greenplumn reloptions_gp 源码

  • 2022-08-18
  • 浏览 (564)

greenplumn reloptions_gp 代码

文件路径:/src/backend/access/common/reloptions_gp.c

/*-------------------------------------------------------------------------
 *
 * reloptions_gp.c
 *	  GPDB-specific relation options.
 *
 * These are in a separate file from reloptions.c, in order to reduce
 * conflicts when merging with upstream code.
 *
 *
 * Portions Copyright (c) 2017-Present VMware, Inc. or its affiliates.
 * Portions Copyright (c) 1996-2009, PostgreSQL Global Development Group
 * Portions Copyright (c) 1994, Regents of the University of California
 *
 *
 * IDENTIFICATION
 *	  src/backend/access/common/reloptions_gp.c
 *
 *-------------------------------------------------------------------------
 */

#include "postgres.h"

#include "access/bitmap.h"
#include "access/reloptions.h"
#include "catalog/pg_type.h"
#include "cdb/cdbappendonlyam.h"
#include "cdb/cdbvars.h"
#include "commands/defrem.h"
#include "utils/array.h"
#include "utils/builtins.h"
#include "utils/formatting.h"
#include "utils/guc.h"
#include "utils/memutils.h"
#include "miscadmin.h"
#include "nodes/makefuncs.h"

/*
 * Helper macro used for validation
 */
#define KIND_IS_APPENDOPTIMIZED(kind) (((kind) & RELOPT_KIND_APPENDOPTIMIZED) != 0)

/*
 * GPDB reloptions specification.
 */

static relopt_bool boolRelOpts_gp[] =
{
	{
		{
			SOPT_CHECKSUM,
			"Append table checksum",
			RELOPT_KIND_APPENDOPTIMIZED,
			AccessExclusiveLock
		},
		AO_DEFAULT_CHECKSUM
	},
	/* list terminator */
	{{NULL}}
};

static relopt_int intRelOpts_gp[] =
{
	{
		{
			SOPT_FILLFACTOR,
			"Packs bitmap index pages only to this percentage",
			RELOPT_KIND_BITMAP,
			ShareUpdateExclusiveLock	/* since it applies only to later
										 * inserts */
		},
		BITMAP_DEFAULT_FILLFACTOR, BITMAP_MIN_FILLFACTOR, 100
	},
	{
		{
			SOPT_BLOCKSIZE,
			"AO tables block size in bytes",
			RELOPT_KIND_APPENDOPTIMIZED,
			AccessExclusiveLock
		},
		AO_DEFAULT_BLOCKSIZE, MIN_APPENDONLY_BLOCK_SIZE, MAX_APPENDONLY_BLOCK_SIZE
	},
	{
		{
			SOPT_COMPLEVEL,
			"AO table compression level",
			RELOPT_KIND_APPENDOPTIMIZED,
			ShareUpdateExclusiveLock	/* since it applies only to later
										 * inserts */
		},
		AO_DEFAULT_COMPRESSLEVEL, AO_MIN_COMPRESSLEVEL, AO_MAX_COMPRESSLEVEL
	},
	/* list terminator */
	{{NULL}}
};

static relopt_real realRelOpts_gp[] =
{
	/* list terminator */
	{{NULL}}
};

static relopt_string stringRelOpts_gp[] =
{
	{
		{
			SOPT_COMPTYPE,
			"AO tables compression type",
			RELOPT_KIND_APPENDOPTIMIZED,
			AccessExclusiveLock
		},
		0, true, NULL, ""
	},
	/* list terminator */
	{{NULL}}
};

static void free_options_deep(relopt_value *options, int num_options);
static relopt_value *get_option_set(relopt_value *options, int num_options, const char *opt_name);
static bool reloption_is_default(const char *optstr, int optlen);

/*
 * initialize_reloptions_gp
 * 		initialization routine for GPDB reloptions
 *
 * We use the add_*_option interface in reloptions.h to add GPDB-specific options.
 */
void
initialize_reloptions_gp(void)
{
	int			i;
	static bool	initialized = false;

	/* only add these on first call. */
	if (initialized)
		return;
	initialized = true;

	/* Set GPDB specific options */
	for (i = 0; boolRelOpts_gp[i].gen.name; i++)
	{
		add_bool_reloption(boolRelOpts_gp[i].gen.kinds,
						   (char *) boolRelOpts_gp[i].gen.name,
						   (char *) boolRelOpts_gp[i].gen.desc,
						   boolRelOpts_gp[i].default_val);
		set_reloption_lockmode(boolRelOpts_gp[i].gen.name, boolRelOpts_gp[i].gen.lockmode);
	}

	for (i = 0; intRelOpts_gp[i].gen.name; i++)
	{
		add_int_reloption(intRelOpts_gp[i].gen.kinds,
						  (char *) intRelOpts_gp[i].gen.name,
						  (char *) intRelOpts_gp[i].gen.desc,
						  intRelOpts_gp[i].default_val,
						  intRelOpts_gp[i].min,
						  intRelOpts_gp[i].max);
		set_reloption_lockmode(intRelOpts_gp[i].gen.name, intRelOpts_gp[i].gen.lockmode);
	}

	for (i = 0; realRelOpts_gp[i].gen.name; i++)
	{
		add_real_reloption(realRelOpts_gp[i].gen.kinds,
						   (char *) realRelOpts_gp[i].gen.name,
						   (char *) realRelOpts_gp[i].gen.desc,
						   realRelOpts_gp[i].default_val,
						   realRelOpts_gp[i].min, realRelOpts_gp[i].max);
		set_reloption_lockmode(realRelOpts_gp[i].gen.name, realRelOpts_gp[i].gen.lockmode);
	}

	for (i = 0; stringRelOpts_gp[i].gen.name; i++)
	{
		add_string_reloption(stringRelOpts_gp[i].gen.kinds,
							 (char *) stringRelOpts_gp[i].gen.name,
							 (char *) stringRelOpts_gp[i].gen.desc,
							 NULL,
							 stringRelOpts_gp[i].validate_cb);
		set_reloption_lockmode(stringRelOpts_gp[i].gen.name, stringRelOpts_gp[i].gen.lockmode);
	}
}

/*
 * This is set whenever the GUC gp_default_storage_options is set.
 */
static StdRdOptions ao_storage_opts;
static bool ao_storage_opts_changed = false;

/*
 * Accumulate a new datum for one AO storage option.
 */
static void
accumAOStorageOpt(char *name, char *value,
				  ArrayBuildState *astate)
{
	text	   *t;
	bool		boolval;
	int			intval;
	StringInfoData buf;

