nopaste.diff

commit b8de8d5d3dbce28d12b5d946c2f6f1cbf7becfbb
Author: Thomas Sibley <tsibley@fredhutch.org>
Date:   Mon Oct 30 13:46:58 2023 -0700

    wip! version extractor
    
    XXX FIXME: how far to take @ as not a given?  i.e. update canonicalizeDataset to make the caller add the version?
    XXX FIXME: default to extracting a version?
    XXX FIXME: move core version extractor into a function we pass?
    XXX FIXME: use route matching instead to make versionExtractor like: req => req.params.versionDescriptor?
    XXX FIXME: adjust setDatasetFromPrefix

diff --git a/src/app.js b/src/app.js
index 51a198ca..db337741 100644
--- a/src/app.js
+++ b/src/app.js
@@ -213,7 +213,7 @@ const coreBuildRoutes = coreBuildPaths.map(path => [
 app.use([coreBuildRoutes, "/narratives/*"], setSource(req => new CoreSource())); // eslint-disable-line no-unused-vars
 
 app.routeAsync(coreBuildRoutes)
-  .all(setDataset(req => req.path, true), canonicalizeDataset(path => `/${path}`))
+  .all(setDataset(req => req.path, req => req.path.match(/@(.*)$/)?.[1]), canonicalizeDataset(path => `/${path}`))
   .getAsync(getDataset)
   .putAsync(putDataset)
   .deleteAsync(deleteDataset)
diff --git a/src/endpoints/sources.js b/src/endpoints/sources.js
index 8b12a388..5d61543b 100644
--- a/src/endpoints/sources.js
+++ b/src/endpoints/sources.js
@@ -33,12 +33,12 @@ const setSource = (sourceExtractor) => (req, res, next) => {
  * stashes it in the request context.
  *
  * @param {pathExtractor} pathExtractor - Function to extract a dataset path from the request
- * @param {Boolean} extractVersion - (Attempt to) extract a version descriptor from the extracted path
+ * @param {versionExtractor} [versionExtractor] - Function to extract a version descriptor from the request
  * @returns {expressMiddleware}
  */
-const setDataset = (pathExtractor, extractVersion=false) => (req, res, next) => {
+const setDataset = (pathExtractor, versionExtractor = req => LATEST) => (req, res, next) => {
   console.log("middleware::setDataset")
-  req.context.dataset = req.context.source.dataset(...pathParts(pathExtractor(req), extractVersion));
+  req.context.dataset = req.context.source.dataset(pathParts(pathExtractor(req)), versionExtractor(req));
   next();
 };
 
@@ -202,11 +202,10 @@ const optionsNarrative = options.forAuthzObject(req => req.context.narrative);
  * and stashes it in the request context.
  *
  * @param {pathExtractor} pathExtractor - Function to extract a narrative path from the request
- * @param {Boolean} extractVersion - (Attempt to) extract a version descriptor from the extracted path
  * @returns {expressMiddleware}
  */
-const setNarrative = (pathExtractor, extractVersion=false) => (req, res, next) => {
-  req.context.narrative = req.context.source.narrative(...pathParts(pathExtractor(req), extractVersion));
+const setNarrative = (pathExtractor) => (req, res, next) => {
+  req.context.narrative = req.context.source.narrative(pathParts(pathExtractor(req)));
   next();
 };
 
@@ -261,45 +260,27 @@ const putNarrative = contentTypesConsumed([
 
 
 /**
- * Split a dataset or narrative `path` into an array of parts and a version
- * descriptor.
+ * Split a dataset or narrative `path` into an array of parts.
  *
  * If `path` is a tangletree path (i.e. refers to two datasets), returns only
  * the parts for the first dataset.
  *
- * By default the returned version descriptor is LATEST. If `extractVersion` is
- * true, we attempt to extract a version descriptor from the provided path,
- * falling back to LATEST if one is not present.
+ * If `path` contains a version descriptor (i.e. @…), it is removed.
  *
  * @param {String} path
- * @param {Boolean} extractVersion
- * @returns {[String[], String]}
+ * @returns {String[]}
  */
-function pathParts(path = "", extractVersion = false) {
-  // Use only the first dataset in a tangletree (dual dataset) path.
-  let normalizedPath = path.split(":")[0]          
+function pathParts(path = "") {
+  const normalizedPath = path
+    .split(":")[0]          // Use only the first dataset in a tangletree (dual dataset) path.
+    .replace(/@.*$/, "")    // Ignore version descriptor.
+    .replace(/^\/+/, "")    // Ignore leading slashes
+    .replace(/\/+$/, "")    //   …and trailing slashes.
+  ;
 
-  /* A part of the path starting with "@" is the version descriptor - this
-  will later be translated to the appropriate S3 version ID / GitHub SHA,
-  but we use human readable descriptors in the URL path. */
-  let versionDescriptor = LATEST; // default;
-  if (extractVersion) {
-    const match = normalizedPath.match(/^(.+)@(.+)$/)
-    if (match) {
-      [normalizedPath, versionDescriptor] = match.slice(1,3);
-    }
-  }
+  if (!normalizedPath) return [];
 
-  // Ignore leading & trailing slashes (after version descriptor removal)
-  normalizedPath = normalizedPath
-    .replace(/\/+$/, "")
-    .replace(/^\/+/, "");
-
-  if (!normalizedPath) return [[], LATEST];
-  
-  console.log(`function::pathParts(${path}, ${extractVersion}) -> `, normalizedPath.split("/"), versionDescriptor)
-
-  return [normalizedPath.split("/"), versionDescriptor]
+  return normalizedPath.split("/");
 }
 
 
@@ -369,6 +350,12 @@ function receiveSubresource(subresourceExtractor) {
  * @returns {String} Path for {@link Source#dataset} or {@link Source#narrative}
  */
 
+/**
+ * @callback versionExtractor
+ * @param {express.request} req
+ * @returns {?String} Version descriptor for {@link Source#dataset} or {@link Source#narrative}
+ */
+
 /* Confused about the duplication below?  It's the documented way to handle
  * overloaded (e.g. arity-dependent) function signatures.¹  Note that it relies
  * on the "nestled" or "cuddled" end and start comment markers.