Merge pull request #203 from nulib/4546-pdf

Auxiliary & Access file download

docs/docs/spec/data-types.yaml

@@ -132,6 +132,9 @@ components:
         description:
           type: string
           nullable: true
+        download_url:
+          type: string
+          nullable: true
         duration:
           type: number
           nullable: true
@@ -172,6 +175,7 @@ components:
         - id
         - accession_number
         - description
+        - download_url
         - duration
         - height
         - label
@@ -497,6 +501,9 @@ components:
         embedding_model:
           type: string
           description: The name of the inference model used to generate the `embedding` from the resource's content.
+        download_url:
+          type: string
+          nullable: true
         file_sets:
           type: array
           description: File sets associated with the resource.
@@ -759,6 +766,7 @@ components:
         - description
         - embedding
         - embedding_model
+        - download_url
         - file_sets
         - folder_name
         - folder_number

node/src/api/response/iiif/manifest.js

@@ -5,7 +5,9 @@ const {
   buildAnnotationBody,
   buildImageResourceId,
   buildImageService,
+  isAltFormat,
   isAudioVideo,
+  isPDF,
   buildSupplementingAnnotation,
 } = require("./presentation-api/items");
 const { metadataLabelFields } = require("./presentation-api/metadata");
@@ -108,6 +110,24 @@ function transform(response) {
           },
         });
 
+        /** Add rendering */
+        let renderings = [];
+        source.file_sets
+          .filter((fileSet) => fileSet.role === "Auxiliary")
+          .filter((fileSet) => isPDF(fileSet.mime_type))
+          .forEach((fileSet) => {
+            const rendering = {
+              id: fileSet.download_url || null,
+              type: "Text",
+              label: {
+                en: [fileSet.label || "Download PDF"],
+              },
+              format: "application/pdf",
+            };
+            renderings.push(rendering);
+          });
+        manifest.setRendering(renderings);
+
         /** Add rights using rights statement */
         source.rights_statement?.id &&
           manifest.setRights(source.rights_statement.id);
@@ -203,6 +223,7 @@ function transform(response) {
 
         source.file_sets
           .filter((fileSet) => fileSet.role === "Auxiliary")
+          .filter((fileSet) => !isAltFormat(fileSet.mime_type))
           .forEach((fileSet, index) => {
             buildCanvasFromFileSet(fileSet, index, true);
           });

node/src/api/response/iiif/presentation-api/items.js

@@ -57,6 +57,15 @@ function buildSupplementingAnnotation({ canvasId, fileSet }) {
   };
 }
 
+function isAltFormat(mimeType) {
+  const acceptedTypes = [
+    "application/pdf",
+    "application/zip",
+    "application/zip-compressed",
+  ];
+  return acceptedTypes.includes(mimeType);
+}
+
 function isAudioVideo(type) {
   return ["Audio", "Video", "Sound"].includes(type);
 }
@@ -65,13 +74,19 @@ function isImage(workType) {
   return workType === "Image";
 }
 
+function isPDF(mimeType) {
+  return mimeType === "application/pdf";
+}
+
 module.exports = {
   annotationType,
   buildAnnotationBody,
   buildAnnotationBodyId,
   buildImageResourceId,
   buildImageService,
   buildSupplementingAnnotation,
+  isAltFormat,
   isAudioVideo,
   isImage,
+  isPDF,
 };

node/src/handlers/get-file-set-download.js

@@ -2,44 +2,75 @@ const { SFNClient, StartExecutionCommand } = require("@aws-sdk/client-sfn");
 const { wrap } = require("./middleware");
 const { getFileSet } = require("../api/opensearch");
 const { videoTranscodeSettings } = require("./transcode-templates");
-
+const { getSignedUrl } = require("@aws-sdk/s3-request-presigner");
+const { S3Client, GetObjectCommand } = require("@aws-sdk/client-s3");
+const { apiTokenName } = require("../environment");
+const ApiToken = require("../api/api-token");
+const axios = require("axios").default;
+const cookie = require("cookie");
+const mime = require("mime-types");
 const opensearchResponse = require("../api/response/opensearch");
 const path = require("path");
 
 /**
- * Handler for download file set endpoint (currently only handles VIDEO)
+ * Handler for download file set endpoint
  */
 exports.handler = wrap(async (event) => {
   const id = event.pathParameters.id;
   const email = event.queryStringParameters?.email;
-  if (!email) {
-    return invalidRequest(400, "Query string must include email address");
-  }
 
-  if (!event.userToken.isSuperUser()) {
-    return invalidRequest(401, "Unauthorized");
-  }
+  const allowPrivate =
+    event.userToken.isSuperUser() ||
+    event.userToken.isReadingRoom() ||
+    event.userToken.hasEntitlement(id);
+  const allowUnpublished =
+    event.userToken.isSuperUser() || event.userToken.hasEntitlement(id);
   const esResponse = await getFileSet(id, {
-    allowPrivate: true,
-    allowUnpublished: true,
+    allowPrivate,
+    allowUnpublished,
   });
 
   if (esResponse.statusCode == "200") {
     const doc = JSON.parse(esResponse.body);
-    if (downloadAvailable(doc)) {
-      return await processDownload(doc, email);
+    if (isVideoDownload(doc)) {
+      if (!email) {
+        return invalidRequest(400, "Query string must include email address");
+      }
+      if (!event.userToken.isSuperUser()) {
+        return invalidRequest(401, "Unauthorized");
+      }
+      return await processAVDownload(doc, email);
+    } else if (isImageDownload(doc)) {
+      return await IIIFImageRequest(doc);
+    } else if (isAltFileDownload(doc)) {
+      const url = await getDownloadLink(doc);
+      return {
+        statusCode: 302,
+        headers: { Location: url },
+      };
     } else {
-      return invalidRequest(
-        405,
-        "Download only allowed for role: Access, work_type: Video, with a valid streaming_url"
-      );
+      return invalidRequest(405, "Download not allowed for role + work_type");
     }
   } else {
     return await opensearchResponse.transform(esResponse);
   }
 });
 
-function downloadAvailable(doc) {
+function isAltFileDownload(doc) {
+  const acceptedTypes = [
+    "application/pdf",
+    "application/zip",
+    "application/zip-compressed",
+  ];
+  return (
+    doc.found &&
+    doc._source.role === "Auxiliary" &&
+    doc._source.mime_type != null &&
+    acceptedTypes.includes(doc._source.mime_type)
+  );
+}
+
+function isVideoDownload(doc) {
   // Note - audio is not currently implemented due to an issue with AWS
   // & MediaConvert and our .m3u8 files
   return (
@@ -51,7 +82,94 @@ function downloadAvailable(doc) {
   );
 }
 
-async function processDownload(doc, email) {
+function isImageDownload(doc) {
+  return (
+    doc.found &&
+    ["Access", "Auxiliary"].includes(doc._source.role) &&
+    doc._source.mime_type != null &&
+    ["image"].includes(doc._source.mime_type.split("/")[0])
+  );
+}
+
+function derivativeKey(doc) {
+  const id = doc._id;
+  let prefix =
+    id.slice(0, 2) +
+    "/" +
+    id.slice(2, 4) +
+    "/" +
+    id.slice(4, 6) +
+    "/" +
+    id.slice(6, 8);
+  return "derivatives/" + prefix + "/" + id;
+}
+
+async function getDownloadLink(doc) {
+  const clientParams = {};
+  const bucket = process.env.PYRAMID_BUCKET;
+  const key = derivativeKey(doc);
+
+  const getObjectParams = {
+    Bucket: bucket,
+    Key: key,
+    ResponseContentDisposition: `attachment; filename=${
+      doc._source.label
+    }.${mime.extension(doc._source.mime_type)}`,
+  };
+
+  const client = new S3Client(clientParams);
+  const command = new GetObjectCommand(getObjectParams);
+  const url = await getSignedUrl(client, command, { expiresIn: 3600 * 24 * 3 }); // 3 days
+  return url;
+}
+
+const getAxiosResponse = (url, config) => {
+  return new Promise((resolve) => {
+    axios
+      .get(url, config)
+      .then((response) => resolve(response))
+      .catch((error) => resolve(error.response));
+  });
+};
+
+const IIIFImageRequest = async (doc) => {
+  const dimensions = "/full/max/0/default.jpg";
+  const iiifImageBaseUrl = doc._source.representative_image_url;
+  const url = `${iiifImageBaseUrl}${dimensions}`;
+  const { status, headers, data } = await getAxiosResponse(url, {
+    headers: {
+      cookie: cookie.serialize(
+        apiTokenName(),
+        new ApiToken().superUser().sign(),
+        {
+          domain: "library.northwestern.edu",
+          path: "/",
+          secure: true,
+        }
+      ),
+    },
+    responseType: "arraybuffer",
+  });
+
+  if (status != 200) {
+    return {
+      statusCode: status,
+      body: data.toString(),
+      headers: headers,
+    };
+  }
+
+  return {
+    statusCode: status,
+    isBase64Encoded: true,
+    body: data.toString("base64"),
+    headers: {
+      "content-type": headers["content-type"],
+    },
+  };
+};
+
+async function processAVDownload(doc, email) {
   const stepFunctionConfig = process.env.STEP_FUNCTION_ENDPOINT
     ? { endpoint: process.env.STEP_FUNCTION_ENDPOINT }
     : {};