|
| 1 | +import argparse |
| 2 | +import sys |
| 3 | +from datetime import timedelta |
| 4 | +from time import time |
| 5 | +from typing import Sequence |
| 6 | + |
| 7 | +import pandas as pd |
| 8 | + |
| 9 | +from exabel import ExabelClient |
| 10 | +from exabel.client.api.data_classes.derived_signal import DerivedSignal |
| 11 | +from exabel.scripts.base_script import BaseScript |
| 12 | + |
| 13 | +_EXTENSION_TO_FILE_FORMAT = { |
| 14 | + "csv": "csv", |
| 15 | + "xlsx": "excel", |
| 16 | + "json": "json", |
| 17 | + "feather": "feather", |
| 18 | + "parquet": "parquet", |
| 19 | +} |
| 20 | + |
| 21 | + |
| 22 | +def _validate_filename(filename: str) -> str: |
| 23 | + """Ensure the output file has one of the server-supported extensions.""" |
| 24 | + extension = filename.rsplit(".", 1)[-1].lower() |
| 25 | + if extension not in _EXTENSION_TO_FILE_FORMAT: |
| 26 | + raise argparse.ArgumentTypeError( |
| 27 | + f"Unknown file extension {extension!r}. Supported: " |
| 28 | + + ", ".join(f".{e}" for e in _EXTENSION_TO_FILE_FORMAT) |
| 29 | + ) |
| 30 | + return filename |
| 31 | + |
| 32 | + |
| 33 | +def _parse_label_expression(value: str) -> DerivedSignal: |
| 34 | + """Parse a ``LABEL=EXPRESSION`` CLI value into a DerivedSignal. |
| 35 | +
|
| 36 | + The label must be non-empty; the expression is everything after the first |
| 37 | + ``=`` so expressions containing ``=`` themselves (e.g. comparisons) still |
| 38 | + parse cleanly. |
| 39 | + """ |
| 40 | + if "=" not in value: |
| 41 | + raise argparse.ArgumentTypeError( |
| 42 | + f"--expression value must be formatted as LABEL=EXPRESSION, got: {value!r}" |
| 43 | + ) |
| 44 | + label, expression = value.split("=", 1) |
| 45 | + label = label.strip() |
| 46 | + expression = expression.strip() |
| 47 | + if not label or not expression: |
| 48 | + raise argparse.ArgumentTypeError( |
| 49 | + f"--expression value must have non-empty LABEL and EXPRESSION, got: {value!r}" |
| 50 | + ) |
| 51 | + return DerivedSignal(name=None, label=label, expression=expression) |
| 52 | + |
| 53 | + |
| 54 | +class ExportSignalsV2(BaseScript): |
| 55 | + """Export time series using the v2 signal export endpoint and write the raw |
| 56 | + server response to a file. |
| 57 | +
|
| 58 | + Unlike ``export_signals``, this supports multi-time-series signals (e.g. |
| 59 | + signals whose expression uses ``for_type(...)`` and returns one series per |
| 60 | + sub-entity). Entities must be specified by resource name or tag; |
| 61 | + bloomberg_ticker / factset_id are not supported by the v2 endpoint. |
| 62 | +
|
| 63 | + The output file contains the server's wire response verbatim (no pandas |
| 64 | + round-trip, no timezone normalization). Parquet and feather preserve the |
| 65 | + full multi-level column headers; csv, excel, and json are flattened per the |
| 66 | + server's serialization of those formats. Running this script does not |
| 67 | + require ``pyarrow`` — only the SDK and a network connection. |
| 68 | + """ |
| 69 | + |
| 70 | + def __init__(self, argv: Sequence[str], description: str): |
| 71 | + super().__init__(argv, description) |
| 72 | + self.parser.add_argument( |
| 73 | + "--signal", |
| 74 | + nargs="+", |
| 75 | + required=False, |
| 76 | + default=[], |
| 77 | + type=str, |
| 78 | + help=( |
| 79 | + "Signal label(s) to export from the library. " |
