Merge pull request #55 from dhvcc/feat/pydantic-v2

Pydantic v2, move old models to legacy, py3.14

Author: dhvcc

.github/workflows/ci.yml

@@ -7,15 +7,16 @@ on:
     paths-ignore:
       - ".gitignore"
       - "README.md"
-  pull_request:
+      - "pre-commit-config.yaml"
+      - "LICENSE"
 
 jobs:
   test:
     strategy:
       max-parallel: 6
       matrix:
         os: [ "ubuntu-latest", "windows-latest", "macos-latest" ]
-        python-version: [ "3.9", "3.10", "3.11", "3.12", "3.13" ]
+        python-version: [ "3.9", "3.10", "3.11", "3.12", "3.13", "3.14" ]
 
     runs-on: ${{ matrix.os }}
 

.pre-commit-config.yaml

@@ -31,6 +31,7 @@ repos:
           - run
           - mypy
           - rss_parser
+        pass_filenames: false
         language: system
         stages: [ push ]
 

README.md

@@ -37,6 +37,31 @@ pip install dist/*.whl
 - Models for RSS-specific schemas have been moved from `rss_parser.models` to `rss_parser.models.rss`. Generic types remain unchanged
 - Date parsing has been improved and now uses pydantic's `validator` instead of `email.utils`, producing better datetime objects where it previously defaulted to `str`
 
+## V2 -> V3 Migration
+
+`rss-parser` 3.x upgrades the runtime models to [Pydantic v2](https://docs.pydantic.dev/latest/migration/). Highlights:
+
+- **New default models** now inherit from `pydantic.BaseModel` v2 and use `model_validate`/`model_dump`. If you extend our classes, switch from `dict()`/`json()` to `model_dump()`/`model_dump_json()`.
+- **Legacy compatibility** lives under `rss_parser.models.legacy`. Point your custom parser at the legacy schema if you must stay on the v1 API surface.
+- **Collections**: list-like XML fields now use `OnlyList[...]` directly with an automatic `default_factory` so that attributes are always lists (no more `Optional[OnlyList[T]] = Field(..., default=[])`). Update custom schemas accordingly.
+- **Custom hooks**: if you relied on `rss_parser.pydantic_proxy`, import it from `rss_parser.models.legacy.pydantic_proxy`. The top-level module only re-exports it for backwards compatibility.
+
+See the “Legacy Models” section below for sample snippets showing how to stay on the older types. Tests in this repo cover both tracks to guarantee matching output.
+
+## Legacy Models
+
+Pydantic v1-based models are still available under `rss_parser.models.legacy`. They retain the previous behaviour and re-export the `import_v1_pydantic` helper as `rss_parser.models.legacy.pydantic_proxy.import_v1_pydantic`. You can continue to use them by pointing your parser at the legacy schema:
+
+```python
+from rss_parser import RSSParser
+from rss_parser.models.legacy.rss import RSS as LegacyRSS
+
+class LegacyRSSParser(RSSParser):
+    schema = LegacyRSS
+```
+
+Tests in this repository run against both the v2 and legacy models to ensure parity.
+
 ## Usage
 
 ### Quickstart
@@ -163,18 +188,17 @@ If you don't want to deal with these conditions and want to parse something **al
 ```python
 from typing import Optional
 
+from pydantic import Field
+
 from rss_parser.models.rss.item import Item
 from rss_parser.models.types.only_list import OnlyList
 from rss_parser.models.types.tag import Tag
-from rss_parser.pydantic_proxy import import_v1_pydantic
-
-pydantic = import_v1_pydantic()
 ...
 
 
 class OptionalChannelElementsMixin(...):
     ...
-    items: Optional[OnlyList[Tag[Item]]] = pydantic.Field(alias="item", default=[])
+    items: Optional[OnlyList[Tag[Item]]] = Field(alias="item", default_factory=list)
 ```
 
 ### Tag Field

poetry.lock

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "rss-parser"
-version = "2.1.1"
+version = "3.0.0a2"
 description = "Typed pythonic RSS/Atom parser"
 authors = ["dhvcc <[email protected]>"]
 license = "GPL-3.0"
@@ -28,6 +28,8 @@ classifiers = [
     "Programming Language :: Python :: 3.10",
     "Programming Language :: Python :: 3.11",
     "Programming Language :: Python :: 3.12",
+    "Programming Language :: Python :: 3.13",
+    "Programming Language :: Python :: 3.14",
 ]
 packages = [{ include = "rss_parser" }, { include = "rss_parser/py.typed" }]
 
@@ -39,7 +41,7 @@ packages = [{ include = "rss_parser" }, { include = "rss_parser/py.typed" }]
 
 [tool.poetry.dependencies]
 python = "^3.9"
-pydantic = ">1.9"
+pydantic = "<3.0"
 xmltodict = "^0.13.0"
 types-xmltodict = "^0.14.0.20241009"
 
@@ -85,6 +87,9 @@ select = [
     "RUF", # ruff
 ]
 
+[tool.ruff.lint.pep8-naming]
+ignore-names = ["LegacyRSS"]
+
 [tool.ruff.per-file-ignores]
 "tests/**.py" = [
     "S101",   # Use of assert detected
@@ -95,7 +100,6 @@ select = [
 "**/__init__.py" = ["F401"]
 "rss_parser/models/atom/**" = ["A003"]
 
