hdf5/doc/html/ddl.html
Ruey-Hsia Li bb4d5ebbb5 [svn-r486] DDL description
C: ----------------------------------------------------------------------
1998-07-14 14:00:46 -05:00

240 lines
7.5 KiB
HTML

<html>
<head>
<TITLE>DDL for HDF5</TITLE>
</head>
<body>
<B> DDL in BNF </B> <br>
<P>
The following describes the data description language for HDF5 file.
The description is in Backus-Naur Form. A quick explanation
of the some symbols used.
<PRE>
::= defined as
&lt;tname&gt; a token with the name tname
&lt;a&gt; | &lt;b&gt; one of &lt;a&gt; or &lt;b&gt;
&lt;a&gt;<FONT SIZE=1.7>opt</FONT> zero or one occurrence of &lt;a&gt;
&lt;a&gt;* zero or more occurrence of &lt;a&gt;
&lt;a&gt;+ one or more occurrence of &lt;a&gt;
TBD To Be Decided
&lt;file&gt; ::= hdf5 "&lt;file_name&gt;" { &lt;file_boot_block&gt;<FONT SIZE=1.7>opt</FONT> &lt;root_group&gt; }
&lt;file_name&gt; ::= &lt;identifier&gt;
&lt;file_boot_block&gt; ::= boot_block { &lt;boot_block_content&gt; }
&lt;boot_block_content&gt; ::= TBD
&lt;root_group&gt; ::= group "/" { &lt;group_attribute&gt;* &lt;group_member&gt;* }
// Attributes and members can be in any order.
&lt;group_attribute&gt; ::= &lt;attribute&gt;
&lt;attribute&gt; ::= attribute "attr_name" { &lt;datatype&gt;
&lt;dataspace&gt;
&lt;data&gt;<FONT SIZE=1.7>opt</FONT> }
// &lt;datatype&gt;, &lt;dataspace&gt; and &lt;data&gt; can be in any order
// as long as &lt;data&gt; comes the last.
&lt;attr_name&gt; ::= &lt;identifier&gt;
&lt;datatype&gt; ::= datatype "&lt;datatype_name&gt;" ; | // shared data type
datatype { &lt;scalar_datatype&gt; } ; |
datatype { &lt;dt_definition&gt;+ } ;
&lt;datatype_name&gt; ::= &lt;identifier&gt;
&lt;scalar_datatype&gt; ::= int8 | uint8 | ... // data types supported by HDF5
&lt;dt_definition&gt; ::= "&lt;datatype_name&gt;" &lt;field_name&gt; ; | &lt;scalar_datatype&gt; &lt;field_name&gt; ;
&lt;field_name&gt; ::= one or more field names separated by ","
&lt;dataspace&gt; ::= dataspace "&lt;dataspace_name&gt;" ; |
dataspace {array &lt;current_dims&gt; &lt;max_dims&gt;} ; |
dataspace {other &lt;ds_definition&gt;+ } ;
&lt;dataspace_name&gt; ::= &lt;identifier&gt;
&lt;current_dims&gt; ::= a list of integers represented of the form ( , , .. , )
&lt;max_dims&gt; ::= a list of integers or H5S_UNLIMITED represented of the form ( , , .. , )
&lt;ds_definition&gt; ::= TBD
&lt;data&gt; ::= data {TBD} ;
&lt;group_member&gt; ::= &lt;named_datatype&gt; | &lt;named_dataspace&gt; | &lt;group&gt; | &lt;dataset&gt; | &lt;link&gt;
&lt;named_datatype&gt; ::= datatype "&lt;datatype_name&gt;" { &lt;scalar_datatype&gt; } |
datatype "&lt;datatype_name&gt;" { &lt;dt_definition&gt;+ }
&lt;named_dataspace&gt; ::= dataspace "&lt;dataspace_name&gt;" {array &lt;current_dims&gt; &lt;max_dims&gt; } |
dataspace "&lt;dataspace_name&gt;" { other &lt;ds_definition&gt;+ }
&lt;group&gt; ::= group "&lt;group_name&gt;" { &lt;group_attribute&gt;* &lt;group_member&gt;* }
// Attributes and members can be in any order.
