refactor(link): add URL validation for link status and result

CHANGELOG.md

@@ -1,3 +1,9 @@
+## v0.9.27 (2025-12-10)
+
+### Refactor
+
+- **link**: add URL validation for link status and result
+
 ## v0.9.26 (2025-12-09)
 
 ### Refactor

README.md

@@ -1,6 +1,6 @@
 # Strain Authentication and Identification Methods - saim
 
-[![release: 0.9.26](https://img.shields.io/badge/rel-0.9.26-blue.svg?style=flat-square)](https://github.com/LeibnizDSMZ/saim.git)
+[![release: 0.9.27](https://img.shields.io/badge/rel-0.9.27-blue.svg?style=flat-square)](https://github.com/LeibnizDSMZ/saim.git)
 [![MIT LICENSE](https://img.shields.io/badge/License-MIT-brightgreen.svg?style=flat-square)](https://choosealicense.com/licenses/mit/)
 [![Documentation Status](https://img.shields.io/badge/docs-GitHub-blue.svg?style=flat-square)](https://LeibnizDSMZ.github.io/saim/)
 

configs/dev/requirements.dev.txt

@@ -323,9 +323,9 @@ typing-extensions==4.15.0 \
     --hash=sha256:0cea48d173cc12fa28ecabc3b837ea3cf6f38c6d1136f85cbaaf598984861466 \
     --hash=sha256:f0fa19c6845758ab08074a0cfa8b7aecb71c999ca73d62883bc25cc018c4e548
     # via mypy
-urllib3==2.6.0 \
-    --hash=sha256:c90f7a39f716c572c4e3e58509581ebd83f9b59cced005b7db7ad2d22b0db99f \
-    --hash=sha256:cb9bcef5a4b345d5da5d145dc3e30834f58e8018828cbc724d30b4cb7d4d49f1
+urllib3==2.6.1 \
+    --hash=sha256:5379eb6e1aba4088bae84f8242960017ec8d8e3decf30480b3a1abdaa9671a3f \
+    --hash=sha256:e67d06fe947c36a7ca39f4994b08d73922d40e6cca949907be05efa6fd75110b
     # via types-requests
 vulture==2.14 \
     --hash=sha256:cb8277902a1138deeab796ec5bef7076a6e0248ca3607a3f3dee0b6d9e9b8415 \

configs/dev/requirements.docs.txt

@@ -258,9 +258,9 @@ six==1.17.0 \
     --hash=sha256:4721f391ed90541fddacab5acf947aa0d3dc7d27b2e1e8eda2be8970586c3274 \
     --hash=sha256:ff70335d468e7eb6ec65b95b99d3a2836546063f63acc5171de367e834932a81
     # via python-dateutil
-urllib3==2.6.0 \
-    --hash=sha256:c90f7a39f716c572c4e3e58509581ebd83f9b59cced005b7db7ad2d22b0db99f \
-    --hash=sha256:cb9bcef5a4b345d5da5d145dc3e30834f58e8018828cbc724d30b4cb7d4d49f1
+urllib3==2.6.1 \
+    --hash=sha256:5379eb6e1aba4088bae84f8242960017ec8d8e3decf30480b3a1abdaa9671a3f \
+    --hash=sha256:e67d06fe947c36a7ca39f4994b08d73922d40e6cca949907be05efa6fd75110b
     # via requests
 watchdog==6.0.0 \
     --hash=sha256:07df1fdd701c5d4c8e55ef6cf55b8f0120fe1aef7ef39a1c6fc6bc2e606d517a \

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "saim"
-version = "0.9.26"
+version = "0.9.27"
 description = """A library for identifying strains."""
 readme = "README.md"
 authors = [
@@ -23,7 +23,7 @@ dependencies = [
   "python-keycloak ==5.8.1",
   "requests ==2.32.5",
   "requests-cache ==1.2.1",
-  "cafi @ git+https://github.com/LeibnizDSMZ/cafi.git@v0.9.13",
+  "cafi @ git+https://github.com/LeibnizDSMZ/cafi.git@v0.9.14",
 ]
 
