Check PREMIS validity before preservation

[skip-codecov]

.dockerignore

@@ -5,3 +5,4 @@
 !go.mod
 !go.sum
 !main.go
+!hack/xsd/premis.xsd

Dockerfile

@@ -3,6 +3,9 @@
 ARG TARGET=enduro
 ARG GO_VERSION
 
+FROM alpine:3.20 AS build-libxml
+RUN apk add --no-cache libxml2-utils
+
 FROM golang:${GO_VERSION}-alpine AS build-go
 WORKDIR /src
 ENV CGO_ENABLED=0
@@ -61,10 +64,22 @@ FROM base AS enduro-a3m-worker
 COPY --from=build-enduro-a3m-worker --link /out/enduro-a3m-worker /home/enduro/bin/enduro-a3m-worker
 COPY --from=build-enduro-a3m-worker --link /src/enduro.toml /home/enduro/.config/enduro.toml
 CMD ["/home/enduro/bin/enduro-a3m-worker", "--config", "/home/enduro/.config/enduro.toml"]
+COPY hack/xsd/premis.xsd /home/enduro/premis.xsd
+COPY --from=build-libxml /usr/bin/xmllint /usr/bin/xmllint
+COPY --from=build-libxml /usr/lib/libxml2.so.2 /usr/lib/libxml2.so.2
+COPY --from=build-libxml /lib/ld-musl-x86_64.so.1 /lib/ld-musl-x86_64.so.1
+COPY --from=build-libxml /lib/libz.so.1 /lib/libz.so.1
+COPY --from=build-libxml /usr/lib/liblzma.so.5 /usr/lib/liblzma.so.5
 
 FROM base AS enduro-am-worker
 COPY --from=build-enduro-am-worker --link /out/enduro-am-worker /home/enduro/bin/enduro-am-worker
 COPY --from=build-enduro-am-worker --link /src/enduro.toml /home/enduro/.config/enduro.toml
 CMD ["/home/enduro/bin/enduro-am-worker", "--config", "/home/enduro/.config/enduro.toml"]
+COPY hack/xsd/premis.xsd /home/enduro/premis.xsd
+COPY --from=build-libxml /usr/bin/xmllint /usr/bin/xmllint
+COPY --from=build-libxml /usr/lib/libxml2.so.2 /usr/lib/libxml2.so.2
+COPY --from=build-libxml /lib/ld-musl-x86_64.so.1 /lib/ld-musl-x86_64.so.1
+COPY --from=build-libxml /lib/libz.so.1 /lib/libz.so.1
+COPY --from=build-libxml /usr/lib/liblzma.so.5 /usr/lib/liblzma.so.5
 
 FROM ${TARGET}

cmd/enduro-a3m-worker/main.go

@@ -19,6 +19,7 @@ import (
 	"github.com/artefactual-sdps/temporal-activities/bagvalidate"
 	"github.com/artefactual-sdps/temporal-activities/bucketupload"
 	"github.com/artefactual-sdps/temporal-activities/removepaths"
+	"github.com/artefactual-sdps/temporal-activities/xmlvalidate"
 	"github.com/hashicorp/go-cleanhttp"
 	"github.com/oklog/run"
 	"github.com/prometheus/client_golang/prometheus/promhttp"
@@ -252,6 +253,10 @@ func main() {
 			archiveextract.New(cfg.ExtractActivity).Execute,
 			temporalsdk_activity.RegisterOptions{Name: archiveextract.Name},
 		)
+		w.RegisterActivityWithOptions(
+			xmlvalidate.New(xmlvalidate.NewXMLLintValidator()).Execute,
+			temporalsdk_activity.RegisterOptions{Name: xmlvalidate.Name},
+		)
 		w.RegisterActivityWithOptions(
 			activities.NewClassifyPackageActivity().Execute,
 			temporalsdk_activity.RegisterOptions{Name: activities.ClassifyPackageActivityName},

cmd/enduro-am-worker/main.go

@@ -20,6 +20,7 @@ import (
 	"github.com/artefactual-sdps/temporal-activities/bagvalidate"
 	"github.com/artefactual-sdps/temporal-activities/bucketupload"
 	"github.com/artefactual-sdps/temporal-activities/removepaths"
+	"github.com/artefactual-sdps/temporal-activities/xmlvalidate"
 	"github.com/hashicorp/go-cleanhttp"
 	"github.com/jonboulle/clockwork"
 	"github.com/oklog/run"
@@ -329,6 +330,10 @@ func main() {
 			bucketupload.New(failedPIPs).Execute,
 			temporalsdk_activity.RegisterOptions{Name: activities.SendToFailedPIPsName},
 		)
+		w.RegisterActivityWithOptions(
+			xmlvalidate.New(xmlvalidate.NewXMLLintValidator()).Execute,
+			temporalsdk_activity.RegisterOptions{Name: xmlvalidate.Name},
+		)
 
 		g.Add(
 			func() error {

enduro.toml

@@ -232,6 +232,10 @@ namespace = "default"
 taskQueue = "preprocessing"
 workflowName = "preprocessing"
 
+[validatePremis]
+enabled = true
+xsdPath = "/home/enduro/premis.xsd"
+
 [failedSips]
 endpoint = "http://minio.enduro-sdps:9000"
 pathStyle = true

go.mod

@@ -12,7 +12,7 @@ require (
 	github.com/XSAM/otelsql v0.29.0
 	github.com/alicebob/miniredis/v2 v2.32.1
 	github.com/artefactual-labs/bagit-gython v0.2.0
-	github.com/artefactual-sdps/temporal-activities v0.0.0-20240821162351-47302711bc7b
+	github.com/artefactual-sdps/temporal-activities v0.0.0-20241018212855-8ea34d29bdf4
 	github.com/coreos/go-oidc/v3 v3.10.0
 	github.com/cyphar/filepath-securejoin v0.2.4
 	github.com/dolmen-go/contextio v1.0.0

go.sum

@@ -442,8 +442,8 @@ github.com/apparentlymart/go-textseg/v15 v15.0.0 h1:uYvfpb3DyLSCGWnctWKGj857c6ew
 github.com/apparentlymart/go-textseg/v15 v15.0.0/go.mod h1:K8XmNZdhEBkdlyDdvbmmsvpAG721bKi0joRfFdHIWJ4=
 github.com/artefactual-labs/bagit-gython v0.2.0 h1:Zje4Lb1goZVUPoxpc/k65sWtYpNgK9Rvphvaok5cYzE=
 github.com/artefactual-labs/bagit-gython v0.2.0/go.mod h1:C+hFZQMDnji1hjGt3nrlMK3BahaBhvo/hU2uqd+Q9Z4=
-github.com/artefactual-sdps/temporal-activities v0.0.0-20240821162351-47302711bc7b h1:kTOc2pbkdII6/Z84Bus1q52z5KAOaT8vLpfRoOs1l1I=
-github.com/artefactual-sdps/temporal-activities v0.0.0-20240821162351-47302711bc7b/go.mod h1:FVh79rCGNlUU1QnioAU+lrSjLqrA1PJFYKIhWPsmyug=
+github.com/artefactual-sdps/temporal-activities v0.0.0-20241018212855-8ea34d29bdf4 h1:WF95IOkZRVSCST/26SAqPYsUrtUuJpavBht6lvdeKl0=
+github.com/artefactual-sdps/temporal-activities v0.0.0-20241018212855-8ea34d29bdf4/go.mod h1:FVh79rCGNlUU1QnioAU+lrSjLqrA1PJFYKIhWPsmyug=
 github.com/aws/aws-sdk-go v1.55.5 h1:KKUZBfBoyqy5d3swXyiC7Q76ic40rYcbqH7qjh59kzU=
 github.com/aws/aws-sdk-go v1.55.5/go.mod h1:eRwEWoyTWFMVYVQzKMNHWP5/RV4xIUGMQfXQHfHkpNU=
 github.com/aws/aws-sdk-go-v2 v1.30.3 h1:jUeBtG0Ih+ZIFH0F4UkmL9w3cSpaMv9tYYDbzILP8dY=

hack/kube/base/enduro.yaml

@@ -70,6 +70,8 @@ spec:
               value: "grafana-alloy.enduro-sdps:4317"
             - name: ENDURO_TELEMETRY_TRACES_SAMPLING_RATIO
               value: "1.0"
+            - name: ENDURO_VALIDATEPREMIS_ENABLED
+              value: "false"
           ports:
             - containerPort: 9000
             - containerPort: 9002