More structure and tests around splitting the page into chunks' DOMs.

[paperdoorknob] / paperdoorknob_test.py

diff --git a/paperdoorknob_test.py b/paperdoorknob_test.py
index c8968108472b675bb127bd29a003064de83650c8..249b6e42e936fa5f60a18e021c133dc824ca15a6 100644 (file)
--- a/paperdoorknob_test.py
+++ b/paperdoorknob_test.py
@@ -5,88 +5,107 @@
 # Free Software Foundation, version 3.
 
 
+from abc import ABC, abstractmethod
 import unittest
 import io
 import subprocess
-import requests
-import requests_cache
+
 import paperdoorknob
+
 from testing.fakeserver import FakeGlowficServer
+from domfilter import ApplyDOMFilters
+from fetch import DirectFetcher, FakeFetcher, Fetcher
+from images import ImageStore
+from spec import Spec
+from texify import DirectTexifier, PandocTexifier, VerifyingTexifier
 
 TIMEOUT = 8
 
 
-class TestFetch(unittest.TestCase):
-    def setUp(self) -> None:
-        self._server = self.enterContext(FakeGlowficServer())
-        self._port = self._server.port()
+class BaseTestProcess(ABC):
 
-    def testFetch(self) -> None:
-        with requests.session() as s:
-            paperdoorknob.fetch(f"http://localhost:{self._port}", s, TIMEOUT)
-            self.assertEqual(self._server.request_count(), 1)
-            paperdoorknob.fetch(f"http://localhost:{self._port}", s, TIMEOUT)
-            self.assertEqual(self._server.request_count(), 2)
-
-    def testFetchCaching(self) -> None:
-        with requests_cache.CachedSession() as s:
-            paperdoorknob.fetch(f"http://localhost:{self._port}", s, TIMEOUT)
-            self.assertEqual(self._server.request_count(), 1)
-            paperdoorknob.fetch(f"http://localhost:{self._port}", s, TIMEOUT)
-            self.assertEqual(self._server.request_count(), 1)
-
-    def testFetchPersistentCaching(self) -> None:
-        with requests_cache.CachedSession() as s:
-            paperdoorknob.fetch(f"http://localhost:{self._port}", s, TIMEOUT)
-            self.assertEqual(self._server.request_count(), 1)
-        with requests_cache.CachedSession() as s:
-            paperdoorknob.fetch(f"http://localhost:{self._port}", s, TIMEOUT)
-            self.assertEqual(self._server.request_count(), 1)
-
-    def testReplies(self) -> None:
-        with requests.session() as s:
-            replies = paperdoorknob.replies(
-                paperdoorknob.clean(
-                    paperdoorknob.fetch(
-                        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)
+    @abstractmethod
+    def url(self) -> str:
+        raise NotImplementedError()
+
+    @abstractmethod
+    def fetcher(self) -> Fetcher:
+        raise NotImplementedError()
 
     def testProcess(self) -> None:
-        with requests.session() as s:
-            buf = io.BytesIO()
-            paperdoorknob.process(
-                f"http://localhost:{self._port}",
-                s,
-                TIMEOUT,
-                buf,
-                'pandoc')
-            self.assertEqual(buf.getvalue(), b'''\\documentclass{article}
+        buf = io.BytesIO()
+        spec = Spec(
+            self.url(),
+            self.fetcher(),
+            ImageStore('is', self.fetcher()),
+            lambda x: x,
+            lambda x: ApplyDOMFilters('NoEdit,NoFooter', x),
+            PandocTexifier('pandoc'),
+            'margin=20mm',
+            buf)
+        paperdoorknob.process(spec)
+        assert buf.getvalue() == b'''\\documentclass{article}
+\\usepackage[margin=20mm]{geometry}
 \\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(),
+            ImageStore('is', self.fetcher()),
+            lambda x: x,
+            lambda x: ApplyDOMFilters('NoEdit,NoFooter', x),
+            texifier,
+            None,
+            buf)
+        paperdoorknob.process(spec)
 
     def testPDF(self) -> None:
-        with requests.session() as s:
-            with open("test.tex", 'wb') as out:
-                paperdoorknob.process(
-                    f"http://localhost:{self._port}", s, TIMEOUT, out, 'pandoc')
-            subprocess.run(['pdflatex', 'test.tex'],
-                           stdin=subprocess.DEVNULL, check=True)
+        with open("test.tex", 'wb') as out:
+            spec = Spec(
+                self.url(),
+                self.fetcher(),
+                ImageStore('is', self.fetcher()),
+                lambda x: x,
+                lambda x: ApplyDOMFilters('NoEdit,NoFooter', x),
+                PandocTexifier('pandoc'),
+                None,
+                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__':