	Assert(astate);

	initStringInfo(&buf);

	if (pg_strcasecmp(SOPT_BLOCKSIZE, name) == 0)
	{
		if (!parse_int(value, &intval, 0 /* unit flags */ ,
					   NULL /* hint message */ ))
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("invalid integer value \"%s\" for storage option \"%s\"",
							value, name)));
		appendStringInfo(&buf, "%s=%d", SOPT_BLOCKSIZE, intval);
	}
	else if (pg_strcasecmp(SOPT_COMPTYPE, name) == 0)
	{
		appendStringInfo(&buf, "%s=%s", SOPT_COMPTYPE, value);
	}
	else if (pg_strcasecmp(SOPT_COMPLEVEL, name) == 0)
	{
		if (!parse_int(value, &intval, 0 /* unit flags */ ,
					   NULL /* hint message */ ))
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("invalid integer value \"%s\" for storage option \"%s\"",
							value, name)));
		appendStringInfo(&buf, "%s=%d", SOPT_COMPLEVEL, intval);
	}
	else if (pg_strcasecmp(SOPT_CHECKSUM, name) == 0)
	{
		if (!parse_bool(value, &boolval))
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("invalid bool value \"%s\" for storage option \"%s\"",
							value, name)));
		appendStringInfo(&buf, "%s=%s", SOPT_CHECKSUM, boolval ? "true" : "false");
	}
	else
	{
		/*
		 * Provide a user friendly message in case that the options are
		 * appendonly and its variants
		 */
		if (!pg_strcasecmp(name, "appendonly") ||
			!pg_strcasecmp(name, "appendoptimized") ||
			!pg_strcasecmp(name, "orientation"))
		{
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("invalid storage option \"%s\"", name),
					 errhint("For table access methods use \"default_table_access_method\" instead.")));
		}
		else
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("invalid storage option \"%s\"", name)));
	}

	t = cstring_to_text(buf.data);

	accumArrayResult(astate, PointerGetDatum(t), /* disnull */ false,
					 TEXTOID, CurrentMemoryContext);
	pfree(t);
	pfree(buf.data);
}

/*
 * Reset appendonly storage options to factory defaults.  Callers must
 * free ao_opts->compresstype before calling this method.
 */
inline void
resetAOStorageOpts(StdRdOptions *ao_opts)
{
	ao_opts->blocksize = AO_DEFAULT_BLOCKSIZE;
	ao_opts->checksum = AO_DEFAULT_CHECKSUM;
	ao_opts->compresslevel = AO_DEFAULT_COMPRESSLEVEL;
	ao_opts->compresstype[0] = '\0';
}

/*
 * This needs to happen whenever gp_default_storage_options GUC is
 * reset.
 */
void
resetDefaultAOStorageOpts(void)
{
	resetAOStorageOpts(&ao_storage_opts);
	ao_storage_opts_changed = false;
}

const StdRdOptions *
currentAOStorageOptions(void)
{
	return (const StdRdOptions *) &ao_storage_opts;
}

/*
 * Set global appendonly storage options.
 */
void
setDefaultAOStorageOpts(StdRdOptions *copy)
{
	Assert(copy);

	memcpy(&ao_storage_opts, copy, sizeof(ao_storage_opts));
	if (pg_strcasecmp(copy->compresstype, "none") == 0)
	{
		/* Represent compresstype=none as an empty string (MPP-25073). */
		ao_storage_opts.compresstype[0] = '\0';
	}

	ao_storage_opts_changed = true;
}

static int	setDefaultCompressionLevel(char *compresstype);

/*
 * Accept a string of the form "name=value,name=value,...".  Space
 * around ',' and '=' is allowed.  Parsed values are stored in
 * corresponding fields of StdRdOptions object.  The parser is a
 * finite state machine that changes states for each input character
 * scanned.
 */
Datum
parseAOStorageOpts(const char *opts_str)
{
	int			dims[1];
	int			lbs[1];
	Datum		result;
	ArrayBuildState *astate;

	const char *cp;
	const char *name_st = NULL;
	const char *value_st = NULL;
	char	   *name = NULL,
			   *value = NULL;

	enum state
	{
		/*
		 * Consume whitespace at the beginning of a name token.
		 */
		LEADING_NAME,

		/*
		 * Name token is being scanned.  Allowed characters are alphabets,
		 * whitespace and '='.
		 */
		NAME_TOKEN,

		/*
		 * Name token was terminated by whitespace.  This state scans the
		 * trailing whitespace after name token.
		 */
		TRAILING_NAME,

		/*
		 * Whitespace after '=' and before value token.
		 */
		LEADING_VALUE,

		/*
		 * Value token is being scanned.  Allowed characters are alphabets,
		 * digits, '_'.  Value should be delimited by a ',', whitespace or end
		 * of string '\0'.
		 */
		VALUE_TOKEN,

		/*
		 * Whitespace after value token.
		 */
		TRAILING_VALUE,

		/*
		 * End of string.  This state can only be entered from VALUE_TOKEN or
		 * TRAILING_VALUE.
		 */
		EOS
	};
	enum state	st = LEADING_NAME;

	/*
	 * Initialize ArrayBuildState ourselves rather than leaving it to
	 * accumArrayResult().  This aviods the catalog lookup (pg_type) performed
	 * by accumArrayResult().
	 */
	astate = (ArrayBuildState *) palloc(sizeof(ArrayBuildState));
	astate->mcontext = CurrentMemoryContext;
	astate->alen = 10;			/* Initial number of name=value pairs. */
	astate->dvalues = (Datum *) palloc(astate->alen * sizeof(Datum));
	astate->dnulls = (bool *) palloc(astate->alen * sizeof(bool));
	astate->nelems = 0;
	astate->element_type = TEXTOID;
	astate->typlen = -1;
	astate->typbyval = false;
	astate->typalign = 'i';

	cp = opts_str - 1;
	do
	{
		++cp;
		switch (st)
		{
			case LEADING_NAME:
				if (isalpha(*cp))
				{
					st = NAME_TOKEN;
					name_st = cp;
				}
				else if (!isspace(*cp))
				{
					ereport(ERROR,
							(errcode(ERRCODE_SYNTAX_ERROR),
							 errmsg("invalid storage option name in \"%s\"",
									opts_str)));
				}
				break;
			case NAME_TOKEN:
				if (isspace(*cp))
					st = TRAILING_NAME;
				else if (*cp == '=')
					st = LEADING_VALUE;
				else if (!isalpha(*cp))
					ereport(ERROR,
							(errcode(ERRCODE_SYNTAX_ERROR),
							 errmsg("invalid storage option name in \"%s\"",
									opts_str)));
				if (st != NAME_TOKEN)
				{
					name = palloc(cp - name_st + 1);
					strncpy(name, name_st, cp - name_st);
					name[cp - name_st] = '\0';
					for (name_st = name; *name_st != '\0'; ++name_st)
						*(char *) name_st = pg_tolower(*name_st);
				}
				break;
			case TRAILING_NAME:
				if (*cp == '=')
					st = LEADING_VALUE;
				else if (!isspace(*cp))
					ereport(ERROR,
							(errcode(ERRCODE_SYNTAX_ERROR),
							 errmsg("invalid value for option \"%s\", expected \"=\"", name)));
				break;
			case LEADING_VALUE:
				if (isalnum(*cp))
				{
					st = VALUE_TOKEN;
					value_st = cp;
				}
				else if (!isspace(*cp))
					ereport(ERROR,
							(errcode(ERRCODE_SYNTAX_ERROR),
							 errmsg("invalid value for option \"%s\"", name)));
				break;
			case VALUE_TOKEN:
				if (isspace(*cp))
					st = TRAILING_VALUE;
				else if (*cp == '\0')
					st = EOS;
				else if (*cp == ',')
					st = LEADING_NAME;
				/* Need to check '_' for rle_type */
				else if (!(isalnum(*cp) || *cp == '_'))
					ereport(ERROR,
							(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
							 errmsg("invalid value for option \"%s\"", name)));
				if (st != VALUE_TOKEN)
				{
					value = palloc(cp - value_st + 1);
					strncpy(value, value_st, cp - value_st);
					value[cp - value_st] = '\0';
					for (value_st = value; *value_st != '\0'; ++value_st)
						*(char *) value_st = pg_tolower(*value_st);
					Assert(name);
					accumAOStorageOpt(name, value, astate);
					pfree(name);
					name = NULL;
					pfree(value);
					value = NULL;
				}
				break;
			case TRAILING_VALUE:
				if (*cp == ',')
					st = LEADING_NAME;
				else if (*cp == '\0')
					st = EOS;
				else if (!isspace(*cp))
					ereport(ERROR,
							(errcode(ERRCODE_SYNTAX_ERROR),
							 errmsg("syntax error after \"%s\"", value)));
				break;
			case EOS:

