g2index.F90 File Reference

Subroutines for dealing with indexes. More...

Go to the source code of this file.

Functions/Subroutines
subroutine	g2_create_index (lugb, lugi, idxver, filename, iret)
	Create a version 1 or 2 index file for a GRIB2 file. More...
subroutine	g2_write_index_headers (lugi, nlen, nnum, idxver, filename)
	Write index headers. More...
subroutine	getg2i (lugi, cbuf, nlen, nnum, iret)
	Read a version 1 index file and return its contents. More...
subroutine	getg2i2 (lugi, cbuf, idxver, nlen, nnum, iret)
	Read a version 1 or 2 index file and return its contents. More...
subroutine	getg2i2r (lugb, msk1, msk2, mnum, idxver, cbuf, nlen, nnum, nmess, iret)
	Generate a version 1 or 2 index record for each message in a GRIB2 file. More...
subroutine	getg2ir (lugb, msk1, msk2, mnum, cbuf, nlen, nnum, nmess, iret)
	Generate a version 1 index record for each message in a GRIB2 file. More...
subroutine	getgb2s (cbuf, nlen, nnum, j, jdisc, jids, jpdtn, jpdt, jgdtn, jgdt, k, gfld, lpos, iret)
	Find information about a GRIB field from the index and fill a grib_mod::gribfield. More...
subroutine	getgb2s2 (cbuf, idxver, nlen, nnum, j, jdisc, jids, jpdtn, jpdt, jgdtn, jgdt, k, gfld, lpos, iret)
	Find information about a GRIB field from the index and fill a grib_mod::gribfield. More...
subroutine	getidx (lugb, lugi, cindex, nlen, nnum, iret)
	Find, read or generate a version 1 GRIB2 index for a GRIB2 file (which must be < 2 GB). More...
subroutine	getidx2 (lugb, lugi, idxver, cindex, nlen, nnum, iret)
	Find, read or generate a version 1 or 2 GRIB2 index for a GRIB2 file (which may be > 2 GB). More...
subroutine	gf_finalize (iret)
	Free all memory associated with the library. More...
subroutine	ix2gb2 (lugb, lskip8, idxver, lgrib8, cbuf, numfld, mlen, iret)
	Generate a version 1 or 2 index record for each field in a GRIB2 message. More...
subroutine	ixgb2 (lugb, lskip, lgrib, cbuf, numfld, mlen, iret)
	Generate a version 1 index record for each field in a GRIB2 message. More...

Detailed Description

Subroutines for dealing with indexes.

Author

Edward Hartnett

Date

Jan 31, 2024

Definition in file g2index.F90.

Function/Subroutine Documentation

g2_create_index()

subroutine g2_create_index	(	integer, intent(in)	lugb,
		integer, intent(in)	lugi,
		integer, intent(in)	idxver,
		character*(*)	filename,
		integer, intent(out)	iret
	)

Create a version 1 or 2 index file for a GRIB2 file.

Parameters

[in]	lugb	Logical unit of opened GRIB2 file.
[in]	lugi	Logical unit file opened to write index to.
[in]	idxver	Index version.
[in]	filename	Name of GRIB2 file. This file must already be open, with logical unit passed to the lugb parameter. The filename is also encoded in one of the index header records.
[out]	iret	Return code: 0 success 90 problem opening GRIB2 file. 91 problem opening index file. 92 no messages found in GRIB2 file.

Author

Ed Hartnett, Mark Iredell

Date

Feb 15, 2024

Definition at line 20 of file g2index.F90.

References g2_write_index_headers(), and getg2i2r().

g2_write_index_headers()

subroutine g2_write_index_headers	(	integer, intent(in)	lugi,
		integer, intent(in)	nlen,
		integer, intent(in)	nnum,
		integer, intent(in)	idxver,
		character, dimension(*), intent(in)	filename
	)

Write index headers.

Parameters

[in]	lugi	integer logical unit of output index file
[in]	nlen	integer total length of index records
[in]	nnum	integer number of index records
[in]	idxver	Index version, 1 for legacy, 2 for files that may be > 2 GB.
[in]	filename	character name of GRIB file

Author

Iredell, Ed Hartnett

Date

93-11-22

Definition at line 102 of file g2index.F90.

