DAS-2446 - Allow HOSS to return an unexecuted OPeNDAP URL when requested.

CHANGELOG.md

@@ -1,3 +1,10 @@
+## [v1.2.0] - 2025-12-11
+
+### Added
+
+- Allow HOSS to format and return an unexecuted OPeNDAP URL instead of a subset
+  file when requested by the user.
+
 ## [v1.1.17] - 2025-11-25
 
 ### Changed

docker/service_version.txt

@@ -1 +1 @@
-1.1.17
+1.2.0

hoss/adapter.py

@@ -37,7 +37,11 @@
 from hoss.dimension_utilities import is_index_subset
 from hoss.harmony_log_context import set_logger
 from hoss.subset import subset_granule
-from hoss.utilities import get_file_mimetype, raise_from_hoss_exception
+from hoss.utilities import (
+    get_file_mimetype,
+    raise_from_hoss_exception,
+    unexecuted_url_requested,
+)
 
 
 class HossAdapter(BaseHarmonyAdapter):
@@ -105,32 +109,39 @@ def process_item(self, item: Item, source: Source):
 
             self.logger.info(f'Collection short name: {source.shortName}')
 
-            # Invoke service logic to retrieve subset of file from OPeNDAP
-            output_file_path = subset_granule(
+            # Invoke service logic to retrieve request output from OPeNDAP.
+            output_url = subset_granule(
                 asset.href, source, workdir, self.message, self.config
             )
 
-            # Stage the output file with a conventional filename
-            mime, _ = get_file_mimetype(output_file_path)
-            staged_filename = generate_output_filename(
-                asset.href,
-                variable_subset=source.variables,
-                ext='.nc4',
-                is_subsetted=(
-                    is_index_subset(self.message) or len(source.variables) > 0
-                ),
-            )
-            url = stage(
-                output_file_path,
-                staged_filename,
-                mime,
-                location=self.message.stagingLocation,
-                logger=self.logger,
-            )
+            # Check if an unexecuted OPeNDAP URL or subset file was requested.
+            if unexecuted_url_requested(self.message.format.mime):
+                asset_name = 'OPeNAP Request URL'
+                url = output_url
+                mime = self.message.format.mime
+            # Otherwise, stage the returned output file using a conventional
+            # filename.
+            else:
+                asset_name = generate_output_filename(
+                    asset.href,
+                    variable_subset=source.variables,
+                    ext='.nc4',
+                    is_subsetted=(
+                        is_index_subset(self.message) or len(source.variables) > 0
+                    ),
+                )
+                mime, _ = get_file_mimetype(output_url)
+                url = stage(
+                    output_url,
+                    asset_name,
+                    mime,
+                    location=self.message.stagingLocation,
+                    logger=self.logger,
+                )
 
             # Update the STAC record
             result.assets['data'] = Asset(
-                url, title=staged_filename, media_type=mime, roles=['data']
+                url, title=asset_name, media_type=mime, roles=['data']
             )
 
         except NoDataException as no_data_exception:

hoss/subset.py

@@ -37,7 +37,7 @@
 
 
 def subset_granule(
-    opendap_url: str,
+    input_granule_url: str,
     harmony_source: Source,
     output_dir: str,
     harmony_message: Message,
@@ -70,7 +70,7 @@ def subset_granule(
 
     # Produce map of variable dependencies with `earthdata-varinfo` and `.dmr`.
     varinfo = get_varinfo(
-        opendap_url,
+        input_granule_url,
         output_dir,
         harmony_source.shortName,
         harmony_message.accessToken,
@@ -96,7 +96,7 @@ def subset_granule(
     if request_is_index_subset:
         # Prefetch all dimension variables in full:
         dimensions_path = get_prefetch_variables(
-            opendap_url,
+            input_granule_url,
             varinfo,
             required_variables,
             output_dir,
@@ -159,21 +159,23 @@ def subset_granule(
         'variables_with_ranges: ' f'{format_variable_set_string(variables_with_ranges)}'
     )
 
-    # Retrieve OPeNDAP data including only the specified variables in the
-    # specified ranges.
-    output_path = get_opendap_nc4(
-        opendap_url,
+    # Retrieve requested OPeNDAP output including only the specified variables
+    # in the specified ranges. Depending on the user input, this will either
+    # be the subset data or a link to the OPeNDAP request.
+    output_url = get_opendap_nc4(
+        input_granule_url,
         variables_with_ranges,
         output_dir,
         harmony_message.accessToken,
         config,
+        harmony_message.format.mime,
     )
 
     # Fill the data outside the requested ranges for variables that cross a
     # dimensional discontinuity (for example longitude and the anti-meridian).
-    fill_variables(output_path, varinfo, required_variables, index_ranges)
+    fill_variables(output_url, varinfo, required_variables, index_ranges)
 
-    return output_path
+    return output_url
 
 
 def get_varinfo(

hoss/utilities.py

@@ -46,17 +46,25 @@ def get_opendap_nc4(
     output_dir: str,
     access_token: str,
     config: Config,
+    mimetype=None,
 ) -> str:
     """Construct a semi-colon separated string of the required variables and
     use as a constraint expression to retrieve those variables from
     OPeNDAP.
 
-    Returns the path of the downloaded granule containing those variables.
+    Returns either the unexecuted OPeNDAP request URL or the path of the
+    downloaded granule containing those variables, depending on the requested
+    mimetype.
 
     """
     constraint_expression = get_constraint_expression(required_variables)
     netcdf4_url = f'{url}.dap.nc4'
 
+    # Check if the user requested an unexecuted OPeNDAP URL.
+    if unexecuted_url_requested(mimetype):
+        get_logger().info('Returning unexecuted OPeNAP URL.')
+        return format_request_url(netcdf4_url, constraint_expression)
+
     if constraint_expression != '':
         request_data = {'dap4.ce': constraint_expression}
     else:
@@ -75,6 +83,24 @@ def get_opendap_nc4(
     return move_downloaded_nc4(output_dir, downloaded_nc4)
 
 
+def unexecuted_url_requested(mimetype: str) -> bool:
+    """Return True if the user requests the unexecuted OPeNDAP URL."""
+    opendap_url_only_mimetype = [
+        'application/x-netcdf4; profile="opendap_url"',
+        'application/x-netcdf4;profile="opendap_url"',
+        'application/x-netcdf4;profile=opendap_url',
+    ]
+    return mimetype in opendap_url_only_mimetype
+
+
+def format_request_url(url: str, constraint_expression: str):
+    """This function formats the request NetCDF4 URL to return a dap4
+    OPeNDAP request URL.
+
+    """
+    return f'{url}?dap4.ce={constraint_expression}'
+
+
 def get_constraint_expression(variables: Set[str]) -> str:
     """Take a set of variables and return a URL encoded, semi-colon separated
     DAP4 constraint expression to retrieve those variables. Each variable