				/*
				 * We better get out of the loop right after entering this
				 * state.  Therefore, we should never get here.
				 */
				elog(ERROR, "invalid value \"%s\" for GUC", opts_str);
				break;
		};
	} while (*cp != '\0');
	if (st != EOS)
		elog(ERROR, "invalid value \"%s\" for GUC", opts_str);

	lbs[0] = 1;
	dims[0] = astate->nelems;
	result = makeMdArrayResult(astate, 1, dims, lbs, CurrentMemoryContext, false);
	pfree(astate->dvalues);
	pfree(astate->dnulls);
	pfree(astate);
	return result;
}

/*
 * Return a datum that is array of "name=value" strings for each
 * appendonly storage option in opts.  This datum is used to populate
 * pg_class.reloptions during relation creation.
 *
 * To avoid catalog bloat, we only create "name=value" item for those
 * values in opts that are not specified in WITH clause and are
 * different from their initial defaults.
 */
Datum
transformAOStdRdOptions(StdRdOptions *opts, Datum withOpts)
{
	char	   *strval;
	Datum	   *withDatums = NULL;
	Datum		d;
	text	   *t;
	int			i,
				withLen,
				soptLen,
				nWithOpts = 0;
	ArrayType  *withArr;
	ArrayBuildState *astate = NULL;
	bool		foundBlksz = false,
				foundComptype = false,
				foundComplevel = false,
				foundChecksum = false;

	/*
	 * withOpts must be parsed to see if an option was spcified in WITH()
	 * clause.
	 */
	if (DatumGetPointer(withOpts) != NULL)
	{
		withArr = DatumGetArrayTypeP(withOpts);
		Assert(ARR_ELEMTYPE(withArr) == TEXTOID);
		deconstruct_array(withArr, TEXTOID, -1, false, 'i', &withDatums,
						  NULL, &nWithOpts);

		/*
		 * Include options specified in WITH() clause in the same order as
		 * they are specified.  Otherwise we will end up with regression
		 * failures due to diff with respect to answer file.
		 */
		for (i = 0; i < nWithOpts; ++i)
		{
			t = DatumGetTextP(withDatums[i]);
			strval = VARDATA(t);

			/*
			 * Text datums are usually not null terminated.  We must never
			 * access beyond their length.
			 */
			withLen = VARSIZE(t) - VARHDRSZ;

			/*
			 * withDatums[i] may not be used directly.  It may be e.g.
			 * "bLoCksiZe=3213".  Therefore we don't set it as reloptions as
			 * is.
			 */
			soptLen = strlen(SOPT_BLOCKSIZE);
			if (withLen > soptLen &&
				pg_strncasecmp(strval, SOPT_BLOCKSIZE, soptLen) == 0)
			{
				foundBlksz = true;
				d = CStringGetTextDatum(psprintf("%s=%d",
												 SOPT_BLOCKSIZE,
												 opts->blocksize));
				astate = accumArrayResult(astate, d, false, TEXTOID,
										  CurrentMemoryContext);
			}
			soptLen = strlen(SOPT_COMPTYPE);
			if (withLen > soptLen &&
				pg_strncasecmp(strval, SOPT_COMPTYPE, soptLen) == 0)
			{
				foundComptype = true;

				/*
				 * Record "none" as compresstype in reloptions if it was
				 * explicitly specified in WITH clause.
				 */
				d = CStringGetTextDatum(psprintf("%s=%s",
												 SOPT_COMPTYPE,
												 (opts->compresstype[0] ? opts->compresstype : "none")));
				astate = accumArrayResult(astate, d, false, TEXTOID,
										  CurrentMemoryContext);
			}
			soptLen = strlen(SOPT_COMPLEVEL);
			if (withLen > soptLen &&
				pg_strncasecmp(strval, SOPT_COMPLEVEL, soptLen) == 0)
			{
				foundComplevel = true;
				d = CStringGetTextDatum(psprintf("%s=%d",
												 SOPT_COMPLEVEL,
												 opts->compresslevel));
				astate = accumArrayResult(astate, d, false, TEXTOID,
										  CurrentMemoryContext);
			}
			soptLen = strlen(SOPT_CHECKSUM);
			if (withLen > soptLen &&
				pg_strncasecmp(strval, SOPT_CHECKSUM, soptLen) == 0)
			{
				foundChecksum = true;
				d = CStringGetTextDatum(psprintf("%s=%s",
												 SOPT_CHECKSUM,
												 (opts->checksum ? "true" : "false")));
				astate = accumArrayResult(astate, d, false, TEXTOID,
										  CurrentMemoryContext);
			}
		}
	}

	if ((opts->blocksize != AO_DEFAULT_BLOCKSIZE) && !foundBlksz)
	{
		d = CStringGetTextDatum(psprintf("%s=%d",
										 SOPT_BLOCKSIZE,
										 opts->blocksize));
		astate = accumArrayResult(astate, d, false, TEXTOID,
								  CurrentMemoryContext);
	}

	/*
	 * Record compression options only if compression is enabled.  No need to
	 * check compresstype here as by the time we get here, "opts" should have
	 * been set by default_reloptions() correctly.
	 */
	if (opts->compresslevel > AO_DEFAULT_COMPRESSLEVEL &&
		opts->compresstype[0])
	{
		if (!foundComptype && (
							   (pg_strcasecmp(opts->compresstype, AO_DEFAULT_COMPRESSTYPE) == 0
								&& opts->compresslevel == 1 && !foundComplevel) ||
							   pg_strcasecmp(opts->compresstype,
											 AO_DEFAULT_COMPRESSTYPE) != 0))
		{
			d = CStringGetTextDatum(psprintf("%s=%s",
											 SOPT_COMPTYPE,
											 opts->compresstype));
			astate = accumArrayResult(astate, d, false, TEXTOID,
									  CurrentMemoryContext);
		}
		/* When compression is enabled, default compresslevel is 1. */
		if ((opts->compresslevel != 1) &&
			!foundComplevel)
		{
			d = CStringGetTextDatum(psprintf("%s=%d",
											 SOPT_COMPLEVEL,
											 opts->compresslevel));
			astate = accumArrayResult(astate, d, false, TEXTOID,
									  CurrentMemoryContext);
		}
	}

	if ((opts->checksum != AO_DEFAULT_CHECKSUM) && !foundChecksum)
	{
		d = CStringGetTextDatum(psprintf("%s=%s",
										 SOPT_CHECKSUM,
										 (opts->checksum ? "true" : "false")));
		astate = accumArrayResult(astate, d, false, TEXTOID,
								  CurrentMemoryContext);
	}
	return astate ?
		makeArrayResult(astate, CurrentMemoryContext) :
		PointerGetDatum(NULL);
}

