📋 Prioritize custom resolver jats urls over pmc/doi

.changeset/selfish-poems-compete.md

@@ -0,0 +1,8 @@
+---
+'jats-convert': patch
+'jats-fetch': patch
+'jats-cli': patch
+'jats-xml': patch
+---
+
+Bump doi-utils version

.changeset/spotty-donuts-knock.md

@@ -0,0 +1,5 @@
+---
+'jats-fetch': patch
+---
+
+Prioritize custom resolver jats urls over pmc/doi

packages/jats-cli/package.json

@@ -43,7 +43,7 @@
     "url": "https://github.com/continuous-foundation/jats/issues"
   },
   "dependencies": {
-    "doi-utils": "^2.0.0",
+    "doi-utils": "^2.0.6",
     "fair-principles": "^2.0.0",
     "jats-convert": "^1.0.16",
     "jats-fetch": "^1.0.16",

packages/jats-convert/package.json

@@ -36,7 +36,7 @@
     "url": "https://github.com/continuous-foundation/jats/issues"
   },
   "dependencies": {
-    "doi-utils": "^2.0.3",
+    "doi-utils": "^2.0.6",
     "dotenv": "^16.4.5",
     "jats-fetch": "^1.0.16",
     "jats-tags": "^1.0.16",

packages/jats-fetch/package.json

@@ -36,7 +36,7 @@
   },
   "dependencies": {
     "@aws-sdk/client-s3": "3.686.0",
-    "doi-utils": "^2.0.0",
+    "doi-utils": "^2.0.6",
     "node-fetch": "^3.3.1",
     "myst-cli-utils": "^2.0.11",
     "xml-js": "^1.6.11"

packages/jats-fetch/src/download.ts

@@ -116,12 +116,24 @@ export async function downloadJatsFromUrl(
     const data = fs.readFileSync(urlOrDoi).toString();
     return { success: true, source: urlOrDoi, data };
   }
-  const expectedUrls = (
-    await Promise.all([
-      constructJatsUrlFromPubMedCentral(session, urlOrDoi, opts),
-      getJatsUrlFromDoi(session, urlOrDoi, opts),
-    ])
-  ).filter((u): u is string => !!u);
+  const expectedUrls: string[] = [];
+  try {
+    // Custom resolvers are prioritized; in these cases we will know more about the "flavor" of the JATS
+    expectedUrls.push(await customResolveJatsUrlFromDoi(session, urlOrDoi, opts));
+  } catch {
+    session.log.debug(`No custom resolvers match ${urlOrDoi}`);
+  }
+  expectedUrls.push(
+    ...(
+      await Promise.all([
+        constructJatsUrlFromPubMedCentral(session, urlOrDoi, opts),
+        getJatsUrlFromDoi(session, urlOrDoi, opts),
+      ])
+    ).filter((u): u is string => !!u),
+  );
+  if (isUrl(urlOrDoi)) {
+    expectedUrls.push(urlOrDoi);
+  }
   if (expectedUrls.length > 0) {
     session.log.debug(['Trying URLs:\n', ...expectedUrls.map((url) => ` ${url}\n`)].join('  - '));
     for (let index = 0; index < expectedUrls.length; index++) {
@@ -133,27 +145,9 @@ export async function downloadJatsFromUrl(
         session.log.debug((error as Error).message);
       }
     }
-    // If there are expected URLs that don't work, report them and do not try other resolvers
     return { success: false, source: expectedUrls[0] };
   }
-  try {
-    if (doi.validate(urlOrDoi)) {
-      const jatsUrl = await customResolveJatsUrlFromDoi(session, urlOrDoi, opts);
-      const data = await downloadFromUrl(session, jatsUrl, opts);
-      return { success: true, source: jatsUrl, data };
-    }
-    if (isUrl(urlOrDoi)) {
-      session.log.debug(
-        "No resolver matched, and the URL doesn't look like a DOI. We will attempt to download it directly.",
-      );
-      const data = await downloadFromUrl(session, urlOrDoi, opts);
-      return { success: true, source: urlOrDoi, data };
-    }
-  } catch (error) {
-    session.log.debug((error as Error).message);
-    return { success: false, source: urlOrDoi };
-  }
-  session.log.debug(`Could not find ${urlOrDoi} locally, and it doesn't look like a URL or DOI`);
+  session.log.debug(`Could not find ${urlOrDoi} locally or resolve it to a valid JATS url`);
   return { success: false, source: urlOrDoi };
 }
 

packages/jats-xml/package.json

@@ -39,7 +39,7 @@
   },
   "dependencies": {
     "adm-zip": "^0.5.10",
-    "doi-utils": "^2.0.3",
+    "doi-utils": "^2.0.6",
     "jats-tags": "^1.0.16",
     "jats-utils": "^1.0.16",
     "myst-cli-utils": "^2.0.11",