The disk free/total values are optional ones, wrapped in a Maybe, so
we shouldn't directly serialise them. In order to simplify the
embedded extraction, we add a small helper function.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Agata Murawska <agatamurawska@google.com>

We need to only query the default (first enabled) hypervisor, not all
hypervisors. For this, we need to add a manual check to ensure that we
don't have a corrupt config (there's no "NonEmptyList" type…).

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

The String parameter to 'nodeLiveFieldExtract' is the query2 field
name, not the RPC-layer field name. Grrr for not having a real data
type for this.

Furthermore, we add some safety check that we don't return JSNull via
rsNormal…

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Agata Murawska <agatamurawska@google.com>

This patch makes regex-pcre optional, allowing its disable via a
preprocessor define NO_REGEX_PCRE. This define will be added to
config.ac/Makefile.am in the next patch.

The patch also changes multi-line strings into string concatenation,
due to limitations with the CPP language extension.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

This patch adds a NiceSort equivalent and the corresponding unittest
(partially copied from Python unittest). The difference between the
Python version and this one is that this implementation doesn't use
regular expressions, and as such it doesn't have the 8-groups
limitation.

The key-based version is separate from the non-key one (since we don't
have default arguments in Haskell), and is tested less in its absolute
properties but only that it is identical to the non-key version under
some transformations (the non-key version is much more tested).

This will be needed later in query name sorting.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

This allows the command completion to treat them as separate option
types; this is not useful in build-bash-completion right now, but
might become so in the future.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

This defines the arguments supported and then modifies the
--help-completion output to include them too.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

Beside parsing the arguments as well, we also change the serialisation
format for choices/suggests, to make the Python code simpler.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

This extends the current option-only support with support for
arguments too. They are serialised without a without the prefix (so
just with the type) and the min/max ranges (max="none" for unlimited).

A few new option types have been added as well.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

This was missing so far…

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

This is, I believe, the last non-htools specific file that still lived
in the htools directory; it's already widely used in non-htools code,
so let's move it before we add more functionality to this module.

All changes are related to the name change, imports fixup, etc.; there
are no other changes in this patch.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

unsafePerformIO is required to go from the IO monad to pure code.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Since ERpcError is still not a monad, a simple foldr will have to
do for now.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

parseHttpResponse is such a simple function that we can actually
do the entire unpacking in rpcResultParse.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

For now ERpcError is still not a monad, but we can still simplify
rpcResultFill implementation for different calls simply by abstracting
it into two helpers.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

There is no need (at least right now) for some of the resulting
types to be in additional monad, so let us just have more
"normal" functions.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

As per Iustin's suggestion, this patch aligns parts of typeclass
instance on "=".

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

As per Iustin's suggestion, show is not a good solution - so instead
the name is now explainRpcError.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This patch implements single instance info call - somewhat similar to
all_instances_info, except we give a specific instance name.

Current implementation of reading the InstanceInfo value is somewhat
counter-intuitive because when we query a node on thich there is
no instance with such name, this is not considered an error - but
it does produce an empty answer. Therefore the resulting type is
optional.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

For storage_list call, the result type depends on the call parameters.
Therefore, we have to add call as an argument for rpcResultFill - and
by extension, to the typeclass.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Implementation of node version query.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Added support for NodeInfo RPC call as source of data for node query.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

We make use of the parameter added in the previous patch and add
option to add live parameters to the query.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

The tests we currently have assume, that all the data required for
running the query is available - once we add live data, this will no
longer be the case.

This patch adds boolean parameter to query function, which tells it
whether to ignore live parameters gathering.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Simple conversion for queries that use live fields gathered from
RPC calls.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This is a helper function that gets the JSON encoding of the first
element of the list (if possible).

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This patch creates ResultT, a monad transformation used later in
RPC <-> query integration.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Implementation of nodegroup queries in Haskell. This is not yet
complete as we are missing merged disk parameters and option
want_diskparams is not implemented.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

For some results in nodegroup queries we need agregation similar to
one that is done for nodes.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Previous version of RPC calls implementation in Haskell did not take
into account that the actual result type for queries is a list, not
a dictionary.

This patch aims at fixing the problem "for now" - it is not a pretty
solution, but it does work. Note that parsing of the result is now
split into two parts - first, we check if server's aswer is positive,
then if it is, we procede with decoding the actual result.

Values and order of some fields in the result type were changed to
reflect actual order of arguments from server responses.

AllInstancesInfo call was particularly tricky, because it returns a
dictionary where keys are instance names - and the response from
a given node is correct if all the instances were deserialized, not
just some.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This was repeated a lot, so we should abstract this into a type.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Curl error messages are self contained and don't need "error" prefix;
added RpcResult Error which will be used once proper deserialization
of rpc response is in place.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

We now have Hypervisor type in Objects, might just as well use it.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

I assumed this was a copy-paste+forgetting to change the header comment.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This patch adds missing Path entry and uses it in Rpc. It was not in
place before due to different build settings - i.e. if curl was
disabled, there were no problems.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This patch adds support for a --help-completion option, which will
display the defined options and their completion information, in a
format designed to be parsed easily from Python, for integration into
build-bash-completion.

The alternative would have been to duplicate the build-bash-completion
functionality on the Haskell side, but that would have been lots of
duplication for not enough gain (since that is only run at build
time).

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Agata Murawska <agatamurawska@google.com>

This is a quite boring patch, just adding annotation information to
all existing options. Some of the annotations are not very good; but
we don't have support for more precise completion in
build-bash-completion, so this is good enough.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Agata Murawska <agatamurawska@google.com>

This can be used for either option or argument completion (although
OptComplNone doesn't make sense for an argument :).

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Agata Murawska <agatamurawska@google.com>

Currently, we test and require that each individual program (hbal,
etc.) defines/supports the generic options (currently --help and
--version). Even with the test, this is not optimal, since it requires
changes in many places whenever we modify the list of generic options,
hence we move these out of the individual programs and into the
generic CLI handling code.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Agata Murawska <agatamurawska@google.com>