fix: fix incorrectly updated descriptions (#116)

* add failing tests * fix tests themselves then resolve all failing * fix linting * ignore call order in test
meltano · Jan 29, 2024 · 42c2fc6 · 42c2fc6
1 parent 69e7e5f
commit 42c2fc6
Show file tree

Hide file tree

Showing 3 changed files with 32 additions and 14 deletions.
diff --git a/hub_utils/meltano_util.py b/hub_utils/meltano_util.py
@@ -377,7 +377,16 @@ def _split_sentence_endings(word_list):
                     continue
                 if not any(
                     keyword in word
-                    for keyword in ("http", "ssh", "ssl", "e.g.", '"', "`")
+                    for keyword in (
+                        "http",
+                        "ssh",
+                        "ssl",
+                        "e.g.",
+                        '"',
+                        "`",
+                        "()",
+                        ".com",
+                    )
                 ):
                     desc_list_clean.extend(word.replace(".", ". ").split())
                     continue
@@ -403,6 +412,9 @@ def _capitalize(cleaned_sentence):
                 or sentence_list[0][0] == "'"
                 or last_elem.endswith("e.g")
                 or last_elem.endswith("i.e")
+                or sentence_list[0].startswith("tap-")
+                or sentence_list[0].startswith("target-")
+                or sentence_list[0].startswith("http")
             ):
                 clean_capital_list.append(" ".join(sentence_list))
             else:

diff --git a/tests/test_main.py b/tests/test_main.py
@@ -40,18 +40,21 @@ def test_download_metadata_ignore(patch):
         ignore_list_str="extractors/tap-hubspot/hotgluexyz"
     )
     assert patch.call_count == 2
-    patch.assert_has_calls([
-        call(
-            expected_bucket,
-        "extractors/tap-hubspot/meltanolabs",
-        f"{local_path}/extractors/tap-hubspot/meltanolabs.json"
-        ),
-        call(
-            expected_bucket,
-        "extractors/tap-github/meltanolabs",
-        f"{local_path}/extractors/tap-github/meltanolabs.json"
-        )
-    ])
+    patch.assert_has_calls(
+        [
+            call(
+                expected_bucket,
+            "extractors/tap-hubspot/meltanolabs",
+            f"{local_path}/extractors/tap-hubspot/meltanolabs.json"
+            ),
+            call(
+                expected_bucket,
+            "extractors/tap-github/meltanolabs",
+            f"{local_path}/extractors/tap-github/meltanolabs.json"
+            )
+        ],
+        any_order=True
+    )
 
 @patch.object(S3, "download_latest")
 def test_download_metadata_list(patch):

diff --git a/tests/test_meltano_utils.py b/tests/test_meltano_utils.py
@@ -529,7 +529,10 @@ def test_get_quality(input, expected):
         ["Path to .duckdb file", "Path to .duckdb file"],
         ["Foo .env file.", "Foo .env file."],
         ["By (e.g. myFolder/thisTable/) sit, S3. This is replicate.", "By (e.g. myFolder/thisTable/) sit, S3. This is replicate."],
-        [["foo.Test."], ""],
+        ["Request timeout used when not overridden in Session.execute().", "Request timeout used when not overridden in Session.execute()."],
+        ['For example, "from:[email protected] rfc822msgid:<[email protected]> is:unread"."', 'For example, "from:[email protected] rfc822msgid:<[email protected]> is:unread"."'],
+        ["tap-saasoptics <api_user_email@your_company.com>.", "tap-saasoptics <api_user_email@your_company.com>."],
+        ["https://api.totango.com", "https://api.totango.com"]
     ]
 )
 def test_clean_description(input, expected):