Referenced by g2_create_index().

getg2i()

subroutine getg2i	(	integer, intent(in)	lugi,
		character(len=1), dimension(:), pointer	cbuf,
		integer, intent(out)	nlen,
		integer, intent(out)	nnum,
		integer, intent(out)	iret
	)

Read a version 1 index file and return its contents.

The index file may be generated by the grb2index utility of the NCEPLIBS-grib_util project.

For the contents of the index file, see getgi2().

This subroutine is maintained for backward compatibility, and may only be used with version 1 of the index files. To handle both version 1 and 2 index files, use getg2i2().

Parameters

[in]	lugi	Integer unit of the GRIB index file. Must be opened by baopen() or baopenr().
[out]	cbuf	Pointer to a buffer that will get the index records. Memory will be allocated within this function, so callers must free the memory that cbuf points to, using deallocate(cbuf) when cbuf is no longer needed.
[out]	nlen	Total length of all index records.
[out]	nnum	Number of index records.
[out]	iret	Return code. 0 No error. 2 not enough memory to hold index buffer 3 error reading index file buffer 4 error reading index file header 5 index format 2 detected

Author

Mark Iredell, Ed Hartnett

Date

2000-05-26

Definition at line 429 of file g2index.F90.

References getg2i2().

getg2i2()

subroutine getg2i2	(	integer, intent(in)	lugi,
		character(len=1), dimension(:), pointer	cbuf,
		integer, intent(out)	idxver,
		integer, intent(out)	nlen,
		integer, intent(out)	nnum,
		integer, intent(out)	iret
	)

Read a version 1 or 2 index file and return its contents.

The index file may be generated by the grb2index utility of the NCEPLIBS-grib_util project.

The index file has two header records:

1. an 81-byte header with 'GB2IX1' in columns 42-47
2. an 81-byte header with the index version number, the number of bytes to skip before index records, total length in bytes of the index records, number of index records, and the GRIB file basename.

Each record in the index table contains the following fields. All integers are in big-endian format in the file. The only difference between index version 1 and index version 2 is the size of the field containing the number of bytes to skip in file before message. To accomodate files > 2 GB, this must be a 64-bit int.

Index Version 1	Index Version 2	Contents
001 - 004	001 - 004	length of index record (4 bytes)
005 - 008	005 - 012	bytes to skip in data file before grib message (4/8 bytes)
009 - 012	013 - 020	bytes to skip in message before lus (local use) set = 0, if no local section. (4/8 bytes)
013 - 016	021 - 028	bytes to skip in message before gds (4/8 bytes)
017 - 020	029 - 036	bytes to skip in message before pds (4/8 bytes)
021 - 024	037 - 044	bytes to skip in message before drs (4/8 bytes)
025 - 028	045 - 052	bytes to skip in message before bms (4/8 bytes)
029 - 032	053 - 060	bytes to skip in message before data section (4/8 bytes)
033 - 040	061 - 068	bytes total in the message (8 bytes)
041 - 041	069 - 069	grib version number (always 2) (1 byte)
042 - 042	070 - 070	message discipline (1 byte)
043 - 044	071 - 072	field number within grib2 message (2 bytes)
045 - ii	073 - ii	identification section (ids) (character)
ii+1- jj	ii+1- jj	grid definition section (gds) (character)
jj+1- kk	jj+1- kk	product definition section (pds) (character)
kk+1- ll	kk+1- ll	the data representation section (drs) (character)
ll+1-ll+6	ll+1-ll+6	first 6 bytes of the bit map section (bms) (character)

Note

Subprogram can be called from a multiprocessing environment. Do not engage the same logical unit from more than one processor.

Parameters

[in]	lugi	Integer unit of the unblocked GRIB index file. Must be opened by baopen() or baopenr().
[out]	cbuf	Pointer to a buffer that will get the index records. Memory will be allocated within this function, so callers must free the memory that cbuf points to, using deallocate(cbuf) when cbuf is no longer needed.
[out]	idxver	Index version of this index, will be 1 or 2.
[out]	nlen	Total length of all index records.
[out]	nnum	Number of index records.
[out]	iret	Return code. 0 No error. 2 not enough memory to hold index buffer 3 error reading index file buffer 4 error reading index file header

Author

Ed Hartnett, Mark Iredell

Date

Feb 9, 2024

Definition at line 511 of file g2index.F90.

