feat: Add upstream dataset processing script

datafux.py

@@ -0,0 +1,155 @@
+#! /usr/bin/env python3
+
+import os, glob, re
+import datetime as dt
+
+import click
+import zipfile as zf
+import polars as pl
+
+GAP_RE = re.compile('Gap of (\\d+)s found between (\\d+) and (\\d+)')
+CSV_SCHEMA = {
+    'ts': pl.String,
+    'open': pl.Float64,
+    'high': pl.Float64,
+    'low': pl.Float64,
+    'close': pl.Float64,
+    'volume': pl.UInt64,
+}
+
+@click.group
+def cli():
+    pass
+
+@click.argument('folder')
+@click.option('--only', '-o', multiple=True)
+@click.option('--exclude', '-e', multiple=True)
+@cli.command()
+def extract(folder, only, exclude):
+    pairs = []
+
+    for entry in os.scandir(folder):
+        if entry.is_dir() and len(entry.name) == 6:
+            pairs.append(entry.path)
+
+    for pair in pairs:
+        name = os.path.basename(pair)
+        if len(only) > 0 and name not in only:
+            continue
+        if len(exclude) > 0 and name in exclude:
+            continue
+
+        print(f"Processing {name}:")
+        gaps, ticks = _extract_and_concatenate_pair(pair)
+
+        gaps_path = os.path.join(pair, 'gaps.parquet')
+        print(f" - Writing {gaps_path}")
+        gaps.write_parquet(gaps_path)
+
+        ticks_path = os.path.join(pair, 'ticks.parquet')
+        print(f" - Writing {ticks_path}")
+        ticks.write_parquet(ticks_path)
+
+        # breakpoint()
+
+
+def _parse_gaps(lines):
+    "Parse the status report to extract gaps"
+    data = []
+
+    for line in lines:
+        line = line.decode('utf-8').strip()
+
+        if line == '':
+            continue
+        if 'HistData.com' in line or 'File:' in line or 'tick interval' in line:
+            continue
+
+        md = GAP_RE.match(line)
+        if md:
+            data.append([int(md[1]), md[2], md[3]])
+        else:
+            print(f"Unable to parse line: {line}")
+
+    gaps = pl.DataFrame(
+        schema={"length": pl.UInt64, "start": pl.String, "end": pl.String},
+        data=data,
+        orient='row'
+    ).with_columns(
+        pl.col("start").str.to_datetime(
+            "%Y%m%d%H%M%S"
+        ).dt.replace_time_zone('EST').dt.convert_time_zone('UTC')
+    ).with_columns(
+        pl.col("end").str.to_datetime(
+            "%Y%m%d%H%M%S"
+        ).dt.replace_time_zone('EST').dt.convert_time_zone('UTC')
+    )
+    return gaps
+
+
+@click.argument('folder')
+@click.option('--only', '-o', multiple=True)
+@click.option('--exclude', '-e', multiple=True)
+@cli.command()
+def merge(folder, only, exclude):
+
+    for kind in ['ticks', 'gaps']:
+        print(f'Processing {kind} files: ')
+        files = glob.glob(os.path.join(folder, '*', f'{kind}.parquet'))
+        dataframes = []
+
+        for file in files:
+            print(f" - {file}")
+            name = os.path.basename(os.path.dirname(file))
+            df = pl.read_parquet(file).with_columns(symbols=pl.lit(name))
+            dataframes.append(df)
+
+        result = pl.concat(dataframes)
+        dest = os.path.join(folder, f'all_{kind}.parquet')
+        print(f'-> Writing merged file to {dest}')
+        result.write_parquet(dest)
+
+
+def _extract_and_concatenate_pair(path, name=None):
+    if name is None:
+        name = os.path.basename(path)
+
+    gaps = []
+    ticks = []
+
+    zipfiles = glob.glob(os.path.join(path, "*.zip"))
+    for zipfile in zipfiles:
+        with zf.ZipFile(zipfile) as archive:
+            for entry in archive.namelist():
+                # This is the status report, let's parse it
+                if entry.endswith(".txt"):
+                    with archive.open(entry) as txt_file:
+                        lines = txt_file.readlines()
+                        gaps.append(_parse_gaps(lines))
+
+                # This is the CSV with the ticker data
+                elif entry.endswith(".csv"):
+                    with archive.open(entry) as csv_file:
+                        ticks.append(pl.read_csv(
+                            csv_file,
+                            has_header=False,
+                            separator=";",
+                            ignore_errors=True,
+                            schema=CSV_SCHEMA,
+                        ))
+                else:
+                    print(f"Unknown file {entry} in {zipfile}")
+
+    gaps = pl.concat(gaps).sort(by="start", descending=False)
+    ticks = pl.concat(ticks).with_columns(
+        pl.col("ts").str.to_datetime(
+            "%Y%m%d %H%M%S"
+        ).dt.replace_time_zone('EST').dt.convert_time_zone('UTC')
+    ).sort(by="ts", descending=False)
+
+    # breakpoint()
+
+    return gaps, ticks
+
+if __name__ == '__main__':
+    cli()

requirements.txt

@@ -0,0 +1,2 @@
+polars
+click