/* 
 * Check if the given reloption string has default value.
 */
static bool
reloption_is_default(const char *optstr, int optlen)
{
	char 		*defaultopt = NULL;
	bool 		res;

	if (optlen > strlen(SOPT_BLOCKSIZE) &&
		pg_strncasecmp(optstr, SOPT_BLOCKSIZE, strlen(SOPT_BLOCKSIZE)) == 0)
	{
		defaultopt = psprintf("%s=%d",
										 SOPT_BLOCKSIZE,
										 AO_DEFAULT_BLOCKSIZE);
	}
	else if (optlen > strlen(SOPT_COMPTYPE) &&
		pg_strncasecmp(optstr, SOPT_COMPTYPE, strlen(SOPT_COMPTYPE)) == 0)
	{
		defaultopt = psprintf("%s=%s",
										 SOPT_COMPTYPE,
										 AO_DEFAULT_COMPRESSTYPE);
	}
	else if (optlen > strlen(SOPT_COMPLEVEL) &&
		pg_strncasecmp(optstr, SOPT_COMPLEVEL, strlen(SOPT_COMPLEVEL)) == 0)
	{
		defaultopt = psprintf("%s=%d",
										 SOPT_COMPLEVEL,
										 AO_DEFAULT_COMPRESSLEVEL);
	}
	else if (optlen > strlen(SOPT_CHECKSUM) &&
		pg_strncasecmp(optstr, SOPT_CHECKSUM, strlen(SOPT_CHECKSUM)) == 0)
	{
		defaultopt = psprintf("%s=%s",
										 SOPT_CHECKSUM,
										 AO_DEFAULT_CHECKSUM ? "true" : "false");
	}

	if (defaultopt != NULL)
		res = strlen(defaultopt) == optlen && 
				pg_strncasecmp(optstr, defaultopt, optlen) == 0;
	else
		res = false;

	pfree(defaultopt);
	return res;
}

/* 
 * Check if two string arrays of reloptions are the same.
 *
 * Note that this will not handle the case where the option doesn't contain 
 * the '=' sign in it, e.g. "checksum" vs. "checksum=true". But it seems 
 * that at this point we should always have both options as "x=y" anyways.
 */
bool
relOptionsEquals(Datum oldOptions, Datum newOptions)
{
	ArrayType 	*oldoptarray, *newoptarray;
	Datum 		*opts1, *opts2;
	int		noldoptions = 0, nnewoptions = 0;
	int		i, j;

	/* Deconstruct both options. */
	if (PointerIsValid(DatumGetPointer(oldOptions)))
	{
		oldoptarray = DatumGetArrayTypeP(oldOptions);
		deconstruct_array(oldoptarray, TEXTOID, -1, false, 'i',
						  &opts1, NULL, &noldoptions);
	}
	if (PointerIsValid(DatumGetPointer(newOptions)))
	{
		newoptarray = DatumGetArrayTypeP(newOptions);
		deconstruct_array(newoptarray, TEXTOID, -1, false, 'i',
						  &opts2, NULL, &nnewoptions);
	}

	for (i = 0; i < nnewoptions; i++)
	{
		char 	*newopt_str = VARDATA(opts2[i]);
		int	newopt_len = VARSIZE(opts2[i]) - VARHDRSZ;
		int 	keylen;

		/* Should be "x=y" but better panic here rather than returning wrong result. */
		Assert(strchr(newopt_str, '=') != 0);

		keylen = strchr(newopt_str, '=') - newopt_str;

		/* Search for a match in old options. */
		for (j = 0; j < noldoptions; j++)
		{
			char 	*oldopt_str = VARDATA(opts1[j]);
			int	oldopt_len = VARSIZE(opts1[j]) - VARHDRSZ;

			/* Not the same option. */
			if (oldopt_len <= keylen || 
					pg_strncasecmp(oldopt_str, newopt_str, keylen) != 0)
				continue;

			/* Old option should be as "x=y" too. */
			Assert(oldopt_str[keylen] == '=');

			/* Key found, now they must match exactly otherwise it's a changed option. */
			if (oldopt_len != newopt_len ||
					pg_strncasecmp(oldopt_str, newopt_str, oldopt_len) != 0)
				return false;
			else
				break;
		}

		/* 
		 * If key not found, then it must've changed unless it's a default value 
		 * that doesn't appear in the old reloptions.
		 */
		if (j == noldoptions && !reloption_is_default(newopt_str, newopt_len))
			return false;
	}
	return true;
}

void
validate_and_adjust_options(StdRdOptions *result,
							relopt_value *options,
							int num_options, relopt_kind kind, bool validate)
{
	int			i;
	relopt_value *blocksize_opt;
	relopt_value *comptype_opt;
	relopt_value *complevel_opt;
	relopt_value *checksum_opt;

	/* blocksize */
	blocksize_opt = get_option_set(options, num_options, SOPT_BLOCKSIZE);
	if (blocksize_opt != NULL)
	{
		if (!KIND_IS_APPENDOPTIMIZED(kind) && validate)
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("usage of parameter \"blocksize\" in a non relation object is not supported")));

		result->blocksize = blocksize_opt->values.int_val;

		if (result->blocksize < MIN_APPENDONLY_BLOCK_SIZE ||
			result->blocksize > MAX_APPENDONLY_BLOCK_SIZE ||
			result->blocksize % MIN_APPENDONLY_BLOCK_SIZE != 0)
		{
			if (validate)
				ereport(ERROR,
						(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
						 errmsg("block size must be between 8KB and 2MB and be a multiple of 8KB"),
						 errdetail("Got block size %d.", result->blocksize)));

			result->blocksize = DEFAULT_APPENDONLY_BLOCK_SIZE;
		}

	}

	/* compression type */
	comptype_opt = get_option_set(options, num_options, SOPT_COMPTYPE);
	if (comptype_opt != NULL)
	{
		if (!KIND_IS_APPENDOPTIMIZED(kind) && validate)
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("usage of parameter \"compresstype\" in a non relation object is not supported")));

		if (!compresstype_is_valid(comptype_opt->values.string_val))
			ereport(ERROR,
					(errcode(ERRCODE_UNDEFINED_OBJECT),
					 errmsg("unknown compresstype \"%s\"",
							comptype_opt->values.string_val)));
		for (i = 0; i < strlen(comptype_opt->values.string_val); i++)
			result->compresstype[i] = pg_tolower(comptype_opt->values.string_val[i]);
		result->compresstype[i] = '\0';
	}

	/* compression level */
	complevel_opt = get_option_set(options, num_options, SOPT_COMPLEVEL);
	if (complevel_opt != NULL)
	{
		if (!KIND_IS_APPENDOPTIMIZED(kind) && validate)
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("usage of parameter \"compresslevel\" in a non relation object is not supported")));

		result->compresslevel = complevel_opt->values.int_val;

		if (result->compresstype[0] &&
			pg_strcasecmp(result->compresstype, "none") != 0 &&
			result->compresslevel == 0 && validate)
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("compresstype \"%s\" can\'t be used with compresslevel 0",
							result->compresstype)));
		if (result->compresslevel < 0)
		{
			if (validate)
				ereport(ERROR,
						(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("compresslevel=%d is out of range (should be positive)",
								result->compresslevel)));

			result->compresslevel = setDefaultCompressionLevel(result->compresstype);
		}

		/*
		 * use the default compressor if compresslevel was indicated but not
		 * compresstype. must make a copy otherwise str_tolower below will
		 * crash.
		 */
		if (result->compresslevel > 0 && !result->compresstype[0])
			strlcpy(result->compresstype, AO_DEFAULT_COMPRESSTYPE, sizeof(result->compresstype));

