rapid archive url bug fix (#20)

Author: shashankbrgowda

app/api/resources/rapid_view.py

@@ -11,10 +11,10 @@
     StableIdResolverResponse
 from app.api.utils.commons import build_stable_id_resolver_content, is_json_request
 from app.api.utils.metadata import get_genome_id_from_assembly_accession_id, get_metadata
-from app.api.utils.rapid import format_assembly_accession, construct_rapid_archive_url, construct_url, \
-    generate_rapid_id_page, generate_rapid_page
+from app.api.utils.rapid import format_assembly_accession, construct_url, \
+    generate_rapid_id_page, generate_rapid_page, construct_rapid_archive_url
 from app.api.utils.search import get_search_results
-from app.core.config import ENSEMBL_URL, RAPID_ARCHIVE_URL
+from app.core.config import ENSEMBL_URL
 from app.core.logging import InterceptHandler
 
 logging.getLogger().handlers = [InterceptHandler()]
@@ -27,7 +27,7 @@ async def resolve_rapid_stable_id(request: Request, stable_id: str):
     # Handle only gene stable id for now
     params = SearchPayload(stable_id=stable_id, type="gene", per_page=10)
     search_results = get_search_results(params)
-    rapid_archive_url = f"{RAPID_ARCHIVE_URL}/id/{stable_id}"
+    rapid_archive_url = construct_rapid_archive_url(request)
 
     if not search_results or not search_results.get("matches"):
         if is_json_request(request):
@@ -77,6 +77,7 @@ async def resolve_rapid_help(request: Request, subpath: str = ""):
         response_type=RapidResolverHtmlResponseType.HELP,
         code=308,
         resolved_url=f"{ENSEMBL_URL}/help",
+        rapid_archive_url=construct_rapid_archive_url(request)
     )
     return rapid_resolved_response(response, request)
 
@@ -87,6 +88,7 @@ async def resolve_rapid_blast(request: Request):
         response_type=RapidResolverHtmlResponseType.BLAST,
         code=308,
         resolved_url=f"{ENSEMBL_URL}/blast",
+        rapid_archive_url=construct_rapid_archive_url(request)
     )
     return rapid_resolved_response(response, request)
 
@@ -97,13 +99,15 @@ async def resolve_rapid_blast(request: Request):
 async def resolve_species(
     request: Request, species_url_name: str, subpath: str = "", r: str = Query(None)
 ):
+    rapid_archive_url = construct_rapid_archive_url(request)
     # Check if its blast redirect
     if "tools/blast" in subpath.lower():
         response = RapidResolverResponse(
             response_type=RapidResolverHtmlResponseType.BLAST,
             code=308,
             resolved_url=f"{ENSEMBL_URL}/blast",
             species_name=species_url_name,
+            rapid_archive_url=rapid_archive_url
         )
         return rapid_resolved_response(response, request)
 
@@ -117,6 +121,7 @@ async def resolve_species(
                 resolved_url=f"{ENSEMBL_URL}/species-selector",
                 message="Invalid input accession ID",
                 species_name=species_url_name,
+                rapid_archive_url=rapid_archive_url
             )
             return rapid_resolved_response(input_error_response, request)
 
@@ -130,7 +135,6 @@ async def resolve_species(
             query_params = parse_qs(query_string, separator=";")
 
             url = construct_url(genome_id, subpath, query_params)
-            rapid_archive_url = construct_rapid_archive_url(species_url_name, subpath, query_params)
             response = RapidResolverResponse(
                 response_type=RapidResolverHtmlResponseType.INFO,
                 code=308,
@@ -151,6 +155,7 @@ async def resolve_species(
             resolved_url=f"{ENSEMBL_URL}/species-selector",
             message=e.detail,
             species_name=species_url_name,
+            rapid_archive_url=rapid_archive_url
         )
         return rapid_resolved_response(response, request)
     except Exception as e:
@@ -161,6 +166,7 @@ async def resolve_species(
             code=500,
             resolved_url=f"{ENSEMBL_URL}/species-selector",
             message=str(e),
+            rapid_archive_url=rapid_archive_url
         )
         return rapid_resolved_response(response, request)
 
@@ -171,6 +177,7 @@ async def resolve_home(request: Request):
         response_type=RapidResolverHtmlResponseType.HOME,
         code=308,
         resolved_url=ENSEMBL_URL,
+        rapid_archive_url=construct_rapid_archive_url(request)
     )
     return rapid_resolved_response(response, request)
 

app/api/utils/rapid.py

@@ -2,6 +2,7 @@
 
 from jinja2 import Environment, FileSystemLoader
 from loguru import logger
+from fastapi import Request
 import requests
 import re
 
@@ -90,23 +91,19 @@ def construct_url(genome_id, subpath, query_params):
     raise ValueError("Invalid path")
 
 
-def construct_rapid_archive_url(species_url_name, subpath, query_params):
-    url = f"{RAPID_ARCHIVE_URL}/{species_url_name}"
-    query_params = query_params or {}
-
-    params = []
-    for key, values in query_params.items():
-        value = values[0] if values else ""
-        if value:
-            params.append(f"{key}={value}")
-
-    if subpath:
-        url += f"/{subpath}"
+def construct_rapid_archive_url(request: Request):
+    try:
+        rapid_archive_path = request.url.path.removeprefix("/rapid/")
+        if not rapid_archive_path:
+            return RAPID_ARCHIVE_URL
 
-    if params:
-        url += "?" + ";".join(params) # ; separator for rapid
+        if request.url.query:
+            rapid_archive_path = f"{rapid_archive_path}?{request.url.query}"
 
-    return url
+        return f"{RAPID_ARCHIVE_URL}/{rapid_archive_path}"
+    except Exception as e:
+        logger.warning(f"Error constructing rapid archive url: {e}")
+        return RAPID_ARCHIVE_URL
 
 
 def generate_html_content(response, page):

tests/test_rapid_utils.py

@@ -0,0 +1,29 @@
+import unittest
+from unittest.mock import MagicMock
+from fastapi import Request
+
+from app.api.utils.rapid import construct_rapid_archive_url
+from app.core.config import RAPID_ARCHIVE_URL
+
+
+class TestRapidUtils(unittest.TestCase):
+    def _assert_rapid_url(self, rapid_path, query_string=""):
+        req = MagicMock(spec=Request)
+        req.url.path = f"/rapid/{rapid_path}"
+        req.url.query = query_string
+        res = construct_rapid_archive_url(req)
+        expected_url = RAPID_ARCHIVE_URL if not rapid_path else f"{RAPID_ARCHIVE_URL}/{rapid_path}"
+        if query_string:
+            expected_url = f"{expected_url}?{query_string}"
+        self.assertEqual(res, expected_url)
+
+    def test_construct_rapid_archive_url(self):
+        self._assert_rapid_url("")
+        self._assert_rapid_url("info/index.html")
+        self._assert_rapid_url("Multi/Tools/Blast")
+        self._assert_rapid_url("Homo_sapiens_GCA_009914755.4/Tools/Blast")
+        self._assert_rapid_url("id/ENSG00000221914.11")
+        self._assert_rapid_url("Camarhynchus_parvulus_GCA_902806625.1")
+        self._assert_rapid_url("Camarhynchus_parvulus_GCA_902806625.1/Location/Genome")
+        self._assert_rapid_url("Homo_sapiens_GCA_009914755.4/Location/View", "r=1:56873660-58424710")
+        self._assert_rapid_url("Homo_sapiens_GCA_009914755.4/Gene/Summary", "r=1:56873660-58424710")