Add support for dpid and IPFS for DeSci Labs Support

fuji_server/data/repodois.yaml

@@ -1,5 +1,4 @@
 arch.avanjj: https://doi.org/10.17616/R31NJMYW
-arch.igsn: https://doi.org/10.17616/R31NJMYW
 ardcx.aims: https://doi.org/10.17616/R3X908
 ardcx.curtin: https://doi.org/10.17616/R3WM02
 ardcx.griffith: https://doi.org/10.17616/R3FG92
@@ -58,6 +57,7 @@ ccdc.csd: https://doi.org/10.17616/R36011
 cern.hepdata: https://doi.org/10.17616/R30W2H
 cern.inspire: https://doi.org/10.17616/R3JC9Z
 cern.zenodo: https://doi.org/10.17616/R3QP53
+chvf.pbyfos: https://doi.org/10.17616/R31NJNLQ
 clarin.clarin: https://doi.org/10.17616/R3RP5D
 cngb.cga: https://doi.org/10.17616/R31NJMGL
 cngb.gigadb: https://doi.org/10.17616/R3TG83
@@ -70,7 +70,7 @@ cornell.ndacan: https://doi.org/10.17616/R3N90S
 cos.osf: https://doi.org/10.17616/R3N03T
 crui.ogsts: https://doi.org/10.17616/R3TR3Z
 crui.unibo: https://doi.org/10.17616/R3P19R
-crui.unipd: https://doi.org/10.17616/R31NJMHN
+crui.unipd: https://doi.org/10.17616/R3W05R
 csc.nrd: https://doi.org/10.17616/R33649
 csic.digital: https://doi.org/10.17616/R3P34J
 cul.ciesin: https://doi.org/10.17616/R31605
@@ -114,6 +114,7 @@ figshare.ariz: https://doi.org/10.17616/R31NJMSF
 figshare.ars: https://doi.org/10.17616/R3PK5R
 figshare.iastate: https://doi.org/10.17616/R3GV3W
 figshare.uct: https://doi.org/10.17616/R36R4R
+fzct.bexis: https://doi.org/10.17616/R32P9Q
 fzj.b2share: https://doi.org/10.17616/R3VK72
 fzj.tereno: https://doi.org/10.17616/R39G9T
 gbif.ec: https://doi.org/10.17616/R31NJNFS
@@ -181,6 +182,7 @@ lxkc.dskyfi: https://doi.org/10.17616/R3W331
 mdw.repository: https://doi.org/10.17616/R30M00
 mit.physio: https://doi.org/10.17616/R3D06S
 mlbs.skuxgs: https://doi.org/10.17616/R3ZP8D
+mzhu.ioerdata: https://doi.org/10.17616/R31NJNL7
 nasapds.nasapds: https://doi.org/10.17616/R37593
 nbqj.daks: https://doi.org/10.17616/R31NJMZS
 nkn.nkn: https://doi.org/10.17616/R3JK91
@@ -217,7 +219,6 @@ rpak.zvrzin: https://doi.org/10.17616/R31NJNLV
 sagebio.synapse: https://doi.org/10.17616/R3B934
 si.cda: https://doi.org/10.17616/R3201S
 si.si: https://doi.org/10.17616/R3W49N
-sml.tcia: https://doi.org/10.17616/R3NH0V
 sml.tdar: https://doi.org/10.17616/R3HK56
 snd.bolin: https://doi.org/10.17616/R3PP99
 snd.icos: https://doi.org/10.17616/R3235B