		/* Check upper bound of compresslevel for each compression type */

		if (result->compresstype[0] &&
			(pg_strcasecmp(result->compresstype, "zlib") == 0))
		{
#ifndef HAVE_LIBZ
			ereport(ERROR,
					(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
					 errmsg("zlib compression is not supported by this build"),
					 errhint("Compile without --without-zlib to use zlib compression.")));
#endif
			if (result->compresslevel > 9)
			{
				if (validate)
					ereport(ERROR,
							(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
							 errmsg("compresslevel=%d is out of range for zlib (should be in the range 1 to 9)",
									result->compresslevel)));

				result->compresslevel = setDefaultCompressionLevel(result->compresstype);
			}
		}

		if (result->compresstype[0] &&
			(pg_strcasecmp(result->compresstype, "zstd") == 0))
		{
#ifndef USE_ZSTD
			ereport(ERROR,
					(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
					 errmsg("Zstandard library is not supported by this build"),
					 errhint("Compile with --with-zstd to use Zstandard compression.")));
#endif
			if (result->compresslevel > 19)
			{
				if (validate)
					ereport(ERROR,
							(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
							 errmsg("compresslevel=%d is out of range for zstd (should be in the range 1 to 19)",
									result->compresslevel)));

				result->compresslevel = setDefaultCompressionLevel(result->compresstype);
			}
		}

		if (result->compresstype[0] &&
			(pg_strcasecmp(result->compresstype, "quicklz") == 0))
		{
#ifndef HAVE_LIBQUICKLZ
			ereport(ERROR,
					(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
					 errmsg("QuickLZ library is not supported by this build"),
					 errhint("Compile with --with-quicklz to use QuickLZ compression.")));
#endif
			if (result->compresslevel != 1)
			{
				if (validate)
					ereport(ERROR,
							(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
							 errmsg("compresslevel=%d is out of range for quicklz (should be 1)",
									result->compresslevel)));

				result->compresslevel = setDefaultCompressionLevel(result->compresstype);
			}
		}

		if (result->compresstype[0] &&
			(pg_strcasecmp(result->compresstype, "rle_type") == 0) &&
			(result->compresslevel > 4))
		{
			if (validate)
				ereport(ERROR,
						(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
						 errmsg("compresslevel=%d is out of range for rle_type (should be in the range 1 to 4)",
								result->compresslevel)));

			result->compresslevel = setDefaultCompressionLevel(result->compresstype);
		}
	}

	/* checksum */
	checksum_opt = get_option_set(options, num_options, SOPT_CHECKSUM);
	if (checksum_opt != NULL)
	{
		if (!KIND_IS_APPENDOPTIMIZED(kind) && validate)
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("usage of parameter \"checksum\" in a non relation object is not supported")));

		result->checksum = checksum_opt->values.bool_val;
	}

	if (result->compresstype[0] &&
		result->compresslevel == AO_DEFAULT_COMPRESSLEVEL)
	{
		result->compresslevel = setDefaultCompressionLevel(result->compresstype);
	}
}

void
validate_and_refill_options(StdRdOptions *result, relopt_value *options,
							int numrelopts, relopt_kind kind, bool validate)
{
	if (validate &&
		ao_storage_opts_changed &&
		KIND_IS_APPENDOPTIMIZED(kind))
	{
		if (!(get_option_set(options, numrelopts, SOPT_BLOCKSIZE)))
			result->blocksize = ao_storage_opts.blocksize;

		if (!(get_option_set(options, numrelopts, SOPT_COMPLEVEL)))
			result->compresslevel = ao_storage_opts.compresslevel;

		if (!(get_option_set(options, numrelopts, SOPT_COMPTYPE)))
			strlcpy(result->compresstype, ao_storage_opts.compresstype, sizeof(result->compresstype));

		if (!(get_option_set(options, numrelopts, SOPT_CHECKSUM)))
			result->checksum = ao_storage_opts.checksum;
	}

	validate_and_adjust_options(result, options, numrelopts, kind, validate);
}

void
parse_validate_reloptions(StdRdOptions *result, Datum reloptions,
						  bool validate, relopt_kind kind)
{
	relopt_value *options;
	int			num_options;

	options = parseRelOptions(reloptions, validate, kind, &num_options);

	validate_and_adjust_options(result, options, num_options, kind, validate);
	free_options_deep(options, num_options);
}

/*
 * validateAppendOnlyRelOptions
 *
 *		Various checks for validity of appendonly relation rules.
 */
void
validateAppendOnlyRelOptions(int blocksize,
							 int safewrite,
							 int complevel,
							 char *comptype,
							 bool checksum,
							 bool co)
{
	if (comptype &&
		(pg_strcasecmp(comptype, "quicklz") == 0 ||
		 pg_strcasecmp(comptype, "zlib") == 0 ||
		 pg_strcasecmp(comptype, "rle_type") == 0 ||
		 pg_strcasecmp(comptype, "zstd") == 0))
	{
		if (!co &&
			pg_strcasecmp(comptype, "rle_type") == 0)
		{
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("%s cannot be used with Append Only relations row orientation",
							comptype)));
		}

		if (comptype && complevel == 0)
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("compresstype cannot be used with compresslevel 0")));

		if (comptype && (pg_strcasecmp(comptype, "zlib") == 0))
		{
#ifndef HAVE_LIBZ
			ereport(ERROR,
					(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
					 errmsg("zlib compression is not supported by this build"),
					 errhint("Compile without --without-zlib to use zlib compression.")));
#endif
			if (complevel < 0 || complevel > 9)
			{
				ereport(ERROR,
						(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
						 errmsg("compresslevel=%d is out of range (should be between 0 and 9)",
								complevel)));
			}
		}

		if (comptype && (pg_strcasecmp(comptype, "zstd") == 0))
		{
#ifndef USE_ZSTD
			ereport(ERROR,
					(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
					 errmsg("Zstandard library is not supported by this build"),
					 errhint("Compile without --without-zstd to use Zstandard compression.")));
#endif
			if (complevel < 0 || complevel > 19)
				ereport(ERROR,
						(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
						 errmsg("compresslevel=%d is out of range for zstd (should be in the range 1 to 19)",
								complevel)));
		}

		if (comptype && (pg_strcasecmp(comptype, "quicklz") == 0))
		{
#ifndef HAVE_LIBQUICKLZ
			ereport(ERROR,
					(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
					 errmsg("QuickLZ library is not supported by this build"),
					 errhint("Compile with --with-quicklz to use QuickLZ compression.")));
#endif
			if (complevel != 1)
				ereport(ERROR,
						(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
						 errmsg("compresslevel=%d is out of range for quicklz (should be 1)",
								complevel)));
		}
		if (comptype && (pg_strcasecmp(comptype, "rle_type") == 0) &&
			(complevel < 0 || complevel > 4))
		{
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("compresslevel=%d is out of range for rle_type (should be in the range 1 to 4)",
							complevel)));
		}
	}

	if (blocksize < MIN_APPENDONLY_BLOCK_SIZE ||
		blocksize > MAX_APPENDONLY_BLOCK_SIZE ||
		blocksize % MIN_APPENDONLY_BLOCK_SIZE != 0)
		ereport(ERROR,
				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
				 errmsg("block size must be between 8KB and 2MB and be an 8KB multiple, got %d", blocksize)));

	if (safewrite > MAX_APPENDONLY_BLOCK_SIZE || safewrite % 8 != 0)
		ereport(ERROR,
				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
				 errmsg("safefswrite size must be less than 8MB and be a multiple of 8")));

	if (gp_safefswritesize > blocksize)
		ereport(ERROR,
				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
				 errmsg("block size (%d) is smaller gp_safefswritesize (%d)",
						blocksize, gp_safefswritesize),
				 errhint("Increase blocksize or decrease gp_safefswritesize if it is safe to do so on this file system.")));
}

