Testing - Improve PDF text change detection tests (#1992)

changedetectionio/tests/test_pdf.py

@@ -2,9 +2,8 @@
 
 import time
 from flask import url_for
-from .util import set_original_response, set_modified_response, live_server_setup
+from .util import set_original_response, set_modified_response, live_server_setup, wait_for_all_checks
 
-sleep_time_for_fetch_thread = 3
 
 # `subtractive_selectors` should still work in `source:` type requests
 def test_fetch_pdf(client, live_server):
@@ -22,7 +21,9 @@ def test_fetch_pdf(client, live_server):
 
     assert b"1 Imported" in res.data
 
-    time.sleep(sleep_time_for_fetch_thread)
+
+    wait_for_all_checks(client)
+
     res = client.get(
         url_for("preview_page", uuid="first"),
         follow_redirects=True
@@ -33,8 +34,42 @@ def test_fetch_pdf(client, live_server):
 
     # So we know if the file changes in other ways
     import hashlib
-    md5 = hashlib.md5(open("test-datastore/endpoint-test.pdf", 'rb').read()).hexdigest().upper()
+    original_md5 = hashlib.md5(open("test-datastore/endpoint-test.pdf", 'rb').read()).hexdigest().upper()
     # We should have one
-    assert len(md5) >0
+    assert len(original_md5) >0
     # And it's going to be in the document
-    assert b'Document checksum - '+bytes(str(md5).encode('utf-8')) in res.data
+    assert b'Document checksum - '+bytes(str(original_md5).encode('utf-8')) in res.data
+
+
+    shutil.copy("tests/test2.pdf", "test-datastore/endpoint-test.pdf")
+    changed_md5 = hashlib.md5(open("test-datastore/endpoint-test.pdf", 'rb').read()).hexdigest().upper()
+    res = client.get(url_for("form_watch_checknow"), follow_redirects=True)
+    assert b'1 watches queued for rechecking.' in res.data
+
+    wait_for_all_checks(client)
+
+    # Now something should be ready, indicated by having a 'unviewed' class
+    res = client.get(url_for("index"))
+    assert b'unviewed' in res.data
+
+    # The original checksum should be not be here anymore (cdio adds it to the bottom of the text)
+
+    res = client.get(
+        url_for("preview_page", uuid="first"),
+        follow_redirects=True
+    )
+
+    assert original_md5.encode('utf-8') not in res.data
+    assert changed_md5.encode('utf-8') in res.data
+
+
+    res = client.get(
+        url_for("diff_history_page", uuid="first"),
+        follow_redirects=True
+    )
+
+    assert original_md5.encode('utf-8') in res.data
+    assert changed_md5.encode('utf-8') in res.data
+
+    assert b'here is a change' in res.data
+