&lt;group_name&gt; ::= &lt;identifier&gt;
&lt;dataset&gt; ::= dataset "&lt;dataset_name&gt;" { &lt;datatype&gt;
&lt;dataspace&gt;
&lt;storagelayout&gt;<FONT SIZE=1.7>opt</FONT>
&lt;compression&gt;<FONT SIZE=1.7>opt</FONT>
&lt;dataset_attribute&gt;*
&lt;data&gt;<FONT SIZE=1.7>opt</FONT> }
// Any order is ok as long as &lt;data&gt; and &lt;dataset_attribute&gt; are
// after &lt;datatype&gt; and &lt;dataspace&gt;.
&lt;dataset_name&gt; ::= &lt;identifier&gt;
&lt;storagelayout&gt; :: = storagelayout &lt;contiguous_layout&gt; ; |
storagelayout &lt;chunked_layout&gt; ; |
storagelayout &lt;compact_layout&gt; ; |
storagelayout &lt;external_layout&gt; ;
&lt;contiguous_layout&gt; ::= {contiguous} // default
&lt;chunked_layout&gt; ::= {chunked &lt;dims&gt; }
&lt;dims&gt; ::= a list of integers represented of the form ( , , .. , )
&lt;compact_layout&gt; ::= TBD // not implemented yet
&lt;external_layout&gt; ::= {external &lt;external_file&gt;+ }
&lt;external_file&gt; ::= (&lt;file_name&gt; &lt;offset&gt; &lt;size&gt;)
&lt;offset&gt; ::= an integer
&lt;size&gt; ::= an integer
&lt;compression&gt; :: = compression { TBD } ; //algorithm name and properties?
&lt;dataset_attribute&gt; ::= &lt;attribute&gt; ;
&lt;link&gt; ::= &lt;hardlink&gt; | &lt;softlink&gt;
&lt;hardlink&gt; ::= hardlink "&lt;hardlink_name&gt;" { linktarget "&lt;target&gt;"; }
&lt;hardlink_name&gt; ::= &lt;identifier&gt;
&lt;target&gt; ::= &lt;identifier&gt;
&lt;softlink&gt; ::= softlink "&lt;softlink_name&gt;" { linktarget "&lt;target&gt;"; }
&lt;softlink_name&gt; ::= &lt;identifier&gt;
&lt;identifier&gt; ::= string
// character "/" should be used with care.
// &lt;dataset_name&gt; cannot contain "/", for example.
</PRE>
<HR>
<B> An example of HDF5 File in DDL </B>
<PRE>
Consider an HDF5 file, example.h5, with the following structure:
/
/ | \
group1 group3 dataset3
/ \ | ^
group2 dataset2 hardlink --|
/ \ ^
dataset1 softlink |
|------|
hdf5 "example" {
group "/" {
datatype "shared_datatype1" { // named data type
char name[20], address[40];
int32 id;
}
datatype "shared_datatype2" {
"shared_datatype1" rec[100];
float32 total;
}
dataspace "shared_dataspace1" { array (1000) (1000)} // named data space
attribute "group_attribute1" { //group attribute
datatype {int32};
dataspace "shared_dataspace1";
}
attribute "group_attribute2" {
datatype {int32};
dataspace {array (3,5) (3,5)} ;
data {(0,0,0,0,0) (1,1,1,1,1) (2,2,2,2,2)};
}
group "group1" {
attribute "group1_attribute" {
datatype {char};
dataspace {array (100) (200)};
data {("attribute_of_group_1")};
}
group "group2" {
datatype "shared_datatype3" {
int8 z;
int32 x;
}
dataset "dataset1" {
dataspace "shared_dataspace1";
datatype "shared_datatype3";
attribute "dataset1_attr" {
datatype {char};
dataspace {array (10) (10)};
data {("data1")};
};
}
softlink "sl" {
linktarget "/group1/dataset2";
}
}
dataset "dataset2" {
datatype {int8};
dataspace {array (2,5) (unlimited, 5)};
data {(0,0,0,0,0) (0,0,0,0,0)};
}
}
group "group3" {
hardlink "hl" {
linktarget "/dataset3" ;
}
}
dataset "dataset3" {
datatype { float32 };
dataspace {array (1000,1000) (1000,1000) };
storagelayout {chunked (250,1000)};
}
} // root group
}