update

scripts/check_dataset_time.py

@@ -0,0 +1,53 @@
+import os
+import sys
+from collections import defaultdict
+
+import yaml
+
+
+def main(dataset_dir):
+    dirs = os.listdir(dataset_dir)
+    dirs = list(
+        filter(lambda x: os.path.isdir(os.path.join(dataset_dir, x)), dirs)
+    )
+
+    problem_dirs = set()
+    problem_count = defaultdict(int)
+    for dir in dirs:
+        with open(os.path.join(dataset_dir, dir, "data.yaml"), "r") as f:
+            data = yaml.full_load(f)
+        for query_i, query in enumerate(data["query_data"]):
+            voices = sorted(query["voice"], key=lambda x: x["start"])
+            if voices != query["voice"]:
+                print("-----", dir)
+            if voices[0]["start"] > voices[0]["end"]:
+                print(
+                    "err1: %s 第%s个query的第%d个voice的start大于end: %s"
+                    % (dir, query_i, 0, voices[0]["answer"])
+                )
+                problem_dirs.add(dir)
+            for voice_i in range(1, len(voices)):
+                voice = voices[voice_i]
+                if voice["start"] > voice["end"]:
+                    print(
+                        "err1: %s 第%s个query的第%d个voice的start大于end: %s"
+                        % (dir, query_i, voice_i, voice["answer"])
+                    )
+                    problem_dirs.add(dir)
+                if voice["start"] < voices[voice_i - 1]["end"]:
+                    print(
+                        "err2: %s 第%s个query的第%d个voice的start小于前一个voice的end: %s"
+                        % (dir, query_i, voice_i, voice["answer"])
+                    )
+                    problem_dirs.add(dir)
+                    problem_count[dir] += 1
+    print(len(dirs))
+    print(problem_dirs)
+    print(problem_count)
+
+
+if __name__ == "__main__":
+    if len(sys.argv) < 2:
+        print("指定 测试数据集文件夹")
+        sys.exit(1)
+    main(sys.argv[1])