References g2logging::g2_log(), and g2logging::g2_log_msg.

Referenced by getg2i(), getgb2p2(), and getidx2().

getg2i2r()

subroutine getg2i2r	(	integer, intent(in)	lugb,
		integer (kind = 8), intent(in)	msk1,
		integer (kind = 8), intent(in)	msk2,
		integer, intent(in)	mnum,
		integer, intent(in)	idxver,
		character(len = 1), dimension(:), pointer	cbuf,
		integer, intent(out)	nlen,
		integer, intent(out)	nnum,
		integer, intent(out)	nmess,
		integer, intent(out)	iret
	)

Generate a version 1 or 2 index record for each message in a GRIB2 file.

The index record contains byte offsets to the message, it's length, and byte offsets within the message to each section. The index file record format is documented in subroutine ixgb2().

Note

Subprogram can be called from a multiprocessing environment. Do not engage the same logical unit from more than one processor.

Parameters

[in]	lugb	Unit of the GRIB file. Must be opened by baopen() or baopenr().
[in]	msk1	Number of bytes to search for first message.
[in]	msk2	Number of bytes to search for other messages.
[in]	mnum	Number of GRIB messages to skip (usually 0).
[in]	idxver	Index version number. Use 1 for legacy, 2 for files > 2 GB.
[out]	cbuf	Pointer to a buffer that will get the index records. If any memory is associated with cbuf when this subroutine is called, cbuf will be nullified in the subroutine. Initially cbuf will get an allocation of 5000 bytes. realloc() will be used to increase the size if necessary. Users must free memory that cbuf points to when cbuf is no longer needed.
[out]	nlen	Total length of index record buffer in bytes.
[out]	nnum	Number of index records, zero if no GRIB messages are found.
[out]	nmess	Last GRIB message in file successfully processed
[out]	iret	Return code. 0 No error. 1 Not enough memory available to hold full index buffer. 2 Not enough memory to allocate initial index buffer. 3 Error deallocating memory.

Author

Mark Iredell, Ed Hartnett

Date

1995-10-31

Definition at line 648 of file g2index.F90.

References g2logging::g2_log(), g2logging::g2_log_msg, ix2gb2(), and skgb8().

Referenced by g2_create_index(), getg2ir(), getgb2p2(), and getidx2().

getg2ir()

subroutine getg2ir	(	integer, intent(in)	lugb,
		integer, intent(in)	msk1,
		integer, intent(in)	msk2,
		integer, intent(in)	mnum,
		character(len = 1), dimension(:), pointer	cbuf,
		integer, intent(out)	nlen,
		integer, intent(out)	nnum,
		integer, intent(out)	nmess,
		integer, intent(out)	iret
	)

Generate a version 1 index record for each message in a GRIB2 file.

The index record contains byte offsets to the message, it's length, and byte offsets within the message to each section. The index file record format is documented in subroutine ixgb2().

Note

Subprogram can be called from a multiprocessing environment. Do not engage the same logical unit from more than one processor.

Parameters

[in]	lugb	Unit of the unblocked GRIB file. Must be opened by baopen() or baopenr().
[in]	msk1	Number of bytes to search for first message.
[in]	msk2	Number of bytes to search for other messages.
[in]	mnum	Number of GRIB messages to skip (usually 0).
[out]	cbuf	Pointer to a buffer that will get the index records. If any memory is associated with cbuf when this subroutine is called, cbuf will be nullified in the subroutine. Initially cbuf will get an allocation of 5000 bytes. realloc() will be used to increase the size if necessary. Users must free memory that cbuf points to when cbuf is no longer needed.
[out]	nlen	Total length of index record buffer in bytes.
[out]	nnum	Number of index records, zero if no GRIB messages are found.
[out]	nmess	Last GRIB message in file successfully processed
[out]	iret	Return code. 0 No error. 1 Not enough memory available to hold full index buffer. 2 Not enough memory to allocate initial index buffer. 3 Error deallocating memory.

