Merge branch 'master' into v0.0.34

openreview · Sep 30, 2024 · ba8b4d5 · ba8b4d5
2 parents 621fb59 + b871d7f
commit ba8b4d5
Show file tree

Hide file tree

Showing 5 changed files with 20 additions and 12 deletions.
diff --git a/packages/client/src/tools.js b/packages/client/src/tools.js
@@ -869,9 +869,18 @@ export default class Tools {
   static async extractAbstract(url) {
     const metaExtractionUrl = 'https://meta-extraction-wivlbyt6ga-uc.a.run.app/metadata';
     const queryString = generateQueryString({ url });
-    const result = await fetch(`${metaExtractionUrl}?${queryString}`, {
-      method: 'GET',
-    });
+    let result;
+
+    try {
+      result = await fetch(`${metaExtractionUrl}?${queryString}`, {
+        method: 'GET',
+      });
+    } catch (error) {
+      throw new OpenReviewError({
+        name: 'ExtractAbstractError',
+        message: error,
+      });
+    }
 
     if (result.status === 200) {
       return result.json();
@@ -883,6 +892,5 @@ export default class Tools {
       message: (contentType && contentType.indexOf('application/json') !== -1) ? JSON.stringify(await result.json()) : await result.text(),
       status: result.status || 500
     });
-
   }
 }
diff --git a/packages/meta-extraction/package.json b/packages/meta-extraction/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@openreview/meta-extraction",
-  "version": "0.0.7",
+  "version": "0.0.8",
   "description": "Extract abstracts for DBLP papers",
   "main": "src/index.js",
   "type": "module",

diff --git a/packages/meta-extraction/src/abstractExtractionRules.js b/packages/meta-extraction/src/abstractExtractionRules.js
@@ -764,8 +764,8 @@ const worldscientificComRule = {
 const iscaSpeechOrgRule = {
   shouldApplyRule: (url) => /isca-speech.org/.test(url) || /isca-archive.org/.test(url),
   executeRule: async (html, page) => {
-    const abstract = await selectElemTextEvidence(page, 'div.w3-card>p');
-    const pdf = await selectElemAttrEvidence(page, 'div.w3-card>a', 'href');
+    const abstract = await selectElemTextEvidence(page, 'div#abstract>p');
+    const pdf = await selectElemAttrEvidence(page, 'div.w3-content div.w3-center a', 'href');
     const allEvidence = [
       { type: 'abstract', value: abstract },
       { type: 'pdf', value: pdf }

diff --git a/packages/meta-extraction/src/helpers.js b/packages/meta-extraction/src/helpers.js
@@ -4,7 +4,7 @@ const shouldEnableMultiRedirect = (url) => [/doi.org/, /linkinghub.elsevier.com/
 
 const getTimeout = (url) => {
   const defaultTimeout = 15_000;
-  if ([/doi.org/, /spiedigitallibrary.org/, /iospress.com/].some((regex) => regex.test(url))) return defaultTimeout*3;
+  if ([/doi.org/, /spiedigitallibrary.org/, /iospress.com/,/export.arxiv.org/].some((regex) => regex.test(url))) return defaultTimeout*3;
   return defaultTimeout;
 };
 

diff --git a/packages/meta-extraction/test/test.js b/packages/meta-extraction/test/test.js