postgresql/src/backend/commands/copy.c

1212 lines
27 KiB
C
Raw Normal View History

/*-------------------------------------------------------------------------
*
* copy.c
*
* Portions Copyright (c) 1996-2000, PostgreSQL, Inc
* Portions Copyright (c) 1994, Regents of the University of California
*
*
* IDENTIFICATION
* $Header: /cvsroot/pgsql/src/backend/commands/copy.c,v 1.119 2000/07/14 22:17:42 tgl Exp $
*
*-------------------------------------------------------------------------
*/
1996-11-06 09:21:43 +01:00
#include <unistd.h>
1999-07-16 07:00:38 +02:00
#include <sys/stat.h>
1996-11-06 09:21:43 +01:00
#include "postgres.h"
#include "access/genam.h"
1999-07-16 07:00:38 +02:00
#include "access/heapam.h"
#include "catalog/catname.h"
1999-07-16 07:00:38 +02:00
#include "catalog/index.h"
2000-06-15 05:33:12 +02:00
#include "catalog/pg_index.h"
#include "catalog/pg_shadow.h"
1999-07-16 07:00:38 +02:00
#include "catalog/pg_type.h"
#include "commands/copy.h"
#include "commands/trigger.h"
1999-07-16 07:00:38 +02:00
#include "executor/executor.h"
#include "libpq/libpq.h"
1999-07-16 07:00:38 +02:00
#include "miscadmin.h"
1999-09-27 22:00:44 +02:00
#include "tcop/tcopprot.h"
1999-07-16 07:00:38 +02:00
#include "utils/acl.h"
#include "utils/builtins.h"
#include "utils/relcache.h"
1999-07-16 07:00:38 +02:00
#include "utils/syscache.h"
#ifdef MULTIBYTE
1999-07-16 07:00:38 +02:00
#include "mb/pg_wchar.h"
#endif
#define ISOCTAL(c) (((c) >= '0') && ((c) <= '7'))
#define VALUE(c) ((c) - '0')
/* non-export function prototypes */
static void CopyTo(Relation rel, bool binary, bool oids, FILE *fp, char *delim, char *null_print);
static void CopyFrom(Relation rel, bool binary, bool oids, FILE *fp, char *delim, char *null_print);
static Oid GetOutputFunction(Oid type);
static Oid GetTypeElement(Oid type);
static Oid GetInputFunction(Oid type);
static Oid IsTypeByVal(Oid type);
static void CopyReadNewline(FILE *fp, int *newline);
static char *CopyReadAttribute(FILE *fp, bool *isnull, char *delim, int *newline, char *null_print);
static void CopyAttributeOut(FILE *fp, char *string, char *delim);
static int CountTuples(Relation relation);
/*
* Static communication variables ... pretty grotty, but COPY has
* never been reentrant...
*/
int lineno = 0; /* used by elog() -- dz */
static bool fe_eof;
/*
* These static variables are used to avoid incurring overhead for each
* attribute processed. attribute_buf is reused on each CopyReadAttribute
* call to hold the string being read in. Under normal use it will soon
* grow to a suitable size, and then we will avoid palloc/pfree overhead
* for subsequent attributes. Note that CopyReadAttribute returns a pointer
* to attribute_buf's data buffer!
* encoding, if needed, can be set once at the start of the copy operation.
*/
static StringInfoData attribute_buf;
#ifdef MULTIBYTE
static int encoding;
#endif
1999-05-25 18:15:34 +02:00
/*
* Internal communications functions
*/
static void CopySendData(void *databuf, int datasize, FILE *fp);
static void CopySendString(char *str, FILE *fp);
static void CopySendChar(char c, FILE *fp);
static void CopyGetData(void *databuf, int datasize, FILE *fp);
static int CopyGetChar(FILE *fp);
static int CopyGetEof(FILE *fp);
static int CopyPeekChar(FILE *fp);
static void CopyDonePeek(FILE *fp, int c, int pickup);
/*
* CopySendData sends output data either to the file
1999-05-25 18:15:34 +02:00
* specified by fp or, if fp is NULL, using the standard
* backend->frontend functions
*
* CopySendString does the same for null-terminated strings
* CopySendChar does the same for single characters
*
* NB: no data conversion is applied by these functions
*/
static void
1999-05-25 18:15:34 +02:00
CopySendData(void *databuf, int datasize, FILE *fp)
{
if (!fp)
{
if (pq_putbytes((char *) databuf, datasize))
fe_eof = true;
}
1999-05-25 18:15:34 +02:00
else
{
1999-05-25 18:15:34 +02:00
fwrite(databuf, datasize, 1, fp);
if (ferror(fp))
elog(ERROR, "CopySendData: %m");
}
}
1999-05-25 18:15:34 +02:00
static void
1999-05-25 18:15:34 +02:00
CopySendString(char *str, FILE *fp)
{
CopySendData(str, strlen(str), fp);
}
static void
1999-05-25 18:15:34 +02:00
CopySendChar(char c, FILE *fp)
{
CopySendData(&c, 1, fp);
}
/*
