]> git.scottworley.com Git - paperdoorknob/blobdiff - paperdoorknob_test.py
Extensible, flag-controlled DOM filters
[paperdoorknob] / paperdoorknob_test.py
index 35756cbd0c8897533bcbda12dc8b38b6e8f2e3bd..aa50340ecfd7709ed3437b9ce006d22a5fe1804d 100644 (file)
 # Free Software Foundation, version 3.
 
 
+from abc import ABC, abstractmethod
 import unittest
-import threading
-from http.server import BaseHTTPRequestHandler, HTTPServer
-import requests
-import requests_cache
+import io
+import subprocess
+
 import paperdoorknob
 
+from testing.fakeserver import FakeGlowficServer
+from domfilter import ApplyDOMFilters
+from fetch import DirectFetcher, FakeFetcher, Fetcher
+from spec import Spec
+from texify import DirectTexifier, PandocTexifier, VerifyingTexifier
+
 TIMEOUT = 8
 
 
-class FakeGlowficHTTPRequestHandler(BaseHTTPRequestHandler):
+class BaseTestProcess(ABC):
 
-    def _notify_test(self) -> None:
+    @abstractmethod
+    def url(self) -> str:
         raise NotImplementedError()
 
-    def _response_code(self) -> int:
-        if self.path == "/not_found":
-            return 404
-        if self.path == "/server_error":
-            return 500
-        return 200
-
-    def do_GET(self) -> None:
-        body = b'''<html>
-  <body>
-    <div class="post-container post-post">
-      <div class="post-edit-box">We don't want edit boxes</div>
-      This is glowfic
-      <div class="post-footer">We don't want footers</div>
-    </div>
-    <div class="flat-post-replies">
-      <div class="post-container post-reply">
-        <div class="post-edit-box">We don't want edit boxes</div>
-        You sure?
-        <div class="post-footer">We don't want footers</div>
-      </div>
-      <div class="post-container post-reply">
-        Pretty sure.
-      </div>
-    </div>
-  </body>
-</html>'''
-        self.send_response(self._response_code())
-        self.send_header("Content-type", "text/html")
-        self.send_header("Content-Length", str(len(body)))
-        self.end_headers()
-        self.wfile.write(body)
-        self._notify_test()
-
-
-class TestFetch(unittest.TestCase):
-    def _port(self) -> int:
-        port = self._web_server.socket.getsockname()[1]
-        assert isinstance(port, int)
-        return port
-
-    def _count_request(self) -> None:
-        self._request_counter += 1
+    @abstractmethod
+    def fetcher(self) -> Fetcher:
+        raise NotImplementedError()
 
-    def setUp(self) -> None:
-        self._request_counter = 0
-        handler = type("Handler", (FakeGlowficHTTPRequestHandler,), {
-                       '_notify_test': lambda _: self._count_request()})
-        self._web_server = HTTPServer(('', 0), handler)
-        self._thread = threading.Thread(target=self._web_server.serve_forever)
-        self._thread.start()
-
-    def tearDown(self) -> None:
-        self._web_server.shutdown()
-        self._thread.join()
-        self._web_server.server_close()
-
-    def testFetch(self) -> None:
-        with requests.session() as s:
-            paperdoorknob.fetch(f"http://localhost:{self._port()}", s, TIMEOUT)
-            self.assertEqual(self._request_counter, 1)
-            paperdoorknob.fetch(f"http://localhost:{self._port()}", s, TIMEOUT)
-            self.assertEqual(self._request_counter, 2)
-
-    def testFetchCaching(self) -> None:
-        with requests_cache.CachedSession() as s:
-            paperdoorknob.fetch(f"http://localhost:{self._port()}", s, TIMEOUT)
-            self.assertEqual(self._request_counter, 1)
-            paperdoorknob.fetch(f"http://localhost:{self._port()}", s, TIMEOUT)
-            self.assertEqual(self._request_counter, 1)
-
-    def testFetchPersistentCaching(self) -> None:
-        with requests_cache.CachedSession() as s:
-            paperdoorknob.fetch(f"http://localhost:{self._port()}", s, TIMEOUT)
-            self.assertEqual(self._request_counter, 1)
-        with requests_cache.CachedSession() as s:
-            paperdoorknob.fetch(f"http://localhost:{self._port()}", s, TIMEOUT)
-            self.assertEqual(self._request_counter, 1)
+    def testReplies(self) -> None:
+        replies = list(paperdoorknob.replies(
+            paperdoorknob.parse(self.fetcher().fetch(self.url()))))
+        for r in replies:
+            ApplyDOMFilters('NoEdit,NoFooter', r)
+        assert [r.text.strip() for r in replies] == [
+            "This is glowfic",
+            "You sure?",
+            "Pretty sure."]
 