| 80 | + "At least one of --signal or --expression must be given." |
| 81 | + ), |
| 82 | + ) |
| 83 | + self.parser.add_argument( |
| 84 | + "--expression", |
| 85 | + nargs="+", |
| 86 | + required=False, |
| 87 | + default=[], |
| 88 | + type=_parse_label_expression, |
| 89 | + help=( |
| 90 | + "DSL expression(s) to evaluate, each as LABEL=EXPRESSION. " |
| 91 | + "The LABEL becomes the column name in the output. Example: " |
| 92 | + "--expression " |
| 93 | + '\'brand_sales=data("sales").for_type("brand")\'' |
| 94 | + ), |
| 95 | + ) |
| 96 | + entity_group = self.parser.add_mutually_exclusive_group(required=True) |
| 97 | + entity_group.add_argument( |
| 98 | + "--resource-name", |
| 99 | + nargs="+", |
| 100 | + type=str, |
| 101 | + help="Entity resource name(s) to evaluate the signal(s) for.", |
| 102 | + ) |
| 103 | + entity_group.add_argument( |
| 104 | + "--tag", |
| 105 | + nargs="+", |
| 106 | + type=str, |
| 107 | + help="Tag(s) whose entities should be evaluated.", |
| 108 | + ) |
| 109 | + self.parser.add_argument( |
| 110 | + "--filename", |
| 111 | + required=True, |
| 112 | + type=_validate_filename, |
| 113 | + help=( |
| 114 | + "The filename to write the raw server response to. Supported extensions: " |
| 115 | + + ", ".join(f".{e}" for e in _EXTENSION_TO_FILE_FORMAT) |
| 116 | + ), |
| 117 | + ) |
| 118 | + self.parser.add_argument( |
| 119 | + "--start-date", |
| 120 | + required=False, |
| 121 | + type=pd.Timestamp, |
| 122 | + help="The first date to evaluate the signals for.", |
| 123 | + ) |
| 124 | + self.parser.add_argument( |
| 125 | + "--end-date", |
| 126 | + required=False, |
| 127 | + type=pd.Timestamp, |
| 128 | + help="The last date to evaluate the signals for.", |
| 129 | + ) |
| 130 | + self.parser.add_argument( |
| 131 | + "--known-time", |
| 132 | + required=False, |
| 133 | + type=pd.Timestamp, |
| 134 | + help="The point-in-time at which to retrieve the time series.", |
| 135 | + ) |
| 136 | + |
| 137 | + def run_script(self, client: ExabelClient, args: argparse.Namespace) -> None: |
| 138 | + start_time = time() |
| 139 | + signals: list[str | DerivedSignal] = [*args.signal, *args.expression] |
| 140 | + if not signals: |
| 141 | + self.parser.error("at least one of --signal or --expression must be provided") |
| 142 | + signal_labels = [s if isinstance(s, str) else s.label for s in signals] |
| 143 | + print("Downloading signal(s):", ", ".join(signal_labels)) |
| 144 | + |
| 145 | + extension = args.filename.rsplit(".", 1)[-1].lower() |
| 146 | + file_format = _EXTENSION_TO_FILE_FORMAT[extension] |
| 147 | + |
| 148 | + content = client.export_api.export_signals_v2_bytes( |
| 149 | + signals, |
| 150 | + file_format=file_format, |
| 151 | + resource_name=args.resource_name, |
| 152 | + tag=args.tag, |
| 153 | + start_time=args.start_date, |
| 154 | + end_time=args.end_date, |
| 155 | + version=args.known_time, |
| 156 | + ) |
| 157 | + with open(args.filename, "wb") as f: |
| 158 | + f.write(content) |
| 159 | + |
| 160 | + spent_time = timedelta(seconds=int(time() - start_time)) |
| 161 | + print(f"{len(content)} bytes written to {args.filename}, spent {spent_time}") |
| 162 | + |
| 163 | + |
| 164 | +if __name__ == "__main__": |
| 165 | + ExportSignalsV2( |
| 166 | + sys.argv, |
| 167 | + "Export time series from the Exabel API for specified signals (v2 endpoint)", |
| 168 | + ).run() |
0 commit comments