From 171a685f5d32c7f94b04ac43d603d3e7c95d8dca Mon Sep 17 00:00:00 2001
From: Donald Campbell <125581724+donaldcampbelljr@users.noreply.github.com>
Date: Tue, 17 Sep 2024 16:29:09 -0400
Subject: [PATCH] add bedfile read exception

---
 bedboss/refgenome_validator/utils.py             | 15 ++++++++++++++-
 scripts/ref_genome_validating/validate_genome.py |  9 +++++++--
 2 files changed, 21 insertions(+), 3 deletions(-)

diff --git a/bedboss/refgenome_validator/utils.py b/bedboss/refgenome_validator/utils.py
index 68cb66e..e0c9d29 100644
--- a/bedboss/refgenome_validator/utils.py
+++ b/bedboss/refgenome_validator/utils.py
@@ -3,6 +3,7 @@
 from geniml.io.utils import is_gzipped
 import logging
 
+from bedboss.exceptions import BedBossException
 
 _LOGGER = logging.getLogger("bedboss")
 
@@ -67,7 +68,7 @@ def _read_file_pd(*args, **kwargs) -> pd.DataFrame:
         except (pd.errors.ParserError, pd.errors.EmptyDataError) as _:
             if row_count <= max_rows:
                 row_count += 1
-    return _read_gzipped_file(*args)
+    raise BedfileReadException(reason="Cannot read bed file.")
 
 
 def get_bed_chrom_info(bedfile: str) -> dict:
@@ -84,3 +85,15 @@ def get_bed_chrom_info(bedfile: str) -> dict:
 
     max_end_for_each_chrom = df.groupby(0)[2].max()
     return max_end_for_each_chrom.to_dict()
+
+
+class BedfileReadException(BedBossException):
+    """Exception when there is an exception during refgenome validation"""
+
+    def __init__(self, reason: str = ""):
+        """
+        Optionally provide explanation for exceptional condition.
+
+        :param str reason: some context why error occurred
+        """
+        super(BedfileReadException, self).__init__(reason)
diff --git a/scripts/ref_genome_validating/validate_genome.py b/scripts/ref_genome_validating/validate_genome.py
index ad207f6..4e04517 100644
--- a/scripts/ref_genome_validating/validate_genome.py
+++ b/scripts/ref_genome_validating/validate_genome.py
@@ -25,7 +25,8 @@
     # BEDFILE_DIRECTORY = (
     #     "/home/drc/GITHUB/bedboss/bedboss/scripts/ref_genome_validating/results"
     # )
-    BEDFILE_DIRECTORY = "/home/drc/GITHUB/bedboss/bedboss/scripts/ref_genome_validating/data/bed_small_subset"
+    # BEDFILE_DIRECTORY = "/home/drc/GITHUB/bedboss/bedboss/scripts/ref_genome_validating/data/bed_small_subset"
+    BEDFILE_DIRECTORY = "/home/drc/GITHUB/bedboss/bedboss/scripts/ref_genome_validating/data/test_singles"
 
 try:
     PEP_URL = os.environ["PEP_URL"]
@@ -72,6 +73,11 @@ def main():
     # build genome models
     # for each reference genome in the user's config file, build a genome model
 
+    # from geniml.io import RegionSet
+    #
+    # ff =RegionSet("/home/drc/GITHUB/bedboss/bedboss/scripts/ref_genome_validating/data/test_singles/GSM8196564_435_RUNX3_KO_H3K27AC_v_435_ctrl_IgG_seacr.relaxed.bed")
+    # ff
+
     all_genome_models = []
 
     # for reference_genome in rgc.list():
@@ -213,7 +219,6 @@ def main():
                             tier = {
                                 "tier_rating": {}
                             }  # add this to a column to make comparisons easier for human eyes on pephub
-                            all_vals = {}
                             if compat_vector:
                                 for i in compat_vector.keys():
                                     if i is not None: