def tsplit_df(state, *columns):
"""Split of dataframe by columns (version of split_df expecting a first row with tags)"""
from liquer.parser import parse
state = qtsplit_df(state, *columns)
df = state.get().copy()
query_column = state.vars.get("query_column")
if query_column is None:
query_column = "query"
link_column = state.vars.get("link_column")
if link_column is None:
link_column = "link"
split_link_type = state.vars.get("split_link_type")
if split_link_type is None:
split_link_type = "url"
# df.loc[:,link_column] = [""]+[evaluate(encode(decode(q)+[["link",split_link_type]])).get() for q in list(df[query_column])[1:]]
df.loc[:, link_column] = [""] + [
evaluate(parse(q).with_action("link", split_link_type).encode()).get()
for q in list(df[query_column])[1:]
]
return state.with_data(df)
def split_df(state, *columns):
"""Split of dataframe by columns
Creates a dataframe with unique (combinations of) value from supplied columns and queries
to obtain the corresponding filtered dataframes from the original dataframe.
This behaves like qsplit_df, with two important differenced:
- each generated query is evaluated (and thus eventually cached)
- link is generated and put into link column (state variable link_column)
The split_link_type state variable is used to determine the link type; url by default.
"""
from liquer.parser import parse
state = qsplit_df(state, *columns)
df = state.get().copy()
query_column = state.vars.get("query_column")
if query_column is None:
query_column = "query"
link_column = state.vars.get("link_column")
if link_column is None:
link_column = "link"
split_link_type = state.vars.get("split_link_type")
if split_link_type is None:
split_link_type = "url"
# df.loc[:,link_column] = [evaluate(encode(decode(q)+[["link",split_link_type]])).get() for q in df[query_column]]
df.loc[:, link_column] = [
evaluate(parse(q).with_action("link", split_link_type).encode()).get()
for q in df[query_column]
]
return state.with_data(df)
def apply(self, query, description=None):
self.debug(f"APPLY {query}")
if self.parent_query in (None, "", "/"):
self.debug(f" no parent query in apply {query}")
return self.evaluate(query, description=description)
if isinstance(query, str):
query = parse(query)
if query.absolute:
self.debug(f" absolute link in apply {query}")
return self.evaluate(query, description=description)
tq = query.transform_query()
if tq is None:
raise Exception(
f"Only transform query supported in apply ({query} on {self.parent_query})"
)
q = (parse(self.parent_query) + tq).encode()
self.debug(f"apply {query} on {self.parent_query} yields {q}")
return self.evaluate(q, description=description)
def to_query(cls, query):
if query is None:
return "", Query()
if isinstance(query, str):
return query, parse(query)
elif isinstance(query, Query):
return query.encode(), query
else:
raise Exception(f"Unsupported query type: {type(query)}")
def metadata(self):
metadata = self._metadata.as_dict()
title = self.title
description = self.description
if title is None:
if self.raw_query is None:
title = ""
else:
p = parse(self.raw_query)
if title in ("", None):
title = p.filename() or ""
mimetype = self.mimetype
if mimetype is None:
if self.query is not None:
if self.query.extension() is None:
mimetype = "application/octet-stream"
else:
mimetype = mimetype_from_extension(self.query.extension())
message = self._metadata.message
if message in (None, ""):
log = self._metadata.get("log", [])
if len(log):
message = log[-1]["message"]
if message in (None, ""):
log = self._metadata.get("child_log", [])
if len(log):
message = log[-1]["message"]
metadata.update(
dict(
status=self.status.value,
title=title,
description=description,
mimetype=mimetype,
query=self.raw_query,
parent_query=self.parent_query,
argument_queries=self.argument_queries,
# log=self.log[:],
is_error=self.is_error,
direct_subqueries=self.direct_subqueries[:],
progress_indicators=self.progress_indicators[:],
child_progress_indicators=self.child_progress_indicators[:],
child_log=self.child_log,
message=message,
