Add word-level scrambler function (#72)

## Description
Adds a function to scramble text at the word level, ensuring that only
unique indices are swapped.

Also removed Eric as a code owner so that he won't be tagged in PRs and
updated the handler event to be an SQS Event based on our conversation
with Geo earlier this week. I moved `augmentation.py` into the
`dibbs_text_to_code` directory instead of nesting it under
`data_curation_and_augmentation` because it made the imports nicer but
let me know if you feel strongly about the nesting.

## Related Issues
Closes #66
Fixes #69 

## Additional Notes
Also removed Eric as a code owner so that he won't be tagged in PRs and
updated the handler event to be an SQS Event based on our conversation
with Geo earlier this week.

Author: m-goggins

.github/CODEOWNERS

@@ -1,2 +1,3 @@
 # Define owners for all files in the repository
-*       @m-goggins @bamader @ericbuckley @BradySkylight
+
+-       @m-goggins @bamader @BradySkylight

src/dibbs_text_to_code/augmentation.py

@@ -0,0 +1,33 @@
+import random
+
+
+def scramble_word_order(
+    text: str,
+    max_perms: int,
+    min_perms: int = 1,
+) -> str:
+    """
+    Scrambles the order of words in the input text by moving a specified
+    number of words to new positions.
+
+    :param text: The input text to scramble.
+    :param max_perms: The maximum number of words to move.
+    :param min_perms: The minimum number of words to move.
+    :return: The text with words scrambled.
+    """
+    words = text.split()
+    if len(words) < 2:
+        return text
+
+    # Ensure max_perms does not exceed the number of words
+    num_perms = min(random.randint(min_perms, max_perms), len(words) - 1)
+
+    # Select unique indices to scramble
+    indices_to_move = sorted(random.sample(range(len(words)), num_perms), reverse=True)
+
+    for idx in indices_to_move:
+        new_pos = random.choice([i for i in range(len(words)) if i != idx])
+        word = words.pop(idx)
+        words.insert(new_pos, word)
+
+    return " ".join(words)

src/dibbs_text_to_code/main.py

@@ -6,7 +6,7 @@
 from .s3_handler import get_file_content_from_s3_event
 
 
-def handler(event: lambda_events.EventBridgeEvent, context: lambda_context.Context):
+def handler(event: lambda_events.SQSEvent, context: lambda_context.Context):
     """
     Text to Code lambda entry point
     """

tests/unit/conftest.py

@@ -1,4 +1,5 @@
 import os
+import random
 
 import boto3
 import moto
@@ -26,3 +27,8 @@ def moto_setup(monkeypatch):
         s3.bucket_name = bucket_name
 
         yield s3
+
+
+@pytest.fixture(autouse=True)
+def fixed_random_seed():
+    random.seed(42)

tests/unit/test_augmentation.py

@@ -0,0 +1,26 @@
+import pytest
+
+from dibbs_text_to_code import augmentation
+
+
+@pytest.mark.parametrize(
+    "text, max_perms, expected",
+    [
+        # Empty string
+        ("", 3, ""),
+        # Single word
+        ("Blood", 3, "Blood"),
+        # Multiple words with special characters
+        (
+            "SARS-CoV-2 E gene Resp Ql NAA+probe",
+            5,
+            "E gene Resp SARS-CoV-2 Ql NAA+probe",
+        ),
+        # More deletions than words
+        ("B pert Spt Ql Cult", 10, "Spt pert B Ql Cult"),
+    ],
+)
+class TestScrambleWordOrder:
+    def test_scramble_word_order(self, text, max_perms, expected):
+        result = augmentation.scramble_word_order(text, max_perms=max_perms)
+        assert result == expected