Convert print statements to logger.debug statements

README.rst

@@ -48,7 +48,14 @@ Examples
 --------
 Download Example Data
 `````````````````````
-Let's download some example data from `openSNP <https://opensnp.org>`_:
+First, let's setup logging to get some helpful output:
+
+>>> import logging, sys
+>>> logger = logging.getLogger()
+>>> logger.setLevel(logging.DEBUG)
+>>> logger.addHandler(logging.StreamHandler(sys.stdout))
+
+Now we're ready to download some example data from `openSNP <https://opensnp.org>`_:
 
 >>> from snps.resources import Resources
 >>> r = Resources()

src/snps/__init__.py

@@ -53,6 +53,10 @@
 # set version string with Versioneer
 from snps._version import get_versions
 
+import logging
+
+logger = logging.getLogger(__name__)
+
 __version__ = get_versions()["version"]
 del get_versions
 
@@ -319,7 +323,7 @@ def _assign_par_snps(self):
                                 break
 
                 except Exception as err:
-                    print(err)
+                    logger.warning(err)
 
     def _assign_snp(self, rsid, alleles, chrom):
         # only assign SNP if positions match (i.e., same build)
@@ -607,7 +611,7 @@ def remap_snps(self, target_assembly, complement_bases=True):
         snps = self.snps
 
         if snps.empty:
-            print("No SNPs to remap")
+            logger.debug("No SNPs to remap")
             return chromosomes_remapped, chromosomes_not_remapped
         else:
             chromosomes = snps["chrom"].unique()
@@ -616,7 +620,7 @@ def remap_snps(self, target_assembly, complement_bases=True):
         valid_assemblies = ["NCBI36", "GRCh37", "GRCh38", 36, 37, 38]
 
         if target_assembly not in valid_assemblies:
-            print("Invalid target assembly")
+            logger.debug("Invalid target assembly")
             return chromosomes_remapped, chromosomes_not_remapped
 
         if isinstance(target_assembly, int):
@@ -655,7 +659,7 @@ def remap_snps(self, target_assembly, complement_bases=True):
                     }
                 )
             else:
-                print(
+                logger.debug(
                     "Chromosome {} not remapped; "
                     "removing chromosome from SNPs for consistency".format(chrom)
                 )
@@ -712,11 +716,13 @@ def _remapper(self, task):
             mapped_region = mapping["mapped"]["seq_region_name"]
 
             if orig_region != mapped_region:
-                print("discrepant chroms")
+                logger.debug("discrepant chroms")
                 continue
 
             if orig_range_len != mapped_range_len:
-                print("discrepant coords")  # observed when mapping NCBI36 -> GRCh38
+                logger.debug(
+                    "discrepant coords"
+                )  # observed when mapping NCBI36 -> GRCh38
                 continue
 
             # find the SNPs that are being remapped for this mapping
@@ -898,7 +904,7 @@ def _load_snps_helper(
         discrepant_genotypes_threshold,
         save_output,
     ):
-        print("Loading " + os.path.relpath(file))
+        logger.debug("Loading " + os.path.relpath(file))
         discrepant_positions, discrepant_genotypes = self._add_snps(
             SNPs(file),
             discrepant_snp_positions_threshold,
@@ -1044,12 +1050,12 @@ def _add_snps(
         source = [s.strip() for s in snps._source.split(",")]
 
         if not snps._build_detected:
-            print("build not detected, assuming build {}".format(snps._build))
+            logger.debug("build not detected, assuming build {}".format(snps._build))
 
         if not self._build:
             self._build = build
         elif self._build != build:
-            print(
+            logger.debug(
                 "build / assembly mismatch between current build of SNPs and SNPs being loaded"
             )
 
@@ -1073,7 +1079,7 @@ def _add_snps(
                 prefix = "{}_".format(clean_str(self._name))
 
             if 0 < len(discrepant_positions) < discrepant_snp_positions_threshold:
-                print(
+                logger.debug(
                     "{} SNP positions were discrepant; keeping original positions".format(
                         str(len(discrepant_positions))
                     )
@@ -1089,7 +1095,7 @@ def _add_snps(
                         ),
                     )
             elif len(discrepant_positions) >= discrepant_snp_positions_threshold:
-                print(
+                logger.debug(
                     "too many SNPs differ in position; ensure same genome build is being used"
                 )
                 return discrepant_positions, discrepant_genotypes
@@ -1138,7 +1144,7 @@ def _add_snps(
             ]
 
             if 0 < len(discrepant_genotypes) < discrepant_genotypes_threshold:
-                print(
+                logger.debug(
                     "{} SNP genotypes were discrepant; marking those as null".format(
                         str(len(discrepant_genotypes))
                     )
@@ -1154,7 +1160,7 @@ def _add_snps(
                         ),
                     )
             elif len(discrepant_genotypes) >= discrepant_genotypes_threshold:
-                print(
+                logger.debug(
                     "too many SNPs differ in their genotype; ensure file is for same "
                     "individual"
                 )

src/snps/io.py

@@ -46,6 +46,10 @@
 import snps
 from snps.utils import save_df_as_csv, clean_str
 
+import logging
+
+logger = logging.getLogger(__name__)
+
 
 class Reader:
     """ Class for reading and parsing raw data / genotype files. """
@@ -155,7 +159,7 @@ def __call__(self):
             else:
                 return pd.DataFrame(), ""
         except Exception as err:
-            print(err)
+            logger.warning(err)
             return pd.DataFrame(), ""
 
     @classmethod
@@ -686,7 +690,7 @@ def read_vcf(self, file):
 
             # snps does not yet support multi-sample vcf.
             if len(vcf_reader.samples) > 1:
-                print(
+                logger.debug(
                     "Multiple samples detected in the vcf file, please use a single sample vcf."
                 )
                 return df, "vcf"

src/snps/resources.py

@@ -62,6 +62,10 @@
 from snps.ensembl import EnsemblRestClient
 from snps.utils import create_dir, Singleton
 
+import logging
+
+logger = logging.getLogger(__name__)
+
 
 class Resources(metaclass=Singleton):
     """ Object used to manage resources required by `snps`. """
@@ -131,7 +135,7 @@ def get_reference_sequences(
         valid_assemblies = ["NCBI36", "GRCh37", "GRCh38"]
 
         if assembly not in valid_assemblies:
-            print("Invalid assembly")
+            logger.debug("Invalid assembly")
             return {}
 
         if not self._reference_chroms_available(assembly, chroms):
@@ -297,7 +301,7 @@ def _load_assembly_mapping_data(filename):
 
             return assembly_mapping_data
         except Exception as err:
-            print(err)
+            logger.warning(err)
             return {}
 
     def _get_paths_reference_sequences(
@@ -472,14 +476,14 @@ def _get_path_assembly_mapping_data(
         if not os.path.exists(destination) or not self._all_chroms_in_tar(
             chroms, destination
         ):
-            print("Downloading {}".format(os.path.relpath(destination)))
+            logger.debug("Downloading {}".format(os.path.relpath(destination)))
 
             try:
                 self._download_assembly_mapping_data(
                     destination, chroms, source_assembly, target_assembly, retries
                 )
             except Exception as err:
-                print(err)
+                logger.warning(err)
                 return ""
 
         return destination
@@ -527,7 +531,7 @@ def _all_chroms_in_tar(self, chroms, filename):
                 if chrom + ".json" not in members:
                     return False
         except Exception as err:
-            print(err)
+            logger.warning(err)
             return False
 
         return True
@@ -554,7 +558,7 @@ def _load_codigo46_resources(self, rsid_map, chrpos_map):
 
             return d
         except Exception as err:
-            print(err)
+            logger.warning(err)
             return {}
 
     def _get_path_codigo46_rsid_map(self):
@@ -613,7 +617,7 @@ def _download_file(self, url, filename, compress=False, timeout=30):
                     else:
                         f.write(data)
             except urllib.error.URLError as err:
-                print(err)
+                logger.warning(err)
                 destination = ""
                 # try HTTP if an FTP error occurred
                 if "ftp://" in url:
@@ -624,7 +628,7 @@ def _download_file(self, url, filename, compress=False, timeout=30):
                         timeout=timeout,
                     )
             except Exception as err:
-                print(err)
+                logger.warning(err)
                 return ""
 
         return destination
@@ -638,7 +642,7 @@ def _print_download_msg(path):
         path : str
             path to file being downloaded
         """
-        print("Downloading " + os.path.relpath(path))
+        logger.debug("Downloading " + os.path.relpath(path))
 
 
 class ReferenceSequence:

src/snps/utils.py

@@ -41,6 +41,9 @@
 import pandas as pd
 
 import snps
+import logging
+
+logger = logging.getLogger(__name__)
 
 
 class Parallelizer:
@@ -106,7 +109,7 @@ def create_dir(path):
     try:
         os.makedirs(path, exist_ok=True)
     except Exception as err:
-        print(err)
+        logger.warning(err)
         return False
 
     if os.path.exists(path):
@@ -156,7 +159,7 @@ def save_df_as_csv(
                 destination = filename
             else:
                 destination = os.path.join(path, filename)
-                print("Saving " + os.path.relpath(destination))
+                logger.debug("Saving " + os.path.relpath(destination))
 
             if prepend_info:
                 s = (
@@ -190,10 +193,10 @@ def save_df_as_csv(
 
             return destination
         except Exception as err:
-            print(err)
+            logger.warning(err)
             return ""
     else:
-        print("no data to save...")
+        logger.debug("no data to save...")
         return ""