started=self.started,
updated=self.now(),
created=self.created,
caching=self.caching,
vars=dict(self.vars),
html_preview=self.html_preview,
side_effect=False,
))
return metadata
def resolve_recipe_definition(r, directory, metadata):
if type(r) == str:
try:
query = parse(r)
filename = query.filename()
return dict(type="query",
query=r,
CWD=directory,
filename=filename,
provides=[filename])
except:
metadata.warning(f"Can't resolve recipe '{r}'",
traceback=traceback.format_exc())
print(f"Can't resolve recipe '{r}'")
traceback.print_exc()
return None
elif isinstance(r, dict):
if r.get("type") in (None, "query") and "query" in r:
try:
query = parse(r["query"])
filename = r.get("filename", query.filename())
title = r.get("title", filename)
description = r.get("description",
f'Generated from query: {r["query"]}')
rkey = join_key(directory, filename)
return dict(type="query",
query=r["query"],
title=title,
description=description,
CWD=directory,
filename=filename,
provides=[filename])
except:
metadata.warning(f"Can't resolve query recipe",
traceback=traceback.format_exc())
traceback.print_exc()
else:
print(f"Unsupported recipe type: {type(r)}")
if "filename" in r and "provides" not in r:
r["provides"] = [r["filename"]]
return r
def make_execution_context(self, tmpdir, store, context):
import datafusion as daf
ctx = daf.ExecutionContext()
register = self.data.get("register", [])
store = store.root_store()
path = Path(tmpdir)
for query in register:
context.info(f"Register {query}")
try:
q = parse(query)
except:
context.warning(
f"Could not parse query '{query}' in parquet_sql recipe {self.recipe_name()}",
traceback=traceback.format_exc())
if q.is_resource_query():
key = q.resource_query().path()
if store.is_dir(key):
context.info(f"Registering directory {key}")
for k in store.listdir_keys(key):
if not store.is_dir(k) and key_extension(
k) == "parquet":
(path / key_name(k)).write_bytes(
store.get_bytes(k))
context.info(
f"Registering {key_name_without_extension(k)} from {key}"
)
ctx.register_parquet(key_name_without_extension(k),
str(path / key_name(k)))
else:
(path / key_name(key)).write_bytes(store.get_bytes(key))
context.info(f"Registering resource {key}")
ctx.register_parquet(key_name_without_extension(key),
str(path / key_name(key)))
else:
filename = q.filename()
if filename is None:
context.warning(
f"Skipping '{query}' registering because it is lacking a filename"
)
continue
v = filename.split(".")
context.info(f"Evaluating query {query}")
context.evaluate_and_save(query,
target_directory=str(tmpdir),
target_file=filename)
context.info(f"Registering {v[0]} from query {query}")
ctx.register_parquet(v[0], str(path / filename))
return ctx
def queries_status(include_ready=False, reduce=True):
import liquer.parser as lp
import traceback
try:
cache = get_cache()
data = []
for key in sorted(cache.keys()):
metadata = cache.get_metadata(key)
if metadata is None:
continue
progress = metadata.get("progress_indicators", []) + metadata.get(
"child_progress_indicators", [])
d = dict(
query=key,
short=lp.parse(key).short(),
status=metadata.get("status", "none"),
updated=metadata.get("updated", "?"),
message=metadata.get("message", ""),
progress=progress[:3],
)
if include_ready or d["status"] != Status.READY.value:
data.append((key, d))
data = [x[1] for x in sorted(data)]
if reduce and len(data):
reduced_data = [data[0]]
for d, next_d in zip(data[1:], data[2:]):
previous_d = reduced_data[-1]
if not (previous_d["status"] == Status.EVALUATING_PARENT.value
and d["status"] == Status.EVALUATING_PARENT.value
and d["query"].startswith(previous_d["query"])
and next_d["query"].startswith(d["query"])):
reduced_data.append(d)
reduced_data.append(data[-1])
data = reduced_data
return data
except:
return [
dict(
query="",
status="not available",
updated="",
message=traceback.format_exc(),
progress=[],
)
]
def dr(state, type_identifier=None, extension=None, context=None):
"""Decode resource
Decodes the bytes into a data structure. This is meant to be used in connection to a resource query.
