commonmeta-py 0.112__py3-none-any.whl → 0.114__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__ = "0.112"
+__version__ = "0.114"
 __author__ = "Martin Fenner"
 __license__ = "MIT"
 

commonmeta/doi_utils.py

@@ -302,19 +302,101 @@ def datacite_api_sample_url(number: int = 1, **kwargs) -> str:
     return f"https://api.datacite.org/dois?random=true&page[size]={number}"
 
 
-def is_rogue_scholar_doi(doi: str) -> bool:
-    """Return True if DOI is from Rogue Scholar"""
-    prefix = validate_prefix(doi)
-    return prefix in [
-        "10.34732",  # not managed by Front Matter
+def is_rogue_scholar_doi(doi: str, ra: str = "crossref") -> bool:
+    """Check if a DOI is from Rogue Scholar with specific registration agency"""
+    rogue_scholar_crossref_prefixes = [
+        "10.13003",
         "10.53731",
         "10.54900",
-        "10.57689",  # not managed by Front Matter
-        "10.58079",  # not managed by Front Matter
+        "10.57689",
+        "10.59347",
         "10.59348",
         "10.59349",
         "10.59350",
         "10.63485",
         "10.64000",
+    ]
+    rogue_scholar_datacite_prefixes = [
+        "10.5438",
+        "10.34732",  # not managed by Front Matter
+        "10.57689",  # not managed by Front Matter
+        "10.58079",  # not managed by Front Matter
+        "10.60804",
         "10.71938",  # not managed by Front Matter
+        # "10.83132",
     ]
+
+    prefix = validate_prefix(doi)
+    if not prefix:
+        return False
+
+    is_crossref = prefix in rogue_scholar_crossref_prefixes
+    is_datacite = prefix in rogue_scholar_datacite_prefixes
+
+    if ra == "crossref":
+        return is_crossref
+    elif ra == "datacite":
+        return is_datacite
+    return is_crossref or is_datacite
+
+
+def generate_wordpress_doi(prefix: str, slug: str, guid: str) -> str:
+    """Generate a DOI from a WordPress GUID and slug"""
+    import re
+
+    if not prefix or not guid:
+        return ""
+
+    pattern = re.compile(r"p=(\d+)$")
+    matched = pattern.search(guid)
+
+    if not matched:
+        return ""
+
+    doi = f"https://doi.org/{prefix}/{slug}.{matched.group(1)}"
+    return doi
+
+
+def generate_doi_from_guid(prefix: str, guid: str) -> str:
+    """Validates a GUID that is a DOI"""
+    import base32_lib as base32
+
+    if not prefix:
+        return ""
+
+    doi = normalize_doi(guid)
+    if not doi:
+        return ""
+
+    p = validate_prefix(doi)
+    if not p or p != prefix:
+        return ""
+
+    suffix = doi.split("/")[-1]
+
+    try:
+        number = base32.decode(suffix, checksum=True)
+        if number != 0:
+            return doi
+    except (ValueError, IndexError):
+        pass
+
+    return ""
+
+
+def generate_substack_doi(prefix: str, guid: str) -> str:
+    """Generate a DOI from a Substack GUID"""
+    import base32_lib as base32
+
+    if not prefix or not guid:
+        return ""
+
+    try:
+        i = int(guid)
+    except ValueError:
+        return ""
+
+    # encode the number using base32 with length=4, split_every=8, and checksum=True
+    suffix = base32.encode(i, length=4, split_every=8, checksum=True)
+    doi = f"https://doi.org/{prefix}/{suffix}"
+    return doi

commonmeta/readers/jsonfeed_reader.py

@@ -13,6 +13,9 @@ from ..date_utils import get_date_from_unix_timestamp
 from ..doi_utils import (
     doi_from_url,
     encode_doi,
+    generate_doi_from_guid,
+    generate_substack_doi,
+    generate_wordpress_doi,
     is_rogue_scholar_doi,
     normalize_doi,
     validate_doi,
@@ -55,13 +58,39 @@ def read_jsonfeed(data: Optional[dict], **kwargs) -> Commonmeta:
         "archive_url", None
     ):
         url = normalize_url(meta.get("archive_url", None))
-    _id = normalize_doi(read_options.get("doi", None) or meta.get("doi", None)) or url
-    _type = "BlogPost"
 
-    # optionally generate a DOI if missing but a DOI prefix is provided
-    prefix = read_options.get("prefix", None) or py_.get(meta, "blog.prefix", None)
-    if doi_from_url(_id) is None and prefix is not None:
-        _id = encode_doi(prefix)
+    # generate DOI string for registration if not provided
+    _id = normalize_doi(read_options.get("doi", None) or meta.get("doi", None))
+    if _id is None:
+        if meta.get("guid") and py_.get(meta, "blog.doi_reg", False):
+            # Generate DOI based on blogging platform
+            generator = py_.get(meta, "blog.generator")
+            prefix = py_.get(meta, "blog.prefix")
+            slug = py_.get(meta, "blog.slug")
+            guid = meta.get("guid")
+
+            # Import these functions only when needed to avoid circular imports
+            if generator in ["WordPress", "WordPress.com"] and prefix and slug and guid:
+                _id = generate_wordpress_doi(prefix, slug, guid)
+            elif generator == "Substack" and prefix and guid:
+                _id = generate_substack_doi(prefix, guid)
+            elif prefix and guid:
+                _id = generate_doi_from_guid(prefix, guid)
+
+        # If still no DOI but prefix provided and not registered for DOI generation
+        elif py_.get(meta, "blog.prefix") and not py_.get(meta, "blog.doi_reg", False):
+            prefix = py_.get(meta, "blog.prefix")
+            _id = encode_doi(prefix)
+
+        # If override prefix is provided in read_options, use that
+        elif read_options.get("prefix"):
+            _id = encode_doi(read_options.get("prefix"))
+
+    # fall back to url if no DOI can be generated
+    if _id is None:
+        _id = url
+
+    _type = "BlogPost"
 
     if meta.get("authors", None):
         contributors = get_authors(from_jsonfeed(wrap(meta.get("authors"))))

commonmeta/writers/inveniordm_writer.py

@@ -23,6 +23,7 @@ from ..utils import (
     get_language,
     id_from_url,
     normalize_url,
+    pages_as_string,
     validate_orcid,
     validate_ror,
 )
@@ -68,7 +69,7 @@ def write_inveniordm(metadata):
         ]
     )
     container = metadata.container if metadata.container else {}
-    journal = (
+    journal_title = (
         container.get("title", None)
         if _type not in ["inbook", "inproceedings"]
         and container.get("type") in ["Journal", "Periodical", "Blog"]
@@ -79,6 +80,10 @@ def write_inveniordm(metadata):
         if container.get("identifierType", None) == "ISSN"
         else None
     )
+    volume = container.get("volume", None)
+    issue = container.get("issue", None)
+    pages = pages_as_string(container)
+
     dates = []
     for date in metadata.date.keys():
         if metadata.date.get(date, None) is None:
@@ -141,7 +146,15 @@ def write_inveniordm(metadata):
             ),
             "custom_fields": compact(
                 {
-                    "journal:journal": compact({"title": journal, "issn": issn}),
+                    "journal:journal": compact(
+                        {
+                            "title": journal_title,
+                            "issn": issn,
+                            "volume": volume,
+                            "issue": issue,
+                            "pages": pages,
+                        }
+                    ),
                     "rs:content_html": presence(metadata.content),
                     "rs:image": presence(metadata.image),
                     "rs:generator": container.get("platform", None),
@@ -463,7 +476,6 @@ def push_inveniordm(metadata, host: str, token: str, legacy_key: str):
         # optionally update rogue-scholar legacy record
         if host == "rogue-scholar.org" and legacy_key is not None:
             record = update_legacy_record(record, legacy_key)
-        print("g", record)
     except Exception as e:
         raise InvenioRDMError(f"Unexpected error: {str(e)}")
 
@@ -511,12 +523,11 @@ def create_draft_record(record, host, token, input):
         )
         response.raise_for_status()
         data = response.json()
-        return {
-            "id": data.get("id", None),
-            "created": data.get("created", None),
-            "updated": data.get("updated", None),
-            "status": "updated",
-        }
+        record["id"]: data.get("id", None)
+        record["created"] = data.get("created", None)
+        record["updated"] = data.get("updated", None)
+        record["status"] = "draft"
+        return record
     except requests.exceptions.RequestException as e:
         raise InvenioRDMError(f"Error creating draft record: {str(e)}")
 
