refactor: update remote tag fetching behaviour so it uses api instead of shallow clone

This commit is contained in:
Jonathan Teran Carballo
2026-09-21 09:15:48 -03:00
parent eabef5eb33
commit 22ba8f5e5f
7 changed files with 89 additions and 14 deletions
+30 -2
View File
@@ -1,5 +1,7 @@
from abc import ABC, abstractmethod
import re
import giturlparse
import requests
@@ -72,6 +74,12 @@ class RemoteAPI(ABC):
viejo), sin depender del historial local."""
pass
@abstractmethod
def list_tags(self, suffix: str = "") -> list[str]:
"""Nombres de los tags del repositorio, sin orden garantizado. `suffix` es solo una pista
para que la API filtre; el filtrado definitivo se hace en `get_latest_tag`."""
pass
@abstractmethod
def get_endpoint(self, resource: str) -> str:
pass
@@ -91,13 +99,33 @@ class RemoteAPI(ABC):
return response
def get_all_pages(self, url: str, page_size: int = 50) -> list:
def get_latest_tag(self, suffix: str = "") -> str | None:
"""Tag SemVer más alto (vX.Y.Z + suffix exacto), o None si no hay ninguno."""
pattern = re.compile(rf"^v(\d+)\.(\d+)\.(\d+){re.escape(suffix)}$")
versions = {}
for name in self.list_tags(suffix):
match = pattern.match(name)
if match:
versions[tuple(map(int, match.groups()))] = name
return versions[max(versions)] if versions else None
def get_all_pages(
self,
url: str,
page_size: int = 50,
params: dict | None = None,
size_param: str = "per_page",
) -> list:
"""Recorre un endpoint paginado por número de página (page / per_page)."""
items: list = []
page = 1
while True:
batch = self.get_json(url, {"page": page, "per_page": page_size}).json()
query = {**(params or {}), "page": page, size_param: page_size}
batch = self.get_json(url, query).json()
items += batch
if len(batch) < page_size:
+16
View File
@@ -77,5 +77,21 @@ class BitbucketRemoteAPI(RemoteAPI):
return commits
def list_tags(self, suffix: str = "") -> list[str]:
names: list[str] = []
url: str | None = self.get_endpoint("refs/tags")
params: dict[str, int|str] | None = {"pagelen": 100}
if suffix:
params["q"] = f'name ~ "{suffix}"'
while url:
page = self.get_json(url, params).json()
names += [t["name"] for t in page["values"]]
url = page.get("next")
params = None # "next" ya incluye la query
return names
def get_endpoint(self, resource: str) -> str:
return f"{BitbucketRemoteAPI.API_ENDPOINT}/{self.repository}/{resource}"
+5
View File
@@ -74,6 +74,11 @@ class GiteaRemoteAPI(RemoteAPI):
for c in reversed(items)
]
def list_tags(self, suffix: str = "") -> list[str]:
tags = self.get_all_pages(self.get_endpoint("/tags"), 50, size_param="limit")
return [t["name"] for t in tags]
def get_endpoint(self, resource: str) -> str:
return f"{self.base_url}/api/v1/repos/{self.repository}{resource}"
+3
View File
@@ -63,5 +63,8 @@ class GithubRemoteAPI(RemoteAPI):
for c in reversed(items)
]
def list_tags(self, suffix: str = "") -> list[str]:
return [t["name"] for t in self.get_all_pages(self.get_endpoint("/tags"), 100)]
def get_endpoint(self, resource: str) -> str:
return f"{GithubRemoteAPI.API_ENDPOINT}/{self.repository}{resource}"
+7
View File
@@ -70,6 +70,13 @@ class GitlabRemoteAPI(RemoteAPI):
for c in items
]
def list_tags(self, suffix: str = "") -> list[str]:
# "search" soporta "$" para anclar al final del nombre
params = {"search": suffix + "$"} if suffix else None
tags = self.get_all_pages(self.get_endpoint("/repository/tags"), 100, params)
return [t["name"] for t in tags]
def get_endpoint(self, resource: str) -> str:
project_id = quote(self.repository, safe="")
return f"{self.base_url}/api/v4/projects/{project_id}{resource}"