def _save_df(cls, df: pd.DataFrame) -> DPTmpFile:
fn = DPTmpFile(ArrowFormat.ext)
df = to_df(df)
process_df(df)
ArrowFormat.save_file(fn.name, df)
log.debug(f"Saved df to {fn} ({os.path.getsize(fn.file)} bytes)")
return fn
def run(
name: str,
parameter: Tuple[str],
cache: bool,
wait: bool,
owner: str,
show_output: bool,
):
"""Run a report"""
params = process_cmd_param_vals(parameter)
log.info(f"Running script with parameters {params}")
script = api.Script.get(name, owner=owner)
with api_error_handler("Error running script"):
r = script.run(parameters=params, cache=cache)
if wait:
with click_spinner.spinner():
while not r.is_complete():
time.sleep(2)
r.refresh()
log.debug(f"Run completed with status {r.status}")
if show_output:
click.echo(r.output)
if r.status == "SUCCESS":
if r.result:
success_msg(f"Script result - '{r.result}'")
if r.report:
report = api.Report.by_id(r.report)
success_msg(f"Report generated at {report.web_url}")
else:
failure_msg(
f"Script run failed/cancelled\n{r.error_msg}: {r.error_detail}"
)
else:
success_msg(f"Script run started, view at {script.web_url}")
def setup_script(s: api.Script, env_dir: Path):
"""Setup the script - unpack & install deps"""
# TODO - add local cache check here
if env_dir.exists():
log.debug("Package already exists, not redownloading")
return None
# download and unpack bundle locally into env_dir
sdist = s.download_pkg()
assert tarfile.is_tarfile(sdist), "Invalid sdist file"
shutil.unpack_archive(sdist, extract_dir=env_dir, format="gztar")
sdist.unlink()
comp_r = compileall.compile_dir(env_dir, force=True, workers=1, quiet=1)
if not comp_r:
log.warning("Compiling script bundle failed - errors may occur")
# install deps
if s.requirements:
pip_args = [sys.executable, "-m", "pip", "install"]
if os.getuid() != 0 and not in_venv():
# we're a normal/non-root user outside a venv
pip_args.append("--user")
pip_args.extend(s.requirements)
log.debug(f"Calling pip as '{pip_args}'")
subprocess.run(args=pip_args, check=True)
importlib.invalidate_caches() # ensure new packages are detected
log.info(f"Successfully installed bundle for script {s.id}")
def _gen_report(self, embedded: bool, title: str,
headline: str) -> t.Tuple[str, t.List[Path]]:
"""Build XML report document"""
# convert Blocks to XML
s = BuilderState(embedded)
_s = self.top_block.to_xml(s)
assert len(_s.elements) == 1
# add main structure and Meta
report_doc: Element = E.Report(
E.Meta(
E.Author("Anonymous"), # TODO - get username from config?
E.CreatedOn(timestamp()),
E.Title(title),
E.Headline(headline),
),
E.Main(*_s.elements),
version="1",
)
report_doc.set("{http://www.w3.org/XML/1998/namespace}id",
f"_{uuid.uuid4().hex}")
# post_process and validate
processed_report_doc = local_post_transform(
report_doc, embedded="true()" if embedded else "false()")
validate_report_doc(xml_doc=processed_report_doc)
# convert to string
report_str = etree.tounicode(processed_report_doc, pretty_print=True)
log.debug("Built Report")
log.info(report_str)
return (report_str, _s.attachments)
def run_api(run_config: RunnerConfig) -> RunResult:
"""Bootstrap the recursive calls into run"""
script = api.Script.by_id(run_config.script_id)
# is the script compatible with the client runner/api
if not is_version_compatible(
__version__, script.api_version, raise_exception=False):
log.warning(
f"Script developed for an older version of Datapane ({script.api_version}) - "
+ "this run may fail, please update.")
