chore: refactor translation extraction

scripts/translate_utils.py

@@ -3,10 +3,89 @@
 import shutil
 import yaml
 
+class TranslatableAsset:
+    translatable_attributes = []
+
+    def __init__(self, asset: dict):
+        self.asset = asset
+        for key in ASSET_FOLDER_MAPPING:
+            if key in asset:
+                self.asset_type = ASSET_FOLDER_MAPPING[key]
+                break
+
+    def extract_text(self):
+        """
+        Extract text from an asset.
+        """
+
+    def extract_text(self):
+        """
+        Remove any variables from the content which should be omitted.
+        """
+        strings = []
+        for var_path in self.translatable_attributes:
+            strings.extend(self.translate_var(self.asset, var_path.split(".")))
+
+        return strings
+
+    def translate_var(self, content, var_path: list):
+        """
+        Helper method to remove content from the content dict.
+        """
+        print("Translating var_path: ", var_path)
+        if not content:
+            return []
+        if len(var_path) == 1:
+            # If the content is a list, we need to iterate over each item in the list
+            if isinstance(content, list):
+                 strings = []
+                 for item in content:
+                     strings.append(item.get(var_path[0], ""))
+                 return strings
+             # print("Translating asset: ", self.asset_type, "var_path: ", var_path)
+            string = [content.get(var_path[0], "")]
+            return string or []
+        else:
+            # If the content is a list, we need to iterate over each item in the list
+            if isinstance(content, list):
+                strings = []
+                for item in content:
+                    strings.extend(self.translate_var(item, var_path))
+                return strings
+            if isinstance(content, dict):
+                return self.translate_var(content.get(var_path[0], {}), var_path[1:])
+            else:
+                print("Could not translate var_path: ", var_path)
+                return []
+
+
+class DashboardAsset(TranslatableAsset):
+    translatable_attributes = [
+        "dashboard_title",
+        "metadata.native_filter_configuration.name",
+        "position.meta.text",
+        "position.meta.code",
+    ]
+
+class ChartAsset(TranslatableAsset):
+    translatable_attributes = [
+        "slice_name",
+        "description",
+        "params.x_axis_label",
+        "params.y_axis_label",
+    ]
+
+class DatasetAsset(TranslatableAsset):
+    translatable_attributes = [
+        "metrics.verbose_name",
+        "columns.verbose_name",
+    ]
+
+
 ASSET_FOLDER_MAPPING = {
-    "dashboard_title": "dashboards",
-    "slice_name": "charts",
-    "table_name": "datasets",
+    "dashboard_title": ("dashboards", DashboardAsset),
+    "slice_name": ("charts", ChartAsset),
+    "table_name": ("datasets", DatasetAsset),
 }
 
 
@@ -88,11 +167,11 @@ def extract_text(asset, type):
 
         return strings
 
-    for key, value in ASSET_FOLDER_MAPPING.items():
+    for key, (asset_type, Asset) in ASSET_FOLDER_MAPPING.items():
         if key in asset:
-            strings = extract_text(asset, value)
+            strings = Asset(asset).extract_text()
             print(
-                f"Extracted {len(strings)} strings from {value} {asset.get('uuid')}"
+                f"Extracted {strings} strings from {asset_type} {asset.get('uuid')}"
             )
             return strings
 
@@ -145,7 +224,7 @@ def compile_translations(root_path):
         outfile.write("---\n")
         # If we don't use an extremely large width, the jinja in our translations
         # can be broken by newlines. So we use the largest number there is.
-        yaml.dump(all_translations, outfile, width=math.inf)
+        yaml.dump(all_translations, outfile, width=math.inf, sort_keys=True)
         outfile.write("\n{{ patch('superset-extra-asset-translations')}}\n")
 
     # We remove these files to avoid confusion about where translations are coming
@@ -178,4 +257,5 @@ def extract_translations(root_path):
     with open(translation_file, "w") as file:
         file.write(yaml.dump(translations))
 
+
     print("Done compiling translations.")