utils: Add convert_cal_data utility

This will convert cal data files based on CSV to the new binary format.
In most cases, running

    ./convert_cal_data.py

is sufficient.

1 files changed, 104 insertions, 0 deletions

diff --git a/host/utils/convert_cal_data.py b/host/utils/convert_cal_data.py
new file mode 100755
index 000000000..e304ce68d
--- /dev/null
+++ b/host/utils/convert_cal_data.py
@@ -0,0 +1,104 @@
+#!/usr/bin/env python3
+"""
+Copyright 2020 Ettus Research, A National Instrument Brand
+
+SPDX-License-Identifier: GPL-3.0-or-later
+
+convert_cal_data: A utility to convert CSV-based calibration data into binary
+cal data.
+"""
+
+import os
+import glob
+import argparse
+import uhd
+
+def csv_to_iq_cal(filename):
+    """
+    Opens the file "filename", parses the CSV data, and returns a
+    uhd.usrp.cal.IQCal class.
+    """
+    def parse_metadata(handle):
+        """
+        Return name, serial, and timestamp, and verify version.
+        """
+        num_lines_md = 4
+        metadata = {}
+        for _ in range(num_lines_md):
+            key, val = [x.strip() for x in handle.readline().split(",", 1)]
+            if key == "version":
+                if val != "0, 1":
+                    raise RuntimeError("Unexpected version: " + val)
+            elif key == "timestamp":
+                metadata[key] = int(val)
+            else:
+                metadata[key] = val
+        if handle.readline().strip() != "DATA STARTS HERE":
+            raise RuntimeError("Unexpected header!")
+        handle.readline() # Flush header
+        return metadata
+    with open(filename, 'r') as handle:
+        metadata = parse_metadata(handle)
+        print("Found metadata: ")
+        for key, val in metadata.items():
+            print("* {k}: {v}".format(k=key, v=val))
+        iq_cal = uhd.usrp.cal.IQCal(
+            metadata['name'], metadata['serial'], metadata['timestamp'])
+        for line in handle.readlines():
+            lo_freq, corr_real, corr_imag, supp_abs, supp_delta = \
+                line.split(",")
+            iq_cal.set_cal_coeff(
+                float(lo_freq),
+                complex(float(corr_real.strip()), float(corr_imag.strip())),
+                float(supp_abs.strip()),
+                float(supp_delta.strip()),
+            )
+    return iq_cal
+
+def get_csv_location():
+    """
+    Returns the path to the location of the CSV files
+    """
+    return os.path.expandvars("$HOME/.uhd/cal")
+
+def get_all_files(csv_location=None):
+    """
+    Returns a list of all CSV files in the standard location
+    """
+    csv_path = csv_location or get_csv_location()
+    return glob.glob(os.path.join(csv_path, "*.csv"))
+
+def parse_args():
+    """ Parse args and return args object """
+    parser = argparse.ArgumentParser(
+        description="Convert CSV-Based cal data to binary format",
+    )
+    parser.add_argument(
+        'files',
+        help="List files to convert. If empty, will attempt to auto-detect "
+             "files from the standard calibration location ($HOME/.uhd/cal)",
+        nargs="*",
+    )
+    return parser.parse_args()
+
+
+def main():
+    """Go, go, go!"""
+    args = parse_args()
+    file_list = args.files or get_all_files()
+    for filename in file_list:
+        print("Converting {}...".format(filename))
+        print("Identifying cal data type...")
+        key = os.path.basename(filename).split('_cal')[0]
+        if key not in ('rx_iq', 'tx_iq', 'tx_dc'):
+            print("Unidentified cal data type: {}".format(key))
+            continue
+        iq_cal = csv_to_iq_cal(filename)
+        uhd.usrp.cal.database.write_cal_data(
+            key,
+            iq_cal.get_serial(),
+            iq_cal.serialize())
+    return 0
+
+if __name__ == "__main__":
+    exit(main())