Add HttpCrawler

PFERD/crawler.py

@@ -6,6 +6,7 @@ from pathlib import Path, PurePath
 from typing import (Any, AsyncContextManager, AsyncIterator, Awaitable,
                     Callable, Dict, Optional, TypeVar)
 
+import aiohttp
 from rich.markup import escape
 
 from .authenticator import Authenticator
@@ -263,3 +264,39 @@ class Crawler(ABC):
         """
 
         pass
+
+
+class HttpCrawler(Crawler):
+    COOKIE_FILE = PurePath(".cookies")
+
+    def __init__(
+            self,
+            name: str,
+            section: CrawlerSection,
+            config: Config,
+            conductor: TerminalConductor,
+    ) -> None:
+        super().__init__(name, section, config, conductor)
+
+        self._cookie_jar_path = self._output_dir.resolve(self.COOKIE_FILE)
+        self._output_dir.register_reserved(self.COOKIE_FILE)
+
+    async def run(self) -> None:
+        cookie_jar = aiohttp.CookieJar()
+
+        try:
+            cookie_jar.load(self._cookie_jar_path)
+        except Exception:
+            pass
+
+        async with aiohttp.ClientSession(cookie_jar=cookie_jar) as session:
+            self.session = session
+            try:
+                await super().run()
+            finally:
+                del self.session
+
+        try:
+            cookie_jar.save(self._cookie_jar_path)
+        except Exception:
+            self.print(f"[bold red]Warning:[/] Failed to save cookies to {escape(str(self.COOKIE_FILE))}")

PFERD/output_dir.py

@@ -86,6 +86,9 @@ class OutputDirectory:
 
         self._report = Report()
 
+    def register_reserved(self, path: PurePath):
+        self._report.mark_reserved(path)
+
     def _mark(self, path: PurePath) -> None:
         """
         May throw an OutputDirException
@@ -100,7 +103,7 @@ class OutputDirectory:
             msg = f"Collides with other file: {e.collides_with}"
             raise OutputDirException(msg)
 
-    def _resolve(self, path: PurePath) -> Path:
+    def resolve(self, path: PurePath) -> Path:
         """
         May throw an OutputDirException.
         """

PFERD/report.py

@@ -44,12 +44,16 @@ class Report:
     """
 
     def __init__(self) -> None:
+        self.reserved_files: Set[PurePath] = set()
         self.known_files: Set[PurePath] = set()
 
         self.new_files: Set[PurePath] = set()
         self.changed_files: Set[PurePath] = set()
         self.deleted_files: Set[PurePath] = set()
 
+    def mark_reserved(self, path: PurePath) -> None:
+        self.reserved_files.add(path)
+
     def mark(self, path: PurePath) -> None:
         """
         Mark a previously unknown file as known.
@@ -58,12 +62,12 @@ class Report:
         detail, see the respective exception's docstring.
         """
 
-        for known_path in self.known_files:
-            if path == known_path:
+        for other in self.known_files & self.reserved_files:
+            if path == other:
                 raise MarkDuplicateException(path)
 
-            if is_relative_to(path, known_path) or is_relative_to(known_path, path):
-                raise MarkConflictException(path, known_path)
+            if is_relative_to(path, other) or is_relative_to(other, path):
+                raise MarkConflictException(path, other)
 
         self.known_files.add(path)
 

setup.cfg

@@ -6,6 +6,8 @@ version = 3.0.0
 packages = PFERD
 python_requires = >=3.8
 install_requires =
+  aiohttp>=3.7.4.post0
+  beautifulsoup4>=4.9.3
   rich>=10.1.0
 
 [options.entry_points]