Resource part of the query will typically fetch the data from a store and thus return bytes (together with metadata).
Command dr will convert the bytes (assuming proper metadata are provided) into a data structure.
The metadata must contain type_identifier in metadata or metadata['resource_metadata'], a filename with extension
or extension with known decoding.
"""
from liquer.state_types import state_types_registry
from liquer.parser import parse
if state.data is None:
context.error(
f"Bytes expected, None received in dr from {state.query}")
return
if type_identifier is None:
type_identifier = state.metadata.get(
"type_identifier",
state.metadata.get("resource_metadata", {}).get("type_identifier"),
)
if type_identifier in (None, "bytes"):
type_identifier = state.metadata.get("resource_metadata",
{}).get("type_identifier")
if extension is None:
extension = state.metadata.get("extension")
if extension is None:
query = state.metadata.get("query")
if query is not None:
filename = parse(query).filename()
if filename is not None:
v = filename.split(".")
if len(v) > 1:
extension = v[-1]
context.info(f"Extension: {extension} - from query '{query}'")
else:
key = state.metadata.get("resource_metadata", {}).get("key")
if key is not None:
filename = context.store().key_name(key)
v = filename.split(".")
if len(v) > 1:
extension = v[-1]
context.info(f"Extension: {extension} - from key '{key}'")
if type_identifier in (None, "bytes"):
type_identifier = type_identifier_from_extension(extension)
context.info(
f"Type identifier: {type_identifier} - from extension '{extension}'"
)
if type_identifier is not None:
if extension in ("parquet", "xlsx", "csv",
"tsv") and type_identifier in ("generic",
"dictionary",
"pickle"):
context.warning(
f"Type identifier '{type_identifier}' seems to be inconsistent with the extension '{extension}'"
)
context.warning(
f"This might indicate a problem with executing the partent query '{context.parent_query}'"
)
type_identifier = type_identifier_by_extension.get(extension)
context.warning(
f"To fix the inconsistency, type identifier: {type_identifier} is used from extension '{extension}'"
)
context.info(
f"Type identifier: {type_identifier}, Extension: {extension}")
t = state_types_registry().get(type_identifier)
data = t.from_bytes(state.data, extension=extension)
return state.with_data(data)
else:
context.error(f"Decode resource (dr) command failed")
raise Exception(
f"Failed to resolve type for query {state.metadata.get('query')}")
return state
def update_recipes(self):
import yaml
recipes = {}
for key in self.substore.keys():
spec = None
if self.key_name(
key
) == self.RECIPES_FILE and not self.substore.is_dir(key):
spec = yaml.load(self.substore.get_bytes(key), Loader=Loader)
recipes_key = key
if spec is not None:
parent = self.parent_key(key)
for directory, items in spec.items():
for r in items:
if type(r) == str:
try:
query = parse(r)
filename = query.filename()
parent = self.parent_key(key)
if len(parent) > 0 and not parent.endswith(
"/"):
parent += "/"
rkey = (f"{parent}{filename}"
if directory == self.LOCAL_RECIPES else
f"{parent}{directory}/{filename}")
recipes[rkey] = r
self.recipes_info[rkey] = dict(
query=r,
title=filename,
description="",
recipes_key=recipes_key,
recipes_directory=directory,
)
except:
traceback.print_exc()
elif isinstance(r, dict):
try:
query = parse(r["query"])
filename = r.get("filename", query.filename())
title = r.get("title", filename)
description = r.get("description", r["query"])
parent = self.parent_key(key)
if len(parent) > 0 and not parent.endswith(
"/"):
parent += "/"
rkey = (f"{parent}{filename}"
if directory == self.LOCAL_RECIPES else
f"{parent}{directory}/{filename}")
recipes[rkey] = r["query"]
self.recipes_info[rkey] = dict(
query=r["query"],
title=title,
description=description,
recipes_key=recipes_key,
recipes_directory=directory,
)
except:
traceback.print_exc()
else:
print(f"Unsupported recipe type: {type(r)}")
self._recipes = recipes
return recipes
