def check_shapes(pfeed, *, as_df=False, include_warnings=False):
"""
Analog of :func:`check_frequencies` for ``pfeed.shapes``
"""
table = "shapes"
problems = []
# Preliminary checks
if pfeed.shapes is None:
return problems
f = pfeed.shapes.copy()
problems = check_for_required_columns(problems, table, f)
if problems:
return gk.format_problems(problems, as_df=as_df)
if include_warnings:
problems = check_for_invalid_columns(problems, table, f)
# Check shape_id
problems = gk.check_column(problems, table, f, "shape_id", gk.valid_str)
# Check geometry
v = lambda x: isinstance(x, sg.LineString) and not x.is_empty
problems = gk.check_column(problems, table, f, "geometry", v)
return gk.format_problems(problems, as_df=as_df)
def check_meta(pfeed, *, as_df=False, include_warnings=False):
"""
Analog of :func:`check_frequencies` for ``pfeed.meta``
"""
table = "meta"
problems = []
# Preliminary checks
if pfeed.meta is None:
problems.append(["error", "Missing table", table, []])
else:
f = pfeed.meta.copy()
problems = check_for_required_columns(problems, table, f)
if problems:
return gk.format_problems(problems, as_df=as_df)
if include_warnings:
problems = check_for_invalid_columns(problems, table, f)
if f.shape[0] > 1:
problems.append([
"error", "Meta must have only one row", table,
list(range(1, f.shape[0]))
])
# Check agency_name
problems = gk.check_column(problems, table, f, "agency_name", gk.valid_str)
# Check agency_url
problems = gk.check_column(problems, table, f, "agency_url", gk.valid_url)
# Check agency_timezone
problems = gk.check_column(problems, table, f, "agency_timezone",
gk.valid_timezone)
# Check start_date and end_date
for col in ["start_date", "end_date"]:
problems = gk.check_column(problems, table, f, col, gk.valid_date)
# Check default_route_speed
problems = gk.check_column(problems, table, f, "default_route_speed",
valid_speed)
return gk.format_problems(problems, as_df=as_df)
def check_service_windows(pfeed, *, as_df=False, include_warnings=False):
"""
Analog of :func:`check_frequencies` for ``pfeed.service_windows``
"""
table = "service_windows"
problems = []
# Preliminary checks
if pfeed.service_windows is None:
problems.append(["error", "Missing table", table, []])
else:
f = pfeed.service_windows.copy()
problems = check_for_required_columns(problems, table, f)
if problems:
return gk.format_problems(problems, as_df=as_df)
if include_warnings:
problems = check_for_invalid_columns(problems, table, f)
# Check service window ID
problems = gk.check_column_id(problems, table, f, "service_window_id")
# Check start_time and end_time
for column in ["start_time", "end_time"]:
problems = gk.check_column(problems, table, f, column, gk.valid_time)
# Check weekday columns
v = lambda x: x in range(2)
for col in [
"monday",
"tuesday",
"wednesday",
"thursday",
"friday",
"saturday",
"sunday",
]:
#
problems = gk.check_column(problems, table, f, col, v)
return gk.format_problems(problems, as_df=as_df)
def check_frequencies(pfeed, *, as_df=False, include_warnings=False):
"""
Check that ``pfeed.frequency`` follows the ProtoFeed spec.
Return a list of problems of the form described in
:func:`gk.check_table`;
the list will be empty if no problems are found.
"""
table = "frequencies"
problems = []
# Preliminary checks
if pfeed.frequencies is None:
problems.append(["error", "Missing table", table, []])
else:
f = pfeed.frequencies.copy()
problems = check_for_required_columns(problems, table, f)
if problems:
return gk.format_problems(problems, as_df=as_df)
if include_warnings:
problems = check_for_invalid_columns(problems, table, f)
# Check route_short_name and route_long_name
for column in ["route_short_name", "route_long_name"]:
problems = gk.check_column(problems,
table,
f,
column,
gk.valid_str,
column_required=False)
cond = ~(f["route_short_name"].notnull() | f["route_long_name"].notnull())
problems = gk.check_table(
problems, table, f, cond,
"route_short_name and route_long_name both empty")
# Check route_type
v = lambda x: x in range(8)
problems = gk.check_column(problems, table, f, "route_type", v)
# Check service window ID
problems = gk.check_column_linked_id(problems, table, f,
"service_window_id",
pfeed.service_windows)
# Check direction
v = lambda x: x in range(3)
problems = gk.check_column(problems, table, f, "direction", v)
# Check frequency
v = lambda x: isinstance(x, int)
problems = gk.check_column(problems, table, f, "frequency", v)
# Check speed
problems = gk.check_column(problems,
table,
f,
"speed",
valid_speed,
column_required=False)
# Check shape ID
problems = gk.check_column_linked_id(problems, table, f, "shape_id",
pfeed.shapes)
return gk.format_problems(problems, as_df=as_df)