@@ -610,20 +621,31 @@ def update_legacy_record(record, legacy_key: str):
     legacy_host = "bosczcmeodcrajtcaddf.supabase.co"
 
     if not legacy_key:
-        return record, ValueError("no legacy key provided")
-
+        raise ValueError("no legacy key provided")
     if not record.get("uuid", None):
-        return record, ValueError("no UUID provided")
-
-    now = f"{int(time())}"
-
+        raise ValueError("no UUID provided")
     if not record.get("doi", None):
-        return ValueError("no valid doi to update")
+        raise ValueError("no valid doi to update")
 
-    output = {"indexed_at": now, "indexed": "true", "archived": "true"}
+    now = f"{int(time())}"
+    if record.get("id", None) is not None:
+        output = {
+            "rid": record.get("id"),
+            "indexed_at": now,
+            "indexed": "true",
+            "archived": "true",
+        }
+    elif record.get("doi", None) is not None:
+        output = {
+            "doi": record.get("doi"),
+            "indexed_at": now,
+            "indexed": "true",
+            "archived": "true",
+        }
+    else:
+        return record # nothing to update
 
     request_url = f"https://{legacy_host}/rest/v1/posts?id=eq.{record['uuid']}"
-
     headers = {
         "Content-Type": "application/json",
         "apikey": legacy_key,
@@ -635,13 +657,13 @@ def update_legacy_record(record, legacy_key: str):
         response = requests.patch(request_url, json=output, headers=headers, timeout=30)
         response.raise_for_status()
         if response.status_code != 204:
-            return record, Exception(f"Unexpected status code: {response.status_code}")
+            return Exception(f"Unexpected status code: {response.status_code}")
 
         record["status"] = "updated_legacy"
         return record
 
     except requests.exceptions.RequestException as e:
-        return record, e
+        raise InvenioRDMError(f"Error updating legacy record: {str(e)}")
 
 
 def search_by_slug(slug, type_value, host, token) -> Optional[str]:

{commonmeta_py-0.112.dist-info → commonmeta_py-0.114.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: commonmeta-py
-Version: 0.112
+Version: 0.114
 Summary: Library for conversions to/from the Commonmeta scholarly metadata format
 Project-URL: Homepage, https://python.commonmeta.org
 Project-URL: Repository, https://github.com/front-matter/commonmeta-py