/*
 * if no compressor type was specified, we set to no compression (level 0)
 * otherwise default for both zlib, quicklz, zstd and RLE to level 1.
 */
static int
setDefaultCompressionLevel(char *compresstype)
{
	if (!compresstype || pg_strcasecmp(compresstype, "none") == 0)
		return 0;
	else
		return 1;
}

void
free_options_deep(relopt_value *options, int num_options)
{
	int			i;

	for (i = 0; i < num_options; ++i)
	{
		if (options[i].isset &&
			options[i].gen->type == RELOPT_TYPE_STRING &&
			options[i].values.string_val != NULL)
		{
			pfree(options[i].values.string_val);
		}
	}
	pfree(options);
}

relopt_value *
get_option_set(relopt_value *options, int num_options, const char *opt_name)
{
	int			i;
	int			opt_name_len;
	int			cmp_len;

	opt_name_len = strlen(opt_name);
	for (i = 0; i < num_options; ++i)
	{
		cmp_len = options[i].gen->namelen > opt_name_len ? opt_name_len : options[i].gen->namelen;
		if (options[i].isset && pg_strncasecmp(options[i].gen->name, opt_name, cmp_len) == 0)
			return &options[i];
	}
	return NULL;
}

/* ------------------------------------------------------------------------
 * Attribute Encoding specific functions
 * ------------------------------------------------------------------------
 */

/*
 * Check if the name is one of the ENCODING clauses.
 */
bool
is_storage_encoding_directive(char *name)
{
	/* names we expect to see in ENCODING clauses */
	static char *storage_directive_names[] = {"compresstype", 
						"compresslevel",
						"blocksize"};

	int i = 0;

	for (i = 0; i < lengthof(storage_directive_names); i++)
	{
		if (strcmp(name, storage_directive_names[i]) == 0)
			return true;
	}
	return false;
}

/*
 * Add any missing encoding attributes (compresstype = none,
 * blocksize=...).  The column specific encoding attributes supported
 * today are compresstype, compresslevel and blocksize.  Refer to
 * pg_compression.c for more info.
 *
 */
static List *
fillin_encoding(List *aocoColumnEncoding)
{
	bool foundCompressType = false;
	bool foundCompressTypeNone = false;
	char *cmplevel = NULL;
	bool foundBlockSize = false;
	char *arg;
	List *retList = list_copy(aocoColumnEncoding);
	ListCell *lc;
	DefElem *el;
	const StdRdOptions *ao_opts = currentAOStorageOptions();

	foreach(lc, aocoColumnEncoding)
	{
		el = lfirst(lc);

		if (pg_strcasecmp("compresstype", el->defname) == 0)
		{
			foundCompressType = true;
			arg = defGetString(el);
			if (pg_strcasecmp("none", arg) == 0)
				foundCompressTypeNone = true;
		}
		else if (pg_strcasecmp("compresslevel", el->defname) == 0)
		{
			cmplevel = defGetString(el);
		}
		else if (pg_strcasecmp("blocksize", el->defname) == 0)
			foundBlockSize = true;
	}

	if (foundCompressType == false && cmplevel == NULL)
	{
		/* No compression option specified, use current defaults. */
		arg = ao_opts->compresstype[0] ?
				pstrdup(ao_opts->compresstype) : "none";
		el = makeDefElem("compresstype", (Node *) makeString(arg), -1);
		retList = lappend(retList, el);
		el = makeDefElem("compresslevel",
						 (Node *) makeInteger(ao_opts->compresslevel),
						 -1);
		retList = lappend(retList, el);
	}
	else if (foundCompressType == false && cmplevel)
	{
		if (strcmp(cmplevel, "0") == 0)
		{
			/*
			 * User wants to disable compression by specifying
			 * compresslevel=0.
			 */
			el = makeDefElem("compresstype", (Node *) makeString("none"), -1);
			retList = lappend(retList, el);
		}
		else
		{
			/*
			 * User wants to enable compression by specifying non-zero
			 * compresslevel.  Therefore, choose default compresstype
			 * if configured, otherwise use zlib.
			 */
			if (ao_opts->compresstype[0] &&
				strcmp(ao_opts->compresstype, "none") != 0)
			{
				arg = pstrdup(ao_opts->compresstype);
			}
			else
			{
				arg = AO_DEFAULT_COMPRESSTYPE;
			}
			el = makeDefElem("compresstype", (Node *) makeString(arg), -1);
			retList = lappend(retList, el);
		}
	}
	else if (foundCompressType && cmplevel == NULL)
	{
		if (foundCompressTypeNone)
		{
			/*
			 * User wants to disable compression by specifying
			 * compresstype=none.
			 */
			el = makeDefElem("compresslevel", (Node *) makeInteger(0), -1);
			retList = lappend(retList, el);
		}
		else
		{
			/*
			 * Valid compresstype specified.  Use default
			 * compresslevel if it's non-zero, otherwise use 1.
			 */
			el = makeDefElem("compresslevel",
							 (Node *) makeInteger(ao_opts->compresslevel > 0 ?
												  ao_opts->compresslevel : 1),
							 -1);
			retList = lappend(retList, el);
		}
	}
	if (foundBlockSize == false)
	{
		el = makeDefElem("blocksize", (Node *) makeInteger(ao_opts->blocksize), -1);
		retList = lappend(retList, el);
	}
	return retList;
}

/*
 * Make encoding (compresstype = ..., blocksize=...) based on
 * currently configured defaults.
 * For blocksize, it is impossible for the value to be unset
 * if an appendonly relation, hence the default is always ignored.
 */
static List *
default_column_encoding_clause(Relation rel)
{
	DefElem *e1, *e2, *e3;
	const StdRdOptions *ao_opts = currentAOStorageOptions();
	bool		appendonly;
	int32		blocksize = -1;
	int16		compresslevel = 0;
	char	   *compresstype = NULL;
	NameData	compresstype_nd;

	appendonly = rel && RelationIsAppendOptimized(rel);
	if (appendonly)
	{
		GetAppendOnlyEntryAttributes(RelationGetRelid(rel),
									 &blocksize,
									 NULL,
									 &compresslevel,
									 NULL,
									 &compresstype_nd);
		compresstype = NameStr(compresstype_nd);
	}

	compresstype = compresstype && compresstype[0] ? pstrdup(compresstype) : 
					(ao_opts->compresstype[0] ? pstrdup(ao_opts->compresstype) : "none");
	e1 = makeDefElem("compresstype", (Node *) makeString(pstrdup(compresstype)), -1);

	blocksize = appendonly ? blocksize : 
					(ao_opts->blocksize != 0 ? ao_opts->blocksize : AO_DEFAULT_BLOCKSIZE);
	e2 = makeDefElem("blocksize", (Node *) makeInteger(blocksize), -1);

	compresslevel = appendonly && compresslevel != 0 ? compresslevel :
					(ao_opts->compresslevel != 0 ? ao_opts->compresslevel : AO_DEFAULT_COMPRESSLEVEL);
	e3 = makeDefElem("compresslevel", (Node *) makeInteger(compresslevel), -1);

	return list_make3(e1, e2, e3);
}

/*
 * See if two encodings attempt to set the same parameters.
 */
static bool
encodings_overlap(List *a, List *b)
{
	ListCell *lca;
	foreach(lca, a)
	{
		ListCell *lcb;
		DefElem *ela = lfirst(lca);

		foreach(lcb, b)
		{
			DefElem *elb = lfirst(lcb);
			if (pg_strcasecmp(ela->defname, elb->defname) == 0)
				return true;
		}
	}
	return false;
}