Author

Mark Iredell, Ed Hartnett

Date

1995-10-31

Definition at line 587 of file g2index.F90.

References getg2i2r().

getgb2s()

subroutine getgb2s	(	character(len = 1), dimension(nlen), intent(in)	cbuf,
		integer, intent(in)	nlen,
		integer, intent(in)	nnum,
		integer, intent(in)	j,
		integer, intent(in)	jdisc,
		integer, dimension(*)	jids,
		integer, intent(in)	jpdtn,
		integer, dimension(*)	jpdt,
		integer, intent(in)	jgdtn,
		integer, dimension(*)	jgdt,
		integer, intent(out)	k,
		type(gribfield), intent(out)	gfld,
		integer, intent(out)	lpos,
		integer, intent(out)	iret
	)

Find information about a GRIB field from the index and fill a grib_mod::gribfield.

For a description of the index record see getg2i2().

Users of this routine will need to include the line "use grib_mod" in their calling routine.

The unpacked bitmap and bitmap data field are the only components of the grib_mod::gribfield type not set by this routine.

Note

This subprogram is intended for private use by getgb2() routines only. Note that derived type grib_mod::gribfield contains pointers to many arrays of data. The memory for these arrays is allocated when the values in the arrays are set. Users must free this memory, when it is no longer needed, by a call to subroutine gf_free().

Parameters

[in]	cbuf	Buffer (of size nlen bytes) containing index data.
[in]	nlen	Total length of all index records.
[in]	nnum	Number of index records.
[in]	j	Number of fields to skip (0 to search from beginning).
[in]	jdisc	GRIB2 discipline number of requested field. See GRIB2 - TABLE 0.0 - DISCIPLINE. Use -1 to accept any discipline.
[in]	jids	Array of values in the identification section. (Set to -9999 for wildcard.) jids(1) Identification of originating centre. See [TABLE 0 - NATIONAL/INTERNATIONAL ORIGINATING CENTERS](https://www.nco.ncep.noaa.gov/pmb/docs/on388/table0.html). jids(2) Identification of originating sub-centre. See [TABLE C - NATIONAL SUB-CENTERS](https://www.nco.ncep.noaa.gov/pmb/docs/on388/tablec.html). jids(3) GRIB master tables version number. See [GRIB2 - TABLE 1.0 GRIB Master Tables Version Number](https://www.nco.ncep.noaa.gov/pmb/docs/grib2/grib2_doc/grib2_table1-0.shtml). jids(4) GRIB local tables version number. See [GRIB2 - TABLE 1.1 GRIB Local Tables Version Number](https://www.nco.ncep.noaa.gov/pmb/docs/grib2/grib2_doc/grib2_table1-1.shtml). jids(5) Significance of reference time. See [GRIB2 - TABLE 1.2 - Significance of Reference Time](https://www.nco.ncep.noaa.gov/pmb/docs/grib2/grib2_doc/grib2_table1-2.shtml). jids(6) year (4 digits) jids(7) month jids(8) day jids(9) hour jids(10) minute jids(11) second jids(12) Production status of processed data. See [GRIB2 - TABLE 1.3 - Production Status of Data](https://www.nco.ncep.noaa.gov/pmb/docs/grib2/grib2_doc/grib2_table1-3.shtml). jids(13) Type of processed data. See GRIB2 - TABLE 1.4 - TYPE OF DATA.
[in]	jpdtn	Product Definition Template (PDT) number (n) (if = -1, don't bother matching PDT - accept any).
[in]	jpdt	Array of values defining the Product Definition Template of the field for which to search (=-9999 for wildcard).
[in]	jgdtn	Grid Definition Template (GDT) number (if = -1, don't bother matching GDT - accept any).
[in]	jgdt	array of values defining the Grid Definition Template of the field for which to search (=-9999 for wildcard).
[out]	k	Field number unpacked.
[out]	gfld	Derived type grib_mod::gribfield.
[out]	lpos	Starting position of the found index record within the complete index buffer, CBUF. = 0, if request not found.
[out]	iret	integer return code: 0 No error. 97 Error reading GRIB file. other gf_getfld GRIB2 unpacker return code.

Author

Stephen Gilbert, Ed Hartnett

Date

2002-01-15

Definition at line 814 of file g2index.F90.