{commonmeta_py-0.112.dist-info → commonmeta_py-0.114.dist-info}/RECORD

@@ -1,11 +1,11 @@
-commonmeta/__init__.py,sha256=eKvnKvDHY9vRV-P7jBGueCHolec-BCQP9YOJWd_L0iM,2098
+commonmeta/__init__.py,sha256=G4y5yCVJHisJQAgUH6EKoM6g79z1I5zu8mfvG9tuYpE,2098
 commonmeta/api_utils.py,sha256=y5KLfIOWOjde7LXZ36u-eneQJ-Q53yXUZg3hWpCBS2E,2685
 commonmeta/author_utils.py,sha256=3lYW5s1rOUWNTKs1FP6XLfEUY3yCLOe_3L_VdJTDMp0,8585
 commonmeta/base_utils.py,sha256=-MGy9q2uTiJEkPWQUYOJMdq-3tRpNnvBwlLjvllQ5g8,11164
 commonmeta/cli.py,sha256=pdBpBosLNq3RS9buO-Voqawc9Ay1eSt-xP5O97iOft4,8480
 commonmeta/constants.py,sha256=wSTEUiHeRdXLwjXEQD9AU2hxFyEKi5OTX2iHOKO6nF0,19844
 commonmeta/date_utils.py,sha256=H2cCobX0JREIUOT_cCigGd3MG7prGiQpXk1m4ZNrFwU,6318
-commonmeta/doi_utils.py,sha256=ZztajfOLtnASk1BbQ1Y2Q4B_xxlnbujn7Opx5a1U5vY,9582
+commonmeta/doi_utils.py,sha256=kS9wBoZQHvV-fqFoW9j-_aN_7Kj1I6sQdqnqK3Nno0M,11512
 commonmeta/file_utils.py,sha256=tGvXxScjh-PPo5YvLDyk4sqwY5Q50N0zAmBHVaUOLeU,3268
 commonmeta/metadata.py,sha256=U9uYOkKAdgqLiQwQKXMXjrAFvO6avMWp9heJG4cNAAY,18893
 commonmeta/schema_utils.py,sha256=WGpmMj9cfNMg_55hhgwY9qpO0A1HSvTLQC2equjBftI,1770
@@ -22,7 +22,7 @@ commonmeta/readers/csl_reader.py,sha256=OxzC2AZKfv43BCah4XGYvlK_LUK-5mxXFcjdzB5v
 commonmeta/readers/datacite_reader.py,sha256=M6gznf1kisR1WzDZaoR0pLJC6Q4Rtnerodfs8lU2khI,12094
 commonmeta/readers/datacite_xml_reader.py,sha256=zJSuN9pnWplYFH7V1eneh0OjKTFCNkOLmEMf6fU6_xg,13048
 commonmeta/readers/inveniordm_reader.py,sha256=6LkT6R20jSFqDdZqAzcREHbdAcIPHiYJvxKsK_mpDdw,8374
-commonmeta/readers/jsonfeed_reader.py,sha256=jn7ux79jGD_ChsBH7zU8_CeCEObIbDh3NDjMhHIzE64,14470
+commonmeta/readers/jsonfeed_reader.py,sha256=zcPxxuyAGW8W7w0-VwP9AhpX97qVWHQJUIJ5p4bBbfE,15655
 commonmeta/readers/kbase_reader.py,sha256=KH3loJvuq2bm8zAYIUG7hTsr5-2Anj3NQvoJUDiqmss,6764
 commonmeta/readers/openalex_reader.py,sha256=4HUkBsut_iUjhUcC5c1GHgxnKsYQc-fgY43QILgVZEg,12826
 commonmeta/readers/ris_reader.py,sha256=oQ3G7qQmNwhr4cNp-Gv5UW28J2K1oKpBlPh-tjRtnpQ,3678
@@ -80,11 +80,11 @@ commonmeta/writers/commonmeta_writer.py,sha256=QpfyhG__7o_XpsOTCPWxGymO7YKwZi2LQ
 commonmeta/writers/crossref_xml_writer.py,sha256=d-Rb2Vd_g3UW8GM4APIT7fivSQ5GMssZ6Ubi3OykHaw,33479
 commonmeta/writers/csl_writer.py,sha256=4gDYs1EzK4_L2UIRTfs25wgHmYRwdRP2zmfxF9387oU,2779
 commonmeta/writers/datacite_writer.py,sha256=bcinpwhq7XnVthKHH8-sdXA34dSlvFH4ImYH768iaQU,6428
-commonmeta/writers/inveniordm_writer.py,sha256=EDIw56kIr-58oB1oWnGOCfUBRwIsivjnR8WT5rrcji0,23334
+commonmeta/writers/inveniordm_writer.py,sha256=1WgFPT6no2v-rurNo-02RcgN3Ic88MdbnKBnVQlv9zo,24142
 commonmeta/writers/ris_writer.py,sha256=3SdyEvMRaPRP1SV1MB-MXBlunE7x6og7RF1zuWtetPc,2094
 commonmeta/writers/schema_org_writer.py,sha256=s18_x0ReXwAGBoEAwp2q-HCgFQ-h5qRg6JyAlqCoSFE,5871
-commonmeta_py-0.112.dist-info/METADATA,sha256=8zV_v5vYgwtipHroPPPc1pms1YB4QNKPpWBCH4uRGrk,7652
-commonmeta_py-0.112.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-commonmeta_py-0.112.dist-info/entry_points.txt,sha256=U4w4BoRuS3rN5t5Y-uYSyOeU5Lh_VRVMS9OIDzIgw4w,50
-commonmeta_py-0.112.dist-info/licenses/LICENSE,sha256=wsIvxF9Q9GC9vA_s79zTWP3BkXJdfUNRmALlU8GbW1s,1074
-commonmeta_py-0.112.dist-info/RECORD,,
+commonmeta_py-0.114.dist-info/METADATA,sha256=a0JivRZx0DkvMk3O1na9HIMHGgUZuvBBDk0Smc8H-Qc,7652
+commonmeta_py-0.114.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+commonmeta_py-0.114.dist-info/entry_points.txt,sha256=U4w4BoRuS3rN5t5Y-uYSyOeU5Lh_VRVMS9OIDzIgw4w,50
+commonmeta_py-0.114.dist-info/licenses/LICENSE,sha256=wsIvxF9Q9GC9vA_s79zTWP3BkXJdfUNRmALlU8GbW1s,1074
+commonmeta_py-0.114.dist-info/RECORD,,