From 6eaecb65fac8276328a9226305dee94549f0cd79 Mon Sep 17 00:00:00 2001
From: Matthew McDermott <mattmcdermott8@gmail.com>
Date: Sun, 4 Aug 2024 15:33:33 -0400
Subject: [PATCH] Got pre_MEDS.py working (for MIMIC-IV, not eICU) for
 arbitrary supported input file encodings so that unzipping works. Also made
 the bash arg (I think) work. Also removed rootutils to help address #114

---
 MIMIC-IV_Example/joint_script.sh |  2 +-
 MIMIC-IV_Example/pre_MEDS.py     | 10 ++++++----
 eICU_Example/pre_MEDS.py         |  3 ---
 pyproject.toml                   |  1 -
 4 files changed, 7 insertions(+), 9 deletions(-)

diff --git a/MIMIC-IV_Example/joint_script.sh b/MIMIC-IV_Example/joint_script.sh
index 2b4ea56..a98fee7 100755
--- a/MIMIC-IV_Example/joint_script.sh
+++ b/MIMIC-IV_Example/joint_script.sh
@@ -66,7 +66,7 @@ else
 fi
 
 if [ "$DO_UNZIP" == "true" ]; then
-  echo "Unzipping gunzip csv files."
+  echo "Unzipping csv files."
     for file in "${MIMICIV_RAW_DIR}"/*/*.csv.gz; do gzip -d --force "$file"; done
 else
   echo "Skipping unzipping."
diff --git a/MIMIC-IV_Example/pre_MEDS.py b/MIMIC-IV_Example/pre_MEDS.py
index ed317b6..1530a8b 100755
--- a/MIMIC-IV_Example/pre_MEDS.py
+++ b/MIMIC-IV_Example/pre_MEDS.py
@@ -1,9 +1,6 @@
 #!/usr/bin/env python
 
 """Performs pre-MEDS data wrangling for MIMIC-IV."""
-import rootutils
-
-root = rootutils.setup_root(__file__, dotenv=True, pythonpath=True, cwd=True)
 
 from datetime import datetime
 from functools import partial
@@ -77,7 +74,12 @@ def main(cfg: DictConfig):
     for in_fp in all_fps:
         pfx = get_shard_prefix(raw_cohort_dir, in_fp)
 
-        fp, read_fn = get_supported_fp(raw_cohort_dir, pfx)
+        try:
+            fp, read_fn = get_supported_fp(raw_cohort_dir, pfx)
+        except FileNotFoundError:
+            logger.info(f"Skipping {pfx} @ {str(in_fp.resolve())} as no compatible dataframe file was found.")
+            continue
+
         if fp.suffix in [".csv", ".csv.gz"]:
             read_fn = partial(read_fn, infer_schema_length=100000)
 
diff --git a/eICU_Example/pre_MEDS.py b/eICU_Example/pre_MEDS.py
index 317ecb5..5ebe058 100755
--- a/eICU_Example/pre_MEDS.py
+++ b/eICU_Example/pre_MEDS.py
@@ -4,9 +4,6 @@
 
 See the docstring of `main` for more information.
 """
-import rootutils
-
-root = rootutils.setup_root(__file__, dotenv=True, pythonpath=True, cwd=True)
 
 import gzip
 from collections.abc import Callable
diff --git a/pyproject.toml b/pyproject.toml
index 5bea0a8..e512e7e 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -23,7 +23,6 @@ dependencies = [
 [tool.setuptools_scm]
 
 [project.optional-dependencies]
-examples = ["rootutils"]
 dev = ["pre-commit"]
 tests = ["pytest", "pytest-cov", "rootutils"]
 local_parallelism = ["hydra-joblib-launcher"]