Initial commit

1 files changed, 115 insertions, 0 deletions

diff --git a/build_tcpd.py b/build_tcpd.py
new file mode 100644
index 0000000..9fe75ed
--- /dev/null
+++ b/build_tcpd.py
@@ -0,0 +1,115 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+"""
+Collect and verify all time series that are not packaged in the repository.
+
+Author: Gertjan van den Burg
+License: See LICENSE file.
+Copyright: 2019, The Alan Turing Institute
+
+"""
+
+import argparse
+import platform
+import os
+
+DATASET_DIR = "./datasets"
+
+TARGETS = [
+    ("apple", "get_apple.py"),
+    ("bee_waggle_6", "get_bee_waggle_6.py"),
+    ("bitcoin", "get_bitcoin.py"),
+    ("global_co2", "get_global_co2.py"),
+    ("iceland_tourism", "get_iceland_tourism.py"),
+    ("occupancy", "get_occupancy.py"),
+    ("ratner_stock", "get_ratner_stock.py"),
+    ("robocalls", "get_robocalls.py"),
+    ("scanline_126007", "get_scanline_126007.py"),
+    ("scanline_42049", "get_scanline_42049.py"),
+]
+
+
+def parse_args():
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "-v", "--verbose", help="Enable logging", action="store_true"
+    )
+    parser.add_argument(
+        "-o", "--output-dir", help="Output directory to store all time series"
+    )
+    parser.add_argument(
+        "action",
+        help="Action to perform",
+        choices=["collect", "clean"],
+        default="collect",
+        nargs="?",
+    )
+    return parser.parse_args()
+
+
+def load_dataset_script(module_name, path):
+    """Load the dataset collection script as a module
+
+    This is not a *super* clean way to do this, but it maintains the modularity 
+    of the dataset, where each dataset can be downloaded individually as well 
+    as through this script.
+    """
+    version = platform.python_version_tuple()
+    if version[0] == "2":
+        import imp
+
+        module = imp.load_source(module_name, path)
+    elif version[0] == "3" and version[1] in ["3", "4"]:
+        from importlib.machinery import SourceFileLoader
+
+        module = SourceFileLoader(module_name, path).load_module()
+    else:
+        import importlib.util
+
+        spec = importlib.util.spec_from_file_location(module_name, path)
+        module = importlib.util.module_from_spec(spec)
+        spec.loader.exec_module(module)
+    return module
+
+
+def run_dataset_func(name, script, funcname):
+    dir_path = os.path.join(DATASET_DIR, name)
+    get_path = os.path.join(dir_path, script)
+    module = load_dataset_script("tcpd.%s" % name, get_path)
+    func = getattr(module, funcname)
+    func(output_dir=dir_path)
+
+
+def collect_dataset(name, script):
+    return run_dataset_func(name, script, "collect")
+
+
+def clean_dataset(name, script):
+    return run_dataset_func(name, script, "clean")
+
+
+def main():
+    args = parse_args()
+
+    log = lambda *a, **kw: print(*a, **kw) if args.verbose else None
+
+    if args.action == "collect":
+        func = collect_dataset
+    elif args.action == "clean":
+        func = clean_dataset
+    else:
+        raise ValueError("Unknown action: %s" % args.action)
+
+    for name, script in TARGETS:
+        log(
+            "Running %s action for dataset: %s ... " % (args.action, name),
+            end="",
+            flush=True,
+        )
+        func(name, script)
+        log("ok", flush=True)
+
+
+if __name__ == "__main__":
+    main()