References getgb2s2().

getgb2s2()

subroutine getgb2s2	(	character(len = 1), dimension(nlen), intent(in)	cbuf,
		integer, intent(in)	idxver,
		integer, intent(in)	nlen,
		integer, intent(in)	nnum,
		integer, intent(in)	j,
		integer, intent(in)	jdisc,
		integer, dimension(*)	jids,
		integer, intent(in)	jpdtn,
		integer, dimension(*)	jpdt,
		integer, intent(in)	jgdtn,
		integer, dimension(*)	jgdt,
		integer, intent(out)	k,
		type(gribfield), intent(out)	gfld,
		integer, intent(out)	lpos,
		integer, intent(out)	iret
	)

Find information about a GRIB field from the index and fill a grib_mod::gribfield.

For a description of the index record see getg2i2().

Users of this routine will need to include the line "use grib_mod" in their calling routine.

The unpacked bitmap and bitmap data field are the only components of the grib_mod::gribfield type not set by this routine.

Note

This subprogram is intended for private use by getgb2() routines only. Note that derived type grib_mod::gribfield contains pointers to many arrays of data. The memory for these arrays is allocated when the values in the arrays are set. Users must free this memory, when it is no longer needed, by a call to subroutine gf_free().

Parameters

[in]	cbuf	Buffer (of size nlen bytes) containing index data.
[in]	idxver	Index version, 1 for legacy, 2 if files may be > 2 GB.
[in]	nlen	Total length of all index records.
[in]	nnum	Number of index records.
[in]	j	Number of fields to skip (0 to search from beginning).
[in]	jdisc	GRIB2 discipline number of requested field. See GRIB2 - TABLE 0.0 - DISCIPLINE. Use -1 to accept any discipline.
[in]	jids	Array of values in the identification section. (Set to -9999 for wildcard.) jids(1) Identification of originating centre. See [TABLE 0 - NATIONAL/INTERNATIONAL ORIGINATING CENTERS](https://www.nco.ncep.noaa.gov/pmb/docs/on388/table0.html). jids(2) Identification of originating sub-centre. See [TABLE C - NATIONAL SUB-CENTERS](https://www.nco.ncep.noaa.gov/pmb/docs/on388/tablec.html). jids(3) GRIB master tables version number. See [GRIB2 - TABLE 1.0 GRIB Master Tables Version Number](https://www.nco.ncep.noaa.gov/pmb/docs/grib2/grib2_doc/grib2_table1-0.shtml). jids(4) GRIB local tables version number. See [GRIB2 - TABLE 1.1 GRIB Local Tables Version Number](https://www.nco.ncep.noaa.gov/pmb/docs/grib2/grib2_doc/grib2_table1-1.shtml). jids(5) Significance of reference time. See [GRIB2 - TABLE 1.2 - Significance of Reference Time](https://www.nco.ncep.noaa.gov/pmb/docs/grib2/grib2_doc/grib2_table1-2.shtml). jids(6) year (4 digits) jids(7) month jids(8) day jids(9) hour jids(10) minute jids(11) second jids(12) Production status of processed data. See [GRIB2 - TABLE 1.3 - Production Status of Data](https://www.nco.ncep.noaa.gov/pmb/docs/grib2/grib2_doc/grib2_table1-3.shtml). jids(13) Type of processed data. See GRIB2 - TABLE 1.4 - TYPE OF DATA.
[in]	jpdtn	Product Definition Template (PDT) number (n) (if = -1, don't bother matching PDT - accept any).
[in]	jpdt	Array of values defining the Product Definition Template of the field for which to search (=-9999 for wildcard).
[in]	jgdtn	Grid Definition Template (GDT) number (if = -1, don't bother matching GDT - accept any).
[in]	jgdt	array of values defining the Grid Definition Template of the field for which to search (=-9999 for wildcard).
[out]	k	Field number unpacked.
[out]	gfld	Derived type grib_mod::gribfield.
[out]	lpos	Starting position of the found index record within the complete index buffer, CBUF. = 0, if request not found.
[out]	iret	integer return code: 0 No error. 97 Error reading GRIB file. other gf_getfld GRIB2 unpacker return code.

