Last modified: 23 August 2013
Tool Name: h5dump
Syntax:
h5dump [OPTIONS] files

Purpose:
Displays HDF5 file contents.

Description:
h5dump enables the user to examine the contents of an HDF5 file and dump those contents, in human readable form, to an ASCII file.

h5dump dumps HDF5 file content to standard output. It can display the contents of the entire HDF5 file or selected objects, which can be groups, datasets, a subset of a dataset, links, attributes, or datatypes.

The --header option displays object header information only.

Names are the absolute names of the objects. h5dump displays objects in the order same as the command order. If a name does not start with a slash, h5dump begins searching for the specified object starting at the root group.

If an object is hard linked with multiple names, h5dump displays the content of the object in the first occurrence. Only the link information is displayed in later occurrences.

h5dump assigns a name for any unnamed datatype in the form of #oid1:oid2, where oid1 and oid2 are the object identifiers assigned by the library. The unnamed types are displayed within the root group.

Datatypes are displayed with standard type names. For example, if a dataset is created with H5T_NATIVE_INT type and the standard type name for integer on that machine is H5T_STD_I32BE, h5dump displays H5T_STD_I32BE as the type of the dataset.

h5dump can also dump a subset of a dataset. This feature operates in much the same way as hyperslabs in HDF5; the parameters specified on the command line are passed to the function H5Sselect_hyperslab and the resulting selection is displayed.

The h5dump output is described in detail in the DDL for HDF5, the Data Description Language document.

Note: It is not permissible to specify multiple attributes, datasets, datatypes, groups, or soft links with one flag. For example, one may not issue the command
         WRONG:   h5dump -a /attr1 /attr2 foo.h5
to display both /attr1 and /attr2. One must issue the following command:
         CORRECT:   h5dump -a /attr1 -a /attr2 foo.h5

One byte integer type data is displayed in decimal by default. When displayed in ASCII, a non-printable code is displayed in 3 octal digits preceeded by a back-slash unless there is a C language escape sequence for it. For example, CR and LF are printed as \r and \n. Though the NUL code is represented as \0 in C, it is printed as \000 to avoid ambiguity as illustrated in the following 1 byte char data (since this is not a string, embedded NUL is possible). 

        141 142 143 000 060 061 062 012
          a   b   c  \0   0   1   2  \n
h5dump prints them as "abc\000012\n". But if h5dump prints NUL as \0, the output is "abc\0012\n" which is ambiguous.

Using file drivers:
It is possible to select the file driver with which to open an HDF5 file by using the --filedriver (or -f) command-line option. Valid values for the --filedriver option are  sec2,  family,  split,  and  multi. If the file driver flag is not specified, then the file will be opened with each driver in turn, and in the order specified above, until one driver succeeds in opening the file.

Special file naming restrictions apply when using h5dump with either the split or the multi driver.

To dump a split file, h5dump requires that the metadata and raw data filenames end with -m.h5 and -r.h5, respectively, and that the entire virtual HDF5 file, or the logical HDF5 file, be referred to on the command line by the common portion of the filename preceding the -m and -r.

For example, assume that a split HDF5 file has its metadata in a file named splitfile-m.h5 and its raw data in a file named splitfile-r.h5. The following command would dump the contents of this logical HDF5 file:
     h5dump --filedriver="split" splitfile

Note that the above split filename restrictions are specific to h5dump; HDF5 applications do not necessarily have the same limitations.

To dump a multi file, h5dump requires that the metadata and raw data filenames end with a subset of the following:
     -s.h5   for userblock, superblock, and driver information block data
     -b.h5   for B-tree node information
     -r.h5   for dataset raw data
     -g.h5   for global heap data
     -l.h5   for local heap data (object names)
     -o.h5   for object headers
The entire virtual HDF5 file must also be referred to on the command line by the common portion of the filename preceding those special tags.

For example, assume that a multi HDF5 file has its userblock, superblock, and driver information block data in a file named multifile-s.h5, its B-tree node information in a file named multifile-b.h5, its raw data in a file named multifile-r.h5, its global heap data in a file named multifile-g.h5, et cetera. The following command would dump the contents of this logical HDF5 file:
     h5dump --filedriver="multi" multifile

Note that the above multi filename restrictions are specific to h5dump; HDF5 applications do not necessarily have the same limitations.

To dump a family file, h5dump requires that the logical file’s name on the command line include the printf(3c)-style integer format specifier that specifies the format of the family file member numbers. For example, if an HDF5 family of files consists of the files family_000.h5, family_001.h5, family_002.h5, and family_003.h5, the logical HDF5 file would be specified on the command line as family_%3d.h5.

