fix language parsing

commonmeta/readers/inveniordm_reader.py

@@ -93,8 +93,6 @@ def read_inveniordm(data: dict, **kwargs) -> Commonmeta:
         ]
     )
     language = py_.get(meta, "metadata.language") or py_.get(meta, "metadata.languages[0].id")
-    if language:
-        language = get_language(language).alpha_2
     subjects = [name_to_fos(i) for i in wrap(py_.get(meta, "metadata.keywords"))]
 
     references = get_references(wrap(py_.get(meta, "metadata.related_identifiers")))
@@ -121,7 +119,7 @@ def read_inveniordm(data: dict, **kwargs) -> Commonmeta:
         # recommended and optional properties
         # "additional_type": additional_type,
         "subjects": presence(subjects),
-        "language": language,
+        "language": get_language(language),
         "version": py_.get(meta, "metadata.version"),
         "license": presence(license_),
         "descriptions": descriptions,

commonmeta/readers/schema_org_reader.py

@@ -19,6 +19,7 @@
     normalize_ids,
     normalize_url,
     name_to_fos,
+    get_language,
 )
 from ..readers.crossref_reader import get_crossref
 from ..readers.datacite_reader import get_datacite
@@ -292,7 +293,7 @@ def read_schema_org(data: Optional[dict], **kwargs) -> Commonmeta:
         # recommended and optional attributes
         "additional_type": additional_type,
         "subjects": presence(subjects),
-        "language": language,
+        "language": get_language(language),
         "identifiers": identifiers,
         "sizes": None,
         "formats": None,

commonmeta/utils.py

@@ -1043,9 +1043,10 @@ def issn_as_url(issn: str) -> Optional[str]:
     return f"https://portal.issn.org/resource/ISSN/{issn}"
 
 
-def get_language(lang: str) -> Optional[dict]:
-    """Provide a language object based on ISO 639, with either a name in English,
-    ISO 639-1, or ISO 639-3 code as input.
+def get_language(lang: str, format: str="alpha_2") -> Optional[str]:
+    """Provide a language string based on ISO 639, with either a name in English,
+    ISO 639-1, or ISO 639-3 code as input. Optionally format as alpha_2 (defaul), 
+    alpha_3, or name.
     """
     if not lang:
         return None
@@ -1055,7 +1056,16 @@ def get_language(lang: str) -> Optional[dict]:
         language = pycountry.languages.get(alpha_3=lang)
     else:
         language = pycountry.languages.get(name=lang)
-    return language
+
+    if language is None:
+        return None
+    elif format == "name":
+        return language.name
+    elif format == "alpha_3":
+        return language.alpha_3
+
+    else:
+        return language.alpha_2
 
 
 def start_case(content: str) -> str:

commonmeta/writers/bibtex_writer.py

@@ -3,7 +3,7 @@
 from bibtexparser.bibdatabase import BibDatabase
 from bibtexparser.customization import page_double_hyphen
 
-from ..utils import pages_as_string
+from ..utils import pages_as_string, get_language
 from ..base_utils import compact
 from ..author_utils import authors_as_string
 from ..date_utils import get_month_from_date, get_iso8601_date, MONTH_SHORT_NAMES
@@ -82,7 +82,7 @@ def write_bibtex_item(metadata: Commonmeta) -> dict:
     booktitle = (
         container.get("title", None) if _type in ["inbook", "inproceedings"] else None
     )
-    language = metadata.language
+    language = get_language(metadata.language)
     location = (
         container.get("location", None)
         if _type not in ["article", "phdthesis"]

commonmeta/writers/schema_org_writer.py

@@ -1,6 +1,6 @@
 """Schema.org writer for commonmeta-py"""
 import orjson as json
-from ..utils import to_schema_org_creators, github_as_repo_url
+from ..utils import to_schema_org_creators, github_as_repo_url, get_language
 from ..base_utils import compact, wrap, presence, parse_attributes
 from ..constants import CM_TO_SO_TRANSLATIONS
 