-
 [build-system]
 requires = ["poetry-core"]
 build-backend = "poetry.core.masonry.api"

pyproject.toml

@@ -1,3 +1,3 @@
 from ._parser import AtomParser, BaseParser, RSSParser
 
-__all__ = ("BaseParser", "AtomParser", "RSSParser")
+__all__ = ("AtomParser", "BaseParser", "RSSParser")

rss_parser/__init__.py

@@ -48,6 +48,10 @@ def parse(
         if root_key:
             root = root.get(root_key, root)
 
+        if hasattr(schema, "model_validate"):
+            return schema.model_validate(root)
+
+        # Pydantic v1 only
         return schema.parse_obj(root)
 
 

rss_parser/_parser.py

@@ -1,32 +1,25 @@
-"""
-Models created according to https://www.rssboard.org/rss-specification.
+from __future__ import annotations
 
-Some types and validation may be a bit custom to account for broken standards in some RSS feeds.
-"""
-
-from json import loads
-from typing import TYPE_CHECKING
+from pydantic import BaseModel, ConfigDict
 
 from rss_parser.models.utils import camel_case
-from rss_parser.pydantic_proxy import import_v1_pydantic
-
-if TYPE_CHECKING:
-    from pydantic import v1 as pydantic
-else:
-    pydantic = import_v1_pydantic()
 
 
-class XMLBaseModel(pydantic.BaseModel):
-    class Config:
-        alias_generator = camel_case
+class XMLBaseModel(BaseModel):
+    model_config = ConfigDict(alias_generator=camel_case)
 
-    def json_plain(self, **kw):
+    def json_plain(self, **kwargs) -> str:
         """
-        Run pydantic's json with custom encoder to encode Tags as only content.
+        Serialize the model while flattening Tag instances into their content.
         """
         from rss_parser.models.types.tag import Tag  # noqa: PLC0415
 
-        return self.json(models_as_dict=False, encoder=Tag.flatten_tag_encoder, **kw)
+        return self.model_dump_json(fallback=Tag.flatten_tag_encoder, **kwargs)
+
+    def dict_plain(self, **kwargs):
+        from rss_parser.models.types.tag import Tag  # noqa: PLC0415
+
+        return self.model_dump(mode="json", fallback=Tag.flatten_tag_encoder, **kwargs)
+
 
-    def dict_plain(self, **kw):
-        return loads(self.json_plain(**kw))
+__all__ = ("XMLBaseModel",)

rss_parser/models/__init__.py

@@ -1,15 +1,14 @@
 from typing import Optional
 
+from pydantic import Field
+
 from rss_parser.models import XMLBaseModel
 from rss_parser.models.atom.feed import Feed
 from rss_parser.models.types.tag import Tag
-from rss_parser.pydantic_proxy import import_v1_pydantic
-
-pydantic = import_v1_pydantic()
 
 
 class Atom(XMLBaseModel):
     """Atom 1.0"""
 
-    version: Optional[Tag[str]] = pydantic.Field(alias="@version")
+    version: Optional[Tag[str]] = Field(alias="@version", default=None)
     feed: Tag[Feed]

rss_parser/models/atom/atom.py

@@ -1,13 +1,12 @@
 from typing import Optional
 
+from pydantic import Field
+
 from rss_parser.models import XMLBaseModel
 from rss_parser.models.atom.person import Person
 from rss_parser.models.types.date import DateTimeOrStr
 from rss_parser.models.types.only_list import OnlyList
 from rss_parser.models.types.tag import Tag
-from rss_parser.pydantic_proxy import import_v1_pydantic
-
-pydantic = import_v1_pydantic()
 
 
 class RequiredAtomEntryMixin(XMLBaseModel):
@@ -22,10 +21,10 @@ class RequiredAtomEntryMixin(XMLBaseModel):
 
 
 class RecommendedAtomEntryMixin(XMLBaseModel):
-    authors: Optional[OnlyList[Tag[Person]]] = pydantic.Field(alias="author", default=[])
+    authors: OnlyList[Tag[Person]] = Field(alias="author", default_factory=OnlyList)
     "Entry authors."
 
-    links: Optional[OnlyList[Tag[str]]] = pydantic.Field(alias="link", default=[])
+    links: OnlyList[Tag[str]] = Field(alias="link", default_factory=OnlyList)
     "The URL of the entry."
 
     content: Optional[Tag[str]] = None
@@ -36,10 +35,10 @@ class RecommendedAtomEntryMixin(XMLBaseModel):
 
 
 class OptionalAtomEntryMixin(XMLBaseModel):
-    categories: Optional[OnlyList[Tag[dict]]] = pydantic.Field(alias="category", default=[])
+    categories: OnlyList[Tag[dict]] = Field(alias="category", default_factory=OnlyList)
     "Specifies a categories that the entry belongs to."
 
-    contributors: Optional[OnlyList[Tag[Person]]] = pydantic.Field(alias="contributor", default=[])
+    contributors: OnlyList[Tag[Person]] = Field(alias="contributor", default_factory=OnlyList)
     "Entry contributors."
 
     rights: Optional[Tag[str]] = None