IRI in search working (#78)

* Fix ModuleNotFoundError for url_utils in PyPI distribution

* Fixed search not working respecting languages

* IRI in search doesn't use search but goes directly to concept

* formatting

---------

Co-authored-by: Valentin Starlinger <[email protected]>

Author: vstarlinger

src/py_semantic_taxonomy/adapters/routers/web_router.py

@@ -2,6 +2,7 @@
 from pathlib import Path as PathLib
 from urllib.parse import quote, unquote, urlencode
 
+import rfc3987
 import structlog
 from fastapi import APIRouter, Depends, HTTPException, Path, Request
 from fastapi.responses import HTMLResponse, RedirectResponse
@@ -22,6 +23,15 @@
 router = APIRouter(prefix="/web", include_in_schema=False)
 
 
+def _is_iri(query: str) -> bool:
+    """Check if query string is a valid HTTP/HTTPS IRI."""
+    try:
+        parsed = rfc3987.parse(query.strip(), rule="IRI")
+        return parsed.get("scheme") in ("http", "https")
+    except ValueError:
+        return False
+
+
 def value_for_language(value: list[dict[str, str]], lang: str) -> str:
     """Get the `@value` for a list of multilingual strings with correct `@language` value"""
     for dct in value:
@@ -189,10 +199,16 @@ async def web_concept_scheme_view(
             },
         )
     except de.ConceptSchemeNotFoundError:
-        raise HTTPException(status_code=404, detail=f"Concept Scheme with IRI `{iri}` not found")
+        raise HTTPException(
+            status_code=404, detail=f"Concept Scheme with IRI `{iri}` not found"
+        )
     except de.ConceptSchemesNotInDatabase as e:
-        logger.error("Database error while fetching concept scheme", iri=iri, error=str(e))
-        raise HTTPException(status_code=500, detail="Database error while fetching concept scheme")
+        logger.error(
+            "Database error while fetching concept scheme", iri=iri, error=str(e)
+        )
+        raise HTTPException(
+            status_code=500, detail="Database error while fetching concept scheme"
+        )
 
 
 def concept_view_url(
@@ -264,7 +280,9 @@ async def get_concept_and_link(iri: str) -> (str, de.Concept | str):
             except de.ConceptNotFoundError:
                 return iri, iri
 
-        relationships = await service.relationships_get(iri=decoded_iri, source=True, target=True)
+        relationships = await service.relationships_get(
+            iri=decoded_iri, source=True, target=True
+        )
         broader = [
             (await get_concept_and_link(obj.target))
             for obj in relationships
@@ -277,7 +295,8 @@ async def get_concept_and_link(iri: str) -> (str, de.Concept | str):
         ]
 
         scheme_list = [
-            (request.url_for("web_concept_view", iri=quote(s["@id"])), s) for s in concept.schemes
+            (request.url_for("web_concept_view", iri=quote(s["@id"])), s)
+            for s in concept.schemes
         ]
 
         associations = await service.association_get_all(source_concept_iri=concept.id_)
@@ -286,29 +305,27 @@ async def get_concept_and_link(iri: str) -> (str, de.Concept | str):
             for target in obj.target_concepts:
                 try:
                     url, assoc_concept = await get_concept_and_link(target["@id"])
-                    formatted_associations.append(
-                        {
-                            "url": url,
-                            "obj": assoc_concept,
-                            "conditional": None,
-                            "conversion": target.get(
-                                "http://qudt.org/3.0.0/schema/qudt/conversionMultiplier"
-                            ),
-                        }
-                    )
+                    formatted_associations.append({
+                        "url": url,
+                        "obj": assoc_concept,
+                        "conditional": None,
+                        "conversion": target.get(
+                            "http://qudt.org/3.0.0/schema/qudt/conversionMultiplier"
+                        ),
+                    })
                 except de.ConceptNotFoundError:
-                    formatted_associations.append(
-                        {
-                            "url": target["@id"],
-                            "obj": target["@id"],
-                            "conditional": None,
-                            "conversion": target.get(
-                                "http://qudt.org/3.0.0/schema/qudt/conversionMultiplier"
-                            ),
-                        }
-                    )
-
-        languages = [(request.url, Language.get(language).display_name(language).title())] + [
+                    formatted_associations.append({
+                        "url": target["@id"],
+                        "obj": target["@id"],
+                        "conditional": None,
+                        "conversion": target.get(
+                            "http://qudt.org/3.0.0/schema/qudt/conversionMultiplier"
+                        ),
+                    })
+
+        languages = [
+            (request.url, Language.get(language).display_name(language).title())
+        ] + [
             (
                 concept_view_url(
                     request,
@@ -343,7 +360,9 @@ async def get_concept_and_link(iri: str) -> (str, de.Concept | str):
     except de.ConceptNotFoundError:
         raise HTTPException(status_code=404, detail=f"Concept with IRI `{iri}` not found")
     except de.ConceptSchemesNotInDatabase as e:
-        logger.error("Database error while fetching concept", iri=decoded_iri, error=str(e))
+        logger.error(
+            "Database error while fetching concept", iri=decoded_iri, error=str(e)
+        )
         raise HTTPException(status_code=500, detail="Database error while fetching concept")
 