/*
 * Validate the sanity of column reference storage clauses.
 *
 * 1. Ensure that we only refer to columns that exist.
 * 2. Ensure that each column is referenced either zero times or once.
 * 3. Ensure that the column reference storage clauses do not clash with the
 * 	  gp_default_storage_options
 */
static void
validateColumnStorageEncodingClauses(List *aocoColumnEncoding,
									 List *tableElts)
{
	ListCell *lc;
	struct HTAB *ht = NULL;
	struct colent {
		char colname[NAMEDATALEN];
		int count;
	} *ce = NULL;

	/* Generate a hash table for all the columns */
	foreach(lc, tableElts)
	{
		Node *n = lfirst(lc);

		if (IsA(n, ColumnDef))
		{
			ColumnDef *c = (ColumnDef *)n;
			char *colname;
			bool found = false;
			size_t n = NAMEDATALEN - 1 < strlen(c->colname) ?
							NAMEDATALEN - 1 : strlen(c->colname);

			colname = palloc0(NAMEDATALEN);
			MemSet(colname, 0, NAMEDATALEN);
			memcpy(colname, c->colname, n);
			colname[n] = '\0';

			if (!ht)
			{
				HASHCTL  cacheInfo;
				int      cacheFlags;

				memset(&cacheInfo, 0, sizeof(cacheInfo));
				cacheInfo.keysize = NAMEDATALEN;
				cacheInfo.entrysize = sizeof(*ce);
				cacheFlags = HASH_ELEM;

				ht = hash_create("column info cache",
								 list_length(tableElts),
								 &cacheInfo, cacheFlags);
			}

			ce = hash_search(ht, colname, HASH_ENTER, &found);

			/*
			 * The user specified a duplicate column name. We check duplicate
			 * column names VERY late (under MergeAttributes(), which is called
			 * by DefineRelation(). For the specific case here, it is safe to
			 * call out that this is a duplicate. We don't need to delay until
			 * we look at inheritance.
			 */
			if (found)
			{
				ereport(ERROR,
						(errcode(ERRCODE_DUPLICATE_COLUMN),
						 errmsg("column \"%s\" duplicated",
								colname)));
			}
			ce->count = 0;
		}
	}

	/*
	 * If the table has no columns -- usually in the partitioning case -- then
	 * we can short circuit.
	 */
	if (!ht)
		return;

	/*
	 * All column reference storage directives without the DEFAULT
	 * clause should refer to real columns.
	 */
	foreach(lc, aocoColumnEncoding)
	{
		ColumnReferenceStorageDirective *c = lfirst(lc);

		Assert(IsA(c, ColumnReferenceStorageDirective));

		if (c->deflt)
			continue;
		else
		{
			bool found = false;
			char colname[NAMEDATALEN];
			size_t collen = strlen(c->column);
			size_t n = NAMEDATALEN - 1 < collen ? NAMEDATALEN - 1 : collen;
			MemSet(colname, 0, NAMEDATALEN);
			memcpy(colname, c->column, n);
			colname[n] = '\0';

			ce = hash_search(ht, colname, HASH_FIND, &found);

			if (!found)
				ereport(ERROR,
						(errcode(ERRCODE_WRONG_OBJECT_TYPE),
						 errmsg("column \"%s\" does not exist", colname)));

			ce->count++;

			if (ce->count > 1)
				ereport(ERROR,
						(errcode(ERRCODE_WRONG_OBJECT_TYPE),
						 errmsg("column \"%s\" referenced in more than one COLUMN ENCODING clause",
								colname)));
		}
	}

	hash_destroy(ht);

	foreach(lc, aocoColumnEncoding)
	{
		ColumnReferenceStorageDirective *crsd = lfirst(lc);

		Datum d = transformRelOptions(PointerGetDatum(NULL),
									  crsd->encoding,
									  NULL, NULL,
									  true, false);
		StdRdOptions *stdRdOptions = (StdRdOptions *)default_reloptions(d,
																	true,
																	RELOPT_KIND_APPENDOPTIMIZED);

		validateAppendOnlyRelOptions(stdRdOptions->blocksize,
									 gp_safefswritesize,
									 stdRdOptions->compresslevel,
									 stdRdOptions->compresstype,
									 stdRdOptions->checksum,
									 true);
	}
}

/*
 * Make a default column storage directive from a WITH clause
 * Ignore options in the WITH clause that don't appear in
 * storage_directives for column-level compression.
 */
List *
form_default_storage_directive(List *enc)
{
	List *out = NIL;
	ListCell *lc;

	foreach(lc, enc)
	{
		DefElem *el = lfirst(lc);

		if (!el->defname)
			out = lappend(out, copyObject(el));

		if (pg_strcasecmp("oids", el->defname) == 0)
			continue;
		if (pg_strcasecmp("fillfactor", el->defname) == 0)
			continue;
		if (pg_strcasecmp("tablename", el->defname) == 0)
			continue;
		/* checksum is not a column specific attribute. */
		if (pg_strcasecmp("checksum", el->defname) == 0)
			continue;
		out = lappend(out, copyObject(el));
	}
	return out;
}

/*
 * Transform and validate the actual encoding clauses.
 *
 * We need tell the underlying system that these are AO/CO tables too,
 * hence the concatenation of the extra elements.
 *
 * If 'validate' is true, we validate that the optionsa are valid WITH options
 * for an AO table. Otherwise, any unrecognized options are passed through as
 * is.
 */
List *
transformStorageEncodingClause(List *aocoColumnEncoding, bool validate)
{
	ListCell   *lc;
	DefElem	   *dl;

	foreach(lc, aocoColumnEncoding)
	{
		dl = (DefElem *) lfirst(lc);
		if (pg_strncasecmp(dl->defname, SOPT_CHECKSUM, strlen(SOPT_CHECKSUM)) == 0)
		{
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("\"%s\" is not a column specific option",
							SOPT_CHECKSUM)));
		}
	}

	/* add defaults for missing values */
	aocoColumnEncoding = fillin_encoding(aocoColumnEncoding);

	/*
	 * The following two statements validate that the encoding clause is well
	 * formed.
	 */
	if (validate)
	{
		Datum		d;

		d = transformRelOptions(PointerGetDatum(NULL),
								aocoColumnEncoding,
								NULL, NULL,
								true, false);
		(void) default_reloptions(d, true, RELOPT_KIND_APPENDOPTIMIZED);
	}

	return aocoColumnEncoding;
}

/*
 * Find the column reference storage encoding clause for `column'.
 *
 * This is called by transformColumnEncoding() in a loop but stenc should be
 * quite small in practice.
 */
static ColumnReferenceStorageDirective *
find_crsd(const char *column, List *stenc)
{
	ListCell *lc;

	foreach(lc, stenc)
	{
		ColumnReferenceStorageDirective *c = lfirst(lc);

		if (c->deflt == false && strcmp(column, c->column) == 0)
			return c;
	}
	return NULL;
}