fuji_server/helper/identifier_helper.py

@@ -21,10 +21,12 @@ class IdentifierHelper:
         "bioproject": {"label": "BioProject ID", "source": "identifiers.org"},
         "biosample": {"label": "BioSample ID", "source": "identifiers.org"},
         "doi": {"label": "Digital Object Identifier (DOI)", "source": "datacite.org"},
+        "dpid": {"label": "Decentralized Persistent Identifier (dPID)", "source": "dpid.org"},
         "ensembl": {"label": "Ensembl ID", "source": "identifiers.org"},
         "genome": {"label": "GenBank or RefSeq genome", "source": "identifiers.org"},
         "gnd": {"label": "Gemeinsame Normdatei (GND) ID", "source": "f-uji.net"},
         "handle": {"label": "Handle System ID", "source": "datacite.org"},
+        "ipfs": {"label": "InterPlanetary File System Content Identifier (IPFS CID)", "source": "ipfs.io"},
         "lsid": {"label": "Life Science Identifier", "source": "datacite.org"},
         "pmid": {"label": "PubMed ID", "source": "datacite.org"},
         "pmcid": {"label": "PubMed Central ID", "source": "identifiers.org"},
@@ -36,6 +38,10 @@ class IdentifierHelper:
         "identifiers.org": {"label": "Identifiers.org Identifier", "source": "identifiers.org"},
         "w3id": {"label": "Permanent Identifier for the Web (W3ID)", "source": "identifiers.org"},
     }
+    # IPFS gateway domains for CID resolution (ordered by resolution reliability for DeSci content)
+    IPFS_GATEWAYS = ["ipfs.desci.com", "pub.desci.com", "dweb.link", "ipfs.io", "cloudflare-ipfs.com", "gateway.pinata.cloud"]
+    # dPID resolver domains
+    DPID_DOMAINS = ["dpid.org", "beta.dpid.org", "dev.dpid.org"]
     # identifiers.org pattern
     # TODO: check if this is needed.. if so ..complete and add check to FAIRcheck
     IDENTIFIERS_PIDS = r"https://identifiers.org/[provider_code/]namespace:accession"
@@ -125,6 +131,29 @@ def __init__(self, idstring, logger=None):
                         self.preferred_schema = "w3id"
                         self.identifier_url = self.identifier
                         self.normalized_id = self.identifier
+
+                    # dPID check - support dpid:// scheme and dpid.org URLs
+                    elif self.is_dpid():
+                        dpid_id = self.extract_dpid_id()
+                        if dpid_id:
+                            self.identifier_schemes = ["dpid", "url"]
+                            self.preferred_schema = "dpid"
+                            # Normalize to canonical dpid.org URL
+                            self.identifier_url = f"https://dpid.org/{dpid_id}"
+                            self.normalized_id = f"dpid://{dpid_id}"
+                            self.is_persistent = True
+
+                    # IPFS CID check - support ipfs:// scheme and IPFS gateway URLs
+                    elif self.is_ipfs_cid():
+                        cid = self.extract_ipfs_cid()
+                        if cid:
+                            self.identifier_schemes = ["ipfs", "url"]
+                            self.preferred_schema = "ipfs"
+                            # Use ipfs.io as the canonical gateway for resolution
+                            self.identifier_url = f"https://ipfs.io/ipfs/{cid}"
+                            self.normalized_id = f"ipfs://{cid}"
+                            self.is_persistent = True
+
                     # identifiers.org
 
                     elif idparts.netloc == "identifiers.org":
@@ -206,6 +235,138 @@ def is_hash(self):
         except Exception:
             return False
 
