taplint: Tests TAP services
taplint runs a series of tests on a Table Access Protocol
(TAP)
service and reports the results. Unlike most of the other tools in
this package it is not likely to be of use to normal users;
its intended use is for people developing or operating TAP services
to assess their services, perhaps with a view to improving compliance.
Testing takes place in a number of stages; it is possible to choose
which stages are run in by using the stages parameter.
The default output (format=text)
is line-based text to standard output,
and each report line is of the (fairly greppable) form:
T-SSS-MMMMxN aaaaa...where the parts have the following meanings:
T: Report type,
one of E(rror), W(arning), I(nfo), S(ummary), F(ailure).
See the documentation of the report parameter for
further description of what these mean.
The report parameter can be used to suppress some of these;
only E indicates actual service compliance errors, but
including the others may make it easier to see what's going on.
SSS: Stage abbreviation, as used in the stages
parameter. The stages parameter can be used to select
which stages are run.
MMMM: Message label, which is always the same for
messages generated by the same test, is usually different for
messages generated by different tests, and may be somewhat mnemonic.
x: Continuation indicator,
either "-" or "+".
In most cases it is "-", indicating the first line of
a message, but multi-line messages (rare) use "-" for the first
line and "+" for any continuation lines.
N: Sequence number, which is 1 for the first time
message T-SSS-MMMM is reported, and increases by one
for each subsequent appearance.
After a certain maximum (determined by the maxrepeat
parameter) additional reports with the same code are no longer output
individually, but a summary of the number of reports so discarded
is written at the end of the section with the character "x"
instead of the sequence number.
This behaviour prevents the output being swamped by
multiple reports of the same issue.
If the maxrepeat parameter is increased above 9, more than
one digit will be used here (so e.g. for maxrepeat=999, the format would
be NNN not N).
aaaaa...: Message text, a free text description of
what is being reported.
If you don't like that format, others may be selected using the
format parameter, which currently also supports JSON.
For more flexible interaction with the output you can invoke
taplint programmatically
and supply your own
OutputReporter
instance.
TAP is a complicated beast, referencing many standards
(including
TAP,
UWS,
VODataService,
ADQL,
VOResource,
VOSI,
TAPRegExt,
DALI,
ObsCore,
ObsLocTAP,
EPN-TAP,
VOTable,
UCD,
VOUnits,
SSO,
SoftID,
HTTP,
RDFa Lite),
and it is hard to write a validator which is comprehensive, especially
one which can provide useful output for services with a range of
compliance levels.
This tool tries to make a wide range of tests, but does not claim to
be comprehensive. An idea of what tests it does perform can be gained
from the stages listed in the description of the stages
parameter. It does make a fairly good job of checking that declared
metadata is consistent and matches the data actually returned from queries,
it tests job submission in most of the various ways permitted by the TAP
standard, and it checks all returned VOTables by effectively running them
through votlint.
Things it does not test much include complex ADQL queries,
coordinate/STC-related data types, queries in non-ADQL languages,
and service registration.
HTTP connections made by this validator are flagged in the
User-Agent field with the token
"(IVOA-test)".