commonmeta-py 0.106__py3-none-any.whl → 0.107__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__ = "0.106"
+__version__ = "0.107"
 __author__ = "Martin Fenner"
 __license__ = "MIT"
 
@@ -31,6 +31,18 @@ from .base_utils import (
     unwrap,
     wrap,
 )
+from .crossref_utils import (
+    CrossrefBadRequestError,
+    CrossrefError,
+    CrossrefForbiddenError,
+    CrossrefNoContentError,
+    CrossrefNotFoundError,
+    CrossrefRequestError,
+    CrossrefServerError,
+    CrossrefUnauthorizedError,
+    HttpError,
+    generate_crossref_xml,
+)
 from .date_utils import (
     get_date_from_crossref_parts,
     get_date_from_date_parts,

commonmeta/crossref_utils.py

@@ -231,7 +231,6 @@ def insert_citation_list(metadata, xml):
 
     citation_list = etree.SubElement(xml, "citation_list")
     for i, ref in enumerate(metadata.references):
-        print(i)
         if ref.get("id", None) is None:
             continue
         citation = etree.SubElement(
@@ -581,3 +580,83 @@ def generate_crossref_xml_list(metalist) -> Optional[str]:
         doctype='<?xml version="1.0" encoding="UTF-8"?>',
         pretty_print=True,
     )
+
+
+"""Errors for the Crossref XML API.
+
+Error responses will be converted into an exception from this module.
+"""
+
+
+class HttpError(Exception):
+    """Exception raised when a connection problem happens."""
+
+
+class CrossrefError(Exception):
+    """Exception raised when the server returns a known HTTP error code.
+
+    Known HTTP error codes include:
+
+    * 204 No Content
+    * 400 Bad Request
+    * 401 Unauthorized
+    * 403 Forbidden
+    * 404 Not Found
+    * 410 Gone (deleted)
+    """
+
+    @staticmethod
+    def factory(err_code, *args):
+        """Create exceptions through a Factory based on the HTTP error code."""
+        if err_code == 204:
+            return CrossrefNoContentError(*args)
+        elif err_code == 400:
+            return CrossrefBadRequestError(*args)
+        elif err_code == 401:
+            return CrossrefUnauthorizedError(*args)
+        elif err_code == 403:
+            return CrossrefForbiddenError(*args)
+        elif err_code == 404:
+            return CrossrefNotFoundError(*args)
+        else:
+            return CrossrefServerError(*args)
+
+
+class CrossrefServerError(CrossrefError):
+    """An internal server error happened on the Crossref end. Try later.
+
+    Base class for all 5XX-related HTTP error codes.
+    """
+
+
+class CrossrefRequestError(CrossrefError):
+    """A Crossref request error. You made an invalid request.
+
+    Base class for all 4XX-related HTTP error codes as well as 204.
+    """
+
+
+class CrossrefNoContentError(CrossrefRequestError):
+    """DOI is known to Crossref, but not resolvable.
+
+    This might be due to handle's latency.
+    """
+
+
+class CrossrefBadRequestError(CrossrefRequestError):
+    """Bad request error.
+
+    Bad requests can include e.g. invalid XML, wrong domain, wrong prefix.
+    """
+
+
+class CrossrefUnauthorizedError(CrossrefRequestError):
+    """Bad username or password."""
+
+
+class CrossrefForbiddenError(CrossrefRequestError):
+    """Login problem, record belongs to another party or quota exceeded."""
+
+
+class CrossrefNotFoundError(CrossrefRequestError):
+    """DOI does not exist in the database."""

commonmeta/utils.py

@@ -408,14 +408,36 @@ def openalex_api_url(id: str, identifier_type: str, **kwargs) -> str:
 
 def openalex_api_query_url(query: dict) -> str:
     """Return the OpenAlex API query URL"""
+    # Define allowed types
+    types = [
+        "article", "book-chapter", "dataset", "preprint", "dissertation",
+        "book", "review", "paratext", "libguides", "letter", "other",
+        "reference-entry", "report", "editorial", "peer-review", "erratum",
+        "standard", "grant", "supplementary-materials", "retraction",
+    ]
+
     url = "https://api.openalex.org/works"
     f = furl(url)
-    rows = min(int(query.get("rows", 20)), 1000)
-    queries = []
+
+    # Handle pagination and sample parameters
+    number = max(1, min(1000, int(query.get("number", query.get("rows", 10)))))
+    page = max(1, int(query.get("page", 1)))
+
+    sample = query.get("sample", False)
+    if sample:
+        f.args["sample"] = str(number)
+    else:
+        f.args["per-page"] = str(number)
+        f.args["page"] = str(page)
+        # Sort results by published date in descending order
+        f.args["sort"] = "publication_date:desc"
+
+    # Build filters
     filters = []
+    queries = []
     _query = None
-    _filter = None
 