* CopyGetData reads output data either from the file
1999-05-25 18:15:34 +02:00
* specified by fp or, if fp is NULL, using the standard
* backend->frontend functions
*
* CopyGetChar does the same for single characters
* CopyGetEof checks if it's EOF on the input (or, check for EOF result
* from CopyGetChar)
*
* NB: no data conversion is applied by these functions
*/
static void
1999-05-25 18:15:34 +02:00
CopyGetData(void *databuf, int datasize, FILE *fp)
{
if (!fp)
{
if (pq_getbytes((char *) databuf, datasize))
fe_eof = true;
}
1999-05-25 18:15:34 +02:00
else
fread(databuf, datasize, 1, fp);
}
static int
1999-05-25 18:15:34 +02:00
CopyGetChar(FILE *fp)
{
if (!fp)
{
unsigned char ch;
if (pq_getbytes((char *) &ch, 1))
{
fe_eof = true;
1999-05-25 18:15:34 +02:00
return EOF;
}
1999-05-25 18:15:34 +02:00
return ch;
}
else
return getc(fp);
}
static int
1999-05-25 18:15:34 +02:00
CopyGetEof(FILE *fp)
{
if (!fp)
return fe_eof;
1999-05-25 18:15:34 +02:00
else
return feof(fp);
}
/*
* CopyPeekChar reads a byte in "peekable" mode.
* after each call to CopyPeekChar, a call to CopyDonePeek _must_
* follow, unless EOF was returned.
1999-05-25 18:15:34 +02:00
* CopyDonePeek will either take the peeked char off the steam
* (if pickup is != 0) or leave it on the stream (if pickup == 0)
*/
static int
1999-05-25 18:15:34 +02:00
CopyPeekChar(FILE *fp)
{
if (!fp)
{
int ch = pq_peekbyte();
if (ch == EOF)
fe_eof = true;
return ch;
}
1999-05-25 18:15:34 +02:00
else
return getc(fp);
}
static void
1999-05-25 18:15:34 +02:00
CopyDonePeek(FILE *fp, int c, int pickup)
{
if (!fp)
{
if (pickup)
{
/*
* We want to pick it up - just receive again into dummy
* buffer
*/
char c;
pq_getbytes(&c, 1);
}
/* If we didn't want to pick it up, just leave it where it sits */
}
else
{
if (!pickup)
{
/* We don't want to pick it up - so put it back in there */
ungetc(c, fp);
}
/* If we wanted to pick it up, it's already there */
}
}
1999-05-25 18:15:34 +02:00
/*
* DoCopy executes the SQL COPY statement.
*/
void
DoCopy(char *relname, bool binary, bool oids, bool from, bool pipe,
char *filename, char *delim, char *null_print)
{
/*----------------------------------------------------------------------------
Either unload or reload contents of class <relname>, depending on <from>.
If <pipe> is false, transfer is between the class and the file named
<filename>. Otherwise, transfer is between the class and our regular
input/output stream. The latter could be either stdin/stdout or a
socket, depending on whether we're running under Postmaster control.
Iff <binary>, unload or reload in the binary format, as opposed to the
more wasteful but more robust and portable text format.
If in the text format, delimit columns with delimiter <delim> and print
NULL values as <null_print>.
When loading in the text format from an input stream (as opposed to
a file), recognize a "." on a line by itself as EOF. Also recognize
a stream EOF. When unloading in the text format to an output stream,
write a "." on a line by itself at the end of the data.
Iff <oids>, unload or reload the format that includes OID information.
Do not allow a Postgres user without superuser privilege to read from
or write to a file.
Do not allow the copy if user doesn't have proper permission to access
the class.
----------------------------------------------------------------------------*/
FILE *fp;
Relation rel;
extern char *UserName; /* defined in global.c */
const AclMode required_access = from ? ACL_WR : ACL_RD;
int result;
/*
* Open and lock the relation, using the appropriate lock type.
*/
rel = heap_openr(relname, (from ? RowExclusiveLock : AccessShareLock));
result = pg_aclcheck(relname, UserName, required_access);
if (result != ACLCHECK_OK)
elog(ERROR, "%s: %s", relname, aclcheck_error_strings[result]);
if (!pipe && !superuser())
elog(ERROR, "You must have Postgres superuser privilege to do a COPY "
"directly to or from a file. Anyone can COPY to stdout or "
"from stdin. Psql's \\copy command also works for anyone.");
/*
* This restriction is unfortunate, but necessary until the frontend
* COPY protocol is redesigned to be binary-safe...
*/
if (pipe && binary)
elog(ERROR, "COPY BINARY is not supported to stdout or from stdin");
/*
* Set up variables to avoid per-attribute overhead.
