fix: replace list(set()) with list(dict.fromkeys())

Use list(dict.fromkeys()) instead of list(set()) to ensure the list
oldvis · Aug 24, 2023 · 1ad3268 · 1ad3268
1 parent c894edc
commit 1ad3268
Show file tree

Hide file tree

Showing 4 changed files with 8 additions and 8 deletions.
diff --git a/libprocess/david_rumsey_map_collection/_process_metadata.py b/libprocess/david_rumsey_map_collection/_process_metadata.py
@@ -101,7 +101,7 @@ def get_tags(field_values: List[Dict]) -> List[str]:
         "Data Visualization",
     ]
     tags = [d for d in tags if d not in useless_tags]
-    return list(set(tags))
+    return list(dict.fromkeys(tags))
 
 
 def get_abstract(field_values: List[Dict]) -> Union[str, None]:

diff --git a/libprocess/gallica/_process_metadata.py b/libprocess/gallica/_process_metadata.py
@@ -118,7 +118,7 @@ def get_authors(record: Record) -> Union[List[str], None]:
     creator = get_english_attr(record, "dc:creator")
     if creator is None:
         return None
-    return list(set(get_english_attr(record, "dc:creator")))
+    return list(dict.fromkeys(creator))
 
 
 def get_publish_date(record: Record) -> Union[TimePoint, None]:
@@ -162,7 +162,7 @@ def get_languages(record: Record) -> List[str]:
     }
     languages = [language_map[d] if d in language_map else d for d in languages]
 
-    return list(set(languages))
+    return list(dict.fromkeys(languages))
 
 
 def get_tags(record: Record) -> List[str]:
@@ -180,7 +180,7 @@ def get_tags(record: Record) -> List[str]:
         "view",
     ]
     tags = [d.title() for d in tags if d not in useless_tags]
-    return list(set(tags))
+    return list(dict.fromkeys(tags))
 
 
 def get_abstract(record: Record) -> Union[str, None]:
@@ -209,7 +209,7 @@ def get_rights(record: Record) -> str:
     if rights is None:
         return "unknown"
 
-    rights = list(set(get_english_attr(record, "dc:rights")))
+    rights = list(dict.fromkeys(rights))
     assert len(rights) == 1, f"Unexpected rights with length != 1: {rights}"
 
     if rights == "public domain":

diff --git a/libprocess/internet_archive/_process_metadata.py b/libprocess/internet_archive/_process_metadata.py
@@ -120,7 +120,7 @@ def get_tags(source_data: SourceData) -> List[str]:
         parts = [d.lstrip("* ").rstrip("; ").title() for d in tag.split(";")]
         tags += [d for d in parts if d != "" and not is_name(d)]
 
-    return list(set(tags))
+    return list(dict.fromkeys(tags))
 
 
 def get_languages(source_data: SourceData) -> Union[List[str], None]:

diff --git a/libprocess/library_of_congress/_process_metadata.py b/libprocess/library_of_congress/_process_metadata.py
@@ -144,7 +144,7 @@ def get_languages(source_data: SourceData) -> List[str]:
             parsed += matched
         else:
             parsed.append(d)
-    return [*set(parsed)]
+    return list(dict.fromkeys(parsed))
 
 
 def get_abstract(source_data: SourceData) -> Union[str, None]:
@@ -220,7 +220,7 @@ def get_rights(source_data: SourceData) -> str:
     if "rights_information" in item:
         rights.append(item["rights_information"])
 
-    rights = [*set(rights)]
+    rights = list(dict.fromkeys(rights))
 
     assert len(rights) <= 1, f"Unexpected rights with length > 1: {rights}"