 [project.scripts]

requirements.txt

@@ -23,7 +23,7 @@ attrs==25.4.0 \
     # via
     #   cattrs
     #   requests-cache
-cafi @ git+https://github.com/LeibnizDSMZ/cafi.git@72a82fe8adce0b3ac4e9ced363550541692cf500
+cafi @ git+https://github.com/LeibnizDSMZ/cafi.git@e90398f246bba7539af28bfaaeca0dcbd54c206c
     # via saim
 cattrs==25.3.0 \
     --hash=sha256:1ac88d9e5eda10436c4517e390a4142d88638fe682c436c93db7ce4a277b884a \
@@ -335,9 +335,9 @@ url-normalize==2.2.1 \
     --hash=sha256:3deb687587dc91f7b25c9ae5162ffc0f057ae85d22b1e15cf5698311247f567b \
     --hash=sha256:74a540a3b6eba1d95bdc610c24f2c0141639f3ba903501e61a52a8730247ff37
     # via requests-cache
-urllib3==2.6.0 \
-    --hash=sha256:c90f7a39f716c572c4e3e58509581ebd83f9b59cced005b7db7ad2d22b0db99f \
-    --hash=sha256:cb9bcef5a4b345d5da5d145dc3e30834f58e8018828cbc724d30b4cb7d4d49f1
+urllib3==2.6.1 \
+    --hash=sha256:5379eb6e1aba4088bae84f8242960017ec8d8e3decf30480b3a1abdaa9671a3f \
+    --hash=sha256:e67d06fe947c36a7ca39f4994b08d73922d40e6cca949907be05efa6fd75110b
     # via
     #   requests
     #   requests-cache

src/saim/culture_link/private/container.py

@@ -2,6 +2,7 @@
 from typing import Iterator, final
 
 from cafi.container.links import CatalogueLink
+from pydantic import HttpUrl, ValidationError
 from saim.culture_link.private.constants import (
     CAT_DET_EXP_DAYS,
     CAT_EXP_DAYS,
@@ -89,6 +90,15 @@ class LinkStatus:
     link_type: str
     status: VerificationStatus
 
+    def __post_init__(self) -> None:
+        try:
+            if self.link != "":
+                object.__setattr__(self, "link", str(HttpUrl(self.link)))
+        except ValidationError:
+            object.__setattr__(self, "link", "")
+        if self.link == "":
+            object.__setattr__(self, "status", VerificationStatus.no_url)
+
 
 @final
 @dataclass(frozen=True, slots=True)
@@ -97,6 +107,13 @@ class LinkResult:
     brc_id: int
     found_ccno: CCNoDes
 
+    def __post_init__(self) -> None:
+        try:
+            if self.link != "":
+                object.__setattr__(self, "link", str(HttpUrl(self.link)))
+        except ValidationError:
+            object.__setattr__(self, "link", "")
+
 
 @final
 @dataclass(frozen=True, slots=True)

src/saim/culture_link/validate_cafi.py

@@ -5,7 +5,6 @@
 from cafi.library.loader import load_catalogue_db, load_regex_db
 from cafi.container.acr_db import AcrDbEntry
 from cafi.container.links import LinkLevel
-from cafi.container.fun.format import url_to_str
 from saim.culture_link.create_links import CcnoLinkGenerator, SearchRequest
 from saim.designation.manager import AcronymManager
 
@@ -21,7 +20,7 @@ def _create_search_request_homepage(
             exclude=(LinkLevel.cat,),
         )
         for task_id, (acr_id, acr_entry) in enumerate(cc_db.items())
-        if url_to_str(acr_entry.homepage) != ""
+        if str(acr_entry.homepage) != ""
     ]
 
 

src/saim/shared/data_con/culture.py

@@ -33,7 +33,6 @@
 from saim.shared.data_ops.clean import detect_empty_dict_keys
 from saim.shared.parse.date import check_date_str, date_to_str
 from saim.taxon_name.manager import TaxonManager
-from cafi.container.fun.format import url_to_str
 
 _REQ_KEYS: Final[tuple[str, ...]] = (
     "acronym",
@@ -131,8 +130,7 @@ class CultureCCNo(BaseModel):
     source: CiDSrc
     # optional fields - default
     url: (
-        Annotated[HttpUrl, PlainSerializer(lambda val: url_to_str(val), return_type=str)]
-        | None
+        Annotated[HttpUrl, PlainSerializer(lambda val: str(val), return_type=str)] | None
     ) = None
     cul_id: Annotated[int, Field(ge=1)] | None = Field(default=None, alias="cultureId")
     history: Annotated[str, AfterValidator(clean_text_rm_tags), Field(min_length=2)] = ""

src/saim/shared/misc/constants.py

@@ -2,4 +2,4 @@
 
 
 ENCODING: Final[str] = "utf-8"
-VERSION: Final[str] = "v0.9.26"
+VERSION: Final[str] = "v0.9.27"