 
@@ -357,9 +376,38 @@ async def web_search(
     language: str = "en",
     semantic: bool = True,
     search_service=Depends(get_search_service),
+    graph_service=Depends(get_graph_service),
     settings=Depends(get_settings),
 ) -> HTMLResponse:
     """Search for concepts."""
+    # Check if query is an IRI and attempt direct lookup
+    if query and _is_iri(query):
+        # Try to get concept directly
+        try:
+            concept = await graph_service.concept_get(iri=query)
+            # If found, redirect to concept page
+            return RedirectResponse(
+                url=concept_view_url(
+                    request,
+                    concept.id_,
+                    concept.schemes[0]["@id"],
+                    language,
+                ),
+                status_code=303,  # See Other
+            )
+        except de.ConceptNotFoundError:
+            # Not a concept, try concept scheme
+            try:
+                concept_scheme = await graph_service.concept_scheme_get(iri=query)
+                # If found, redirect to concept scheme page
+                return RedirectResponse(
+                    url=concept_scheme_view_url(request, concept_scheme.id_, language),
+                    status_code=303,  # See Other
+                )
+            except de.ConceptSchemeNotFoundError:
+                # IRI not found in database, fall through to regular search
+                pass
+
     try:
         results = []
         if query:

tests/integration/test_web_ui.py

@@ -85,3 +85,107 @@ async def test_web_search_empty_query(sqlite, anonymous_client):
 
     html_content = response.text
     assert "Start searching" in html_content or "Search" in html_content
+
+
+@pytest.mark.postgres
+async def test_web_search_with_concept_iri_redirects_to_concept(
+    postgres, anonymous_client, cn_db_engine, cn
+):
+    """Test that searching for a concept IRI redirects directly to the concept page."""
+    concept_iri = cn.concept_top["@id"]
+
+    response = await anonymous_client.get(
+        "/web/search/",
+        params={"query": concept_iri, "language": "de"},
+        follow_redirects=False,
+    )
+
+    # Should redirect with 303 See Other
+    assert response.status_code == 303
+
+    # Should redirect to the concept view page with language preserved
+    redirect_url = response.headers["location"]
+    assert "/web/concept/" in redirect_url
+    assert quote(concept_iri) in redirect_url
+    assert "language=de" in redirect_url
+
+
+@pytest.mark.postgres
+async def test_web_search_with_concept_scheme_iri_redirects(
+    postgres, anonymous_client, cn_db_engine, cn
+):
+    """Test that searching for a concept scheme IRI redirects to the concept scheme page."""
+    scheme_iri = cn.scheme["@id"]
+
+    response = await anonymous_client.get(
+        "/web/search/",
+        params={"query": scheme_iri, "language": "en"},
+        follow_redirects=False,
+    )
+
+    # Should redirect with 303 See Other
+    assert response.status_code == 303
+
+    # Should redirect to the concept scheme view page
+    redirect_url = response.headers["location"]
+    assert "/web/concept_scheme/" in redirect_url
+    assert quote(scheme_iri) in redirect_url
+    assert "language=en" in redirect_url
+
+
+@pytest.mark.postgres
+async def test_web_search_with_nonexistent_iri_shows_search_page(
+    postgres, anonymous_client, cn_db_engine
+):
+    """Test that searching for an IRI that doesn't exist falls back to search (or error if not configured)."""
+    nonexistent_iri = "http://example.com/nonexistent/concept/12345"
+
+    response = await anonymous_client.get(
+        "/web/search/",
+        params={"query": nonexistent_iri, "language": "en"},
+        follow_redirects=True,
+    )
+
+    # Should show search page (200) if search engine configured
+    # or error (503) if search engine not configured
+    # Important: should NOT redirect (303) since concept doesn't exist
+    assert response.status_code in (200, 503)
+    if response.status_code == 503:
+        assert "Search engine not available" in response.text or "503" in response.text
+
+
+@pytest.mark.postgres
+@pytest.mark.typesense
+async def test_web_search_with_nonexistent_iri_falls_back_to_search(
+    postgres, typesense, anonymous_client, cn_db_engine, cn
+):
+    """Test that searching for an IRI that doesn't exist falls back to text search."""
+    nonexistent_iri = "http://example.com/nonexistent/concept/12345"
+
+    response = await anonymous_client.get(
+        "/web/search/",
+        params={"query": nonexistent_iri, "language": "en"},
+        follow_redirects=True,
+    )
+
+    # Should show search page (no redirect) since concept doesn't exist
+    assert response.status_code == 200
+    # Should show the search interface (may or may not have results from text search)
+    assert "Search" in response.text or "search" in response.text
+
+
+async def test_web_search_with_regular_text_not_treated_as_iri(anonymous_client):
+    """Test that regular search text is not treated as an IRI (no database required)."""
+    response = await anonymous_client.get(
+        "/web/search/",
+        params={"query": "test query", "language": "en"},
+        follow_redirects=True,
+    )
+
+    # Should get error because search engine not configured (503)
+    # or show search page if engine is configured (200)
+    # The important thing is it doesn't try to treat it as an IRI and redirect
+    assert response.status_code in (200, 503)
+    if response.status_code == 503:
+        assert "Search engine not available" in response.text or "503" in response.text
+