feat: Change LinkId field from int to UUID and generate LinkId determ…

…inistically from URL

common/birdxplorer_common/models.py

@@ -1,10 +1,18 @@
 from abc import ABC, abstractmethod
 from datetime import datetime, timezone
 from enum import Enum
+from random import Random
 from typing import Any, Dict, List, Literal, Optional, Type, TypeAlias, TypeVar, Union
+from uuid import UUID
 
 from pydantic import BaseModel as PydanticBaseModel
-from pydantic import ConfigDict, GetCoreSchemaHandler, HttpUrl, TypeAdapter
+from pydantic import (
+    ConfigDict,
+    GetCoreSchemaHandler,
+    HttpUrl,
+    TypeAdapter,
+    model_validator,
+)
 from pydantic.alias_generators import to_camel
 from pydantic.main import IncEx
 from pydantic_core import core_schema
@@ -677,24 +685,66 @@ class XUser(BaseModel):
 MediaDetails: TypeAlias = List[HttpUrl] | None
 
 
-class LinkId(NonNegativeInt):
+class LinkId(UUID):
     """
-    >>> LinkId.from_int(1)
-    LinkId(1)
+    >>> LinkId("53dc4ed6-fc9b-54ef-1afa-90f1125098c5")
+    LinkId('53dc4ed6-fc9b-54ef-1afa-90f1125098c5')
+    >>> LinkId(UUID("53dc4ed6-fc9b-54ef-1afa-90f1125098c5"))
+    LinkId('53dc4ed6-fc9b-54ef-1afa-90f1125098c5')
     """
 
-    pass
+    def __init__(
+        self,
+        hex: str | None = None,
+        int: int | None = None,
+    ) -> None:
+        if isinstance(hex, UUID):
+            hex = str(hex)
+        super().__init__(hex, int=int)
+
+    @classmethod
+    def from_url(cls, url: HttpUrl) -> "LinkId":
+        """
+        >>> LinkId.from_url("https://example.com/")
+        LinkId('d5d15194-6574-0c01-8f6f-15abd72b2cf6')
+        """
+        random_number_generator = Random()
+        random_number_generator.seed(str(url).encode("utf-8"))
+        return LinkId(int=random_number_generator.getrandbits(128))
+
+    @classmethod
+    def __get_pydantic_core_schema__(cls, _source_type: Any, _handler: GetCoreSchemaHandler) -> core_schema.CoreSchema:
+        return core_schema.no_info_plain_validator_function(
+            cls.validate,
+            serialization=core_schema.plain_serializer_function_ser_schema(cls.serialize, when_used="json"),
+        )
+
+    @classmethod
+    def validate(cls, v: Any) -> "LinkId":
+        return cls(v)
+
+    def serialize(self) -> str:
+        return str(self)
 
 
 class Link(BaseModel):
     """
-    >>> Link.model_validate_json('{"linkId": 1, "canonicalUrl": "https://example.com", "shortUrl": "https://example.com/short"}')
-    Link(link_id=LinkId(1), canonical_url=Url('https://example.com/'), short_url=Url('https://example.com/short'))
+    >>> Link.model_validate_json('{"linkId": "d5d15194-6574-0c01-8f6f-15abd72b2cf6", "url": "https://example.com"}')
+    Link(link_id=LinkId('d5d15194-6574-0c01-8f6f-15abd72b2cf6'), url=Url('https://example.com/'))
+    >>> Link(url="https://example.com/")
+    Link(link_id=LinkId('d5d15194-6574-0c01-8f6f-15abd72b2cf6'), url=Url('https://example.com/'))
+    >>> Link(link_id=UUID("d5d15194-6574-0c01-8f6f-15abd72b2cf6"), url="https://example.com/")
+    Link(link_id=LinkId('d5d15194-6574-0c01-8f6f-15abd72b2cf6'), url=Url('https://example.com/'))
     """  # noqa: E501
 
     link_id: LinkId
-    canonical_url: HttpUrl
-    short_url: HttpUrl
+    url: HttpUrl
+
+    @model_validator(mode="before")
+    def validate_link_id(cls, values: Dict[str, Any]) -> Dict[str, Any]:
+        if "link_id" not in values:
+            values["link_id"] = LinkId.from_url(values["url"])
+        return values
 