The following command would dump the contents of this logical HDF5 file:
     h5dump --filedriver="family" family_%3d.h5

XML Output:
With the --xml option, h5dump generates XML output. This output contains a complete description of the file, marked up in XML. The XML conforms to the HDF5 Document Type Definition (DTD) available at “HDF5 XML Software.”

The XML output is suitable for use with other tools, including the HDF5 Java Products.

Options and Parameters:
-h   or   --help Print a usage message and exit.
-V   or   --version Print version number and exit.
  
Formatting options:
-e   or   --escape Escape non-printing characters.
-r   or   --string Print 1-byte integer datasets as ASCII.
-y   or   --noindex Do not print array indices with data.
-m T   or   --format=T Set the floating point output format.
T is a string defining the floating point format, e.g., '%.3f'.
-q Q   or   --sort_by=Q Sort groups and attributes by the specified index type, Q. Valid values of Q are as follows:
     name              Alpha-numeric index by name (Default)
     creation_order    Index by creation order
-z Z   or   --sort_order=Z Sort groups and attributes in the specified order, Z. Valid values of Z are as follows:
     ascending    Sort in ascending order (Default)
     descending   Sort in descending order
--enable-error-stack Prints messages from the HDF5 error stack as they occur.

Injects error stack information, which is normally suppressed, directly into the output stream. This will disrupt normal h5dump output but is a useful diagnostic tool when data is not being correctly dumped. Consider the case when an h5dump call produces this message: 

    dump error: unable to print data
h5dump can be called again with ‘--enable-error-stack’ plus the original options to reveal error stack messages.
--no-compact-subset Enables recognition of the left square bracket ( [ ) as a character in a dataset name.

This option must disable compact subsetting, which is described at the end of this “Options and Parameters” section.

-w N   or   --width=N Set the number of columns of output.
A value of 0 (zero) sets the number of columns to the maximum (65535).
Default width is 80 columns.
File options:
-n   or   --contents Print a list of the file contents and exit.
-n 1   or   --contents=1 The optional value 1 (one) on the -n, --contents option adds attributes to the output.
-B   or   --superblock Print the content of the superblock.
-H   or   --header Print the header only; no data is displayed.
-f D   or   --filedriver=D Specify which driver to open the file with.
-o F   or   --output=F Output raw data into file F.

The files specified for the -o and -O options must be different files. The output from these options must not be comingled.

To suppress the raw data display, use this option with no filename, as in either of the following examples. This has the effect of sending the output to a NULL file:
     -o
     --output=

-b B   or   --binary=B Output dataset to a binary file using the datatype specified by B.
B must have one of the following values:
     LE        Little-endian
     BE        Big-endian
     MEMORY    Memory datatype
     FILE      File datatype
Recommended usage is with the -d and -o options.
-O F   or   --ddl=F Output DDL text into file F.

The files specified for the -o and -O options must be different files. The output from these options must not be comingled.

h5import can use the files output by the -o and -O when importing HDF5 data. See “Using h5dump to create input for h5import.”

To suppress the DDL display, use this option with no filename, as in either of the following examples. This has the effect of sending the output to a NULL file:
     -O
     --ddl=

Object options:
-a P  or  --attribute=P     Print the specified attribute.
-d P   or   --dataset=P Print the specified dataset.
-g P   or   --group=P Print the specified group and all members.
-l P   or   --soft-link=P Print the value(s) of the specified soft link.
-t P   or   --datatype=P Print the specified named datatype.
-A   or   --onlyattr Print the header and value of attributes; data of datasets is not displayed.
-A 0   or   --onlyattr=0 The optional value 0 (zero) on the -A, --onlyattr option prints everything except attributes.
-N P   or   --any-path=P Print any attribute, dataset, datatype, group, or link whose path matches P.
P may match either the absolute path or any portion of the path.
Object property options:
-i   or   --object-ids Print the object ids.
-p   or   --properties
                            		Print information regarding dataset properties, including filters, storage layout, fill value, and allocation time.
The filter output lists any filters used with a dataset, including the type of filter, its name, and any filter parameters.
The storage layout output specifies the dataset layout (chunked, compact, or contiguous), the size in bytes of the dataset on disk, and, if a compression filter is associated with the dataset, the compression ratio. The compression ratio is computed as (uncompressed size)/(compressed size).
The fill value output includes the fill value datatype and value.
The allocation time output displays the allocation time as specified with H5Pset_alloc_time.
-M L   or   --packedbits=L Print packed bits as unsigned integers, using the mask format L for an integer dataset specified with option -d.
L is a list of offset,length values, separated by commas.
offset is the beginning bit in the data value.
length is the number of bits in the mask.
-R   or   --region Print dataset pointed by region references.
XML options:
-x   or   --xml Output XML using XML schema (default) instead of DDL.
-u   or   --use-dtd Output XML using XML DTD instead of DDL.
-D U   or   --xml-dtd=U In XML output, refer to the DTD or schema at U instead of the default schema/DTD.
-X S   or   --xml-ns=S In XML output, (XML Schema) use qualified names in the XML:
      ":": no namespace, default: "hdf5:"
Subsetting options and compact subsetting:
  Subsetting is available by using the following options with the dataset option, -d or --dataset. Subsetting is accomplished by selecting a hyperslab from the data, so the options mirror those for performing a hyperslab selection.