*/
initStringInfo(&attribute_buf);
#ifdef MULTIBYTE
encoding = pg_get_client_encoding();
#endif
if (from)
{ /* copy from file to database */
if (rel->rd_rel->relkind == RELKIND_SEQUENCE)
2000-06-05 19:07:56 +02:00
elog(ERROR, "You cannot change sequence relation %s", relname);
if (pipe)
{
if (IsUnderPostmaster)
{
ReceiveCopyBegin();
fp = NULL;
}
else
fp = stdin;
}
else
{
fp = AllocateFile(filename, PG_BINARY_R);
if (fp == NULL)
elog(ERROR, "COPY command, running in backend with "
"effective uid %d, could not open file '%s' for "
"reading. Errno = %s (%d).",
(int) geteuid(), filename, strerror(errno), errno);
}
CopyFrom(rel, binary, oids, fp, delim, null_print);
}
else
{ /* copy from database to file */
if (pipe)
{
if (IsUnderPostmaster)
{
SendCopyBegin();
pq_startcopyout();
fp = NULL;
}
else
fp = stdout;
}
else
{
mode_t oumask; /* Pre-existing umask value */
if (*filename != '/')
elog(ERROR, "Relative path not allowed for server side"
" COPY command.");
oumask = umask((mode_t) 022);
fp = AllocateFile(filename, PG_BINARY_W);
umask(oumask);
if (fp == NULL)
elog(ERROR, "COPY command, running in backend with "
"effective uid %d, could not open file '%s' for "
"writing. Errno = %s (%d).",
(int) geteuid(), filename, strerror(errno), errno);
}
CopyTo(rel, binary, oids, fp, delim, null_print);
}
if (!pipe)
FreeFile(fp);
else if (!from)
{
if (!binary)
CopySendData("\\.\n", 3, fp);
if (IsUnderPostmaster)
pq_endcopyout(false);
}
pfree(attribute_buf.data);
/*
* Close the relation. If reading, we can release the AccessShareLock
* we got; if writing, we should hold the lock until end of
* transaction to ensure that updates will be committed before lock is
* released.
*/
heap_close(rel, (from ? NoLock : AccessShareLock));
}
static void
CopyTo(Relation rel, bool binary, bool oids, FILE *fp, char *delim, char *null_print)
{
HeapTuple tuple;
HeapScanDesc scandesc;
int32 attr_count,
i;
#ifdef _DROP_COLUMN_HACK__
bool *valid;
#endif /* _DROP_COLUMN_HACK__ */
1998-09-01 05:29:17 +02:00
Form_pg_attribute *attr;
FmgrInfo *out_functions;
Oid out_func_oid;
Oid *elements;
int32 *typmod;
Datum value;
bool isnull; /* The attribute we are copying is null */
char *nulls;
/*
* <nulls> is a (dynamically allocated) array with one character per
* attribute in the instance being copied. nulls[I-1] is 'n' if
* Attribute Number I is null, and ' ' otherwise.
*
* <nulls> is meaningful only if we are doing a binary copy.
*/
char *string;
int32 ntuples;
TupleDesc tupDesc;
scandesc = heap_beginscan(rel, 0, QuerySnapshot, 0, NULL);
attr_count = rel->rd_att->natts;
attr = rel->rd_att->attrs;
tupDesc = rel->rd_att;
if (!binary)
{
out_functions = (FmgrInfo *) palloc(attr_count * sizeof(FmgrInfo));
elements = (Oid *) palloc(attr_count * sizeof(Oid));
1998-07-15 20:53:40 +02:00
typmod = (int32 *) palloc(attr_count * sizeof(int32));
#ifdef _DROP_COLUMN_HACK__
valid = (bool *) palloc(attr_count * sizeof(bool));
#endif /* _DROP_COLUMN_HACK__ */
for (i = 0; i < attr_count; i++)
{
#ifdef _DROP_COLUMN_HACK__
if (COLUMN_IS_DROPPED(attr[i]))
{
valid[i] = false;
continue;
}
else
valid[i] = true;
#endif /* _DROP_COLUMN_HACK__ */
out_func_oid = (Oid) GetOutputFunction(attr[i]->atttypid);
fmgr_info(out_func_oid, &out_functions[i]);
elements[i] = GetTypeElement(attr[i]->atttypid);
1998-02-10 17:04:38 +01:00
typmod[i] = attr[i]->atttypmod;
}
nulls = NULL; /* meaningless, but compiler doesn't know
* that */
}
else
{
elements = NULL;
1998-02-10 17:04:38 +01:00
typmod = NULL;
out_functions = NULL;
nulls = (char *) palloc(attr_count);
for (i = 0; i < attr_count; i++)
nulls[i] = ' ';
/* XXX expensive */
ntuples = CountTuples(rel);
CopySendData(&ntuples, sizeof(int32), fp);
}
while (HeapTupleIsValid(tuple = heap_getnext(scandesc, 0)))
{
1999-09-27 22:00:44 +02:00
if (QueryCancel)
CancelQuery();
if (oids && !binary)
{
string = DatumGetCString(DirectFunctionCall1(oidout,
ObjectIdGetDatum(tuple->t_data->t_oid)));
CopySendString(string, fp);
1999-05-25 18:15:34 +02:00
CopySendChar(delim[0], fp);
pfree(string);
}
for (i = 0; i < attr_count; i++)
{
value = heap_getattr(tuple, i + 1, tupDesc, &isnull);
if (!binary)
{
#ifdef _DROP_COLUMN_HACK__
if (!valid[i])
{
if (i == attr_count - 1)
CopySendChar('\n', fp);
continue;
}
#endif /* _DROP_COLUMN_HACK__ */
if (!isnull)
{
string = DatumGetCString(FunctionCall3(&out_functions[i],
value,
ObjectIdGetDatum(elements[i]),
Int32GetDatum(typmod[i])));
CopyAttributeOut(fp, string, delim);
pfree(string);
}
else
CopySendString(null_print, fp); /* null indicator */
if (i == attr_count - 1)
CopySendChar('\n', fp);
else
{
/*
* when copying out, only use the first char of the
* delim string
*/
CopySendChar(delim[0], fp);
}
}
else
{
/*
* only interesting thing heap_getattr tells us in this
* case is if we have a null attribute or not.