 
 class Post(BaseModel):

common/birdxplorer_common/storage.py

@@ -5,7 +5,7 @@
 from sqlalchemy import ForeignKey, create_engine, func, select
 from sqlalchemy.engine import Engine
 from sqlalchemy.orm import DeclarativeBase, Mapped, Session, mapped_column, relationship
-from sqlalchemy.types import CHAR, DECIMAL, JSON, Integer, String
+from sqlalchemy.types import CHAR, DECIMAL, JSON, Integer, String, Uuid
 
 from .models import BinaryBool, LanguageIdentifier
 from .models import Link as LinkModel
@@ -36,7 +36,7 @@ def adapt_pydantic_http_url(url: AnyUrl) -> AsIs:
 
 class Base(DeclarativeBase):
     type_annotation_map = {
-        LinkId: Integer,
+        LinkId: Uuid,
         TopicId: Integer,
         TopicLabel: JSON,
         NoteId: String,
@@ -95,8 +95,7 @@ class LinkRecord(Base):
     __tablename__ = "links"
 
     link_id: Mapped[LinkId] = mapped_column(primary_key=True)
-    canonical_url: Mapped[HttpUrl] = mapped_column(nullable=False, index=True)
-    short_url: Mapped[HttpUrl] = mapped_column(nullable=False, index=True)
+    url: Mapped[HttpUrl] = mapped_column(nullable=False, index=True)
 
 
 class PostLinkAssociation(Base):
@@ -216,10 +215,7 @@ def _post_record_to_model(cls, post_record: PostRecord) -> PostModel:
             like_count=post_record.like_count,
             repost_count=post_record.repost_count,
             impression_count=post_record.impression_count,
-            links=[
-                LinkModel(link_id=link.link_id, canonical_url=link.link.canonical_url, short_url=link.link.short_url)
-                for link in post_record.links
-            ],
+            links=[LinkModel(link_id=link.link_id, url=link.link.url) for link in post_record.links],
         )
 
     def get_user_enrollment_by_participant_id(self, participant_id: ParticipantId) -> UserEnrollment:

common/tests/conftest.py

@@ -128,10 +128,10 @@ def topic_samples(topic_factory: TopicFactory) -> Generator[List[Topic], None, N
 @fixture
 def link_samples(link_factory: LinkFactory) -> Generator[List[Link], None, None]:
     links = [
-        link_factory.build(link_id=0, canonical_url="https://t.co/xxxxxxxxxxx/", short_url="https://example.com/sh0"),
-        link_factory.build(link_id=1, canonical_url="https://t.co/yyyyyyyyyyy/", short_url="https://example.com/sh1"),
-        link_factory.build(link_id=2, canonical_url="https://t.co/zzzzzzzzzzz/", short_url="https://example.com/sh2"),
-        link_factory.build(link_id=3, canonical_url="https://t.co/wwwwwwwwwww/", short_url="https://example.com/sh3"),
+        link_factory.build(link_id="9f56ee4a-6b36-b79c-d6ca-67865e54bbd5", url="https://example.com/sh0"),
+        link_factory.build(link_id="f5b0ac79-20fe-9718-4a40-6030bb62d156", url="https://example.com/sh1"),
+        link_factory.build(link_id="76a0ac4a-a20c-b1f4-1906-d00e2e8f8bf8", url="https://example.com/sh2"),
+        link_factory.build(link_id="6c352be8-eca3-0d96-55bf-a9bbef1c0fc2", url="https://example.com/sh3"),
     ]
     yield links
 
@@ -416,7 +416,7 @@ def link_records_sample(
     link_samples: List[Link],
     engine_for_test: Engine,
 ) -> Generator[List[LinkRecord], None, None]:
-    res = [LinkRecord(link_id=d.link_id, canonical_url=d.canonical_url, short_url=d.short_url) for d in link_samples]
+    res = [LinkRecord(link_id=d.link_id, url=d.url) for d in link_samples]
     with Session(engine_for_test) as sess:
         sess.add_all(res)
         sess.commit()