At least one of the START, COUNT, STRIDE, or BLOCK options is mandatory if you do subsetting; the remainder are optional, with default values as follows:
• STRIDE, COUNT, and BLOCK default to 1 in each dimension.
• START defaults to 0 in each dimension.

-s START   or
  --start=START		 Offset of start of subsetting selection.
Default: 0 in all dimensions, specifying the beginning of the dataset.

Each of START, STRIDE, COUNT, and BLOCK must be a comma-separated list of integers with one integer for each dimension of the dataset.

-S STRIDE   or
  --stride=STRIDE		 Hyperslab stride.
Default: 1 in all dimensions.
-c COUNT   or
  --count=COUNT		 Number of blocks to include in the selection.
Default: 1 in all dimensions.
-k BLOCK   or
  --block=BLOCK		 Size of block in hyperslab.
Default: 1 in all dimensions.

Compact subsetting:
Subsetting parameters can also be expressed in a convenient compact form, as follows:
         --dataset="/foo/mydataset[START;STRIDE;COUNT;BLOCK]"

It is not required to use all parameters, but until the last parameter value used, all of the semicolons (;) are required, even when a parameter value is not specified. Example:
         --dataset="/foo/mydataset[START;;COUNT]"
         --dataset="/foo/mydataset[START]"

Each of START, STRIDE, COUNT, and BLOCK must be a comma-separated list of integers with one integer for each dimension of the dataset.

When not specified, default parameter values are used: zeros ( 0 ) for START and ones ( 1 ) for STRIDE, COUNT, and BLOCK.

Option Argument Conventions:
-- Two dashes followed by whitespace.
Indicates that the following argument is not an option.

For example, this structure can be used to dump a file called -f: 

    h5dump -- -f
This option is necessary only when the name of the file to be examined starts with a dash (-), which could confuse the tool’s command-line parser.
Option parameters appearing above are defined as follows:
D File driver to use in opening the file
Valid values are  sec2,  family,  split, and  multi.
Without the file driver option, the file will be opened with each driver in turn, and in the order specified immediately above, until one driver succeeds in opening the file.
P Path to the object
For most options, this must be the absolute path from the root group to the object.
With the -N, --any-path option, this may be either the absolute path or a partial path.
F A filename
N An integer greater than 1
START,  STRIDE,
     COUNT, and BLOCK		 Comma-separated lists of integers
Each of these option parameters must be a list of integers with one integer for each dimension of the dataspace being queried.
U A URI (as defined in [IETF RFC 2396], updated by [IETF RFC 2732]) that refers to the DTD to be used to validate the XML
B The form of binary output:
    MEMORY    for a memory type
    FILE      for the file type
    LE or BE   for pre-existing little- or big-endian types
Files parameter:
files File or files to be examined; one or more files may be listed.

The file name may include a printf(3C) integer format such as %05d to open a file family.

On Unix, Linux, and Mac OS X systems, multiple files can be examined through the use of Unix-style wildcards. For example, assume that we are working with the files FileA.h5, FileB.h5, FileC.h5, and FileAB.h5:

  • File[AB].h5 will pick up any file that begins with File, followed by any one of the characters contained in the square brackets, and ending with .h5.

    In this case, File[AB].h5 will pick up the files FileA.h5 and FileB.h5.

  • File?.h5 will pick up all files whose names are File, followed by exactly 1 character (any character), followed by .h5.

    In this case, File?.h5 will pick up the files FileA.h5 and FileB.h5, and FileC.h5.

  • File*.h5 will pick up all files whose names begin with File and end with .h5.

    In this case, File*.h5 will pick up all four files.

The wildcard capability is not currently available on Windows systems.