/*
 * Parse and validate COLUMN <col> ENCODING ... directives.
 *
 * The 'colDefs', 'stenc' and 'taboptions' arguments are parts of the
 * CREATE TABLE or ALTER TABLE command:
 *
 * 'colDefs' - list of ColumnDefs
 * 'stenc' - list of ColumnReferenceStorageDirectives
 * 'withOptions' - list of WITH options
 *
 * ENCODING options can be attached to column definitions, like
 * "mycolumn integer ENCODING ..."; these go into ColumnDefs. They
 * can also be specified with the "COLUMN mycolumn ENCODING ..." syntax;
 * they go into the ColumnReferenceStorageDirectives. And table-wide
 * defaults can be given in the WITH clause.
 *
 * Normally if any ENCODING clause was given for a non-AO/CO table,
 * we should report an error. However, exception exists in DefineRelation()
 * where we allow that to happen, so we pass in errorOnEncodingClause to
 * indicate whether we should report this error. 
 *
 * This function is called for RELKIND_PARTITIONED_TABLE as well even if we
 * don't store entries in pg_attribute_encoding for rootpartition. The reason
 * is to transformColumnEncoding for parent as need to use them later while
 * creating partitions in GPDB legacy partitioning syntax. Hence, if
 * rootpartition add to the list, only encoding elements specified in command,
 * defaults based on GUCs and such are skipped. Each child partition would
 * independently later run through this logic and that time add those GUC
 * specific defaults if required. Reason to avoid adding defaults for
 * rootpartition is need to first merge partition level user specified options
 * and then need to add defaults only for remaining columns.
 *
 * NOTE: This is *not* performed during the parse analysis phase, like
 * most transformation, but only later in DefineRelation() or ATExecAddColumn(). 
 * This needs access to possible inherited columns, so it can only be done after
 * expanding them.
 */
List* transformColumnEncoding(Relation rel, List *colDefs, List *stenc, List *withOptions, bool rootpartition, bool errorOnEncodingClause)
{
	ColumnReferenceStorageDirective *deflt = NULL;
	ListCell   *lc;
	List	   *result = NIL;

	if (stenc)
		validateColumnStorageEncodingClauses(stenc, colDefs);

	/* get the default clause, if there is one. */
	foreach(lc, stenc)
	{
		ColumnReferenceStorageDirective *c = lfirst(lc);
		Assert(IsA(c, ColumnReferenceStorageDirective));

		if (errorOnEncodingClause)
			ereport(ERROR,
					(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
					errmsg("ENCODING clause only supported with column oriented tables")));
		if (c->deflt)
		{
			/*
			 * Some quick validation: there should only be one default
			 * clause
			 */
			if (deflt)
				elog(ERROR, "only one default column encoding may be specified");

			deflt = copyObject(c);
			deflt->encoding = transformStorageEncodingClause(deflt->encoding, true);

			/*
			 * The default encoding and the with clause better not
			 * try and set the same options!
			 */
			if (encodings_overlap(withOptions, deflt->encoding))
				ereport(ERROR,
						(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
						 errmsg("DEFAULT COLUMN ENCODING clause cannot override values set in WITH clause")));
		}

	}

	/*
	 * If no default has been specified, we might create one out of the
	 * WITH clause.
	 */
	if (!deflt)
	{
		List	   *tmpenc;
		if ((tmpenc = form_default_storage_directive(withOptions)) != NULL)
		{
			deflt = makeNode(ColumnReferenceStorageDirective);
			deflt->deflt = true;
			deflt->encoding = transformStorageEncodingClause(tmpenc, false);
		}
	}

	foreach(lc, colDefs)
	{
		Node	   *elem = (Node *) lfirst(lc);
		ColumnDef *d;
		ColumnReferenceStorageDirective *c;
		List *encoding = NIL;

		Assert(IsA(elem, ColumnDef));

		d = (ColumnDef *) elem;

		/*
		 * Find a storage encoding for this column, in this order:
		 *
		 * 1. An explicit encoding clause in the ColumnDef
		 * 2. A column reference storage directive for this column
		 * 3. A default column encoding in the statement
		 * 4. A default for the type.
		 */
		if (d->encoding)
		{
			encoding = transformStorageEncodingClause(d->encoding, true);
			if (errorOnEncodingClause)
				ereport(ERROR,
						(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
						errmsg("ENCODING clause only supported with column oriented tables")));
		}
		else
		{
			ColumnReferenceStorageDirective *s = find_crsd(d->colname, stenc);

			if (s)
				encoding = transformStorageEncodingClause(s->encoding, true);
			else
			{
				if (deflt)
					encoding = copyObject(deflt->encoding);
				else if (!rootpartition)
				{
					if (d->typeName)
						encoding = get_type_encoding(d->typeName);
					if (!encoding)
						encoding = default_column_encoding_clause(rel);
				}
			}
		}

		if (encoding)
		{
			c = makeNode(ColumnReferenceStorageDirective);
			c->column = pstrdup(d->colname);
			c->encoding = encoding;

			result = lappend(result, c);
		}
	}

	return result;
}

/*
 * GPDB: Convenience function to judge a relation option whether already in opts
 */
bool
reloptions_has_opt(List *opts, const char *name)
{
	ListCell *lc;
	foreach(lc, opts)
	{
		DefElem *de = lfirst(lc);
		if (pg_strcasecmp(de->defname, name) == 0)
			return true;
	}
	return false;
}

/*
 * GPDB: Convenience function to build storage reloptions for a given relation, just for AO table.
 */
List *
build_ao_rel_storage_opts(List *opts, Relation rel)
{
	bool		checksum = true;
	int32		blocksize = -1;
	int16		compresslevel = 0;
	char	   *compresstype = NULL;
	NameData	compresstype_nd;

	GetAppendOnlyEntryAttributes(RelationGetRelid(rel),
								 &blocksize,
								 NULL,
								 &compresslevel,
								 &checksum,
								 &compresstype_nd);
	compresstype = NameStr(compresstype_nd);

	if (!reloptions_has_opt(opts, "blocksize"))
		opts = lappend(opts, makeDefElem("blocksize", (Node *) makeInteger(blocksize), -1));

	if (!reloptions_has_opt(opts, "compresslevel"))
		opts = lappend(opts, makeDefElem("compresslevel", (Node *) makeInteger(compresslevel), -1));

	if (!reloptions_has_opt(opts, "checksum"))
		opts = lappend(opts, makeDefElem("checksum", (Node *) makeInteger(checksum), -1));

	if (!reloptions_has_opt(opts, "compresstype"))
	{
		compresstype = (compresstype && compresstype[0]) ? pstrdup(compresstype) : "none";
		opts = lappend(opts, makeDefElem("compresstype", (Node *) makeString(compresstype), -1));
	}

	return opts;
}

相关信息

greenplumn 源码目录

相关文章

greenplumn bufmask 源码

greenplumn heaptuple 源码

greenplumn indextuple 源码

greenplumn memtuple 源码

greenplumn printsimple 源码

greenplumn printtup 源码

greenplumn relation 源码

greenplumn reloptions 源码

greenplumn scankey 源码

greenplumn session 源码

0  赞