Fixed tests and updated bed hashing

bedboss/bedstat/bedstat.py

@@ -1,13 +1,13 @@
-from hashlib import md5
 from typing import NoReturn
 import json
 import yaml
 import os
 import requests
-import gzip
 import pypiper
 import bbconf
 import logging
+from geniml.io import RegionSet
+
 
 _LOGGER = logging.getLogger("bedboss")
 
@@ -16,30 +16,6 @@
 )
 
 
-def digest_bedfile(filepath: str) -> str:
-    """
-    Generate digest for bedfile
-
-    :param str filepath: path to the bed file
-    :return str: digest of the files
-    """
-    with gzip.open(filepath, "rb") as f:
-        # concate column values
-        chrs = ",".join([row.split()[0].decode("utf-8") for row in f])
-        starts = ",".join([row.split()[1].decode("utf-8") for row in f])
-        ends = ",".join([row.split()[2].decode("utf-8") for row in f])
-        # hash column values
-        chr_digest = md5(chrs.encode("utf-8")).hexdigest()
-        start_digest = md5(starts.encode("utf-8")).hexdigest()
-        end_digest = md5(ends.encode("utf-8")).hexdigest()
-        # hash column digests
-        bed_digest = md5(
-            ",".join([chr_digest, start_digest, end_digest]).encode("utf-8")
-        ).hexdigest()
-
-        return bed_digest
-
-
 def convert_unit(size_in_bytes: int) -> str:
     """
     Convert the size from bytes to other units like KB, MB or GB
@@ -106,7 +82,7 @@ def bedstat(
         pass
     bbc = bbconf.BedBaseConf(config_path=bedbase_config, database_only=True)
 
-    bed_digest = digest_bedfile(bedfile)
+    bed_digest = RegionSet(bedfile).identifier
     bedfile_name = os.path.split(bedfile)[1]
 
     fileid = os.path.splitext(os.path.splitext(bedfile_name)[0])[0]

requirements/requirements-all.txt

@@ -7,4 +7,5 @@ piper>=0.13.3a1
 bbconf>=0.4.0a5
 refgenconf>=0.12.2
 pandas>=1.5.3
-ubiquerg>=0.6.2
+ubiquerg>=0.6.2
+geniml

test/bash_requirements_test.sh

@@ -121,7 +121,7 @@ if is_executable "R"; then
     echo -e "-----------------------------------------------------------"
     echo -e "Checking required R packages for bedstat...                            "
     echo -e "-----------------------------------------------------------"
-    declare -a requiredRPackages=("optparse ""devtools" "ensembldb" "ExperimentHub" "AnnotationHub" "AnnotationFilter" "BSgenome" "GenomicFeatures" "GenomicDistributions" "GenomicDistributionsData" "GenomeInfoDb" "ensembldb" "tools" "R.utils" "LOLA")
+    declare -a requiredRPackages=("optparse ""devtools" "ensembldb" "ExperimentHub" "AnnotationHub" "AnnotationFilter" "BSgenome" "GenomicFeatures" "GenomicDistributions" "GenomicDistributionsData" "GenomeInfoDb" "ensembldb" "tools" "R.utils" "LOLA" "conflicted")
     for package in "${requiredRPackages[@]}"; do
       if ! r_check_req $package; then
         INSTALL_ERROR=$((INSTALL_ERROR+1))

test/test_bedboss.py

@@ -1,5 +1,6 @@
 from bedboss.bedboss import main
 import os
+import warnings
 import subprocess
 import pytest
 from bbconf import BedBaseConf
@@ -14,28 +15,37 @@
 BEDBASE_CONFIG = os.path.join(FILE_DIR, "test_dependencies", "bedbase_config_test.yaml")
 DEPENDENCIES_TEST_SCRIPT = f"{FILE_DIR}/bash_requirements_test.sh"
 
+pytest_db_skip_reason = "Database is not set up... To run this test, set up the database. Go to test/README.md for more information."
 
-def test_dependencies():
+
+def check_dependencies_installed() -> bool:
     # Make sure bedToBigBed etc is in your PATH.
     print("Testing dependencies...")
     key = "PATH"
     value = os.getenv(key)
     test_dep_return_code = subprocess.run([DEPENDENCIES_TEST_SCRIPT], shell=True)
-    assert 1 > test_dep_return_code.returncode
+    if not (1 > test_dep_return_code.returncode):
+        warnings.warn(UserWarning(f"{pytest_db_skip_reason}"))
+        return False
+    return True
+    # return 1 > test_dep_return_code.returncode
+
+
+dependencies_installed = check_dependencies_installed()
 
 
 def db_setup():
     # Check if the database is setup
     try:
         BedBaseConf(BEDBASE_CONFIG)
     except Exception as err:
-        print(f"Error: {err}")
-        BedBaseConf(BEDBASE_CONFIG)
+        warnings.warn(UserWarning(f"{pytest_db_skip_reason}"))
         return False
     return True
 
 
-pytest_db_skip_reason = "Database is not set up... To run this test, set up the database. Go to test/README.md for more information."
+def test_dependencies():
+    assert dependencies_installed
 
 
 @pytest.mark.parametrize(
@@ -55,6 +65,14 @@ def test_qc(bedfile, tmpdir):
     assert qc_passed is None
 
 
+@pytest.mark.skipif(
+    not db_setup() or not dependencies_installed,
+    reason=pytest_db_skip_reason,
+)
+@pytest.mark.skipif(
+    not db_setup() or not dependencies_installed,
+    reason=pytest_db_skip_reason,
+)
 @pytest.mark.parametrize(
     "bedfile",
     [
@@ -80,7 +98,7 @@ def test_make(bedfile, tmpdir):
 
 
 @pytest.mark.skipif(
-    not db_setup(),
+    not db_setup() or not dependencies_installed,
     reason=pytest_db_skip_reason,
 )
 class TestStat:
@@ -142,14 +160,14 @@ def test_check_file_exists(self, file, output_temp_dir):
                 output_temp_dir,
                 "output",
                 "bedstat_output",
-                "c557c915a9901ce377ef724806ff7a2c",
+                "49a72983ca9ddcf6692c5ec8b51c3d92",
                 file,
             )
         )
 
 
 @pytest.mark.skipif(
-    not db_setup(),
+    not db_setup() or not dependencies_installed,
     reason=pytest_db_skip_reason,
 )
 class TestAll:
@@ -212,7 +230,7 @@ def test_check_file_exists(self, file, output_temp_dir):
                 output_temp_dir,
                 "output",
                 "bedstat_output",
-                "c557c915a9901ce377ef724806ff7a2c",
+                "49a72983ca9ddcf6692c5ec8b51c3d92",
                 file,
             )
         )