Merge pull request #99 from USEPA/develop

Release v1.1.1

lciafmt/__init__.py

@@ -8,7 +8,6 @@
 """
 
 import json
-import pkg_resources
 from typing import Union
 
 import pandas as pd
@@ -77,7 +76,7 @@ def get_class(name: str):
 
 def supported_methods() -> list:
     """Return a list of dictionaries of supported method meta data."""
-    json_file = pkg_resources.resource_filename("lciafmt", 'data/methods.json')
+    json_file = util.datapath / 'methods.json'
     with open(json_file, "r", encoding="utf-8") as f:
         return json.load(f)
 
@@ -150,7 +149,15 @@ def map_flows(df: pd.DataFrame, system=None, mapping=None,
     mapper = fmap.Mapper(df, system=system, mapping=mapping,
                          preserve_unmapped=preserve_unmapped,
                          case_insensitive=case_insensitive)
-    return mapper.run()
+    mapped = mapper.run()
+    x = mapped[mapped[['Method', 'Indicator', 'Flowable', 'Flow UUID']
+                      ].duplicated(keep=False)]
+    duplicates = list(set(zip(x.Indicator, x.Flowable)))
+    util.log.warn(f'Identified duplicate factors for {len(duplicates)} '
+                  f'flow/indicator combinations and {len(x)} factors.')
+    util.log.debug(f'{duplicates}')
+    util.log.warn(f'Use collapse_indicators() to drop these duplicates.')
+    return mapped
 
 
 def supported_mapping_systems() -> list:

lciafmt/ipcc.py

@@ -40,3 +40,4 @@ def get() -> pd.DataFrame:
     method = lciafmt.Method.IPCC
     df = lciafmt.get_method(method)
     mapped_df = lciafmt.map_flows(df, system=method.get_metadata().get('mapping'))
+    mapped_df2 = lciafmt.util.collapse_indicators(mapped_df)

lciafmt/util.py

@@ -12,7 +12,6 @@
 import pandas as pd
 import numpy as np
 import yaml
-import pkg_resources
 from pathlib import Path
 from esupy.processed_data_mgmt import Paths, FileMeta, load_preprocessed_output,\
     write_df_to_file, write_metadata_to_file, download_from_remote, \
@@ -22,7 +21,7 @@
 
 
 # set version number of package, needs to be updated with setup.py
-pkg_version_number = '1.1.0'
+pkg_version_number = '1.1.1'
 MODULEPATH = Path(__file__).resolve().parent
 datapath = MODULEPATH / 'data'
 
@@ -36,7 +35,6 @@
 paths.local_path = paths.local_path / 'lciafmt'
 OUTPUTPATH = paths.local_path
 
-pkg = pkg_resources.get_distribution('lciafmt')
 GIT_HASH = get_git_hash()
 
 method_metadata = {
@@ -58,7 +56,7 @@ def set_lcia_method_meta(method_id):
         lcia_method_meta.category = ""
     else:
         lcia_method_meta.name_data = method_id
-    lcia_method_meta.tool = pkg.project_name
+    lcia_method_meta.tool = "lciafmt"
     lcia_method_meta.tool_version = pkg_version_number
     lcia_method_meta.ext = write_format
     lcia_method_meta.git_hash = GIT_HASH
@@ -155,11 +153,13 @@ def collapse_indicators(df) -> pd.DataFrame:
     """
     cols = ['Method', 'Indicator', 'Indicator unit', 'Flow UUID']
     duplicates = df[df.duplicated(subset=cols, keep=False)]
-    cols_to_keep = [c for c in df.columns.values.tolist()]
-    cols_to_keep.remove('Characterization Factor')
-    df2 = df.groupby(cols_to_keep, as_index=False)['Characterization Factor'].mean()
-    log.info(str(len(duplicates))+" duplicate factors consolidated to "
-             + str(len(duplicates)-(len(df)-len(df2))))
+    cols_to_keep = [c for c in df.columns.values.tolist() if 
+                    c not in ('Characterization Factor', 'CAS No')]
+    df2 = (df.groupby(cols_to_keep, as_index=False)
+             .agg({'Characterization Factor': 'mean',
+                   'CAS No': 'first'}))
+    log.info(f'{len(duplicates)} duplicate factors consolidated to '
+             f'{(len(duplicates)-(len(df)-len(df2)))}')
 
     return df2
 

setup.py

@@ -2,7 +2,7 @@
 
 setup(
     name="lciafmt",
-    version="1.1.0",
+    version="1.1.1",
     packages=["lciafmt"],
     package_dir={'lciafmt': 'lciafmt'},
     package_data={'lciafmt': ["data/*.*"]},