*/
if (isnull)
nulls[i] = 'n';
}
}
if (binary)
{
int32 null_ct = 0,
length;
for (i = 0; i < attr_count; i++)
{
if (nulls[i] == 'n')
null_ct++;
}
1998-11-27 20:52:36 +01:00
length = tuple->t_len - tuple->t_data->t_hoff;
CopySendData(&length, sizeof(int32), fp);
if (oids)
CopySendData((char *) &tuple->t_data->t_oid, sizeof(int32), fp);
CopySendData(&null_ct, sizeof(int32), fp);
if (null_ct > 0)
{
for (i = 0; i < attr_count; i++)
{
if (nulls[i] == 'n')
{
CopySendData(&i, sizeof(int32), fp);
nulls[i] = ' ';
}
}
}
1999-05-25 18:15:34 +02:00
CopySendData((char *) tuple->t_data + tuple->t_data->t_hoff,
length, fp);
}
}
heap_endscan(scandesc);
if (binary)
pfree(nulls);
else
{
pfree(out_functions);
pfree(elements);
1998-02-10 17:04:38 +01:00
pfree(typmod);
}
}
static void
CopyFrom(Relation rel, bool binary, bool oids, FILE *fp,
char *delim, char *null_print)
{
HeapTuple tuple;
TupleDesc tupDesc;
1998-09-01 05:29:17 +02:00
Form_pg_attribute *attr;
AttrNumber attr_count;
FmgrInfo *in_functions;
Oid *elements;
int32 *typmod;
int i;
Oid in_func_oid;
Datum *values;
char *nulls;
bool *byval;
bool isnull;
int done = 0;
char *string = NULL,
*ptr;
int32 len,
null_ct,
null_id;
int32 ntuples,
tuples_read = 0;
bool reading_to_eof = true;
RelationInfo *relationInfo;
1999-05-25 18:15:34 +02:00
EState *estate = makeNode(EState); /* for ExecConstraints() */
TupleTable tupleTable;
TupleTableSlot *slot;
Oid loaded_oid = InvalidOid;
bool skip_tuple = false;
1998-09-01 05:29:17 +02:00
tupDesc = RelationGetDescr(rel);
attr = tupDesc->attrs;
attr_count = tupDesc->natts;
/*
* We need a RelationInfo so we can use the regular executor's
* index-entry-making machinery. (There used to be a huge amount
* of code here that basically duplicated execUtils.c ...)
*/
relationInfo = makeNode(RelationInfo);
relationInfo->ri_RangeTableIndex = 1; /* dummy */
relationInfo->ri_RelationDesc = rel;
relationInfo->ri_NumIndices = 0;
relationInfo->ri_IndexRelationDescs = NULL;
relationInfo->ri_IndexRelationInfo = NULL;
ExecOpenIndices(relationInfo);
estate->es_result_relation_info = relationInfo;
/* Set up a dummy tuple table too */
tupleTable = ExecCreateTupleTable(1);
slot = ExecAllocTableSlot(tupleTable);
ExecSetSlotDescriptor(slot, tupDesc);
if (!binary)
{
in_functions = (FmgrInfo *) palloc(attr_count * sizeof(FmgrInfo));
elements = (Oid *) palloc(attr_count * sizeof(Oid));
1998-07-15 20:53:40 +02:00
typmod = (int32 *) palloc(attr_count * sizeof(int32));
for (i = 0; i < attr_count; i++)
{
#ifdef _DROP_COLUMN_HACK__
if (COLUMN_IS_DROPPED(attr[i]))
continue;
#endif /* _DROP_COLUMN_HACK__ */
in_func_oid = (Oid) GetInputFunction(attr[i]->atttypid);
fmgr_info(in_func_oid, &in_functions[i]);
elements[i] = GetTypeElement(attr[i]->atttypid);
1998-02-10 17:04:38 +01:00
typmod[i] = attr[i]->atttypmod;
}
}
else
{
in_functions = NULL;
elements = NULL;
1998-02-10 17:04:38 +01:00
typmod = NULL;
CopyGetData(&ntuples, sizeof(int32), fp);
if (ntuples != 0)
reading_to_eof = false;
}
values = (Datum *) palloc(attr_count * sizeof(Datum));
nulls = (char *) palloc(attr_count * sizeof(char));
byval = (bool *) palloc(attr_count * sizeof(bool));
for (i = 0; i < attr_count; i++)
{
nulls[i] = ' ';
#ifdef _DROP_COLUMN_HACK__
if (COLUMN_IS_DROPPED(attr[i]))
{
byval[i] = 'n';
continue;
}
#endif /* _DROP_COLUMN_HACK__ */
byval[i] = (bool) IsTypeByVal(attr[i]->atttypid);
}
lineno = 0;
fe_eof = false;
while (!done)
{
if (QueryCancel)
{
contrib-array.patch this is an old patch which I have already submitted and never seen in the sources. It corrects the datatype oids used in some iterator functions. This bug has been reported to me by many other people. contrib-datetime.patch some code contributed by Reiner Dassing <dassing@wettzell.ifag.de> contrib-makefiles.patch fixes all my contrib makefiles which don't work with some compilers, as reported to me by another user. contrib-miscutil.patch an old patch for one of my old contribs. contrib-string.patch a small change to the c-like text output functions. Now the '{' is escaped only at the beginning of the string to distinguish it from arrays, and the '}' is no more escaped. elog-lineno.patch adds the current lineno of CopyFrom to elog messages. This is very useful when you load a 1 million tuples table from an external file and there is a bad value somehere. Currently you get an error message but you can't know where is the bad data. The patch uses a variable which was declared static in copy.c. The variable is now exported and initialized to 0. It is always cleared at the end of the copy or at the first elog message or when the copy is canceled. I know this is very ugly but I can't find any better way of knowing where the copy fails and I have this problem quite often. plperl-makefile.patch fixes a typo in a makefile, but the error must be elsewhere because it is a file generated automatically. Please have a look. tprintf-timestamp.patch restores the original 2-digit year format, assuming that the two century digits don't carry much information and that '000202' is easier to read than 20000202. Being only a log file it shouldn't break anything. Please apply the patches before the next scheduled code freeze. I also noticed that some of the contribs don't compile correcly. Should we ask people to fix their code or rename their makefiles so that they are ignored by the top makefile? -- Massimo Dal Zotto
2000-02-13 19:59:53 +01:00
lineno = 0;
1999-09-27 22:00:44 +02:00
CancelQuery();
contrib-array.patch this is an old patch which I have already submitted and never seen in the sources. It corrects the datatype oids used in some iterator functions. This bug has been reported to me by many other people. contrib-datetime.patch some code contributed by Reiner Dassing <dassing@wettzell.ifag.de> contrib-makefiles.patch fixes all my contrib makefiles which don't work with some compilers, as reported to me by another user. contrib-miscutil.patch an old patch for one of my old contribs. contrib-string.patch a small change to the c-like text output functions. Now the '{' is escaped only at the beginning of the string to distinguish it from arrays, and the '}' is no more escaped. elog-lineno.patch adds the current lineno of CopyFrom to elog messages. This is very useful when you load a 1 million