     def testProcess(self) -> None:
-        with requests.session() as s:
-            replies = paperdoorknob.process(
-                f"http://localhost:{self._port()}", s, TIMEOUT)
-            self.assertEqual([r.text.strip() for r in replies],
-                             ["This is glowfic", "You sure?", "Pretty sure."])
-
-    def testFetchErrors(self) -> None:
-        with requests.session() as s:
-            with self.assertRaises(requests.HTTPError):
-                paperdoorknob.fetch(
-                    f"http://localhost:{self._port()}/not_found", s, TIMEOUT)
-            with self.assertRaises(requests.HTTPError):
-                paperdoorknob.fetch(
-                    f"http://localhost:{self._port()}/server_error", s, TIMEOUT)
+        buf = io.BytesIO()
+        spec = Spec(
+            self.url(),
+            self.fetcher(),
+            lambda x: x,
+            lambda x: ApplyDOMFilters('NoEdit,NoFooter', x),
+            PandocTexifier('pandoc'),
+            buf)
+        paperdoorknob.process(spec)
+        assert buf.getvalue() == b'''\\documentclass{article}
+\\begin{document}
+This is glowfic
+You \\emph{sure}?
+Pretty sure.
+\\end{document}
+'''
+
+    def testDirectTexifier(self) -> None:
+        texifier = VerifyingTexifier(
+            PandocTexifier('pandoc'), DirectTexifier())
+        buf = io.BytesIO()
+        spec = Spec(
+            self.url(),
+            self.fetcher(),
+            lambda x: x,
+            lambda x: ApplyDOMFilters('NoEdit,NoFooter', x),
+            texifier,
+            buf)
+        paperdoorknob.process(spec)
+
+    def testPDF(self) -> None:
+        with open("test.tex", 'wb') as out:
+            spec = Spec(
+                self.url(),
+                self.fetcher(),
+                lambda x: x,
+                lambda x: ApplyDOMFilters('NoEdit,NoFooter', x),
+                PandocTexifier('pandoc'),
+                out)
+            paperdoorknob.process(spec)
+        subprocess.run(['pdflatex', 'test.tex'],
+                       stdin=subprocess.DEVNULL, check=True)
+
+
+class TestProcessFromWebserver(BaseTestProcess, unittest.TestCase):
+
+    def setUp(self) -> None:
+        self._fetcher = self.enterContext(DirectFetcher(TIMEOUT))
+        self._server = self.enterContext(FakeGlowficServer())
+        self._port = self._server.port()
+
+    def url(self) -> str:
+        return f"http://localhost:{self._port}"
+
+    def fetcher(self) -> Fetcher:
+        return self._fetcher
+
+
+class TestProcessFromFakeFetcher(BaseTestProcess, unittest.TestCase):
+
+    def url(self) -> str:
+        return 'fic'
+
+    def fetcher(self) -> Fetcher:
+        with open('testdata/this-is-glowfic.html', 'rb') as f:
+            return FakeFetcher({'fic': f.read(9999)})
 
 
 if __name__ == '__main__':