Commit Graph

55 Commits

Author SHA1 Message Date
Giacomo Vercesi dbc734f23a CMake: force use of Python3
Switch from using `find_package(Python ...)` to `find_package(Python3
...)` as it excludes that CMake picks a Python 2 version by accident.
2026-05-14 11:26:41 +02:00
Giacomo Vercesi 18e3501731 pypeline: implement proper config parsing
Overhaul the pipeline configuration logic by collapsing all dynamic
configuration options, both for analyses and pipes into a single
dictionary. Change all the interfaces so that there is no longer
distinction between the configuration of an analysis and of pipes.
Expose these options to the command line via `--{name}-configuration`
options for each pipe/analysis that is applicable to the command-line
invocation.
2026-04-10 11:45:12 +02:00
Giacomo Vercesi 2e6193e77f test_pypeline_daemon: fix flaky SIGSEGV
Avoid triggering a flaky SIGSEGV caused by running the asyncio event
loop in a thread. Switch to using `Process` which avoids the issue
altogether.
2026-03-31 17:00:49 +02:00
Giacomo Vercesi 9b68cf4a39 pypeline-daemon: rework /artifact endpoint
Rework the artifact endpoint, making it similar to the CLI invocation.
It now works on a single artifact and can be made to return both `json`
and `tar` as formats.
2026-03-31 17:00:49 +02:00
Giacomo Vercesi 4c6436f7b2 pypeline: introduce artifact category
Add categories in the pipeline, allow an artifact to have a category
specified which allows it to be shown or hidden by default. Adapt the
CLI tools to hide artifacts of the category that don't
`show_by_default=True`.
2026-03-31 17:00:48 +02:00
Giacomo Vercesi 39f1a9c3a9 pypeline: drop underscores from pipeline yaml
Remove the only instance of underscores in the pipeline yaml from the
use of `analysis_lists`. Substitute it with `analysis-list`.
2026-03-31 17:00:48 +02:00
Giacomo Vercesi 550ac7b700 pypeline-daemon: add put-file endpoint
Add an endpoint to the daemon that allows adding a file to the file
storage.
2026-03-31 17:00:48 +02:00
Giacomo Vercesi 6dc053e55c pypeline-daemon: rework error propagation
Rework how errors are propagated from the daemon code to the HTTP
client, creating an exception hierarchy to handle the most common
cases.
2026-03-31 17:00:48 +02:00
Giacomo Vercesi 14d95a924d pypeline: rework fields in pipeline metadata
Rework which information is transmitted in the pipeline metadata,
avoiding redundancy and moving some information there instead of
returning it every time a request is made.
2026-03-31 17:00:48 +02:00
Giacomo Vercesi c53e615fa0 Cleanup pypeline code
Remove some cruft and apply trivial changes to the existing pypeline
code, especially on the daemon side.
2026-03-31 17:00:48 +02:00
Giacomo Vercesi 773245b893 pypeline/cli: use dataclass for ctx.obj
Add typing to the `ctx.obj` object used to store pypeline-specific data
while parsing the command-line in click.
2026-03-31 17:00:48 +02:00
Giacomo Vercesi ce60d8a1e5 pypeline: integrate model migration infrastructure
When calling `Model.deserialize` the model will be automatically be
migrated if it needs to.
2026-03-05 14:23:06 +01:00
Giacomo Vercesi f4d0573017 pypeline: StorageProvider works with Model
Change the `get_model` and `set_model` interface of `StorageProvider` so
that it is responsibility of the `StorageProvider` to
serialize/deserialize the model before returning to the caller.
This is in preparation to the model migration being implemented, since
it's now a responsibility of the storage provider to deserialize it it
can trivially re-save it to disk if it is migrated.
2026-03-05 14:23:06 +01:00
Giacomo Vercesi a6a3e15e6b pypeline: add env prefixes
Add the `PYPELINE_` prefix to all the pypeline environment variables.
2026-03-05 14:23:06 +01:00
Giacomo Vercesi 9287c51d27 pypeline: fix -C option
Instead of executing `chdir` when using the `-C` option, which is
fragile when other paths are involved, store the option in the click
context and propagate it to the required code paths that require knowing
what the base directory is.