@@ -117,7 +117,7 @@ def write_schema_org(metadata):
                     wrap(metadata.subjects), content="subject", first=False
                 )
             ),
-            "inLanguage": metadata.language,
+            "inLanguage": get_language(metadata.language, format="name"),
             "dateCreated": metadata.date.get("created", None),
             "datePublished": metadata.date.get("published", None),
             "dateModified": metadata.date.get("updated", None),

tests/cassettes/test-schema_org_reader/test_arxiv.yaml

@@ -571,13 +571,13 @@ interactions:
       Content-Length:
       - '45287'
       Date:
-      - Fri, 27 Sep 2024 16:12:50 GMT
+      - Sun, 29 Sep 2024 08:44:43 GMT
       X-Cache:
-      - MISS, MISS
+      - HIT, MISS
       X-Served-By:
-      - cache-lga21969-LGA, cache-fra-etou8220025-FRA
+      - cache-lga21969-LGA, cache-fra-etou8220148-FRA
       X-Timer:
-      - S1727453570.607760,VS0,VE520
+      - S1727599483.468512,VS0,VE242
       cache-control:
       - max-age=3600
       content-security-policy:
@@ -591,7 +591,7 @@ interactions:
       via:
       - 1.1 google, 1.1 varnish, 1.1 varnish
       x-cloud-trace-context:
-      - 70996ce163e93a14aabf2d2281dd1f0b
+      - e33c0a667e8f81f037b820dada490d93
       x-frame-options:
       - SAMEORIGIN
     status:
@@ -621,19 +621,19 @@ interactions:
       CF-Cache-Status:
       - DYNAMIC
       CF-RAY:
-      - 8c9cc70e4bbac7ed-DUS
+      - 8caab1660939c7c7-DUS
       Connection:
       - keep-alive
       Content-Encoding:
       - gzip
       Content-Type:
       - application/json;charset=UTF-8
       Date:
-      - Fri, 27 Sep 2024 16:12:50 GMT
+      - Sun, 29 Sep 2024 08:44:43 GMT
       NEL:
       - '{"success_fraction":0,"report_to":"cf-nel","max_age":604800}'
       Report-To:
-      - '{"endpoints":[{"url":"https:\/\/a.nel.cloudflare.com\/report\/v4?s=M86uOkY%2BHmKkgERQ8fMD6xKIOolzZO7rwHTlrKNdDl3VbkkVWaA%2BAjDbk4hMH4vr6CuewH8WLobEBEl5NeuLZO0RQXuF7lg32XvHtzf%2B3ZZeMV9Av3nokGMpqrLfLcAkldhhRVA%3D"}],"group":"cf-nel","max_age":604800}'
+      - '{"endpoints":[{"url":"https:\/\/a.nel.cloudflare.com\/report\/v4?s=kku5xi6i%2Bc6dW2J%2FBAPkqOS334XTsyPig7NsjqCKrx5a1sXehfLWOIJZkORBFHpcyEHoFL1%2BibmBuQ9zpjoU0T6wOjJMbtFYpe%2BbIBcr%2FiBzPJ5BEstT8EUnS8TKe5vUebqdSgQ%3D"}],"group":"cf-nel","max_age":604800}'
       Server:
       - cloudflare
       Strict-Transport-Security:
@@ -665,7 +665,7 @@ interactions:
   response:
     body:
       string: !!binary |
-        H4sIAILZ9mYAA7xaWXfiyJL+Kzp+qpr2IgnoMp4ndkSz2OzonnrQkhYJ2kYbiDr13+dLpQSIcve4
+        H4sIAHwT+WYAA7xaWXfiyJL+Kzp+qpr2IgnoMp4ndkSz2OzonnrQkhYJ2kYbiDr13+dLpQSIcve4
         7syZhz5lhSJjjy+CVP+4M7VIu3v5cUfNu5c7SXysPtdq4pMWHGnyKNVF+VGUa5Xq3f1dlPoELKZH
         QzxpURRQPY5IyA6D+D+c9gPyTo9XTKCF8TunfcBPTeJG9J2SAAr+9ePqmcn4mHPOLdSCNU3ufn6H
         kXZEAleLiHIj7YMXpcMfHi0rZvKNgGiRl8t0NYcJ6BOq03tBSTQXYhgxl/wK5Z6r2aBaNCHumPPn
@@ -749,9 +749,9 @@ interactions:
       Content-Type:
       - application/json; charset=utf-8
       Date:
-      - Fri, 27 Sep 2024 16:12:50 GMT
+      - Sun, 29 Sep 2024 08:44:44 GMT
       ETag:
-      - W/"2e8303d763630376f4748e71ec92d1f1"
+      - W/"b0031bc1da359a74e851ba83fb66bf91"
       Referrer-Policy:
       - strict-origin-when-cross-origin
       Server:
@@ -773,9 +773,9 @@ interactions:
       X-Powered-By:
       - Phusion Passenger(R) 6.0.23
       X-Request-Id:
-      - 07b0e136-abc9-459c-85be-52d8c912e335
+      - 446a22b2-942e-43de-b566-ad85bc482022
       X-Runtime:
-      - '0.035136'
+      - '0.037905'
       X-XSS-Protection:
       - '0'
     status:

tests/cassettes/test-schema_org_reader/test_blog_posting.yaml

@@ -21,15 +21,15 @@ interactions:
       Accept-Ranges:
       - bytes
       Age:
-      - '285128'
+      - '431042'
       Alt-Svc:
       - clear
       Connection:
       - keep-alive
       Content-Length:
       - '0'
       Date:
-      - Fri, 27 Sep 2024 16:12:44 GMT
+      - Sun, 29 Sep 2024 08:44:38 GMT
       Ghost-Fastly:
       - 'true'
       Vary:
@@ -39,9 +39,9 @@ interactions:
       X-Cache-Hits:
       - 0, 3, 1
       X-Served-By:
-      - cache-ams2100143-AMS, cache-ams2100143-AMS, cache-fra-eddf8230051-FRA
+      - cache-ams2100143-AMS, cache-ams2100143-AMS, cache-fra-eddf8230070-FRA
       X-Timer:
-      - S1727453565.865111,VS0,VE2
+      - S1727599479.935885,VS0,VE2
       cache-control:
       - public, max-age=31536000
       ghost-age:
@@ -57,7 +57,7 @@ interactions:
       via:
       - 1.1 varnish, 1.1 varnish, 1.1 varnish
       x-request-id:
-      - 63d87d30-4cca-4c1b-b9ca-ec18e096d97f
+      - df93d0ff-f07f-46f2-84fe-89538141dd67
     status:
       code: 301
       message: Moved Permanently
@@ -237,27 +237,27 @@ interactions:
       Accept-Ranges:
       - bytes
       Age:
-      - '73779'
+      - '219693'
       Alt-Svc:
       - clear
       Connection:
       - keep-alive
       Content-Length:
       - '8733'
       Date:
-      - Fri, 27 Sep 2024 16:12:44 GMT
+      - Sun, 29 Sep 2024 08:44:38 GMT
       Ghost-Fastly:
       - 'true'
       Vary:
       - Cookie, Accept-Encoding
       X-Cache:
       - MISS, HIT, HIT
       X-Cache-Hits:
-      - 0, 1, 1
+      - 0, 1, 0
       X-Served-By:
-      - cache-ams21021-AMS, cache-ams21021-AMS, cache-fra-eddf8230051-FRA
+      - cache-ams21021-AMS, cache-ams21021-AMS, cache-fra-eddf8230070-FRA
       X-Timer:
-      - S1727453565.882696,VS0,VE2
+      - S1727599479.956337,VS0,VE1
       cache-control:
       - public, max-age=0
       content-encoding:
@@ -277,7 +277,7 @@ interactions:
       via:
       - 1.1 varnish, 1.1 varnish, 1.1 varnish
       x-request-id:
-      - 1abc9be4-6a48-45a2-87e6-4f663b52aabe
+      - d7f2445e-c3bc-44b0-8881-93c69f1f8dee
     status:
       code: 200
       message: OK
@@ -305,27 +305,25 @@ interactions:
       CF-Cache-Status:
       - DYNAMIC
       CF-RAY:
-      - 8c9cc6ecfda41619-DUS
+      - 8caab1484a7ec7e4-DUS
       Connection:
       - keep-alive
       Content-Encoding:
       - gzip
       Content-Type:
       - application/json;charset=UTF-8
       Date:
-      - Fri, 27 Sep 2024 16:12:44 GMT
+      - Sun, 29 Sep 2024 08:44:39 GMT
       NEL:
       - '{"success_fraction":0,"report_to":"cf-nel","max_age":604800}'
       Report-To:
-      - '{"endpoints":[{"url":"https:\/\/a.nel.cloudflare.com\/report\/v4?s=hVehXeHwTwCLua6GHpmjRsOFKzSaRxyZ4%2Bt%2F4meOA9AzsdVPZjpcdSRRCKa0sM%2FlgeV0spfKCSSSR7Hvs2jzKEkwwI7VDHxWorj%2FcHVOvbgwgRYaJBfiyTY1YY0y3Ll7TxPiOPg%3D"}],"group":"cf-nel","max_age":604800}'
+      - '{"endpoints":[{"url":"https:\/\/a.nel.cloudflare.com\/report\/v4?s=g0i7GSsud%2B%2F7oKI5ItkMzYrs633gTRtHpKCzRUh0bSazP48S8SqbSBiHYqgtQBOgI4UjDkq5Ac8GTfJE9km6sbCm1MCHRhcNeIjGDzcBM8d%2F1eacRDow7QCXvuQVi7aaBLNIa50%3D"}],"group":"cf-nel","max_age":604800}'
       Server:
       - cloudflare
       Strict-Transport-Security:
       - max-age=31536000; includeSubDomains; preload
       Transfer-Encoding:
       - chunked
-      alt-svc:
-      - h3=":443"; ma=86400
       permissions-policy:
       - interest-cohort=(),browsing-topics=()
       vary:
@@ -391,7 +389,7 @@ interactions:
       content-type:
       - application/json
       date:
-      - Fri, 27 Sep 2024 16:12:45 GMT
+      - Sun, 29 Sep 2024 08:44:39 GMT
       permissions-policy:
       - interest-cohort=()
       server:

tests/cassettes/test-schema_org_reader/test_dataverse.yaml

@@ -23,19 +23,19 @@ interactions:
       CF-Cache-Status:
       - DYNAMIC
       CF-RAY:
-      - 8c9cc6fbed6bc81e-DUS
+      - 8caab1558f04c7fe-DUS
       Connection:
       - keep-alive
       Content-Encoding:
       - gzip
       Content-Type:
       - application/json;charset=UTF-8
       Date:
-      - Fri, 27 Sep 2024 16:12:47 GMT
+      - Sun, 29 Sep 2024 08:44:41 GMT
       NEL:
       - '{"success_fraction":0,"report_to":"cf-nel","max_age":604800}'
       Report-To:
-      - '{"endpoints":[{"url":"https:\/\/a.nel.cloudflare.com\/report\/v4?s=b%2FIDuMkDoKWrGahd552RyR5G4qnxKfC16TZsRCkckd40ZrKt%2Fj92UdbXlUfhzw34lbAQuoyydkLpRq0Cuw3y73mkVZLBAKWxUN4f2AahJkMRUmYo7qFEoaVgZ5nD9V1GN0vfGps%3D"}],"group":"cf-nel","max_age":604800}'
+      - '{"endpoints":[{"url":"https:\/\/a.nel.cloudflare.com\/report\/v4?s=ZI0tWsrmxkJGS9hcBG4w7WW58K%2Fuq%2Fg73Qo4Rz1FGuCbWAzRRTn2eWCN25O42i7SsLjwwl62CtXGXZiGwZI3P1bsPfq7g7%2F8B5N3O6z8MZot3iqzXiB0wBvsYt8B67UeQRKwoEk%3D"}],"group":"cf-nel","max_age":604800}'
       Server:
       - cloudflare
       Strict-Transport-Security:
@@ -67,7 +67,7 @@ interactions:
   response:
     body:
       string: !!binary |
-        H4sIAH/Z9mYAA5xXXXPiOhL9Ky5edu/OkNgGJkDV1BaBBMwQyBhig2/dB1sStoj8UZYMMan89235
+        H4sIAHkT+WYAA5xXXXPiOhL9Ky5edu/OkNgGJkDV1BaBBMwQyBhig2/dB1sStoj8UZYMMan89235
         A0gyc+vWvuFWn+6jbql1eG1gV7iN/muD4ka/oalXNz1Nvcb76Dra3bzwuPG1IfKEwBqOKYcvV4iU
         epkgXKLA+DtYkpItfTmvgoln29L0zpFiEgm6pSSFkH/+BSmYIGnkCmJ8XEEpcUVcfL02IjeUvIyo
         9KZx5DJlTCIiKOLKYqsMouecxZxGvrJM4gjnjArKlWEc8TgVNAuVfxvjwfIPuSsgxmgRpMwkg39I
@@ -117,9 +117,9 @@ interactions:
       Content-Type:
       - application/json; charset=utf-8
       Date:
-      - Fri, 27 Sep 2024 16:12:47 GMT
+      - Sun, 29 Sep 2024 08:44:41 GMT
       ETag:
-      - W/"e9ee585353da9a606624ed35e3d57a2b"
+      - W/"1e28b954b5d93cbc96d1469a3d00a448"
       Referrer-Policy:
       - strict-origin-when-cross-origin
       Server:
@@ -141,9 +141,9 @@ interactions:
       X-Powered-By:
       - Phusion Passenger(R) 6.0.23
       X-Request-Id:
-      - 8020cc24-cdcc-47cb-b1f2-eca9139e604a
+      - 6cf945bb-0333-4dc9-858d-0bb91e56d28d
       X-Runtime:
-      - '0.028964'
+      - '0.029253'
       X-XSS-Protection:
       - '0'
     status: