Commit bdae12c3 authored by Mickaël Desfrênes's avatar Mickaël Desfrênes
Browse files

OAI-PMH setSpec support

parent 729e38a7

oai/apps.py

+2 −2
Original line number Diff line number Diff line
@@ -2,5 +2,5 @@ from django.apps import AppConfig 




class OaiConfig(AppConfig):

    default_auto_field = 'django.db.models.BigAutoField'

    name = 'oai'
 
    default_auto_field = "django.db.models.BigAutoField"

    name = "oai"

oai/views.py

+59 −9
Original line number Diff line number Diff line
@@ -14,6 +14,19 @@ from django.core.paginator import Paginator 
PAGINATION_SIZE = 100





def _set_spec_to_collection(set_spec: str) -> Collection:

    try:

        last_id = int(set_spec.split(":")[-1])

        collection = Collection.objects.get(pk=last_id)

        return collection

    except (ValueError, Collection.DoesNotExist):

        raise ValueError("no such collection")





def _collection_to_set_spec(collection: Collection) -> str:

    return collection.oai_setspec()





def _validate_date(in_date: str) -> bool:

    try:

        res = bool(datetime.strptime(in_date, "%Y-%m-%d"))
@@ -74,7 +87,9 @@ def _get_record(owner: User, request: HttpRequest) -> HttpResponse: 


def _identify(owner: User, request: HttpRequest) -> HttpResponse:

    earliest_record = (

        Collection.objects.filter(is_oai_record=True).order_by("created_at").first()

        Collection.objects.filter(is_oai_record=True, owner=owner)

        .order_by("created_at")

        .first()

    )

    return HttpResponse(

        render_to_string(
@@ -113,27 +128,56 @@ def _list_records( 
    resumption_token = request.GET.get("resumptionToken") or request.POST.get(

        "resumptionToken"

    )



    #

    # Get page number (ie. resumption token)

    #

    if resumption_token:

        try:

            page_number = abs(int(resumption_token))

        except ValueError:

            pass

    from_date = request.GET.get("from") or request.POST.get("from")

    if not from_date:

        earliest_record = (

            Collection.objects.filter(is_oai_record=True).order_by("created_at").first()

        )

        from_date = earliest_record.created_at.strftime("%Y-%m-%d")

    if not _validate_date(from_date):



    #

    # Get from date

    #

    from_date = request.GET.get("from") or request.POST.get("until")

    if from_date and not _validate_date(from_date):

        return _oai_error(owner, request, {"badArgument": "bad from date format"})



    #

    # Get until date

    #

    until_date = request.GET.get("until") or request.POST.get("until")

    if until_date and not _validate_date(until_date):

        return _oai_error(owner, request, {"badArgument": "bad until date format"})



    #

    # Get setSpec collection

    #

    spec_collection = None

    set_spec = request.GET.get("set") or request.POST.get("set")

    if set_spec:

        spec_collection = _set_spec_to_collection(set_spec=set_spec)



    #

    # Build query

    #

    all_records = Collection.objects.filter(

        is_oai_record=True, created_at__gte=from_date

        is_oai_record=True, owner=owner, public_access=True

    ).order_by("created_at")

    if from_date:

        all_records = all_records.filter(created_at__gte=from_date)

    if until_date:

        all_records = all_records.filter(created_at__lte=until_date)

    if spec_collection:

        all_records = all_records.filter(

            parent__in=spec_collection.descendants_and_self_ids()

        )



    #

    # Paginate, render.

    #

    total_count = all_records.count()

    if total_count == 0:

        return _oai_error(owner, request, {"noRecordsMatch": "no records match"})
@@ -156,6 +200,7 @@ def _list_records( 
                "total_pages": paginator.num_pages,

                "nb_of_already_delivered_identifiers": (page_number - 1)

