Add recordhep.py

modules/store_tools/recordhep.py

+"""recordhep.py
+
+"""
+import pprint
+
+
+class RecordHep(dict):
+    """Base class for JSON record coming from inspirehep.net version v2.
+
+    Schema documentation is defined here:
+    https://inspire-schemas.readthedocs.io/en/latest/schemas/
+
+    Args:
+        recjson (dict):
+            meta data from the JSON record returns by the store
+
+    """
+
+    def __init__(self, recjson):
+
+        super().__init__(recjson)
+
+        # meta data
+        # the authors of my institutes signing the record
+        # string containing a list of name separated by a comma
+        self.my_authors = ""
+
+    @staticmethod
+    def _oai_url(value):
+        """Build the Open Archive Initiative URL.
+
+        Args:
+            value (str):
+                OAI identifier, *e.g.* ``oai:host:id``
+
+        Returns:
+            str:
+                the pattern of the string is `https://host/api/literature/id`
+                The string is empty when it is not defined or when the value
+                is not well formed.
+
+        """
+
+    def debug(self):
+        """Print the record structure on the standard output.
+
+        """
+        pprint.pprint(self)
+
+    def host(self):
+        """The store housing the record.
+
+        Returns:
+            str:
+
+        """
+        return "inspirehep.net"
+
+    def id(self):
+        """The id of the record in the store.
+
+        Returns:
+            int:
+
+        """
+        return self["control_number"]
+
+    def oai(self):
+        """The Open Archive Initiative identifier(s).
+
+        Returns:
+            str:
+                the primary and secondary OAI identifier are separated
+                by a comma. The pattern of the identifier is ``oai:host:id`` or
+                an empty string when it is not defined.
+
+        """
+        lst = [self.primary_oai(), self.secondary_oai()]
+        return ", ".join(lst).strip(", ")
+
+    def oai_url(self):
+        """The Open Archive Initiative identifier URL(s).
+
+        Returns:
+            str:
+                the primary and secondary URLs are separated by a comma.
+                The pattern of the URL is ``https://host/api/literature/id``
+                or an empty string when it is not defined or when the OAI is
+                not well formed.
+
+        """
+        lst = [self.primary_oai_url(), self.secondary_oai_url()]
+        return ", ".join(lst).strip(", ")
+
+    def primary_oai(self):
+        """The primary Open Archive Initiative identifier.
+
+        The primary OAI identifier matches the record identifier.
+
+        Returns:
+            str:
+                the pattern of the string is ``oai:host:id``.
+                It is an empty string when not defined
+
+        """
+        return f"oai:inspirehep.net:{self['control_number']}"
+
+    def primary_oai_url(self):
+        """The Open Archive Initiative URL for the primary OAI.
+
+        Returns:
+            str:
+                the pattern of the string is ``https://host/api/literature/id``.
+                The string is empty when it is not defined or when the OAI
+                is not well formed.
+
+        """
+        recid = self["control_number"]
+        return f"https://inspirehep.net/api/literature/{recid}"
+
+    def secondary_oai(self):
+        """The secondary OAI identifier.
+
+        the secondary OAI identifier corresponds to the record in the
+        store, *cds.cern.ch*.
+
+        Returns:
+            str:
+                the pattern of the string is ``oai:host:id``.
+                It is an empty string when not defined
+
+        """
+        if "external_system_identifiers" not in self:
+            return ""
+
+        for elt in self["external_system_identifiers"]:
+            if elt["schema"] == "CDS":
+                return f"oai:cds.cern.ch:{elt['value']}"
+
+        return ""
+
+    def secondary_oai_url(self):
+        """The Open Archive Initiative URL for the secondary OAI.
+
+        the secondary OAI URL corresponds to the record in the
+        store, *cds.cern.ch*.
+
+        Returns:
+            str:
+                the pattern of the string is ``https://host/record/id``.
+                The string is empty when it is not defined or when the OAI
+                is not well formed.
+
+        """
+        if "external_system_identifiers" not in self:
+            return ""
+
+        for elt in self["external_system_identifiers"]:
+            if elt["schema"] == "CDS":
+                return f"https://cds.cern.ch/record/{elt['value']}"
+
+        return ""