PDF.js fix, rate-limiting, max page size, record deletion fixes

invenio.cfg

@@ -435,8 +435,7 @@ RDM_REQUEST_RECORD_DELETION_CHECKLIST = [
     {
         "label": _("I want to get a Zenodo DOI for this record"),
         "message": _(
-            '<a href="/support">Contact the Zenodo support team</a> '
-            "about this record."
+            '<a href="/support">Contact the Zenodo support team</a> about this record.'
         ),
     },
 ]
@@ -1001,8 +1000,10 @@ RATELIMIT_PER_ENDPOINT = {
     "invenio_app_rdm.frontpage_view_function": "10 per second",
     # Records
     "invenio_app_rdm_records.record_thumbnail": "20 per second",
+    "records.search": "30 per minute",
+    "community-records.search": "30 per minute",
     # OAI-PMH
-    "invenio_oaiserver.response": "120 per minute",
+    "invenio_oaiserver.response": "30 per minute",
     # 'zenodo_redirector.contact': '10 per second',
     # TODO most similar view found was: invenio_redirector.redirect_contact
 }
@@ -1013,7 +1014,7 @@ RATELIMIT_GUEST_USER = "8000 per hour;133 per minute"
 """Rate limits for guest users."""
 
 OAISERVER_RESUMPTION_TOKEN_EXPIRE_TIME = 1 * 120
-OAISERVER_PAGE_SIZE = 100
+OAISERVER_PAGE_SIZE = 50
 OAISERVER_XSL_URL = "/static/oai2.xsl"
 
 RDM_DATACITE_FUNDER_IDENTIFIERS_PRIORITY = ("doi", "ror", "grid", "isni", "gnd")
@@ -1083,6 +1084,17 @@ PREVIEWER_PREFERENCE = [
 ]
 """Preferred previewers."""
 
+PREVIEWER_PDF_JS_DOCUMENT_INIT_PARAMS = {
+    "disableStream": True,
+    "disableRange": True,
+    "disableAutoFetch": True,
+}
+"""PDF.js initialization parameters for document previewer.
+
+We disable range requests and streaming, to avoid overloading the server with
+multiple requests per PDF preview.
+"""
+
 PREVIEWER_MAX_IMAGE_SIZE_BYTES = 15 * (10**6)  # 15 MB
 """Maximum file size in bytes for image files."""
 

pyproject.toml

@@ -1,13 +1,13 @@
 [project]
 name = "zenodo-rdm-app"
-version = "21.1.6"
+version = "21.1.7"
 authors = [
     { name = "CERN" }
 ]
 license = "GPL-3.0"
 requires-python = ">=3.9"
 dependencies = [
-    "invenio-app-rdm[opensearch2]~=14.0.0b1.dev2",
+    "invenio-app-rdm[opensearch2]==14.0.0b1.dev6",
     "zenodo-legacy",
     "zenodo-rdm",
     # "commonmeta-py==0.8.6",
@@ -31,7 +31,6 @@ zenodo-legacy = { workspace = true }
 invenio-swh = { git = "https://github.com/inveniosoftware/invenio-swh", rev = "v0.13.4" }
 # User profile fixes
 invenio-records-resources = { git = "https://github.com/inveniosoftware/invenio-records-resources", branch = "fix-read-many" }
-invenio-previewer = { git = "https://github.com/inveniosoftware/invenio-previewer", branch = "pdfjs-disable-range-stream" }
 
 [tool.uv.workspace]
 members = [

site/zenodo_rdm/params.py

@@ -9,6 +9,8 @@
 
 from functools import partial
 
+from flask import request
+from flask_login import current_user
 from invenio_rdm_records.resources.config import RDMSearchRequestArgsSchema
 from invenio_rdm_records.services.config import RDMSearchOptions
 from invenio_records_resources.services.records.params.base import ParamInterpreter
@@ -49,7 +51,7 @@ def apply(self, identity, search, params):
             try:
                 community = Community.pid.resolve(slug)  # Ensure community's existence
                 community_ids.append(community.id)
-            except:
+            except Exception:
                 # we show only results from communities that resolve
                 pass
 
@@ -99,6 +101,13 @@ class ZenodoArgsSchema(RDMSearchRequestArgsSchema):
     type = fields.String()
     subtype = fields.String()
 
+    @property
+    def max_page_size(self):
+        """Max page size depending on user authentication."""
+        if request and current_user.is_authenticated:
+            return 200
+        return 100
+
     @pre_load
     def load_all_versions(self, data, **kwargs):
         """Load legacy `all_versions` param and feed the new `allversions` one."""