Made review changes

Author: oree-xx

scripts/1-fetch/smithsonian_fetch.py

@@ -219,7 +219,7 @@ def query_smithsonian(args, session):
             " API key is set in .env",
             1,
         )
-    LOGGER.info("Fetch CC0 metrics and units from units from Smithsonain")
+    LOGGER.info("Fetch CC0 metrics and units from units from Smithsonian")
     url = "https://api.si.edu/openaccess/api/v1.0/stats"
     params = {"api_key": DATA_GOV_API_KEY}
     try:

scripts/2-process/smithsonian_process.py

@@ -109,9 +109,6 @@ def process_totals_by_records(args, count_data):
         CC0_records_with_CC0_media = int(row.CC0_RECORDS_WITH_CC0_MEDIA)
         total_objects = int(row.TOTAL_OBJECTS)
 
-        if CC0_records == 0 and CC0_records_with_CC0_media == 0:
-            continue
-
         if unit not in data:
             data[unit] = {
                 "CC0_records": 0,

scripts/3-report/smithsonian_report.py

@@ -179,7 +179,7 @@ def plot_totals_by_top10_units(args):
         " distribution of top 10 institute member across"
         " Smithsonian Institute with an average of"
         f" {average_unit:,} objects across the top 10"
-        "Institute members.",
+        " Institute members.",
     )
 
 
@@ -244,13 +244,15 @@ def plot_totals_by_records(args):
     )
     LOGGER.info(f"data file: {file_path.replace(PATHS['repo'], '.')}")
     name_label = "Unit_name"
+    data_label = "Total_objects"
     stack_labels = [
         "CC0_without_media_percentage",
         "CC0_with_media_percentage",
         "Others_percentage",
     ]
     data = shared.open_data_file(LOGGER, file_path, index_col=name_label)
-    data = data.head(10)
+    data.sort_values(data_label, ascending=True, inplace=True)
+    data = data.tail(10)
     title = "Totals by records"
     plt = plot.stacked_barh_plot(
         args=args,
@@ -275,7 +277,7 @@ def plot_totals_by_records(args):
         title,
         image_path,
         "Plots showing totals by CC0 records. This is the"
-        " breakdown of top 10 records with highest CC0 records"
+        " top 10 records with a breakdown of CC0 records"
         " without media, CC0 records with media and records"
         " that are not associated with CC0.",
     )