookup.h

data/ext/pg_query/include/postgres/common/pg_prng.h

@@ -2,7 +2,7 @@
  *
  * Pseudo-Random Number Generator
  *
- * Copyright (c) 2021-2023, PostgreSQL Global Development Group
+ * Copyright (c) 2021-2024, PostgreSQL Global Development Group
  *
  * src/include/common/pg_prng.h
  *
@@ -51,6 +51,7 @@ extern uint64 pg_prng_uint64(pg_prng_state *state);
 extern uint64 pg_prng_uint64_range(pg_prng_state *state, uint64 rmin, uint64 rmax);
 extern int64 pg_prng_int64(pg_prng_state *state);
 extern int64 pg_prng_int64p(pg_prng_state *state);
+extern int64 pg_prng_int64_range(pg_prng_state *state, int64 rmin, int64 rmax);
 extern uint32 pg_prng_uint32(pg_prng_state *state);
 extern int32 pg_prng_int32(pg_prng_state *state);
 extern int32 pg_prng_int32p(pg_prng_state *state);

data/ext/pg_query/include/postgres/common/relpath.h

@@ -3,7 +3,7 @@
  * relpath.h
  *		Declarations for GetRelationPath() and friends
  *
- * Portions Copyright (c) 1996-2023, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1996-2024, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
  * src/include/common/relpath.h
@@ -50,7 +50,7 @@ typedef enum ForkNumber
 	MAIN_FORKNUM = 0,
 	FSM_FORKNUM,
 	VISIBILITYMAP_FORKNUM,
-	INIT_FORKNUM
+	INIT_FORKNUM,
 
 	/*
 	 * NOTE: if you add a new fork, change MAX_FORKNUM and possibly
@@ -74,7 +74,7 @@ extern int	forkname_chars(const char *str, ForkNumber *fork);
 extern char *GetDatabasePath(Oid dbOid, Oid spcOid);
 
 extern char *GetRelationPath(Oid dbOid, Oid spcOid, RelFileNumber relNumber,
-							 int backendId, ForkNumber forkNumber);
+							 int procNumber, ForkNumber forkNumber);
 
 /*
  * Wrapper macros for GetRelationPath.  Beware of multiple
@@ -88,7 +88,7 @@ extern char *GetRelationPath(Oid dbOid, Oid spcOid, RelFileNumber relNumber,
 
 /* First argument is a RelFileLocator */
 #define relpathperm(rlocator, forknum) \
-	relpathbackend(rlocator, InvalidBackendId, forknum)
+	relpathbackend(rlocator, INVALID_PROC_NUMBER, forknum)
 
 /* First argument is a RelFileLocatorBackend */
 #define relpath(rlocator, forknum) \

data/ext/pg_query/include/postgres/common/scram-common.h

@@ -3,7 +3,7 @@
  * scram-common.h
  *		Declarations for helper functions used for SCRAM authentication
  *
- * Portions Copyright (c) 1996-2023, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1996-2024, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
  * src/include/common/scram-common.h

data/ext/pg_query/include/postgres/common/sha2.h

@@ -3,7 +3,7 @@
  * sha2.h
  *	  Constants related to SHA224, 256, 384 AND 512.
  *
- * Portions Copyright (c) 1996-2023, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1996-2024, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
  * IDENTIFICATION

data/ext/pg_query/include/postgres/common/string.h

@@ -2,7 +2,7 @@
  *	string.h
  *		string handling helpers
  *
- *	Portions Copyright (c) 1996-2023, PostgreSQL Global Development Group
+ *	Portions Copyright (c) 1996-2024, PostgreSQL Global Development Group
  *	Portions Copyright (c) 1994, Regents of the University of California
  *
  *	src/include/common/string.h

data/ext/pg_query/include/postgres/common/unicode_east_asian_fw_table.h

@@ -39,14 +39,13 @@ static const struct mbinterval east_asian_fw[] = {
 	{0x2E80, 0x2E99},
 	{0x2E9B, 0x2EF3},
 	{0x2F00, 0x2FD5},
-	{0x2FF0, 0x2FFB},
-	{0x3000, 0x303E},
+	{0x2FF0, 0x303E},
 	{0x3041, 0x3096},
 	{0x3099, 0x30FF},
 	{0x3105, 0x312F},
 	{0x3131, 0x318E},
 	{0x3190, 0x31E3},
-	{0x31F0, 0x321E},
+	{0x31EF, 0x321E},
 	{0x3220, 0x3247},
 	{0x3250, 0x4DBF},
 	{0x4E00, 0xA48C},