# TODO - we should pull param defaults from script and add in the call
script.call(run_config.env, **run_config.format())
# create the RunResult
script_result = str(api.Result.get()) if api.Result.exists() else None
report_id = None
try:
report = api._report.pop()
log.debug(f"Returning report id {report.id}")
report_id = report.id
except IndexError:
log.debug(
"User script didn't generate report - perhaps result / action only"
)
return RunResult(report_id=report_id, script_result=script_result)
def is_version_compatible(
provider_v_in: Union[str, v.Version],
consumer_v_in: Union[str, v.Version],
raise_exception: bool = True,
) -> bool:
"""
Check provider supports consumer and throws exception if not
Set the spec so that the consumer has to be within a micro/patch release of the provider
NOTE - this isn't semver - breaks when have > v1 release as then treats minor as breaking,
e.g. 2.2.5 is not compat with 2.1.5
"""
consumer_v = v.Version(consumer_v_in) if isinstance(consumer_v_in, str) else consumer_v_in
provider_v = v.Version(provider_v_in) if isinstance(provider_v_in, str) else provider_v_in
provider_spec = SpecifierSet(f"~={provider_v.major}.{provider_v.minor}.0")
log.debug(f"Provider spec {provider_spec}, Consumer version {consumer_v}")
if consumer_v not in provider_spec:
if raise_exception:
raise VersionMismatch(
f"Consumer ({consumer_v}) and Provider ({provider_spec}) API versions not compatible"
)
return False
return True
def script_env(env_dir: Path, env: SSDict) -> t.ContextManager[None]:
"""
Change the local dir and add to site-path so relative files and imports work
TODO
- this is NOT thread-safe - unlikely we can run multiple concurrent scripts atm
- this doesn't save env's as a call-stack directly - however handled implictly via Python stack anyway
"""
cwd = os.getcwd()
log.debug(f"[cd] {cwd} -> {env_dir}")
if not env_dir.exists():
env_dir.mkdir(parents=True)
full_env_dir = str(env_dir.resolve())
sys.path.insert(0, full_env_dir)
os.chdir(full_env_dir)
os.environ.update(env)
try:
yield
finally:
for env_key in env.keys():
with suppress(KeyError):
os.environ.pop(env_key)
try:
sys.path.remove(full_env_dir)
except ValueError as e:
raise CodeError(
"sys.path not as expected - was it modified?") from e
os.chdir(cwd)
log.debug(f"[cd] {cwd} <- {env_dir}")
def _save_obj(cls, data: t.Any, is_json: bool) -> DPTmpFile:
# import here as a very slow module due to nested imports
from ..files import save
fn = save(data, default_to_json=is_json)
log.debug(f"Saved object to {fn} ({os.path.getsize(fn.file)} bytes)")
return fn
def load_defaults(self, config_fn: NPath = DATAPANE_YAML) -> None:
if not by_datapane:
log.debug(f"loading parameter defaults from {config_fn}")
# TODO - move dp-server parameter handling into common to use runnerconfig.format
# NOTE - this is a bit hacky as we don't do any type formatting
cfg = DatapaneCfg.create_initial(config_file=Path(config_fn))
defaults = {p["name"]: p["default"] for p in cfg.parameters if "default" in p}
self.update(defaults)
else:
log.debug("Ignoring call to load_defaults as by datapane")
def setup_api(dp_token: str,
dp_host: str,
debug: bool = False,
logs: TextIO = None):
"""Init the Datapane API for automated use"""
# login, ping, and create the default env file for CMD usage
api.login(token=dp_token, server=dp_host, cli_login=False)
# setup input and config, logging, login, etc.
verbosity = 2 if debug else 0
_setup_dp_logging(verbosity=verbosity, logs_stream=logs)
log.debug("Running DP on DP")
def setup_api(dp_host: str,
dp_token: str,
debug: bool = False,
logs: TextIO = None):
"""Init the Datapane API for automated use"""
# setup input and config, logging, login, etc.
config = c.Config(server=dp_host, token=dp_token, analytics=False)
verbosity = 2 if debug else 0
_setup_dp_logging(verbosity=verbosity, logs_stream=logs)
c.init(config=config)
# check can login/ping
api.ping(config=config)
log.debug("Running DP on DP")
def _gen_report(self,
embedded: bool,
title: str,
description: str = "Description",
author: str = "Anonymous") -> t.Tuple[str, t.List[Path]]:
"""Build XML report document"""
# convert Pages to XML
s = BuilderState(embedded)
_s = reduce(lambda _s, p: p._to_xml(_s), self.pages, s)
# add main structure and Meta
report_doc: Element = E.Report(
E.Meta(
E.Author(author), # TODO - get username from config?
E.CreatedOn(timestamp()),
E.Title(title),
E.Description(description),
),
E.Main(*_s.elements, type=self.report_type.value),
version="1",
)
report_doc.set("{http://www.w3.org/XML/1998/namespace}id",
f"_{uuid.uuid4().hex}")
# post_process and validate
processed_report_doc = local_post_transform(
report_doc, embedded="true()" if embedded else "false()")
validate_report_doc(xml_doc=processed_report_doc)
# check for any unsupported local features, e.g. DataTable
# NOTE - we could eventually have different validators for local and published reports
if embedded:
uses_datatable: bool = processed_report_doc.xpath(
"boolean(/Report/Main//DataTable)")
if uses_datatable:
raise UnsupportedFeature(
"DataTable component not supported when saving locally, please publish to a Datapane Server or use dp.Table"
)
# convert to string
report_str = etree.tounicode(processed_report_doc, pretty_print=True)
log.debug("Built Report")
log.debug(report_str)
return (report_str, _s.attachments)
def __init__(self,
*arg_blocks: BlockOrPrimitive,
blocks: t.Union[BlockDict, BlockList] = None,
**kw_blocks: BlockOrPrimitive):
"""
Blocks can be created with the `name` parameter, if not set, one can be provided here using keyword args.
Use the blocks dict parameter to add a dynamically generated set of named blocks, useful when working in Jupyter
Args:
*arg_blocks: List of blocks to add to document - if a name is not present it will be auto-generated
blocks: Allows providing the document blocks as a single dictionary of named blocks
**kw_blocks: Keyword argument set of blocks, whose block name will be that given in the keyword
Returns:
A `TextReport` object containing assets that can be uploaded for use with your online TextReport
.. tip:: Blocks can be passed using either arg parameters or the `blocks` kwarg as a dictionary, e.g.