tuples table from an external file and there is a bad value somehere. Currently you get an error message but you can't know where is the bad data. The patch uses a variable which was declared static in copy.c. The variable is now exported and initialized to 0. It is always cleared at the end of the copy or at the first elog message or when the copy is canceled. I know this is very ugly but I can't find any better way of knowing where the copy fails and I have this problem quite often. plperl-makefile.patch fixes a typo in a makefile, but the error must be elsewhere because it is a file generated automatically. Please have a look. tprintf-timestamp.patch restores the original 2-digit year format, assuming that the two century digits don't carry much information and that '000202' is easier to read than 20000202. Being only a log file it shouldn't break anything. Please apply the patches before the next scheduled code freeze. I also noticed that some of the contribs don't compile correcly. Should we ask people to fix their code or rename their makefiles so that they are ignored by the top makefile? -- Massimo Dal Zotto
2000-02-13 19:59:53 +01:00
}
1999-09-27 22:00:44 +02:00
lineno++;
if (!binary)
{
int newline = 0;
if (oids)
{
string = CopyReadAttribute(fp, &isnull, delim, &newline, null_print);
if (string == NULL)
done = 1;
else
{
loaded_oid = DatumGetObjectId(DirectFunctionCall1(oidin,
CStringGetDatum(string)));
if (loaded_oid == InvalidOid)
elog(ERROR, "COPY TEXT: Invalid Oid");
}
}
for (i = 0; i < attr_count && !done; i++)
{
#ifdef _DROP_COLUMN_HACK__
if (COLUMN_IS_DROPPED(attr[i]))
{
values[i] = PointerGetDatum(NULL);
nulls[i] = 'n';
continue;
}
#endif /* _DROP_COLUMN_HACK__ */
string = CopyReadAttribute(fp, &isnull, delim, &newline, null_print);
if (isnull)
{
values[i] = PointerGetDatum(NULL);
nulls[i] = 'n';
}
else if (string == NULL)
done = 1;
else
{
values[i] = FunctionCall3(&in_functions[i],
CStringGetDatum(string),
ObjectIdGetDatum(elements[i]),
Int32GetDatum(typmod[i]));
}
}
if (!done)
CopyReadNewline(fp, &newline);
}
else
{ /* binary */
CopyGetData(&len, sizeof(int32), fp);
if (CopyGetEof(fp))
done = 1;
else
{
if (oids)
{
CopyGetData(&loaded_oid, sizeof(int32), fp);
if (loaded_oid == InvalidOid)
elog(ERROR, "COPY BINARY: Invalid Oid");
}
CopyGetData(&null_ct, sizeof(int32), fp);
if (null_ct > 0)
{
for (i = 0; i < null_ct; i++)
{
CopyGetData(&null_id, sizeof(int32), fp);
nulls[null_id] = 'n';
}
}
string = (char *) palloc(len);
CopyGetData(string, len, fp);
ptr = string;
for (i = 0; i < attr_count; i++)
{
if (byval[i] && nulls[i] != 'n')
{
switch (attr[i]->attlen)
{
case sizeof(char):
values[i] = (Datum) *(unsigned char *) ptr;
ptr += sizeof(char);
break;
case sizeof(short):
ptr = (char *) SHORTALIGN(ptr);
values[i] = (Datum) *(unsigned short *) ptr;
ptr += sizeof(short);
break;
case sizeof(int32):
ptr = (char *) INTALIGN(ptr);
values[i] = (Datum) *(uint32 *) ptr;
ptr += sizeof(int32);
break;
default:
elog(ERROR, "COPY BINARY: impossible size");
break;
}
}
else if (nulls[i] != 'n')
{
1999-05-25 18:15:34 +02:00
ptr = (char *) att_align(ptr, attr[i]->attlen, attr[i]->attalign);
values[i] = (Datum) ptr;
ptr = att_addlength(ptr, attr[i]->attlen, ptr);
}
}
}
}
if (done)
continue;
tuple = heap_formtuple(tupDesc, values, nulls);
if (oids)
1998-11-27 20:52:36 +01:00
tuple->t_data->t_oid = loaded_oid;
skip_tuple = false;
/* BEFORE ROW INSERT Triggers */
if (rel->trigdesc &&
rel->trigdesc->n_before_row[TRIGGER_EVENT_INSERT] > 0)
{
HeapTuple newtuple;
newtuple = ExecBRInsertTriggers(rel, tuple);
if (newtuple == NULL) /* "do nothing" */
skip_tuple = true;
else if (newtuple != tuple) /* modified by Trigger(s) */
{
heap_freetuple(tuple);
tuple = newtuple;
}
}
if (!skip_tuple)
{
/* ----------------
* Check the constraints of the tuple
* ----------------
*/
if (rel->rd_att->constr)
ExecConstraints("CopyFrom", rel, tuple, estate);
/* ----------------
* OK, store the tuple and create index entries for it
* ----------------
*/
heap_insert(rel, tuple);
if (relationInfo->ri_NumIndices > 0)
{
ExecStoreTuple(tuple, slot, InvalidBuffer, false);
ExecInsertIndexTuples(slot, &(tuple->t_self), estate, false);
}
/* AFTER ROW INSERT Triggers */
if (rel->trigdesc &&
rel->trigdesc->n_after_row[TRIGGER_EVENT_INSERT] > 0)
ExecARInsertTriggers(rel, tuple);
}
if (binary)
pfree(string);
for (i = 0; i < attr_count; i++)
{
if (!byval[i] && nulls[i] != 'n')
{
if (!binary)
pfree((void *) values[i]);
}
/* reset nulls[] array for next time */
nulls[i] = ' ';
}