Exit Status:
0 Succeeded.
> 0     An error occurred.

Examples:
  1. Dump the group /GroupFoo/GroupBar in the file quux.h5:
         h5dump -g /GroupFoo/GroupBar quux.h5

  2. Dump the dataset Fnord, which is in the group /GroupFoo/GroupBar in the file quux.h5:
         h5dump -d /GroupFoo/GroupBar/Fnord quux.h5

  3. Dump the attribute metadata of the dataset Fnord, which is in the group /GroupFoo/GroupBar in the file quux.h5:
         h5dump -a /GroupFoo/GroupBar/Fnord/metadata quux.h5

  4. Dump the attribute metadata which is an attribute of the root group in the file quux.h5:
         h5dump -a /metadata quux.h5

  5. Produce an XML listing of the file bobo.h5, saving the listing in the file bobo.h5.xml:
         h5dump --xml bobo.h5 > bobo.h5.xml

  6. Dump a subset of the dataset /GroupFoo/databar/ in the file quux.h5:
         h5dump -d /GroupFoo/databar --start="1,1" --stride="2,3"
                --count="3,19" --block="1,1" quux.h5

  7. The same example, using the short form to specify the subsetting parameters:
         h5dump -d "/GroupFoo/databar[1,1;2,3;3,19;1,1]" quux.h5

  8. Dump a binary copy of the dataset /GroupD/FreshData/ in the file quux.h5, with data written in little-endian form, to the output file FreshDataD.bin:
         h5dump -d "/GroupD/FreshData" -b LE
                -o "FreshDataD.bin" quux.h5

  9. Display two sets of packed bits (bits 0-1 and bits 4-6) in the dataset /dset of the file quux.h5:
         h5dump -d /dset -M 0,1,4,3 quux.h5

  10. Dump the dataset /GroupFoo/GroupBar/Fnord to the file quux.h5 and output the DDL into the file ddl.txt and the raw data into the file data.txt:
         h5dump -d /GroupFoo/GroupBar/Fnord --ddl=ddl.txt -y
                -o data.txt quux.h5

  11. Dump the dataset /GroupFoo/GroupBar/Fnord to the file quux.h5, suppress the DDL output, and output the raw data into the file data.txt:
         h5dump -d /GroupFoo/GroupBar/Fnord --ddl= -y
                -o data.txt quux.h5

Current Status:
The current version of h5dump displays the following information:

See Also:
  • HDF5 Data Description Language syntax at DDL for HDF5
  • HDF5 XML Schema at http://www.hdfgroup.org/HDF5/XML/schema/HDF5-File.xsd
  • HDF5 XML information at http://www.hdfgroup.org/HDF5/XML/

    • History:
    Release     Change
    1.10.1 File space information (strategy, persist, threshold, page size) was added when printing the contents of the superblock with the -B option.
    1.8.12 Optional value of 0 (zero) for the -A, --onlyattr option added in this release.

    Option added in this release:
       -N P or --any-path=P

    1.8.11 Option added in this release:
       -O F or --ddl=F
    This option can be used to suppress the DDL output.
    This option, combined with the '--output=F' (or '-o F') option will generate files that can be used as input to h5import.

    h5dump updated in this release to display the compression ratio for user-defined filters.

    In h5dump output, UNKNOWN_FILTER has been changed to USER_DEFINED_FILTER.

    1.8.9 Option added in this release:
       --no-compact-subset

    h5dump output has been updated for this release to improve compliance with the HDF5 DDL specification and to improve readablilty. This output is now fully compliant with the HDF5 DDL specification, but these changes may affect some user scripts:

    • Whitespace has been modified.
    • Display errors, such as misplaced brackets, have been fixed.
    • When printing superblock content, user block content is now properly displayed within the superblock.
    1.8.7 Option added in this release:
       --enable-error-stack

    Tool updated in this release to correctly display reference type:
       H5T_REFERENCE {H5T_STD_REF_OBJ} for object references
       H5T_REFERENCE {H5T_STD_REF_DSETREG} for dataset region references

    1.8.5 Bitfield display fixed in this release.

    Option added in this release for packed bits data display:
       -M or --packedbits option

    1.8.4 Option added in this release for region reference display:
       -R or --region option
    1.8.1 Compression ratio added to output of -p or --properties option in this release.
    1.8.0 Options added in this release:
       -q or --sort_by
       -z or --sort_order
    1.6.5 Options added in this release:
       -n or --contents
       -e or --escape
       -y or --noindex
       -p or --properties
       -b or --binary