yogesh pr addressed

examples/zenodo_data_store_preload.ipynb

@@ -31,8 +31,8 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "CPU times: user 2.91 s, sys: 193 ms, total: 3.1 s\n",
-      "Wall time: 1.24 s\n"
+      "CPU times: user 2.5 s, sys: 422 ms, total: 2.92 s\n",
+      "Wall time: 2.65 s\n"
      ]
     }
    ],
@@ -65,8 +65,8 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "CPU times: user 33.9 ms, sys: 7.05 ms, total: 40.9 ms\n",
-      "Wall time: 40.2 ms\n"
+      "CPU times: user 40.8 ms, sys: 10.6 ms, total: 51.4 ms\n",
+      "Wall time: 61.2 ms\n"
      ]
     },
     {
@@ -90,7 +90,7 @@
        "type": "object"
       },
       "text/plain": [
-       "<xcube.util.jsonschema.JsonObjectSchema at 0x78c92d5b39e0>"
+       "<xcube.util.jsonschema.JsonObjectSchema at 0x7069ba9b39e0>"
       ]
      },
      "execution_count": 2,
@@ -113,15 +113,15 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 3,
+   "execution_count": 4,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "CPU times: user 5.04 ms, sys: 4 μs, total: 5.05 ms\n",
-      "Wall time: 4.98 ms\n"
+      "CPU times: user 8.1 ms, sys: 43 μs, total: 8.14 ms\n",
+      "Wall time: 7.66 ms\n"
      ]
     }
    ],
@@ -140,15 +140,15 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 4,
+   "execution_count": 5,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "CPU times: user 30 μs, sys: 2 μs, total: 32 μs\n",
-      "Wall time: 34.1 μs\n"
+      "CPU times: user 68 μs, sys: 11 μs, total: 79 μs\n",
+      "Wall time: 83.7 μs\n"
      ]
     },
     {
@@ -172,10 +172,10 @@
        "type": "object"
       },
       "text/plain": [
-       "<xcube.util.jsonschema.JsonObjectSchema at 0x78c92d652720>"
+       "<xcube.util.jsonschema.JsonObjectSchema at 0x7069baa53bf0>"
       ]
      },
-     "execution_count": 4,
+     "execution_count": 5,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -195,7 +195,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": 6,
    "metadata": {},
    "outputs": [
     {

xcube_zenodo/_utils.py

@@ -66,19 +66,19 @@ def get_attrs_from_record(
     return attrs
 
 
-def estimate_file_format(data_id: str) -> Union[str, None]:
+def identify_file_format(data_id: str) -> Union[str, None]:
     for key, val in MAP_FILE_EXTENSION_FORMAT.items():
         if data_id.endswith(key.lower()):
             return val
     return None
 
 
 def is_supported_file_format(data_id: str) -> bool:
-    return estimate_file_format(data_id) is not None
+    return identify_file_format(data_id) is not None
 
 
 def is_supported_compressed_file_format(data_id: str) -> bool:
-    return estimate_file_format(data_id) in COMPRESSED_FORMATS
+    return identify_file_format(data_id) in COMPRESSED_FORMATS
 
 
 def translate_data_id2fs_path(data_id: str) -> str:

xcube_zenodo/preload.py

@@ -37,7 +37,7 @@
 from xcube.core.store import DataStoreError
 from xcube.core.store import MutableDataStore
 
-from ._utils import estimate_file_format
+from ._utils import identify_file_format
 from ._utils import translate_data_id2uri
 
 LOG = logging.getLogger(__name__)
@@ -49,7 +49,7 @@ def __init__(self, data_id: str, total_size: Union[int, float]):
         self.data_id = data_id
         self.status = "Not started"
         self.progress = 0.0
-        self.message = "Preloading not started jet."
+        self.message = "Preloading not started yet."
         self.total_size = total_size
         self._callback = None
 
@@ -114,7 +114,7 @@ def close(self):
                     self._cache_store.delete_data(event.data_id)
                 else:
                     record, filename = event.data_id.split("/")
-                    format_id = estimate_file_format(event.data_id)
+                    format_id = identify_file_format(event.data_id)
                     dirname = filename.replace(f".{format_id}", "")
                     data_id_mod = f"{record}/{dirname}"
                     list_data_ids = self._cache_store.list_data_ids()
@@ -247,7 +247,7 @@ def prepare():
                     time.sleep(1)
                 self._events[i].update("File processing started", np.nan, "")
                 record, filename = data_id.split("/")
-                format_id = estimate_file_format(data_id)
+                format_id = identify_file_format(data_id)
                 dirname = filename.replace(f".{format_id}", "")
                 extract_dir = os.path.join(self._download_folder, record, dirname)
                 dss = []

xcube_zenodo/store.py

@@ -42,7 +42,7 @@
 from .constants import COMPRESSED_FORMATS
 from .constants import PRELOAD_CACHE_FOLDER
 from .preload import PreloadHandle
-from ._utils import estimate_file_format
+from ._utils import identify_file_format
 from ._utils import get_attrs_from_record
 from ._utils import is_supported_file_format
 from ._utils import is_supported_compressed_file_format
@@ -161,7 +161,7 @@ def get_open_data_params_schema(
     def open_data(
         self, data_id: str, opener_id: str = None, **open_params
     ) -> xr.Dataset:
-        format_id = estimate_file_format(data_id)
+        format_id = identify_file_format(data_id)
         if format_id in COMPRESSED_FORMATS:
             if not self.cache_store.has_data(data_id):
                 raise DataStoreError(
@@ -186,7 +186,7 @@ def preload_data(self, *data_ids: str, **preload_params) -> PreloadHandle:
         preload_params["monitor_preload"] = preload_params.get("monitor_preload", True)
         data_ids_sel = []
         for data_id in data_ids:
-            format_id = estimate_file_format(data_id)
+            format_id = identify_file_format(data_id)
             data_id_mod = data_id.replace(f".{format_id}", "/")
             list_data_ids = self.cache_store.list_data_ids()
             list_data_ids_mod = [i for i in list_data_ids if data_id_mod in i]