heap_freetuple(tuple);
tuples_read++;
if (!reading_to_eof && ntuples == tuples_read)
done = true;
}
/*
* Done, clean up
*/
contrib-array.patch this is an old patch which I have already submitted and never seen in the sources. It corrects the datatype oids used in some iterator functions. This bug has been reported to me by many other people. contrib-datetime.patch some code contributed by Reiner Dassing <dassing@wettzell.ifag.de> contrib-makefiles.patch fixes all my contrib makefiles which don't work with some compilers, as reported to me by another user. contrib-miscutil.patch an old patch for one of my old contribs. contrib-string.patch a small change to the c-like text output functions. Now the '{' is escaped only at the beginning of the string to distinguish it from arrays, and the '}' is no more escaped. elog-lineno.patch adds the current lineno of CopyFrom to elog messages. This is very useful when you load a 1 million tuples table from an external file and there is a bad value somehere. Currently you get an error message but you can't know where is the bad data. The patch uses a variable which was declared static in copy.c. The variable is now exported and initialized to 0. It is always cleared at the end of the copy or at the first elog message or when the copy is canceled. I know this is very ugly but I can't find any better way of knowing where the copy fails and I have this problem quite often. plperl-makefile.patch fixes a typo in a makefile, but the error must be elsewhere because it is a file generated automatically. Please have a look. tprintf-timestamp.patch restores the original 2-digit year format, assuming that the two century digits don't carry much information and that '000202' is easier to read than 20000202. Being only a log file it shouldn't break anything. Please apply the patches before the next scheduled code freeze. I also noticed that some of the contribs don't compile correcly. Should we ask people to fix their code or rename their makefiles so that they are ignored by the top makefile? -- Massimo Dal Zotto
2000-02-13 19:59:53 +01:00
lineno = 0;
pfree(values);
1998-02-10 17:04:38 +01:00
pfree(nulls);
pfree(byval);
1999-05-25 18:15:34 +02:00
if (!binary)
1998-02-10 17:04:38 +01:00
{
pfree(in_functions);
1998-02-10 17:04:38 +01:00
pfree(elements);
pfree(typmod);
}
ExecDropTupleTable(tupleTable, true);
ExecCloseIndices(relationInfo);
}
static Oid
GetOutputFunction(Oid type)
{
HeapTuple typeTuple;
typeTuple = SearchSysCacheTuple(TYPEOID,
ObjectIdGetDatum(type),
0, 0, 0);
if (HeapTupleIsValid(typeTuple))
1998-09-01 05:29:17 +02:00
return (int) ((Form_pg_type) GETSTRUCT(typeTuple))->typoutput;
elog(ERROR, "GetOutputFunction: Cache lookup of type %u failed", type);
1998-09-01 05:29:17 +02:00
return InvalidOid;
}
static Oid
GetTypeElement(Oid type)
{
HeapTuple typeTuple;
typeTuple = SearchSysCacheTuple(TYPEOID,
ObjectIdGetDatum(type),
0, 0, 0);
if (HeapTupleIsValid(typeTuple))
1998-09-01 05:29:17 +02:00
return (int) ((Form_pg_type) GETSTRUCT(typeTuple))->typelem;
elog(ERROR, "GetOutputFunction: Cache lookup of type %u failed", type);
1998-09-01 05:29:17 +02:00
return InvalidOid;
}
static Oid
GetInputFunction(Oid type)
{
HeapTuple typeTuple;
typeTuple = SearchSysCacheTuple(TYPEOID,
ObjectIdGetDatum(type),
0, 0, 0);
if (HeapTupleIsValid(typeTuple))
1998-09-01 05:29:17 +02:00
return (int) ((Form_pg_type) GETSTRUCT(typeTuple))->typinput;
elog(ERROR, "GetInputFunction: Cache lookup of type %u failed", type);
1998-09-01 05:29:17 +02:00
return InvalidOid;
}
static Oid
IsTypeByVal(Oid type)
{
HeapTuple typeTuple;
typeTuple = SearchSysCacheTuple(TYPEOID,
ObjectIdGetDatum(type),
0, 0, 0);
if (HeapTupleIsValid(typeTuple))
1998-09-01 05:29:17 +02:00
return (int) ((Form_pg_type) GETSTRUCT(typeTuple))->typbyval;
elog(ERROR, "GetInputFunction: Cache lookup of type %u failed", type);
1998-09-01 05:29:17 +02:00
return InvalidOid;
}
/*
* Reads input from fp until an end of line is seen.
*/
static void
CopyReadNewline(FILE *fp, int *newline)
{
if (!*newline)
{
elog(NOTICE, "CopyReadNewline: extra fields ignored");
while (!CopyGetEof(fp) && (CopyGetChar(fp) != '\n'));
}
*newline = 0;
}
/*
* Read the value of a single attribute.
*
* Result is either a string, or NULL (if EOF or a null attribute).
* Note that the caller should not pfree the string!
*
* *isnull is set true if a null attribute, else false.
* delim is the string of acceptable delimiter characters(s).
* *newline remembers whether we've seen a newline ending this tuple.