Author

Stephen Gilbert, Ed Hartnett

Date

Feb 9 2024

Definition at line 927 of file g2index.F90.

References g2_gbytec1(), g2logging::g2_log(), g2logging::g2_log_msg, gf_unpack1(), gf_unpack3(), gf_unpack4(), and gf_unpack5().

Referenced by getgb2i2(), getgb2p2(), and getgb2s().

getidx()

subroutine getidx	(	integer, intent(in)	lugb,
		integer, intent(in)	lugi,
		character(len = 1), dimension(:), pointer	cindex,
		integer, intent(out)	nlen,
		integer, intent(out)	nnum,
		integer, intent(out)	iret
	)

Find, read or generate a version 1 GRIB2 index for a GRIB2 file (which must be < 2 GB).

If the index already exists in library memory, it is returned, otherwise, the index is read from an existing index file associated with unit lugi or generated from the GRIB2 file lugb.

Users can force a regeneration of an index: if lugi equals lugb, the index will be regenerated from the data in file lugb. If lugi is less than zero, then the index is re-read from index file abs(lugi).

This subroutine allocates memory and stores the resulting pointers in an array that is a Fortran "save" variable. The result is that the memory will not be freed by the library and cannot be reached by the caller. To free this memory call gf_finalize() after all library operations are complete.

Note

The file unit numbers must be in range 1 - 9999.

Parameters

[in]	lugb	integer unit of the GRIB2 data file. File must have been opened with baopen() or baopenr() before calling this routine. If 0, then all saved memory will be released (necessary for g2_finalize()).
[in]	lugi	integer unit of the GRIB2 index file. If nonzero, file must have been opened with baopen() or baopenr() before calling this routine. Set to 0 to get index information from the GRIB2 file.
[in,out]	cindex	character*1 Pointer to a buffer that will get index records.
[out]	nlen	integer Total length of all index records.
[out]	nnum	integer Number of index records.
[out]	iret	integer Return code: 0 No error. 90 Unit number out of range. 96 Error reading/creating index file. 97 Index version 2 detected.

Author

Stephen Gilbert, Ed Hartnett

Date

2005-03-15

Definition at line 197 of file g2index.F90.

References getidx2().

Referenced by gf_finalize().

getidx2()

subroutine getidx2	(	integer, intent(in)	lugb,
		integer, intent(in)	lugi,
		integer, intent(inout)	idxver,
		character(len = 1), dimension(:), pointer	cindex,
		integer, intent(out)	nlen,
		integer, intent(out)	nnum,
		integer, intent(out)	iret
	)

Find, read or generate a version 1 or 2 GRIB2 index for a GRIB2 file (which may be > 2 GB).

If the index already exists in library memory, it is returned, otherwise, the index is read from an existing index file associated with unit lugi or generated from the GRIB2 file lugb.

Users can force a regeneration of an index: if lugi equals lugb, the index will be regenerated from the data in file lugb. If lugi is less than zero, then the index is re-read from index file abs(lugi).

This subroutine allocates memory and stores the resulting pointers in an array that is a Fortran "save" variable. The result is that the memory will not be freed by the library and cannot be reached by the caller. To free this memory call gf_finalize() after all library operations are complete.

Note

The file unit numbers must be in range 1 - 9999.

Parameters

[in]	lugb	integer unit of the GRIB2 data file. File must have been opened with baopen() or baopenr() before calling this routine. If 0, then all saved memory will be released (necessary for g2_finalize()).
[in]	lugi	integer unit of the GRIB2 index file. If nonzero, file must have been opened with baopen() or baopenr() before calling this routine. Set to 0 to get index information from the GRIB2 file.
[in]	idxver	Index version, 1 for legacy, 2 if files may be > 2 GB. index records.
[in,out]	cindex	character*1 Pointer to a buffer that will get index records.
[out]	nlen	integer Total length of all index records.
[out]	nnum	integer Number of index records.
[out]	iret	integer Return code: 0 No error. 90 Unit number out of range. 96 Error reading/creating index file.

Author

Stephen Gilbert, Ed Hartnett

Date

Feb 9, 2024

Definition at line 265 of file g2index.F90.

