First pass at adding options to parsing.

gcgc/cli.py

@@ -29,22 +29,40 @@ def version():
 
 
 def to_path(ctx, param, value) -> pathlib.Path:
+    """
+    Simple callback to convert the string in the option to a command.
+    """
+
     return pathlib.Path(value)
 
 
 @main.command()
 @click.argument("filename", callback=to_path)
 @click.argument("format")
-def convert_file_to_tf_records(filename, format):
+@click.option("-e", "--encapsulate", is_flag=True)
+@click.option("-c", "--conform-to-length", "conform_to", type=int)
+@click.option("-p", "--pad-to-length", "pad_to", type=int)
+def convert_file_to_tf_records(filename, format, encapsulate, conform_to, pad_to):
 
-    output_file = filename.with_suffix(".tf_records")
+    output_file = filename.with_suffix(".tf-records")
     logger.info(f"Reading from {filename} in format {format} and writing to {output_file}.")
 
     writer = tf.python_io.TFRecordWriter(str(output_file))
     try:
         with open(filename, "rU") as handle:
             for seq_record in SeqIO.parse(handle, format):
+
                 encoded_seq = EncodedSeq.from_seq(seq_record.seq)
+
+                if encapsulate:
+                    encoded_seq = encoded_seq.encapsulate()
+
+                if conform_to:
+                    encoded_seq = encoded_seq.conform(conform_to)
+
+                if pad_to:
+                    encoded_seq = encoded_seq.pad(pad_to)
+
                 example = record.to_tensorflow_record(encoded_seq)
                 writer.write(example.SerializeToString())
     finally:

gcgc/encoded_seq/encoded_seq.py

@@ -85,7 +85,6 @@ def __add__(self, other) -> "EncodedSeq":
         return self.from_seq(added_seq)
 
     def __getitem__(self, index) -> "EncodedSeq":
-
         got_item = super().__getitem__(index)
         if isinstance(index, int):
             return got_item

gcgc/tests/third_party/tensorflow_utils/test_gcgc_record.py

@@ -8,7 +8,6 @@
 import tensorflow as tf
 
 from gcgc.encoded_seq import EncodedSeq
-from gcgc.alphabet.iupac import ExtendedIUPACDNAEncoding
 from gcgc.third_party.tensorflow_utils import record as gcgc_record
 from gcgc.tests.fixtures import P53_HUMAN
 

gcgc/third_party/tensorflow_utils/record.py

@@ -1,13 +1,13 @@
 # (c) Copyright 2018 Trent Hauck
 # All Rights Reserved
 
-from typing import NamedTuple, Sequence
+from typing import NamedTuple
 
 import tensorflow as tf
 import numpy as np
 
 
-def to_tensorflow_record(encoded_seq):
+def to_tensorflow_record(encoded_seq) -> tf.train.Example:
     """
     Convert the sequence to a tensorflow record.
     """
@@ -28,13 +28,18 @@ def to_tensorflow_record(encoded_seq):
     return example
 
 
-# TODO(trent): Fully functional EncodedSeq?
+# TODO: Should this be SequenceRecordExample?? I.e. probably want to have the actual id of the
+# sequence for later use.
 class ParsedEncodedSequence(NamedTuple):
     integer_encoded: np.ndarray
     alphabet_letters: np.ndarray
 
 
-def from_tensorflow_example(example):
+def from_tensorflow_example(example: tf.train.Example) -> ParsedEncodedSequence:
+    """
+    Given the example return an encoded sequence with the integer encoding and alphabet.
+    """
+
     features = {
         "integer_encoded": tf.FixedLenSequenceFeature((), tf.int64, allow_missing=True),
         "alphabet_letters": tf.FixedLenSequenceFeature((), tf.string, allow_missing=True),

mypy.ini

@@ -0,0 +1,2 @@
+[mypy]
+ignore_missing_imports = True