+    # Handle query parameters
     if query.get("query", None) is not None:
         queries += [query.get("query")]
     for key, value in query.items():
@@ -428,23 +450,60 @@ def openalex_api_query_url(query: dict) -> str:
             queries += [f"{key}:{value}"]
     if queries:
         _query = ",".join(queries)
+        f.args["query"] = _query
+
+    # Member/IDs filter
+    ids = query.get("ids", query.get("member", ""))
+    if ids:
+        filters.append(f"member:{ids}")
+
+    # Type filter
+    type_ = query.get("type_", query.get("type", ""))
+    if type_ and type_ in types:
+        filters.append(f"type:{type_}")
 
+    # ROR filter
+    ror = query.get("ror", "")
+    if ror:
+        r = validate_ror(ror)
+        if r:
+            filters.append(f"authorships.institutions.ror:{r}")
+
+    # ORCID filter
+    orcid = query.get("orcid", "")
+    if orcid:
+        o = validate_orcid(orcid)
+        if o:
+            filters.append(f"authorships.author.id:{o}")
+
+    # Year filter
+    year = query.get("year", query.get("publication_year", ""))
+    if year:
+        filters.append(f"publication_year:{year}")
+
+    # Other filters from the original function
     for key, value in query.items():
         if key in [
             "prefix",
-            "member",
-            "type",
             "has-full-text",
-            "has-references",
-            "has-orcid",
             "has-funder",
             "has-license",
         ]:
-            filters += [f"{key}:{value}"]
-    if filters:
-        _filter = ",".join(filters)
+            filters.append(f"{key}:{value}")
+
+    # Boolean filters
+    # if query.get("hasORCID", query.get("has-orcid", False)):
+    #     filters.append("has-orcid:true")
+
+    # if query.get("hasReferences", query.get("has-references", False)):
+    #     filters.append("has-references:true")
 
-    f.args.update(compact({"rows": rows, "query": _query, "filter": _filter}))
+    # if query.get("hasAbstract", query.get("has-abstract", False)):
+    #     filters.append("has-abstract:true")
+
+    # Add filters to params if any exist
+    if filters:
+        f.args["filter"] = ",".join(filters)
 
     return f.url
 

commonmeta/writers/citation_writer.py

@@ -1,13 +1,18 @@
 """Citation writer for commonmeta-py"""
 
-import orjson as json
 import re
-from pydash import py_
-from citeproc import CitationStylesStyle, CitationStylesBibliography
-from citeproc import Citation, CitationItem
-from citeproc import formatter
+
+import orjson as json
+from citeproc import (
+    Citation,
+    CitationItem,
+    CitationStylesBibliography,
+    CitationStylesStyle,
+    formatter,
+)
 from citeproc.source.json import CiteProcJSON
 from citeproc_styles import get_style_filepath
+from pydash import py_
 
 
 def write_citation(metadata):

commonmeta/writers/commonmeta_writer.py

@@ -3,6 +3,7 @@
 import orjson as json
 import orjsonl
 import pydash as py_
+
 from ..base_utils import compact
 
 

commonmeta/writers/crossref_xml_writer.py

@@ -1,6 +1,7 @@
 """Crossref XML writer for commonmeta-py"""
 
 from typing import Optional
+
 from ..constants import Commonmeta
 from ..crossref_utils import generate_crossref_xml, generate_crossref_xml_list
 

commonmeta/writers/csl_writer.py

@@ -1,13 +1,14 @@
 """CSL-JSON writer for commonmeta-py"""
 
-import orjson as json
 from typing import Optional
 
-from ..utils import pages_as_string, to_csl
-from ..base_utils import wrap, presence, parse_attributes, compact
+import orjson as json
+
+from ..base_utils import compact, parse_attributes, presence, wrap
+from ..constants import CM_TO_CSL_TRANSLATIONS, Commonmeta
 from ..date_utils import get_date_parts
 from ..doi_utils import doi_from_url
-from ..constants import CM_TO_CSL_TRANSLATIONS, Commonmeta
+from ..utils import pages_as_string, to_csl
 
 
 def write_csl(metadata: Commonmeta) -> Optional[str]:

commonmeta/writers/datacite_writer.py

@@ -1,19 +1,20 @@
 """DataCite writer for commonmeta-py"""
 
-import orjson as json
 from typing import Optional, Union
 
-from ..base_utils import wrap, compact
-from ..doi_utils import doi_from_url, normalize_doi
+import orjson as json
+
+from ..base_utils import compact, wrap
 from ..constants import (
     CM_TO_BIB_TRANSLATIONS,
-    CM_TO_CSL_TRANSLATIONS,
     CM_TO_CR_TRANSLATIONS,
+    CM_TO_CSL_TRANSLATIONS,
     CM_TO_DC_TRANSLATIONS,
     CM_TO_RIS_TRANSLATIONS,
     CM_TO_SO_TRANSLATIONS,
     Commonmeta,
 )
+from ..doi_utils import doi_from_url, normalize_doi
 
 
 def write_datacite(metadata: Commonmeta) -> Optional[Union[str, dict]]:

commonmeta/writers/inveniordm_writer.py

@@ -1,23 +1,24 @@
 """InvenioRDM writer for commonmeta-py"""
 
-import orjson as json
 from typing import Optional
 
-from ..base_utils import compact, wrap, parse_attributes, presence
-from ..date_utils import get_iso8601_date
-from ..doi_utils import doi_from_url, normalize_doi
+import orjson as json
+
+from ..base_utils import compact, parse_attributes, presence, wrap
 from ..constants import (
     CM_TO_INVENIORDM_TRANSLATIONS,
-    INVENIORDM_IDENTIFIER_TYPES,
     CROSSREF_FUNDER_ID_TO_ROR_TRANSLATIONS,
+    INVENIORDM_IDENTIFIER_TYPES,
 )
+from ..date_utils import get_iso8601_date
+from ..doi_utils import doi_from_url, normalize_doi
 from ..utils import (
+    FOS_MAPPINGS,
     get_language,
-    validate_orcid,
-    validate_ror,
     id_from_url,
     normalize_url,
-    FOS_MAPPINGS,
+    validate_orcid,
+    validate_ror,
 )
 
 

{commonmeta_py-0.106.dist-info → commonmeta_py-0.107.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: commonmeta-py
-Version: 0.106
+Version: 0.107
 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
@@ -22,6 +22,7 @@ Requires-Dist: edtf<6,>=5.0.0
 Requires-Dist: furl<3,>=2.1.3
 Requires-Dist: jsonschema~=4.21
 Requires-Dist: lxml>=4.8
+Requires-Dist: marshmallow-utils>=0.10.0
 Requires-Dist: nameparser<2,>=1.1.1
 Requires-Dist: nh3<0.3,>=0.2.14
 Requires-Dist: orjson<4,>=3.9.14