                * PAGINATION_SIZE,

                "spec_collection": spec_collection,

            },

        ),

        content_type="text/xml",
@@ -163,12 +208,16 @@ def _list_records( 




def _list_sets(owner: User, request: HttpRequest) -> HttpResponse:

    collections = Collection.objects.filter(

        owner=owner, is_oai_record=False, public_access=True

    )

    return HttpResponse(

        render_to_string(

            "oai/list_sets.xml",

            {

                "owner": owner,

                "base_url": settings.JAMA_SITE,

                "collections": collections,

            },

        ),

        content_type="text/xml",
@@ -201,5 +250,6 @@ def oai(request: HttpRequest, owner_id: int) -> HttpResponse: 
            return _list_records(owner, request)

        if oai_verb == "ListSets":

            return _list_sets(owner, request)

    # each user/owner has his own endpoint

    except User.DoesNotExist:

        raise Http404()

resources/models.py

+18 −0
Original line number Diff line number Diff line
@@ -335,6 +335,12 @@ class Collection(models.Model): 
        """

        return Resource.objects.raw(sql, [self.id])



    def descendants_and_self_ids(self) -> List[int]:

        ids = [self.pk]

        for descendant in self.descendants():

            ids.append(descendant.pk)

        return ids



    def descendants_count(self) -> int:

        with connection.cursor() as cursor:

            cursor.execute(
@@ -448,6 +454,18 @@ class Collection(models.Model): 
        except MetadataSet.DoesNotExist:

            return []



    def oai_setspec(self) -> str:

        """

        Get setSpec value for OAI-PMH xml.

        (see https://www.openarchives.org/OAI/openarchivesprotocol.html#Set)

        """

        ids = []

        ancestors = self.ancestors()

        for ancestor in ancestors[1:]:

            ids.append(str(ancestor.pk))

        ids.append(str(self.pk))

        return ":".join(ids)





class MetadataCollectionValue(models.Model):

    metadata = models.ForeignKey(Metadata, on_delete=models.CASCADE)

templates/oai/list_identifiers.xml

+6 −8
Original line number Diff line number Diff line
@@ -5,11 +5,9 @@ 
         http://www.openarchives.org/OAI/2.0/OAI-PMH.xsd">

    <responseDate>{% now "c" %}</responseDate>

    <request verb="ListIdentifiers"

             from="{{ from }}"

             {% if until %}

             until="{{ until }}"

             {% endif %}

             set="physics:hep"

             {% if from %}from="{{ from }}"{% endif %}

             {% if until %}until="{{ until }}"{% endif %}

             {% if spec_collection %}set="{{ spec_collection.oai_setspec }}"{% endif %}

             metadataPrefix="oai_dc">

        {{ base_url }}oai/{{ owner.pk }}

    </request>
@@ -19,9 +17,9 @@ 
                <header>

                    <identifier>{{ collection.pk }}</identifier>

                    <datestamp>{{ collection.created_at | date:"Y-m-d" }}</datestamp>

                    {% for ancestor in collection.ancestors|slice:"1:" %}

                        <setSpec>{{ ancestor.title }}</setSpec>

                    {% endfor %}

                    {% if collection.parent %}

                    <setSpec>{{ collection.parent.oai_setspec }}</setSpec>

                    {% endif %}

                </header>

            </record>

        {% endfor %}

templates/oai/list_records.xml

+8 −4
Original line number Diff line number Diff line
@@ -5,11 +5,15 @@ 
         http://www.openarchives.org/OAI/2.0/OAI-PMH.xsd">

    <responseDate>{% now "c" %}</responseDate>

    <request verb="ListRecords"

             {% if from %}

             from="{{ from }}"

             {% endif %}

             {% if until %}

             until="{{ until }}"

             {% endif %}

             set="physics:hep"

             {% if spec_collection %}

             set="{{ spec_collection.oai_setspec }}"

             {% endif %}

             metadataPrefix="oai_dc">

        {{ base_url }}oai/{{ owner.pk }}

    </request>
@@ -19,9 +23,9 @@ 
                <header>

                    <identifier>{{ collection.pk }}</identifier>

                    <datestamp>{{ collection.created_at | date:"Y-m-d" }}</datestamp>

                    {% for ancestor in collection.ancestors|slice:"1:" %}

                        <setSpec>{{ ancestor.title }}</setSpec>

                    {% endfor %}

                    {% if collection.parent %}

                    <setSpec>{{ collection.parent.oai_setspec }}</setSpec>

                    {% endif %}

                </header>

                <metadata>

                    <oai_dc:dc