feat: add source imagery type detection and update prepare_data function

Author: kshitijrajsharma

backend/core/models.py

@@ -1,3 +1,5 @@
+import re
+
 from django.contrib.gis.db import models as geomodels
 from django.contrib.postgres.fields import ArrayField
 from django.core.exceptions import ValidationError
@@ -112,6 +114,36 @@ class Training(models.Model):
     freeze_layers = models.BooleanField(default=False)
     centroid = geomodels.PointField(srid=4326, null=True, blank=True)
 
+    def get_source_imagery_type(self):
+        """
+        Returns: 'TILEJSON', 'XYZ', 'TMS', or 'UNKNOWN'
+        """
+        ## source ref here : https://github.com/hotosm/fAIr/blob/develop/frontend/src/utils/regex-utils.ts
+        if not self.source_imagery:
+            return "UNKNOWN"
+        xyz_pattern = re.compile(
+            r"^https?:\/\/[^\/]+(?:\/[^\/]+)*\/\{z\}\/\{x\}\/\{y\}(?:@[0-9a-z]+)?(?:\.(jpg|png|jpeg|webp))?(?:\?.*)?$",
+            re.IGNORECASE,
+        )
+        tms_pattern = re.compile(
+            r"^https?:\/\/[^\/]+(?:\/[^\/]+)*\/\{z\}\/\{x\}\/\{-y\}(?:@[0-9a-z]+)?(?:\.(jpg|png|jpeg|webp))?(?:\?.*)?$",
+            re.IGNORECASE,
+        )
+        tilejson_pattern = re.compile(
+            r"^https?:\/\/[^\/]+(?:\/[^\/?#]+)*\/[^\/?#]+\.json(?:\?.*)?$",
+            re.IGNORECASE,
+        )
+
+        url = self.source_imagery
+        if tilejson_pattern.match(url):
+            return "TILEJSON"
+        elif xyz_pattern.match(url):
+            return "XYZ"
+        elif tms_pattern.match(url):
+            return "TMS"
+        else:
+            return "UNKNOWN"
+
 
 class Feedback(models.Model):
     ACTION_CHOICES = (

backend/core/tasks.py

@@ -316,7 +316,9 @@ def upload_to_s3(path, parent=None):
     ).upload(path)
 
 
-def prepare_data(inst, dataset_id, feedback, zoom_level, imagery, input_path):
+def prepare_data(
+    inst, dataset_id, feedback, zoom_level, imagery, input_path, imagery_type
+):
     from geomltoolkits.downloader import tms as TMSDownloader
 
     safe_rmtree(input_path)
@@ -349,6 +351,7 @@ def prepare_data(inst, dataset_id, feedback, zoom_level, imagery, input_path):
                     georeference=False,
                     crs="3857",
                     extension="png",
+                    is_tilejson=("TILEJSON" == imagery_type),
                 )
             )
     input_path = pathlib.Path(input_path)
@@ -457,7 +460,13 @@ def train_model(
             )
 
             input_path, aoi_ser, labels = prepare_data(
-                inst, dataset_id, feedback, zoom_level, source_imagery, input_path
+                inst,
+                dataset_id,
+                feedback,
+                zoom_level,
+                source_imagery,
+                input_path,
+                inst.get_source_imagery_type,
             )
             output_path = os.path.join(
                 settings.TRAINING_WORKSPACE,