References g2logging::g2_log(), g2logging::g2_log_msg, getg2i2(), and getg2i2r().

Referenced by getgb2i2(), and getidx().

gf_finalize()

subroutine gf_finalize

(

integer, intent(out)

iret

)

Free all memory associated with the library.

Parameters

[out]

iret

integer Return code: 0 No error. otherwise Error freeing internal resources.

Author

Ed Hartnett

Date

7/16/23

Definition at line 1664 of file g2index.F90.

References getidx().

ix2gb2()

subroutine ix2gb2	(	integer, intent(in)	lugb,
		integer (kind = 8), intent(in)	lskip8,
		integer, intent(in)	idxver,
		integer (kind = 8), intent(in)	lgrib8,
		character(len = 1), dimension(:), intent(inout), pointer	cbuf,
		integer, intent(out)	numfld,
		integer, intent(out)	mlen,
		integer, intent(out)	iret
	)

Generate a version 1 or 2 index record for each field in a GRIB2 message.

The index records are written to index buffer pointed to by cbuf. All integers in the index are in big-endian format.

This subroutine is called by getg2ir(), which packages the index records into an index file.

See getg2i2() for thr format of the index buffer records.

Parameters

[in]	lugb	Unit of the unblocked GRIB file. Must be opened by baopen() or baopenr().
[in]	lskip8	Number of bytes to skip before GRIB message.
[in]	idxver	Index version, use 1 for legacy, 2 for GRIB2 files > 2 GB.
	lgrib8	Number of bytes in GRIB message. When subroutine is called, if this is < 5000, it will be used as the number of bytes initially read from the file. So for GRIB2 messages of < 5000 in size, this should be set to the size of the GRIB2 message.
[in,out]	cbuf	Pointer to a buffer that will get the index records. If any memory is associated with cbuf when this subroutine is called, cbuf will be nullified in the subroutine. Initially cbuf will get an allocation of 5000 bytes. realloc() will be used to increase the size if necessary. Users must free memory that cbuf points to when cbuf is no longer needed.
[out]	numfld	Number of index records created.
[out]	mlen	Total length of all index records.
[out]	iret	Return code 0 No error 1 Not enough memory available to hold full index buffer. 2 I/O error in read. 3 GRIB message is not edition 2. 4 Not enough memory to allocate extent to index buffer. 5 Unidentified GRIB section encountered.

Author

Ed Hartnett, Mark Iredell

Date

Feb 5, 2024

Definition at line 1255 of file g2index.F90.

References g2_gbytec1(), g2logging::g2_log(), g2logging::g2_log_msg, g2_sbytec1(), and g2_sbytec81().

Referenced by getg2i2r(), and ixgb2().

ixgb2()

subroutine ixgb2	(	integer	lugb,
		integer	lskip,
		integer	lgrib,
		character(len = 1), dimension(:), pointer	cbuf,
		integer	numfld,
		integer	mlen,
		integer	iret
	)

Generate a version 1 index record for each field in a GRIB2 message.

The index records are written to index buffer pointed to by cbuf. All integers in the index are in big-endian format.

This subroutine is called by getg2ir(), which packages the index records into an index file.

See getg2i2() for thr format of the index buffer records.

Parameters

lugb	Unit of the unblocked GRIB file. Must be opened by baopen() or baopenr().
lskip	Number of bytes to skip before GRIB message.
lgrib	Number of bytes in GRIB message. When subroutine is called, this must be set to the size of the cbuf buffer.
cbuf	Pointer to a buffer that will get the index records. If any memory is associated with cbuf when this subroutine is called, cbuf will be nullified in the subroutine. Initially cbuf will get an allocation of 5000 bytes. realloc() will be used to increase the size if necessary. Users must free memory that cbuf points to when cbuf is no longer needed.
numfld	Number of index records created.
mlen	Total length of all index records.
iret	Return code 0 No error 1 Not enough memory available to hold full index buffer. 2 I/O error in read. 3 GRIB message is not edition 2. 4 Not enough memory to allocate extent to index buffer. 5 Unidentified GRIB section encountered.

Author

Mark Iredell, Ed Hartnett

Date

1995-10-31

Definition at line 1191 of file g2index.F90.

References ix2gb2().