+    def is_dpid(self):
+        """Check if the identifier is a dPID (Decentralized Persistent Identifier).
+
+        Supports:
+        - dpid:// scheme (e.g., dpid://500)
+        - dpid.org URLs (e.g., https://dpid.org/500, https://beta.dpid.org/500)
+        """
+        if not self.identifier:
+            return False
+        try:
+            # Check for dpid:// scheme (e.g., dpid://500)
+            if self.identifier.startswith("dpid://"):
+                # Extract and validate the numeric ID
+                path = self.identifier[7:].split("/")[0]
+                return path.isdigit()
+
+            # Check for dpid.org URLs
+            idparts = urllib.parse.urlparse(self.identifier)
+            netloc = idparts.netloc.lower()
+            # Remove port if present for comparison
+            netloc_no_port = netloc.split(":")[0]
+
+            if netloc_no_port in self.DPID_DOMAINS or netloc_no_port.endswith(".dpid.org"):
+                # Check that there's a path with an ID
+                path = idparts.path.strip("/")
+                if path and (path.isdigit() or re.match(r"^\d+(/v\d+)?$", path)):
+                    return True
+            return False
+        except Exception:
+            return False
+
+    def extract_dpid_id(self):
+        """Extract the dPID number from a dPID URL or scheme.
+
+        Returns the dPID ID (e.g., "500" from "https://dpid.org/500" or "dpid://500")
+        """
+        if not self.identifier:
+            return None
+        try:
+            # Handle dpid:// scheme (e.g., dpid://500)
+            if self.identifier.startswith("dpid://"):
+                # Extract the first path component which should be the numeric ID
+                path = self.identifier[7:].split("/")[0]
+                return path if path.isdigit() else None
+
+            # Handle HTTP URLs
+            idparts = urllib.parse.urlparse(self.identifier)
+            path = idparts.path.strip("/")
+            # Extract numeric ID, handling version suffixes like /v1
+            match = re.match(r"^(\d+)(?:/v\d+)?$", path)
+            if match:
+                return match.group(1)
+            return path if path.isdigit() else None
+        except Exception:
+            return None
+
+    def is_ipfs_cid(self):
+        """Check if the identifier is an IPFS Content Identifier (CID).
+
+        Supports:
+        - ipfs:// scheme (e.g., ipfs://bafybeic...)
+        - IPFS gateway URLs (e.g., https://ipfs.io/ipfs/bafybeic...)
+        - Raw CIDv0 (Qm...) and CIDv1 (bafy...) identifiers
+        """
+        if not self.identifier:
+            return False
+        try:
+            # CIDv0 pattern: starts with 'Qm' followed by 44 base58 chars (total 46 chars)
+            cidv0_pattern = r"^Qm[1-9A-HJ-NP-Za-km-z]{44}$"
+            # CIDv1 pattern: starts with 'bafy' or 'bafk' followed by base32 chars
+            cidv1_pattern = r"^baf[yk][a-z2-7]{50,}$"
+
+            # Check for ipfs:// scheme
+            if self.identifier.startswith("ipfs://"):
+                cid = self.identifier[7:].split("/")[0]
+                return bool(re.match(cidv0_pattern, cid) or re.match(cidv1_pattern, cid))
+
+            # Check for IPFS gateway URLs
+            idparts = urllib.parse.urlparse(self.identifier)
+            netloc = idparts.netloc.lower()
+            path = idparts.path
+
+            # Check if it's an IPFS gateway URL
+            if any(gateway in netloc for gateway in self.IPFS_GATEWAYS) or "/ipfs/" in path:
+                # Extract CID from path
+                if "/ipfs/" in path:
+                    cid = path.split("/ipfs/")[1].split("/")[0]
+                    return bool(re.match(cidv0_pattern, cid) or re.match(cidv1_pattern, cid))
+
+            # Check if raw identifier is a CID
+            raw_id = self.identifier.strip()
+            return bool(re.match(cidv0_pattern, raw_id) or re.match(cidv1_pattern, raw_id))
+
+        except Exception:
+            return False
+
+    def extract_ipfs_cid(self):
+        """Extract the IPFS CID from an IPFS URL, scheme, or raw identifier.
+
+        Returns the CID (e.g., "bafybeic..." from "https://ipfs.io/ipfs/bafybeic...")
+        """
+        if not self.identifier:
+            return None
+        try:
+            # CIDv0 pattern
+            cidv0_pattern = r"(Qm[1-9A-HJ-NP-Za-km-z]{44})"
+            # CIDv1 pattern
+            cidv1_pattern = r"(baf[yk][a-z2-7]{50,})"
+
+            # Handle ipfs:// scheme
+            if self.identifier.startswith("ipfs://"):
+                cid = self.identifier[7:].split("/")[0]
+                return cid
+
+            # Handle IPFS gateway URLs
+            if "/ipfs/" in self.identifier:
+                parts = self.identifier.split("/ipfs/")
+                if len(parts) > 1:
+                    cid = parts[1].split("/")[0]
+                    return cid
+
+            # Check if raw identifier is a CID
+            raw_id = self.identifier.strip()
+            if re.match(cidv0_pattern, raw_id):
+                return re.match(cidv0_pattern, raw_id).group(1)
+            if re.match(cidv1_pattern, raw_id):
+                return re.match(cidv1_pattern, raw_id).group(1)
+
+            return None
+        except Exception:
+            return None
+
     def verify_handle(self, val, includeparams=True):
         # additional checks for handles since the syntax is very generic
         try: