| Quick-Start Introduction to PDS Archiving |
 |
PDS Data Files and Labels
Content
Users of archived data need the data bits themselves, but they also need
information describing the format and content of the data files. PDS
data
are always accompanied by labels which provide this information. In
general each data file has its own label; a single label may be used
to reference several distinct files in cases where those files are
intimately related (chapters of a document, for example).
Labels may be attached
(stored as part of the data file) or detached (the label stands as a
file in its own right).
The label provides a field-by-field description of the contents of the
associated file; it may also provide information describing the
circumstances of the recorded observation (start time, for example) and
may provides links to other related PDS files, through various
identification fields. Labels are constructed so they can be read
by both people and machines.
PDS labels are written in the Object Description Language (ODL), as are
all
PDS catalog files.
ODL consists of a series of lines of the form "keyword =
value",
with certain keywords (for example,OBJECT) being used to delimit
structures within the label.
PDS prefers to use non-cryptic keywords, so that the data labels are more
easily interpreted by users browsing through them.
- More about ODL (ODL is defined and described
in full in
Chapter 12 of the
PDS Standards Reference, which is available for on-line browsing.
- Label file requirements are described in full in
Chapter 5 of the
PDS Standards Reference, which is available for on-line browsing.
Format
The keywords in the data label are generally laid out in the following
order:
- Standards Identifiers: one or two keywords defining
which version of PDS standards were observed during creation of
the label and data set
- File Characteristics: physical attributes of the data file,
such as whether the records are fixed-length or variable-length,
record length (if applicable), and number of records.
- Data Pointers: links to the start of the actual data, either
past the end of the label in the same file, or the name of another file
where the data reside
- Identification Parameters: numbers and codes used to link this
data file with other related data and with descriptions in the PDS
catalog.
These
include things like the target being observed, the mission, the
instrument recording the data, the identification number for the data
set, etc.
- Observational Parameters: filter names, pointing information, or
any other descriptive keywords relevant to the data
- Data Structure Definition: a definition of both the physical
organization and logical significance of the bytes in the data file
Comments, beginning with "/*", may be included in the label file as
desired for internal documentation and clarification.
Data Structure Definition
The physical and logical data structure are included in the PDS label using
the OBJECT and END_OBJECT keywords as delimiters. (The "object"
concept anticipates the development of software which can treat
a complex data structure as a single basic entity.) OBJECT
definitions may be nested where that is appropriate.
To promote uniformity of description and to simplify development of
general-purpose tools, PDS has defined several standard OBJECTs which should be
used to describe the structure of data files wherever that is possible.
It is nearly always preferable to reformat the data to conform with an
existing OBJECT definition than to develop a new OBJECT specific to the
pathology of a particular data set.
The defined PDS OBJECTS are reasonably flexible and generic, however, so any
reformatting required tends to be minimal for non-proprietary data formats.
The type of PDS OBJECT indicated by the OBJECT keyword defines the
general data structure (for example, data table vs. image). Within that
OBJECT definition, keywords are used to provide physical specifications,
such as field widths and byte locations, as well as logical
specifications,
such as units of measure or field names.
|