`dp.TextReport(my_plot=plot, my_table=table)` or `dp.TextReport(blocks={"my_plot": plot, "my_table":table})`
.. tip:: Create a dictionary first to hold your blocks to edit them dynmically, for instance when using Jupyter, and use the `blocks` parameter
"""
super().__init__()
# set the blocks
def _conv_block(name: str, block: BlockOrPrimitive) -> Block:
x = wrap_block(block)
x._set_name(name)
return x
_blocks: BlockList
if isinstance(blocks, dict):
_blocks = [_conv_block(k, v) for (k, v) in blocks.items()]
elif isinstance(blocks, list):
_blocks = blocks
else:
# use arg and kw blocks
_blocks = list(arg_blocks)
_blocks.extend([_conv_block(k, v) for (k, v) in kw_blocks.items()])
if not _blocks:
log.debug("No blocks provided - creating empty report")
# set the pages
self.pages = [Page(blocks=[Group(blocks=_blocks, name="top-group")])]
def deploy(name: Optional[str], script: Optional[str], config: Optional[str],
visibility: str):
"""Package and deploy a Python script or Jupyter notebook as a Datapane Script bundle"""
script = script and Path(script)
config = config and Path(config)
init_kwargs = dict(visibility=visibility,
name=name,
script=script,
config_file=config)
kwargs = {k: v for k, v in init_kwargs.items() if v is not None}
# if not (script or config or sc.DatapaneCfg.exists()):
# raise AssertationError(f"Not valid project dir")
dp_cfg = scripts.DatapaneCfg.create_initial(**kwargs)
log.debug(f"Packaging and uploading Datapane project {dp_cfg.name}")
# start the build process
with scripts.build_bundle(dp_cfg) as sdist:
if EXTRA_OUT:
tf: tarfile.TarFile
log.debug("Bundle from following files:")
with tarfile.open(sdist) as tf:
for n in tf.getnames():
log.debug(f" {n}")
r: api.Script = api.Script.upload_pkg(sdist, dp_cfg)
success_msg(
f"Uploaded {click.format_filename(str(dp_cfg.script))} to {r.web_url}"
)
def _gen_report(
self,
embedded: bool,
title: str = "Title",
description: str = "Description",
author: str = "Anonymous",
) -> t.Tuple[str, t.List[Path]]:
"""Generate a report for saving/uploading"""
report_doc, attachments = self._to_xml(embedded, title, description,
author)
# post_process and validate
processed_report_doc = local_post_transform(
report_doc, embedded="true()" if embedded else "false()")
validate_report_doc(xml_doc=processed_report_doc)
self._report_status_checks(processed_report_doc, embedded)
# convert to string
report_str = etree.tounicode(processed_report_doc)
log.debug("Successfully Built Report")
# log.debug(report_str)
return (report_str, attachments)
def create_initial(cls,
config_file: Path = None,
script: Path = None,
**kw) -> "DatapaneCfg":
raw_config = {}
if config_file:
assert config_file.exists()
else:
config_file = DATAPANE_YAML
if config_file.exists():
# read config from the yaml file
log.debug(f"Reading datapane config file at {config_file}")
with config_file.open("r") as f:
raw_config = yaml.safe_load(f)
elif PYPROJECT_TOML.exists():
# TODO - implement pyproject parsing
log.warning(
"pyproject.toml found but not currently supported - ignoring")
raw_config = {}
elif script:
# we don't have a default config - perhaps in the script file
# TODO - try read config from source-code
abs_script = config_file.parent / script
if script.suffix == ".ipynb":
log.debug("Converting notebook")
mod_code = extract_py_notebook(abs_script)
else:
mod_code = abs_script.read_text()
log.debug("Reading config from python script/notebook")
log.debug(mod_code)
# overwrite config with command-line options
if script:
raw_config.update(script=script)
raw_config.update(kw)
readme = config_file.parent / "README.md"
if readme.exists():
raw_config["description"] = readme.read_text()
elif "description" not in raw_config:
raw_config["description"] = cls.description
dp_cfg = dacite.from_dict(cls,
data=raw_config,
config=dacite.Config(cast=[Path]))
return dp_cfg
def __getattr__(self, attr):
if self.has_dto and not attr.startswith("__"):
log.debug(f"Proxying '{attr}' lookup to DTO")
return getattr(self._dto, attr)
# Default behaviour
return self.__getattribute__(attr)
def save(obj: Any, default_to_json: bool = False) -> DPTmpFile:
fn = get_wrapper(obj, default_to_json=default_to_json,
error_msg=None).write(obj)
log.debug(f"Saved object to {fn} ({os.path.getsize(fn.file)} bytes)")
return fn
def refresh(self):
"""Update the local representation of the object"""
self.dto = self.res.get()
log.debug(f"Refreshed {self.url}")
def delete(self):
self.res.delete()
log.debug(f"Deleted object {self.url}")
def update(self, **kwargs):
# filter None values
kwargs = {k: v for (k, v) in kwargs.items() if v is not None}
self.res.patch(**kwargs)
self.refresh()
log.debug(f"Updated object {self.url}")
