From 2e9f73309da4ec3cc11ab87d8f46a17523af0c0d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Ricardo=20Montan=CC=83ana?= <rmontanana@gmail.com>
Date: Thu, 10 Feb 2022 11:57:13 +0100
Subject: [PATCH] Enhance list with sort key and summary

---
 src/Results.py | 32 +++++++++++++++++++++++++-------
 src/list.py    | 14 +++++++++++---
 src/summary.py |  1 +
 3 files changed, 37 insertions(+), 10 deletions(-)

diff --git a/src/Results.py b/src/Results.py
index 4c8440f..e85812c 100644
--- a/src/Results.py
+++ b/src/Results.py
@@ -700,14 +700,22 @@ class Summary:
             self.datasets[result] = report.lines
             self.data.append(entry)
 
-    def list_results(self, score=None, model=None) -> None:
+    def list_results(
+        self, score=None, model=None, input_data=None, sort_key="date"
+    ) -> None:
         """Print the list of results"""
-        data = self.data.copy()
+        data = self.data.copy() if input_data is None else input_data
         if score:
             data = [x for x in data if x["score"] == score]
         if model:
             data = [x for x in data if x["model"] == model]
-        data = sorted(data, key=lambda x: x["date"], reverse=True)
+        data = sorted(
+            data,
+            key=lambda x: 0.0
+            if type(x[sort_key]) is float and math.isnan(x[sort_key])
+            else x[sort_key],
+            reverse=True,
+        )
         max_file = max(len(x["file"]) for x in data)
         max_title = max(len(x["title"]) for x in data)
         print(TextColor.LINE1, end="")
@@ -776,9 +784,7 @@ class Summary:
         print("*" + whites(length - 2))
         print("*" * length)
 
-    def best_result(
-        self, criterion=None, value=None, score="accuracy"
-    ) -> dict:
+    def best_results(self, criterion=None, value=None, score="accuracy", n=10):
         # First filter the same score results (accuracy, f1, ...)
         haystack = [x for x in self.data if x["score"] == score]
         haystack = (
@@ -791,11 +797,16 @@ class Summary:
                 haystack,
                 key=lambda x: -1.0 if math.isnan(x["metric"]) else x["metric"],
                 reverse=True,
-            )[0]
+            )[:n]
             if len(haystack) > 0
             else {}
         )
 
+    def best_result(
+        self, criterion=None, value=None, score="accuracy"
+    ) -> dict:
+        return self.best_results(criterion, value, score)[0]
+
     def best_results_datasets(self, score="accuracy") -> dict:
         """Get the best results for each dataset"""
         dt = Datasets()
@@ -814,3 +825,10 @@ class Summary:
                         entry["title"],
                     )
         return best_results
+
+    def show_top(self, score="accuracy", n=10):
+        self.list_results(
+            score=score,
+            input_data=self.best_results(score=score, n=n),
+            sort_key="metric",
+        )
diff --git a/src/list.py b/src/list.py
index eabcf74..9f118ea 100755
--- a/src/list.py
+++ b/src/list.py
@@ -29,13 +29,21 @@ def parse_arguments():
         required=False,
         help="model used in experiment",
     )
+    ap.add_argument(
+        "-k",
+        "--key",
+        type=str,
+        required=False,
+        default="date",
+        help="key to sort results",
+    )
     args = ap.parse_args()
 
-    return (args.excel, args.score, args.model)
+    return (args.excel, args.score, args.model, args.key)
 
 
-(excel, score, model) = parse_arguments()
+(excel, score, model, key) = parse_arguments()
 
 data = Summary()
 data.acquire()
-data.list_results(score, model)
+data.list_results(score=score, model=model, sort_key=key)
diff --git a/src/summary.py b/src/summary.py
index 3f3e4a7..e78408e 100755
--- a/src/summary.py
+++ b/src/summary.py
@@ -60,5 +60,6 @@ for metric in metrics:
     summary.show_result(
         summary.best_result(score=metric), title=f"BEST RESULT of {metric}"
     )
+    summary.show_top(score=metric, n=10)
 if list_results:
     summary.list_results()