query change

8Knot/cache_manager/db_init.py

@@ -281,7 +281,8 @@ def _create_application_tables() -> None:
                 name text,
                 current_release_date text,
                 latest_release_date text,
-                libyear float4
+                libyear float4,
+                dep_age text
             )
             """
         )

8Knot/pages/repo_overview/visualizations/package_version.py

@@ -110,53 +110,11 @@ def package_version_graph(repolist):
         logging.warning(f"{VIZ_ID} - NO DATA AVAILABLE")
         return nodata_graph
 
-    # function for all data pre processing
-    df = process_data(df)
-
-    fig = create_figure(df)
-
-    logging.warning(f"{VIZ_ID} - END - {time.perf_counter() - start}")
-    return fig
-
-
-def process_data(df: pd.DataFrame):
-
-    # get max time difference to use for grouping and round for correct format
-    max_time = round(df["libyear"].max(), 3)
-
-    # create intervals for df grouper
-    intervals = [0.0, 0.001, 0.5, 1.0, max_time]
-    intervals = [val for val in intervals if val <= max_time]
-
-    # round up for formating and corner case ease
-    intervals[-1] = intervals[-1] + 0.001
-
-    # group by the time intervals
-    depend_pie = pd.DataFrame(df["libyear"].groupby(pd.cut(df["libyear"], intervals, right=False)).count()).rename(
-        columns={"libyear": "packages"}
-    )
-    depend_pie = depend_pie.reset_index().rename(columns={"libyear": "date_range"})
-    depend_pie["date_range"] = depend_pie["date_range"].astype(str)
-
-    # update column names dynamically based on specific ranges
-    for x in range(len(intervals) - 1):
-        updated_date_range = "Up to date"
-        if intervals[x] > 0 and intervals[x] < 0.5:
-            updated_date_range = "Less than 6 months"
-        if intervals[x] >= 0.5 and intervals[x] < 1:
-            updated_date_range = "6 months to year"
-        if intervals[x] == 1:
-            updated_date_range = "Greater than a year"
-        current_date_range = "[" + str(intervals[x]) + ", " + str(intervals[x + 1]) + ")"
-        depend_pie.loc[depend_pie["date_range"] == current_date_range, "date_range"] = updated_date_range
-
-    return depend_pie
-
-
-def create_figure(df: pd.DataFrame):
+    # count the number of each package grouping
+    df = pd.DataFrame(df["dep_age"].value_counts().reset_index())
 
     # graph generation
-    fig = px.pie(df, names="date_range", values="packages", color_discrete_sequence=color_seq)
+    fig = px.pie(df, names="index", values="dep_age", color_discrete_sequence=color_seq)
     fig.update_traces(
         textposition="inside",
         textinfo="percent+label",
@@ -166,4 +124,5 @@ def create_figure(df: pd.DataFrame):
     # add legend title
     fig["layout"]["legend_title"] = "Date Range"
 
+    logging.warning(f"{VIZ_ID} - END - {time.perf_counter() - start}")
     return fig

8Knot/queries/package_version_query.py

@@ -35,7 +35,14 @@ def package_version_query(self, repos):
                     rdl.name,
                     rdl.current_release_date,
                     rdl.latest_release_date,
-                    rdl.libyear
+                    rdl.libyear,
+                    case -- categorize the dependency out-of-date-edness;
+                            WHEN rdl.libyear >= 1.0 THEN 'Greater than a year'
+                            WHEN rdl.libyear < 1.0 and rdl.libyear > 0.5 THEN '6 months to year'
+                            when rdl.libyear < 0.5 and rdl.libyear > 0 then 'Less than 6 months'
+                            when rdl.libyear = 0 then 'Up to date'
+                            else 'Unclear version history'
+                    END as dep_age
                 FROM
                     repo_deps_libyear rdl
                 WHERE