2026-03-04 14:58:02 +01:00
Giacomo Vercesi e14d1cee6f Add native runner tests 2026-02-16 10:34:39 +01:00
Giacomo Vercesi df4f8b76ac pypeline: re-organize tests
Split the tests into two subdirectories: those that work in pure python
and those that use the C++ bindings.
2026-02-16 10:34:39 +01:00
Giacomo Vercesi f1878f1194 pypeline: drop TraceRunner
Drop the `pypeline-trace-run` executable, as its purposed has been
superseded by the combination of `--debug` and native pipe/analysis
runners.
2026-02-16 10:34:39 +01:00
Giacomo Vercesi d7e605fe63 pypeline: add debug mode
Add the functionality the the pipeline infrastructure and CLI to run
individual pipe and analyses as subcommands instead of in-process. This
allow better debuggability of individual pipes.
2026-01-29 15:00:12 +01:00
Giacomo Vercesi fd14580619 pipebox: improve signal handling
Improve the handling of signals by leveraging the `Py_AtExit`
functionality to trigger cleanup when the interpreter exits.
When receiving SIGINT handle it specially because exceptions in python
are only thrown when the interpreter is running and not when C code is.
2026-01-20 12:29:12 +01:00
Giacomo Vercesi 7f7f9571a8 pypeline: implement container discarding
Add infrastructure to pypeline that allows containers to be notified
when they are being used last, this allows two things:
* `Pipe`s eagerly clearing those containers once they are done reading
  their contents
* `ScheduledTask`s clearing those out at the end of their execution in
  case the pipe did not do it
This overall should improve memory usage as container no longer take up
memory if they are no longer used as part of a `Schedule`.
2025-12-19 10:29:28 +01:00
Giacomo Vercesi a2a7e6442b pypeline: move PipelineNode.run
Inline the body of `PipelineNode.run` into `ScheduledTask.run`, making
`PipelineNode` a pure data structure. This also allows eliding all the
arguments since all of them as properties of `ScheduledTask`.
2025-12-19 10:18:10 +01:00
Giacomo Vercesi 9338ac3486 pypeline: expand ScheduledTask's constructor
Move the arguments with a longer lifetime from the arguments of
`ScheduledTask`'s `run` function to its constructor.
2025-12-19 10:17:37 +01:00
Giacomo Vercesi 8af5e1d05c Implement revng2 quick artifact 2025-12-10 16:16:14 +01:00
Giacomo Vercesi 3cd011c146 Introduce CliftFunctionMixin
In equivalence with `LLVMFunctionMixin`, add a mixin that allows
writing a pipe that instead of defining the `runOnFunction` method it
defines the `runOnCliftFunction` which receives both the
`module::Function` and the `FunctionOp`.
2025-12-10 16:16:14 +01:00
Giacomo Vercesi 76d9eae64b Add test for custom invalidation
Add a test to the pypeline tests that checks that the pipeline actually
invalidates the objects communicated by the custom invalidation of a
pipe.
2025-11-26 16:13:28 +01:00
Giacomo Vercesi b1b5778822 Add infrastructure for custom invalidation
Add infrastructure to pypeline to allow for pipes to provide custom
invalidation, by implementing an `invalidate` function.
2025-11-26 16:13:28 +01:00
Giacomo Vercesi 2f7a055121 pypeline: convert pipe names to kebab-case 2025-11-26 16:13:28 +01:00
Giacomo Vercesi e3a47de0f1 Analysis: drop name argument in constructor
In the same vein as the `Pipe`'s counterpart, this feature is unused,
drop it for now.
2025-11-26 16:13:28 +01:00
Giacomo Vercesi 4a1edec018 get_registry: use name for selected classes
Use the `name` class attribute to build the registry when discovering
subclasses of `Pipe`, `Analysis` and `Container`.
2025-11-26 14:59:43 +01:00
Tommaso Fontana b7cf7d3266 Pypeline: add pype tests
Add some basic tests for the pype CLI command.
2025-11-24 16:41:15 +01:00
Tommaso Fontana 91299b0c18 Pypeline: change daemon pipeline format
Now the dameon's pipeline endpoint doesn't return just the parsed yaml
but it returns a format that's easier to use for the UI.
2025-11-24 16:41:15 +01:00
Tommaso Fontana 9f5ca81401 Pypeline: introduce analysis lists
Now in the pipeline.yml you can specify lists of analysis that are run
sequentially.
2025-11-24 16:41:15 +01:00
Giacomo Vercesi 57d9c3deb8 PipeboxCommon: add Access enum
Add an enum to explicitly specify the `TaskArgumentAccess` of a given
`Pipe`. This is needed because in some cases it's required to manually
specify the `READ` or `WRITE` access.
2025-11-17 10:04:13 +01:00
Giacomo Vercesi 1c07656448 pipebox.py: add initialize function
Add the `initialize` function to `pipebox.py`. This allows passing
command-line arguments to the pipebox.
2025-11-17 10:04:13 +01:00
Giacomo Vercesi 4f93c9e7ad Container: add mime type
Add the `MimeType` field to the C++ classes. Forward it to python via
nanobind.
2025-11-17 10:04:13 +01:00
Giacomo Vercesi 964abca598 pypeline: allow multiple root nodes
Change the `Pipeline` class and `pypeline_parser.py` to allow multiple
root nodes to be specified when creating a `Pipeline` object.
2025-11-17 10:04:12 +01:00
Giacomo Vercesi 32fd86d33a Pipe: drop name argument in constructor
Having a pipe's name different from its class name is useless, drop the
feature entirely.
2025-11-17 10:04:12 +01:00
Tommaso Fontana 229d12e538 Pypeline: add tests for the new daemon 2025-11-11 17:15:28 +01:00
Tommaso Fontana b38832e251 Pypeline: move epoch to storage_provider
Move the epoch inside the storage_provider, this allows us to make the
storage_provider_factory to return an async context manager.
This way we can simplify the locking for the daemon, but forces the
CLI project commands to deal with async code.
2025-11-11 17:15:28 +01:00
Tommaso Fontana 036411b441 Pypeline: add pipebox initialize and -- args
The pipebox now has to have an initialize function that is called with
all arguments passed to the CLI after --.
2025-11-11 17:15:28 +01:00
Tommaso Fontana a7f77b1704 Pypeline: propagate invalidated artifacts objects
Changed the API of storage_provider to treturn the invalidated object
ids and propagated this info to the CLI and daemon.
2025-11-11 17:15:28 +01:00
Tommaso Fontana 639593164a Pypeline: add autocomplete and refactor CLI
Now pype has an `autocomplete` command usable to enable autocompletion
and revng2 now is based on `pype`, modifying its defalults and
injecting new commands.
2025-11-11 17:15:27 +01:00
Tommaso Fontana 2daa9f938e Pypeline: add mime_type and model_name to Model
Now the model has a mime_type that we expect to return to the UI, and
model_name so the name of the model on disk can be impl-specific.
Moreover, now Model and Container use the same logic to detect if
a mimetype is textual.
2025-11-11 17:15:27 +01:00
Giacomo Vercesi b622fe8f19 Implement FileProvider interface
Implement the `FileProvider` interface, this allows pipes to request
files given a `FileRequest` (hash and optional size/name).
2025-10-16 17:48:45 +02:00
Giacomo Vercesi 0f7dc7047d pypeline.Container: fix signature of (de)serialize
Change the signature of `Container.serialize` and
`Container.deserialize` to use `Buffer` instead of `bytes`, as this
makes the functions compatible with the `revng::pypeline::Buffer`
classes.
2025-10-16 16:08:05 +02:00
Giacomo Vercesi ca0f312b1e Pipebox: remove const from ArrayRef
Remove the `const` from `ArrayRef<const char>` as the `ArrayRef` already
implies the underlying type will be accessed only in `const`.
2025-10-16 16:07:09 +02:00
Giacomo Vercesi d08c490993 pypeline::Model: expose children function
Expose the `revng::pypeline::Model.children` function to python, as it
was not previously exposed.
2025-10-16 12:51:55 +02:00
Giacomo Vercesi f165f99392 pypeline.Model: make deserialize to static
Change the `Model.deserialize` to a static function, which will return
a new instance of the Model.
2025-10-16 10:30:01 +02:00
Giacomo Vercesi bad8d0f622 Add test for pypeline's StorageProviders
Add a test that checks that the newly implemented invalidation logic
actually works and the objects stored as children or parent objectids
are also dropped when invalidating.
2025-10-15 17:43:09 +02:00