{commonmeta_py-0.106.dist-info → commonmeta_py-0.107.dist-info}/RECORD

@@ -1,16 +1,16 @@
-commonmeta/__init__.py,sha256=4QbwqizFuHpgUA-IbFXoocikH5j9AXSEAHpyKfF3lOk,1933
+commonmeta/__init__.py,sha256=Vz7vv934Mmzs1qxlSSx2bkuMVYtdkFJvvikHMihyzAA,2220
 commonmeta/api_utils.py,sha256=jkM1d4jgvdZ5k0gkLeg8NDXsizPEesiHNB9sF7P-gFA,2687
 commonmeta/author_utils.py,sha256=3lYW5s1rOUWNTKs1FP6XLfEUY3yCLOe_3L_VdJTDMp0,8585
 commonmeta/base_utils.py,sha256=TCaMENjyySStUxFowg6R08Vt5eLkHfW7WjdEHVcAZWo,3752
 commonmeta/cli.py,sha256=_sMYNzfiEgQ1qx6tZPMbOp1GZ9BBiWe0fYF3jJlamKU,6251
 commonmeta/constants.py,sha256=NuugDPtaAlYx7d-0WKPHNO-qnjt4rzcj-EbzL2vSi2Q,19259
-commonmeta/crossref_utils.py,sha256=mEXNP7LQXcJyEmCqcpWxi_VI3GU1ZD_XvarYjUMxPL0,22284
+commonmeta/crossref_utils.py,sha256=odkcP_Ivk0Fs_BvUzxdPQ4F2098RIsKQT_NbZGQXo4A,24354
 commonmeta/date_utils.py,sha256=H2cCobX0JREIUOT_cCigGd3MG7prGiQpXk1m4ZNrFwU,6318
 commonmeta/doi_utils.py,sha256=ZztajfOLtnASk1BbQ1Y2Q4B_xxlnbujn7Opx5a1U5vY,9582
 commonmeta/metadata.py,sha256=qgqbrvxPEbYn0pKIArhpaGjIeiHeuubc6nF37QgWp2Y,16702
 commonmeta/schema_utils.py,sha256=yEb8Bt7WqjfQW5y15TV1uaZZdK_k9Ufz6V7Nq7Ywhi0,917
 commonmeta/translators.py,sha256=CBMK4jrXRmGZiAhCh6wsJjhbDJWbcsda8UvXFXxccAw,1363
-commonmeta/utils.py,sha256=APDXdeR2oWyn15vcQPJjPisRHn14hzt1q_21vPxkFCE,48589
+commonmeta/utils.py,sha256=G4BL_YQ-Ug0XwBHU4L83OViRQCiyOqZXg5uJC7WQZ2A,50462
 commonmeta/readers/__init__.py,sha256=vOf7UsOKNoh_ZCuyexxhAmPMt8wjB-pF_CfpWRaN8pk,45
 commonmeta/readers/bibtex_reader.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 commonmeta/readers/cff_reader.py,sha256=kIG-OKGQz37oWL6cU3Cz7HQ83DnRCmQ2PKSjIDodSp8,6152
@@ -61,16 +61,16 @@ commonmeta/resources/styles/modern-language-association.csl,sha256=HI2iU4krze1aH
 commonmeta/resources/styles/vancouver.csl,sha256=lun3_i2oTilgsANk4LjFao2UDPQlGj_hgFgKAWC_DF8,12878
 commonmeta/writers/__init__.py,sha256=47-snms6xBHkoEXKYV1DBtH1npAtlVtvY29Z4Zr45qI,45
 commonmeta/writers/bibtex_writer.py,sha256=RjxpFt19N2PjX7cNtvN8kMYTxLWp_KhLYTsQNxYiC-4,4972
-commonmeta/writers/citation_writer.py,sha256=7CIC6zuT87dkqTDx-r7xvdbXauAFopNiBvR4CBThr5U,2357
-commonmeta/writers/commonmeta_writer.py,sha256=mu-yXQvQs_xdYM4MxbTTxxYB5n0lJA5BqiYfRPV562Q,1870
-commonmeta/writers/crossref_xml_writer.py,sha256=ckEOL1q7x8obWLzoMrLuOiDuRKiXHammI1m_GA85BVg,493
-commonmeta/writers/csl_writer.py,sha256=_5tmPFLzEIaRgYI9Wz00nPwtfXmBuukCrTVUEFuf83c,2808
-commonmeta/writers/datacite_writer.py,sha256=rygkSNUWdF4lvGd4BsjrlBfPHHsSLwkFp-aYLGOWNiU,6305
-commonmeta/writers/inveniordm_writer.py,sha256=M_VmBZ_E5x_BHSgMjt2u70c34Eij2ngoaDwYWOrJzkg,11429
+commonmeta/writers/citation_writer.py,sha256=qs_4X3BjrSqHexmJFPvPDTp0mRIqzb0F70_Wuc7S9x0,2343
+commonmeta/writers/commonmeta_writer.py,sha256=erIIcAGBy9cZb6FWSR6P0qzxr1c6iS3oCgKbGAsA94o,1871
+commonmeta/writers/crossref_xml_writer.py,sha256=REORiWPyT2x13Y66IT1RNtOwB9NWutrzNCt_HLeQrsY,494
+commonmeta/writers/csl_writer.py,sha256=IU70HwS77DD-NhQjc4TEGQVSaF-1TgIvyVR9E_zOqy0,2809
+commonmeta/writers/datacite_writer.py,sha256=vijXscY411UMC72WL9VUYJhBtH6ZCPQTk8xtCo1jYtQ,6306
+commonmeta/writers/inveniordm_writer.py,sha256=vGCItVrxkzPgvV-_m-cYnxTSux6mc3eer4JpI-10JhU,11430
 commonmeta/writers/ris_writer.py,sha256=CXd9MiSHOMLReeGmr1moqteiq4x8AtLNYVS60s-aq8I,2094
 commonmeta/writers/schema_org_writer.py,sha256=J2Y-krzyhgBZ6n__S_YvdjUtwiCm-RUtpGA_esFZUwg,5742
-commonmeta_py-0.106.dist-info/METADATA,sha256=JsAnwb_wOkMWQs7f4oc7VB9Ig6XPOY6b78n6YUqqwDE,7511
-commonmeta_py-0.106.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-commonmeta_py-0.106.dist-info/entry_points.txt,sha256=U4w4BoRuS3rN5t5Y-uYSyOeU5Lh_VRVMS9OIDzIgw4w,50
-commonmeta_py-0.106.dist-info/licenses/LICENSE,sha256=wsIvxF9Q9GC9vA_s79zTWP3BkXJdfUNRmALlU8GbW1s,1074
-commonmeta_py-0.106.dist-info/RECORD,,
+commonmeta_py-0.107.dist-info/METADATA,sha256=1jO_8bgd6pHyUSLCr_LKKRwEYy73kXIgJJaEBowdalw,7552
+commonmeta_py-0.107.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+commonmeta_py-0.107.dist-info/entry_points.txt,sha256=U4w4BoRuS3rN5t5Y-uYSyOeU5Lh_VRVMS9OIDzIgw4w,50
+commonmeta_py-0.107.dist-info/licenses/LICENSE,sha256=wsIvxF9Q9GC9vA_s79zTWP3BkXJdfUNRmALlU8GbW1s,1074
+commonmeta_py-0.107.dist-info/RECORD,,