From 9c88bc4278725ca9faae37a7026c023374b4c96b Mon Sep 17 00:00:00 2001
From: Charles Tapley Hoyt <cthoyt@gmail.com>
Date: Mon, 30 Oct 2023 16:26:45 +0100
Subject: [PATCH] Add PubMed and PMC contact and update PMC information (#968)

Closes #965
Closes #967
Closes #966
---
 src/bioregistry/data/bioregistry.json         | 40 ++++++++++++++-----
 .../data/external/miriam/processed.json       |  9 -----
 src/bioregistry/external/miriam.py            |  1 +
 tests/test_data.py                            |  6 ++-
 4 files changed, 37 insertions(+), 19 deletions(-)

diff --git a/src/bioregistry/data/bioregistry.json b/src/bioregistry/data/bioregistry.json
index 89fd9f241..a8b9fcc6a 100644
--- a/src/bioregistry/data/bioregistry.json
+++ b/src/bioregistry/data/bioregistry.json
@@ -79079,6 +79079,18 @@
       "prefix": "PMC",
       "uri_format": "http://europepmc.org/articles/PMC$1"
     },
+    "comment": "Note that the Europe PMC instance redirects canonical local unique identifiers from https://europepmc.org/articles/PMC3084216 to https://europepmc.org/article/PMC/3084216. However, this is an implementation detail and does not reflect the official format of a PMC identifier, which indeed does include `PMC` at the beginning of the local unique identifier",
+    "contact": {
+      "email": "beck@ncbi.nlm.nih.gov",
+      "github": "jeffbeckncbi",
+      "name": "Jeff Beck",
+      "orcid": "0000-0002-1798-9797"
+    },
+    "example_extras": [
+      "PMC7418728",
+      "PMC7418728.1",
+      "PMC7418728.2"
+    ],
     "fairsharing": {
       "abbreviation": "PMC",
       "description": "PubMed Central (PMC) is the U.S. National Institutes of Health (NIH) digital archive of biomedical and life sciences journal literature.",
@@ -79135,15 +79147,6 @@
       "namespaceEmbeddedInLui": false,
       "pattern": "PMC\\d+",
       "prefix": "pmc",
-      "providers": [
-        {
-          "code": "ncbi",
-          "description": "PubMed Central",
-          "homepage": "https://www.ncbi.nlm.nih.gov/pmc/",
-          "name": "PubMed Central",
-          "uri_format": "https://www.ncbi.nlm.nih.gov/pmc/articles/$1/?tool=pubmed"
-        }
-      ],
       "sampleId": "PMC3084216",
       "uri_format": "http://europepmc.org/articles/$1"
     },
@@ -79158,6 +79161,7 @@
       "uri_format": "http://europepmc.org/articles/$1"
     },
     "name": "PubMed Central",
+    "pattern": "^PMC\\d+(\\.\\d+)?$",
     "prefixcommons": {
       "description": "PubMed Central (PMC) is the U.S. National Institutes of Health (NIH) digital archive of biomedical and life sciences journal literature.",
       "example": "PMC3084216",
@@ -79173,6 +79177,15 @@
       ],
       "uri_format": "http://www.ncbi.nlm.nih.gov/pmc/articles/$1/?tool=pmcentrez"
     },
+    "providers": [
+      {
+        "code": "ncbi.pmc",
+        "description": "PubMed Central",
+        "homepage": "https://www.ncbi.nlm.nih.gov/pmc/",
+        "name": "PubMed Central",
+        "uri_format": "https://www.ncbi.nlm.nih.gov/pmc/articles/$1"
+      }
+    ],
     "publications": [
       {
         "doi": "10.1093/nar/gkr1184",
@@ -79189,6 +79202,9 @@
         "year": 2010
       }
     ],
+    "references": [
+      "https://github.com/biopragmatics/bioregistry/issues/965"
+    ],
     "synonyms": [
       "PMC",
       "PMCID"
@@ -82914,6 +82930,12 @@
       "obsolete": false,
       "prefix": "000302"
     },
+    "contact": {
+      "email": "beck@ncbi.nlm.nih.gov",
+      "github": "jeffbeckncbi",
+      "name": "Jeff Beck",
+      "orcid": "0000-0002-1798-9797"
+    },
     "edam": {
       "description": "PubMed unique identifier of an article.",
       "name": "PubMed",
diff --git a/src/bioregistry/data/external/miriam/processed.json b/src/bioregistry/data/external/miriam/processed.json
index c17b60c01..e887730e8 100644
--- a/src/bioregistry/data/external/miriam/processed.json
+++ b/src/bioregistry/data/external/miriam/processed.json
@@ -7967,15 +7967,6 @@
     "namespaceEmbeddedInLui": false,
     "pattern": "PMC\\d+",
     "prefix": "pmc",
-    "providers": [
-      {
-        "code": "ncbi",
-        "description": "PubMed Central",
-        "homepage": "https://www.ncbi.nlm.nih.gov/pmc/",
-        "name": "PubMed Central",
-        "uri_format": "https://www.ncbi.nlm.nih.gov/pmc/articles/$1/?tool=pubmed"
-      }
-    ],
     "sampleId": "PMC3084216",
     "uri_format": "http://europepmc.org/articles/$1"
   },
diff --git a/src/bioregistry/external/miriam.py b/src/bioregistry/external/miriam.py
index acfbfdcd5..d03861b97 100644
--- a/src/bioregistry/external/miriam.py
+++ b/src/bioregistry/external/miriam.py
@@ -60,6 +60,7 @@ def get_miriam(force_download: bool = False, force_process: bool = False):
     ("ega.study", "omicsdi"),
     # see discussion at https://github.com/biopragmatics/bioregistry/pull/944
     ("bioproject", "ebi"),
+    ("pmc", "ncbi"),
 }
 
 
diff --git a/tests/test_data.py b/tests/test_data.py
index b6243fa68..5d472c131 100644
--- a/tests/test_data.py
+++ b/tests/test_data.py
@@ -681,7 +681,11 @@ def test_providers(self):
             for provider in resource.providers:
                 with self.subTest(prefix=prefix, code=provider.code):
                     self.assertNotEqual(provider.code, prefix)
-                    self.assertNotIn(provider.code, self.metaregistry)
+                    self.assertNotIn(
+                        provider.code,
+                        set(self.metaregistry),
+                        msg="Provider code is duplicate of metaregistry prefix.",
+                    )
                     self.assertNotIn(provider.code, {"custom", "default"})
                     self.assertEqual(
                         provider.code.lower(),