1001.0039.txt

arXiv:1001.0039v1  [astro-ph.IM]  30 Dec 2009Astronomical Data Analysis Software and Systems XIX P45
ASP Conference Series, Vol. XXX, 2009
Y. Mizumoto, K.-I. Morita, and M. Ohishi, eds.
TGCat, The Chandra Transmission Grating Catalog and
Archive: Systems, Design and Accessibility
Arik W. Mitschang1, David P. Huenemoerder2, Joy S. Nichols1
Abstract. The recently released Chandra Transmission Grating Catalog and
Archive, TGCat, presents a fully dynamic on-line catalog allowing users to
browse and categorize Chandra gratings observations quickly and easily, gen-
erate custom plots of resulting response corrected spectra on- line without the
need for special software and to download analysis ready product s from multi-
ple observations in one convenient operation. TGCathas been registered as
a VO resource with the NVO providing direct access to the catalogs in terface.
The catalog is supported by a back-end designed to automatically fe tch newly
public data, process , archive and catalog them, At the same time ut ilizing an
advanced queue system integrated into the archive’s MySQL datab ase allowing
large processing projects to take advantage of an unlimited numbe r of CPUs
across a network for rapid completion. A unique feature of the cat alog is that
all of the high level functions used to retrieve inputs from the Chan dra archive
and to generate the final data products are available to the user in an ISIS writ-
ten library with detailed documentation. Here we present a structu ral overview
of the Systems, Design, and Accessibility features of the catalog a nd archive.
1. Introduction
TGCataims to be the definitive end-user source for all Chandra HETG S and
LETGS observations. In order to achieve this goal the catalo g must both have
the absolute best processed and calibrated data, as well as h ave an interface
which makes it easy for users to find, review and download thei r observations
of choice. The science requirements for accurately process ing gratings data are
discussed elsewhere ( Huenemoerder et. al. 2010 ). This writ ing will focus on
the automated system that collects and processes new public data, the interfaces
used for administrative review, and the interfaces and syst ems provided for user
access.
2. Database, Archive, and Subsystems
TheTGCatprocessing system is comprised of three major components:
•MySQL database storing meta-data
•File archive
•Processing Software
1Smithsonian Astrophysical Observatory, Cambridge, MA, US A
2MIT Kavli Institute for Space Research, Cambridge, MA
12 MITSCHANG, HUENEMOERDER, NICHOLS
Figure 1. TGCatdata flow
These components are described in the context of TGCatprocessing in this
section.
Tables: Thetables relevant toprocessingincludetheextractions, source, spec-
tral properties, files, and queue tables. The extractions ta ble has one entry per
processed extraction, where extraction is taken to mean a si ngle source in a
uniqueChandra observation ID (ObsID). Any one ObsIDcan have many sources
and any one source can be in many ObsIDs, the extractions tabl e will store one
entry for each combination thereof. In order to consolidate all extractions of a
single source, there is a source table indexed on SIMBAD1identifier, a TGCat
identifier, and coordinates, that associates entries in the extractions table. The
files table tracks processing output products and summary im ages allowing the
webpagestoeasily displayinformation onfilesavailable fo rdownload. Thespec-
tral properties table stores and indexes spectral properti es in several different
wavebands, these data are also available in a fits table for do wnload.
Data Flow: The flow of data through the system is rather simple, as illus-
trated in Figure 1. A bash script run via cronat regular intervals downloads a
list of gratings observations from the public chandra archi ve2and compares with
the list of ObsIDs that have been submitted to TGCat. Any not in TGCat
will be added to the queue table to be processed in line. Daemo n processes,
written in python, run on any number of network connected hos ts continuously
requesting entries from the queue in a FIFOmanner, the first process to ask
entries from the queue table will retrieve the first entry ( by time of creation
). They then parse the queue entry which can contain a number o f custom
processing parameters, setup work spaces and logging, fork off the processing
which is implemented in ISIS ( Houck & Denicola 2000 ) interac tive library
functions, which are available for download3and custom use, and finally ingest
1http://simbad.u-strasbg.fr/simbad
2http://cda.harvard.edu/chaser/
3http://space.mit.edu/cxc/analysis/tgcatTGCat 3
data into the database and archive. During the ingest step, s everal checks are
done to evaluate whether or not the resultant processed data are a complete
set and worth being manually reviewed by TGCatscientists. If so, meta data
are added to the database’s main extractions table which ret urns a unique iden-
tifier that is then used to tag the file based products. At this t ime linking is
done between the source table and extractions table. If not, the queue entry is
marked as an error and notification sent. TGCatoperators have the choice to
investigate the existing processing workspace or simply re -evaluate parameters
and re-queue the extraction as new.
Validation and Verification: Each extraction available for browsing has
been reviewed by a member of the TGCatscience team in order to confirm
zeroth order placement, proper masking, etc. This is done vi a an internal web-
site nearly mimicking the public interface for reviewing ex tractions but with
the addition of forms for queueing processing and for markin g the extraction as
“good”, “bad”, or “warning” and optional comment fields whic h are available
for review by the end user. “bad” extractions are never shown on the public
site and administrators have the option of rejecting any oth er extraction for
whatever reason. Because each extraction is tagged with the date of processing,
the version of TGCatused for processing, and a so called group ID which is
unique per group of extractions intended to be of the same obj ect for a single
ObsID, keeping track of accepted sources and avoiding dupli cates is made easy.
3. Access
TGCathas several interfaces for data access. Currently three are operational
includingthewebbrowserinterface, andtwoVirtualObserv atoryaccessprotocol
interfaces. Each oftheseaccess interfaces areimplemente dasa“plug-in”written
in php. Plug-in functions to create output are called by a cat alog independent
generalized query library written in php under TGCatdevelopment dubbed
“queryLib”. This modular approach makes it trivial to add ac cess interfaces to
TGCat.
queryLib: queryLib stores state and type information of an individual query
in a database table. Each time a query is performed, fields in t his table such as
awhereclause, sort fields, indexed IDs and columns are populated wi th infor-
mation specific to that query. As a user interacts with the par ticular interface,
information in this table entry are updated to reflect the new state of the query
through object method calls. queryLib initialized a query b y creating a new en-
try in the table with two unique IDs: one is the primary key ide ntifier assigned
at creation and can be used to reference the query, the other i s created on the fly
using a combination of the primary key and a random hash strin g that is very
unlikely to be guessed. This is done so that a malicious user c annot change the
state of anyone’s query by simply guessing an ID. The queryLi b speeds up the
process of redrawing a table of results by storing the indexe d primary keys of
all entries selected rather than having to rerun a potential ly complicated where
clause.4 MITSCHANG, HUENEMOERDER, NICHOLS
Web Catalog Interface: The web interface4toTGCatallows users browse
extraction and object entries in a way that is most suitable t o the catalog data
structure. Each extraction has an associated preview page w here a set of stan-
dard plots produced at the time of processing is available fo r viewing. A table
of spectral properties providing quick general categoriza tion, and a table of each
individual file available for the extraction along with a lin k to download that file
are provided. Perhaps one of the most important features of t he web interface
is the interactive plotting of individual or combined spect ra directly on the site.
This is implemented server side by taking the POST request fo r plotting param-
eters along with a unique file name, creating a small script of the commands for
loading data and plotting, and piping this script into an ISI S process. Malicious
useisprevented by checking all parameter inputsfor approp riatevalues, running
the ISIS process as an unprivileged user, and checking the te mp file name for
validity before piping. Since the file name is known at the tim e of the request,
the page simply needs to reload to show the new plot. The comma nds, any error
output, and an ASCII dump of the plot data are available for do wnload as well.
Virtual Observatory: TGCatis a registered VO service providing both
the Simple Cone Search5and Simple Image Access6Protocols. These are both
implemented as an XML output plug-ins taking input from a php script that
provides appropriate GET parameters, then creates a query o bject after parsing
the input data, finally running the query in exactly the same w ay as for the web
and ASCII interfaces. Error handling is done by the calling s cript and meta
data is provided by running a query with parameters known to h ave no catalog
entries ( indexed ID=0 ). VO queries, and any other type, can b e tracked in
the query table using the type column. In this way we can track statistics on
requests coming from services such as datascope7.
Data Package Downloads: Generally, users will want to download more
than one file at a time for analysis, such as such as spectra and responses or
productsfrom multiple extractions of the same target. To th is endTGCatruns
apackager processthatparsesaqueuetablemuchlikethepro cessingqueuetable
described above. Requested packages are added to the packag e queue table,
validated and read by the packager which then fetches data to a temp space
placing them in a directory hierarchy tagged with ObsID and TGCatID. The
entire hierarchy is then tarred, compressed and provided to the user along with
file checksums via HTTP.
Acknowledgements. We would like to thank Dan Dewey for extensive input
on data organization and interface layout, and Mike Nowak fo r ISIS plotting
routines and advice on the interactive plotting interface. This work is supported
4http://tgcat.mit.edu
5http://tgcat.mit.edu/tgCli.php?OUTPUT=V
6http://tgcat.mit.edu/tgSia.php
7http://heasarc.gsfc.nasa.gov/cgi-bin/vo/datascope/i nit.plTGCat 5
by the Chandra X-ray Center (CXC) NASA contract NAS8-03060. DPH was
supportedby NASA through the Smithsonian Astrophysical Ob servatory (SAO)
contract SV3-73016 for the Chandra X-Ray Center and Science Instruments.
References
Huenemoerder, et. al. 2010, ( in preparation )
Houck, J. C., Denicola, L. A. 2000, in ASP Conf. Ser. 216, ADASS IX, ed. N. Manset,
C. Veillet, & D. Crabtree (San Francisco: ASP), 591
Williams, Roy, Hanisch, Robert, Szalay, Alex, Plante, Ray 2009, Simple Cone Search
Version1.03,http://www.ivoa.net/Documents/REC/DAL/ConeSea rch-20080222.html
Tody, Doug & Plante, Ray 2009, Simple Image Access Specification Ve rsion 1.0,
http://www.ivoa.net/Documents/SIA/20091008/PR-SIA-1.0-20 091008.html