Test converting the UVA Library Beeldbank MARC XML data to simple RDF

beeldbank_marc_to_rdf.py

from xml.etree import ElementTree as ET
import marcalyx
import pyoxigraph as px
import os
from rich.progress import track

MARCXML_FILENAME = os.environ.get("MARCXML_FILENAME", "uva_alma_beeldbank_marc_new.xml")


def main():
    print("Parsing", MARCXML_FILENAME)
    doc = ET.parse(MARCXML_FILENAME).getroot()
    coll = marcalyx.Collection(doc)
    G = px.Store()

    def tu(uid, field, val):
        if not val:
            return
        if type(val) == px.BlankNode:
            G.add(px.Quad(uid, px.NamedNode(field), val))
        else:
            G.add(px.Quad(uid, px.NamedNode(field), px.NamedNode(val)))

    def tl(uid, field, val, language="nl", datatype=None):
        if not val:
            return
        val = val.replace("<p>", "").replace("</p>", "\n")
        if datatype:
            G.add(
                px.Quad(
                    (
                        uid,
                        px.NamedNode(field),
                        px.Literal(val, datatype=px.NamedNode(datatype)),
                    )
                )
            )
        else:
            G.add(
                px.Quad(
                    uid,
                    px.NamedNode(field),
                    px.Literal(val, language=language),
                )
            )

    places = {}

    for marc in track(coll.records()):
        mid = marc["001"][0].value
        uid = px.NamedNode(marc["024"][0]["a"][0].value)
        tu(
            uid,
            "http://www.w3.org/1999/02/22-rdf-syntax-ns#type",
            "http://www.cidoc-crm.org/cidoc-crm/E22_Human-Made_Object",
        )
        tu(
            uid,
            "http://www.w3.org/ns/prov#wasDerivedFrom",
            f"https://lib.uva.nl/permalink/31UKB_UAM1_INST/2nbb0e/alma{mid}",
        )
        tl(
            uid,
            "http://www.w3.org/2000/01/rdf-schema#label",
            "\n".join(x.value for x in marc[("245", "a")])
            + " "
            + "\n".join(x.value for x in marc[("245", "b")])
            + " "
            + "\n".join(x.value for x in marc[("245", "c")]),
        )
        tl(uid, "http://purl.org/dc/terms/identifier", mid, language=None)

        for code in ("500", "520"):
            tl(
                uid,
                "http://purl.org/dc/terms/description",
                "\n".join(x.value for x in marc[(code, "a")]),
            )

        for code in ("510", "581"):
            for x in marc[(code, "a")]:
                tl(uid, "http://purl.org/dc/terms/references", x.value)

        for x in marc[("600", "0")]:
            if x.value.startswith("http"):
                tu(uid, "http://purl.org/dc/terms/creator", x.value)
            else:
                tl(uid, "http://purl.org/dc/terms/creator", x.value, language=None)
        for code in ("650", "651", "655"):
            for x in marc[(code, "0")]:
                if x.value.startswith("http"):
                    tu(uid, "http://purl.org/dc/terms/subject", x.value)
                else:
                    tl(uid, "http://purl.org/dc/terms/subject", x.value, language=None)

        # Add the place-of-origin in 694f with a blank node and a label
        for x in marc[("694", "f")]:
            place = places.get(x.value)
            if not place:
                place = px.BlankNode()
                tl(place, "http://www.w3.org/2000/01/rdf-schema#label", x.value)
                places[x.value] = place
            tu(uid, "http://purl.org/dc/terms/spatial", place)

        # Note, is very interesting to use 100a 100d and 100e as combo to see how person is referenced and related see: https://hdl.handle.net/11245/3.3707
    G.dump(open("ap.nt", "wb"), "application/n-triples")


if __name__ == "__main__":
    main()