commonmeta-py 0.63__py3-none-any.whl → 0.65__py3-none-any.whl

commonmeta/__init__.py

@@ -10,7 +10,7 @@ commonmeta-py is a Python library to convert scholarly metadata
 """
 
 __title__ = "commonmeta-py"
-__version__ = "0.63"
+__version__ = "0.65"
 __author__ = "Martin Fenner"
 __license__ = "MIT"
 
@@ -94,4 +94,5 @@ from .doi_utils import (
     normalize_doi,
     validate_doi,
     validate_prefix,
+    is_rogue_scholar_doi,
 )

commonmeta/utils.py

@@ -1096,22 +1096,48 @@ def from_curie(id: Optional[str]) -> Optional[str]:
     if id is None:
         return None
     _type = id.split(":")[0]
-    if _type == "DOI":
+    if _type.upper() == "DOI":
         return doi_as_url(id.split(":")[1])
-    elif _type == "ROR":
+    elif _type.upper() == "ROR":
         return "https://ror.org/" + id.split(":")[1]
-    elif _type == "ISNI":
+    elif _type.upper() == "ISNI":
         return "https://isni.org/isni/" + id.split(":")[1]
-    elif _type == "ORCID":
+    elif _type.upper() == "ORCID":
         return normalize_orcid(id.split(":")[1])
-    elif _type == "URL":
+    elif _type.upper() == "URL":
         return normalize_url(id.split(":")[1])
-    elif _type == "JDP":
+    elif _type.upper() == "JDP":
         return id.split(":")[1]
     # TODO: resolvable url for other identifier types
     return None
 
 
+def extract_curie(string: Optional[str]) -> Optional[str]:
+    """Extract CURIE"""
+    if string is None:
+        return None
+    match = re.search(r"((?:doi|DOI):\s?([\w.,@?^=%&:\/~+#-]*[\w@?^=%&\/~+#-]))", string)
+    if match is None:
+        return None
+    return doi_as_url(match.group(2))
+
+    
+def extract_url(string: str) -> list:
+    """Extract urls from string, including markdown and html."""
+
+    match = re.search(r"((?:http|https):\/\/(?:[\w_-]+(?:(?:\.[\w_-]+)+))(?:[\w.,@?^=%&:\/~+#-]*[\w@?^=%&\/~+#-]))", string)
+    if match is None:
+        return None
+    return normalize_url(match.group(1))
+
+
+def extract_urls(string: str) -> list:
+    """Extract urls from string, including markdown and html."""
+
+    urls = re.findall(r"((?:http|https):\/\/(?:[\w_-]+(?:(?:\.[\w_-]+)+))(?:[\w.,@?^=%&:\/~+#-]*[\w@?^=%&\/~+#-]))", string)
+    return py_.uniq(urls)
+
+
 def issn_as_url(issn: str) -> Optional[str]:
     """ISSN as URL"""
     if normalize_issn(issn) is None:

commonmeta/writers/inveniordm_writer.py

@@ -257,7 +257,8 @@ def to_inveniordm_reference(reference: dict) -> dict:
         identifier = reference.get("id", None)
         scheme = "url"
     else:
-        return None
+        identifier = None
+        scheme = None
 
     if reference.get("unstructured", None) is None:
         # use title as unstructured reference

{commonmeta_py-0.63.dist-info → commonmeta_py-0.65.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: commonmeta-py
-Version: 0.63
+Version: 0.65
 Summary: Library for conversions to/from the Commonmeta scholarly metadata format
 Home-page: https://python.commonmeta.org
 License: MIT

{commonmeta_py-0.63.dist-info → commonmeta_py-0.65.dist-info}/RECORD

@@ -1,4 +1,4 @@
-commonmeta/__init__.py,sha256=Vn41OSNVhIRsKaPeFCbK6z9ms3fO0VD2e_p8lSoBe90,1795
+commonmeta/__init__.py,sha256=LqrBGNgQuAGQiRF9HrhX5iHHu14t44M7nTTOYmg4Rmc,1821
 commonmeta/api_utils.py,sha256=-ZHGVZZhJqnjnsLtp4-PoeHYbDqL0cQme7W70BEjo4U,2677
 commonmeta/author_utils.py,sha256=zBIPTgP5n7Zx57xomJ2h7x0dvC0AV8gJ2gPoYeDy5Lo,8348
 commonmeta/base_utils.py,sha256=AsUElA5kT2fw_Osy7Uaj2F6MKeq9yB7d5f2V-h2lh7c,3750
@@ -58,7 +58,7 @@ commonmeta/resources/styles/modern-language-association.csl,sha256=HI2iU4krze1aH
 commonmeta/resources/styles/vancouver.csl,sha256=lun3_i2oTilgsANk4LjFao2UDPQlGj_hgFgKAWC_DF8,12878
 commonmeta/schema_utils.py,sha256=gg3l1jd_lFtRkQlO1DYGMVbC10nEmVTN4AWacxC4AAE,915
 commonmeta/translators.py,sha256=RpGJtKNLjmz41VREZDY7KyyE2eXOi8j7m-da4jHmknI,1362
-commonmeta/utils.py,sha256=0ky8xyDQWVND5nJWApPgyVhbjXdPPzfpx4fJpX9ivyw,43674
+commonmeta/utils.py,sha256=lIH7VejIn_gReLsuXsAZxE-RiMCRGECA_6aPrhGsBFc,44596
 commonmeta/writers/__init__.py,sha256=47-snms6xBHkoEXKYV1DBtH1npAtlVtvY29Z4Zr45qI,45
 commonmeta/writers/bibtex_writer.py,sha256=s3hIJIgWvSG7TAriZMRQEAyuitw6ebwWSI1YcYFQ-do,4971
 commonmeta/writers/citation_writer.py,sha256=RjaNh9EALxq6gfODLRWVJxGxPArGd6ZiHUlkYnCT6MA,2355
@@ -66,11 +66,11 @@ commonmeta/writers/commonmeta_writer.py,sha256=2qlttCfYpGhfVjrYkjzbIra7AywssRLT3
 commonmeta/writers/crossref_xml_writer.py,sha256=0Ds494RnXfdfjWw5CLX1kwV2zP7gqffdVqO-X74Uc6c,492
 commonmeta/writers/csl_writer.py,sha256=6N-93R1emcOsZrUTIhPBVd_Fv1C8Z5EAFYI0mYjoYaY,2797
 commonmeta/writers/datacite_writer.py,sha256=G7Lr0aZ4sAEdbfXe3dG4Y6AyGUKA9UWr_iiaQRDnV24,6233
-commonmeta/writers/inveniordm_writer.py,sha256=oVcXdHYtuugbfDgKO8JwTRXmP7AK2U96uK4YF0eKXAY,11399
+commonmeta/writers/inveniordm_writer.py,sha256=YXLfiMkWDMMd7ZlOzhp0zNieQFfHKZ4m5FQLIl_XuWI,11427
 commonmeta/writers/ris_writer.py,sha256=AcnCszS3WY9lF594NbFBtLylsA8ownnYp_XLQJ84Ios,2093
 commonmeta/writers/schema_org_writer.py,sha256=5j002uCNLdlScZMNQmPjodcVWqaBh2z38zL1H4lo2hY,5741
-commonmeta_py-0.63.dist-info/LICENSE,sha256=746hEF2wZCKkcckk5-_DcBLtHewfaEMS4iXTlA1PVwk,1074
-commonmeta_py-0.63.dist-info/METADATA,sha256=RQXKld5RnUdijPEzxt4KFE8qoZ9i0MkWTB3hmk2f12k,8279
-commonmeta_py-0.63.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
-commonmeta_py-0.63.dist-info/entry_points.txt,sha256=vbcDw3_2lMTKdcAL2VUF4DRYRpKuzXVYLMCdgKVf88U,49
-commonmeta_py-0.63.dist-info/RECORD,,
+commonmeta_py-0.65.dist-info/LICENSE,sha256=746hEF2wZCKkcckk5-_DcBLtHewfaEMS4iXTlA1PVwk,1074
+commonmeta_py-0.65.dist-info/METADATA,sha256=th0VmBY3Kk5evcQrUOk55lTyqjeZ6CEuXnvumG_Duvc,8279
+commonmeta_py-0.65.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
+commonmeta_py-0.65.dist-info/entry_points.txt,sha256=vbcDw3_2lMTKdcAL2VUF4DRYRpKuzXVYLMCdgKVf88U,49
+commonmeta_py-0.65.dist-info/RECORD,,