* null_print says how NULL values are represented
*/
static char *
CopyReadAttribute(FILE *fp, bool *isnull, char *delim, int *newline, char *null_print)
{
int c;
#ifdef MULTIBYTE
int mblen;
unsigned char s[2];
char *cvt;
int j;
s[1] = 0;
#endif
/* reset attribute_buf to empty */
attribute_buf.len = 0;
attribute_buf.data[0] = '\0';
/* if last delimiter was a newline return a NULL attribute */
if (*newline)
{
*isnull = (bool) true;
1998-09-01 05:29:17 +02:00
return NULL;
}
*isnull = (bool) false; /* set default */
for (;;)
{
c = CopyGetChar(fp);
if (c == EOF)
goto endOfFile;
if (c == '\n')
{
*newline = 1;
break;
}
if (strchr(delim, c))
break;
if (c == '\\')
{
c = CopyGetChar(fp);
if (c == EOF)
goto endOfFile;
switch (c)
{
case '0':
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
{
int val;
val = VALUE(c);
c = CopyPeekChar(fp);
if (ISOCTAL(c))
{
val = (val << 3) + VALUE(c);
1999-05-25 18:15:34 +02:00
CopyDonePeek(fp, c, 1); /* Pick up the
* character! */
c = CopyPeekChar(fp);
1999-05-25 18:15:34 +02:00
if (ISOCTAL(c))
{
CopyDonePeek(fp, c, 1); /* pick up! */
val = (val << 3) + VALUE(c);
}
else
{
if (c == EOF)
goto endOfFile;
1999-05-25 18:15:34 +02:00
CopyDonePeek(fp, c, 0); /* Return to stream! */
}
}
else
{
if (c == EOF)
goto endOfFile;
1999-05-25 18:15:34 +02:00
CopyDonePeek(fp, c, 0); /* Return to stream! */
}
c = val & 0377;
}
break;
/*
* This is a special hack to parse `\N' as
* <backslash-N> rather then just 'N' to provide
* compatibility with the default NULL output. -- pe
*/
case 'N':
appendStringInfoCharMacro(&attribute_buf, '\\');
c = 'N';
break;
case 'b':
c = '\b';
break;
case 'f':
c = '\f';
break;
case 'n':
c = '\n';
break;
case 'r':
c = '\r';
break;
case 't':
c = '\t';
break;
case 'v':
c = '\v';
break;
case '.':
c = CopyGetChar(fp);
if (c != '\n')
elog(ERROR, "CopyReadAttribute: end of record marker corrupted");
goto endOfFile;
}
}
appendStringInfoCharMacro(&attribute_buf, c);
#ifdef MULTIBYTE
/* get additional bytes of the char, if any */
s[0] = c;
mblen = pg_encoding_mblen(encoding, s);
for (j = 1; j < mblen; j++)
{
c = CopyGetChar(fp);
if (c == EOF)
goto endOfFile;
appendStringInfoCharMacro(&attribute_buf, c);
}
#endif
}
#ifdef MULTIBYTE
cvt = (char *) pg_client_to_server((unsigned char *) attribute_buf.data,
attribute_buf.len);
if (cvt != attribute_buf.data)
{
/* transfer converted data back to attribute_buf */
attribute_buf.len = 0;
attribute_buf.data[0] = '\0';
appendBinaryStringInfo(&attribute_buf, cvt, strlen(cvt));
pfree(cvt);
}
#endif
if (strcmp(attribute_buf.data, null_print) == 0)
*isnull = true;
return attribute_buf.data;
endOfFile:
return NULL;
}
static void
CopyAttributeOut(FILE *fp, char *server_string, char *delim)
{
char *string;
1998-07-27 21:38:40 +02:00
char c;
#ifdef MULTIBYTE
char *string_start;
int mblen;
int i;
#endif
#ifdef MULTIBYTE
string = (char *) pg_server_to_client((unsigned char *) server_string,
strlen(server_string));
string_start = string;
#else
string = server_string;
#endif
#ifdef MULTIBYTE
for (; (mblen = pg_encoding_mblen(encoding, string)) &&
((c = *string) != '\0'); string += mblen)
#else
for (; (c = *string) != '\0'; string++)
#endif
{
if (c == delim[0] || c == '\n' || c == '\\')
CopySendChar('\\', fp);
#ifdef MULTIBYTE
for (i = 0; i < mblen; i++)
CopySendChar(*(string + i), fp);
#else
CopySendChar(c, fp);
#endif
}
#ifdef MULTIBYTE
if (string_start != server_string)
pfree(string_start); /* pfree pg_server_to_client result */
#endif
}
/*
* Returns the number of tuples in a relation. Unfortunately, currently
* must do a scan of the entire relation to determine this.
*
* relation is expected to be an open relation descriptor.
*/
static int
CountTuples(Relation relation)
{
HeapScanDesc scandesc;
HeapTuple tuple;
int i;
scandesc = heap_beginscan(relation, 0, QuerySnapshot, 0, NULL);
i = 0;
while (HeapTupleIsValid(tuple = heap_getnext(scandesc, 0)))
i++;
heap_endscan(scandesc);
1998-09-01 05:29:17 +02:00
return i;
}