commonmeta-py 0.52__tar.gz → 0.53__tar.gz

{commonmeta_py-0.52 → commonmeta_py-0.53}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: commonmeta-py
-Version: 0.52
+Version: 0.53
 Summary: Library for conversions to/from the Commonmeta scholarly metadata format
 Home-page: https://python.commonmeta.org
 License: MIT

{commonmeta_py-0.52 → commonmeta_py-0.53}/commonmeta/__init__.py

@@ -10,7 +10,7 @@ commonmeta-py is a Python library to convert scholarly metadata
 """
 
 __title__ = "commonmeta-py"
-__version__ = "0.52"
+__version__ = "0.53"
 __author__ = "Martin Fenner"
 __license__ = "MIT"
 

{commonmeta_py-0.52 → commonmeta_py-0.53}/commonmeta/readers/json_feed_reader.py

@@ -49,7 +49,7 @@ def read_json_feed_item(data: Optional[dict], **kwargs) -> Commonmeta:
     read_options = kwargs or {}
     url = None
     if py_.get(meta, "blog.status", None) == "active":
-         url = normalize_url(meta.get("url", None))
+        url = normalize_url(meta.get("url", None))
     elif py_.get(meta, "blog.status", None) == "archived":
         url = normalize_url(meta.get("archive_url", None))
     _id = normalize_doi(read_options.get("doi", None) or meta.get("doi", None)) or url
@@ -189,9 +189,14 @@ def get_references(references: list) -> list:
             )
 
         else:
-            return {
-                "id": reference.get("id", None),
-            }
+            id_ = normalize_url(reference.get("id", None))
+            if id_ is None:
+                return None
+            return compact(
+                {
+                    "id": id_,
+                }
+            )
 
     def number_reference(reference: dict, index: int) -> dict:
         """number reference"""

{commonmeta_py-0.52 → commonmeta_py-0.53}/commonmeta/utils.py

@@ -183,12 +183,18 @@ def normalize_ids(ids: list, relation_type=None) -> list:
     return [format_id(i) for i in ids]
 
 
-def normalize_url(url: Optional[str], secure=False, lower=False) -> Optional[str]:
+def normalize_url(
+    url: Optional[str], secure=False, fragments=False, lower=False
+) -> Optional[str]:
     """Normalize URL"""
     if url is None or not isinstance(url, str):
         return None
+    url = url.strip()
     if url.endswith("/"):
         url = url.strip("/")
+    scheme = urlparse(url).scheme
+    if not scheme or scheme not in ["http", "https"]:
+        return None
     if secure is True and url.startswith(HTTP_SCHEME):
         url = url.replace(HTTP_SCHEME, HTTPS_SCHEME)
     if lower is True:
@@ -196,6 +202,50 @@ def normalize_url(url: Optional[str], secure=False, lower=False) -> Optional[str
     return url
 
 
+# def normalize_url(url: Optional[str], secure=False, fragments=False, lower=False) -> Optional[str]:
+#     """Normalize URL"""
+#     if url is None or not isinstance(url, str):
+#         return None
+#     try:
+#         f = furl(url.strip())
+#         f.path.normalize()
+
+#         # only allow http and https schemes
+#         if f.scheme not in ["http", "https"]:
+#             return None
+#         if secure and f.scheme == "http":
+#             f.set(scheme="https")
+
+#         # remove index.html
+#         if f.path.segments and f.path.segments[-1] in ["index.html"]:
+#             f.path.segments.pop(-1)
+
+#         # remove fragments
+#         if fragments:
+#             f.remove(fragment=True)
+
+#         # remove specific query parameters
+#         f.remove(
+#             [
+#                 "origin",
+#                 "ref",
+#                 "referrer",
+#                 "source",
+#                 "utm_content",
+#                 "utm_medium",
+#                 "utm_campaign",
+#                 "utm_source",
+#             ]
+#         )
+
+#         if lower:
+#             return f.url.lower().strip("/")
+#         return f.url.strip("/")
+#     except ValueError:
+#         print(f"Error normalizing url {url}")
+#         return None
+
+
 def normalize_cc_url(url: Optional[str]):
     """Normalize Creative Commons URL"""
     if url is None or not isinstance(url, str):

{commonmeta_py-0.52 → commonmeta_py-0.53}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name="commonmeta-py"
-version="0.52"
+version="0.53"
 repository = "https://github.com/front-matter/commonmeta-py"
 homepage = "https://python.commonmeta.org"
 documentation = "https://python.commonmeta.org"