Upload folder using huggingface_hub

Dockerfile

@@ -1,5 +1,5 @@
-# Use Node.js 20 Alpine for closer parity with local (reduces HF vs local behavior drift)
-FROM node:20-alpine
+# Use Node.js 18 Alpine for smaller image size
+FROM node:18-alpine
 
 # Set working directory
 WORKDIR /app
@@ -7,9 +7,8 @@ WORKDIR /app
 # Copy package files
 COPY package*.json ./
 
-# Install dependencies deterministically from lockfile
-# (prevents HF vs local drift in docx/diff behavior)
-RUN npm ci --omit=dev
+# Install dependencies
+RUN npm install --only=production
 
 # Copy server source code
 COPY . ./
@@ -30,6 +29,7 @@ HEALTHCHECK --interval=60s --timeout=10s --start-period=120s --retries=5 \
   CMD curl -f http://localhost:7860/api/health || exit 1
 
 # Start the application
+# rebuild trigger 2025-09-01T14:18:50Z
 CMD npm start
 
-# rebuild trigger 2025-12-19T00:40:00Z
+# rebuild trigger 2025-09-02T03:17:52Z

routes/refinity.js

@@ -49,20 +49,6 @@ function safeDiffTokens(a, b) {
 }
 const { Document, Packer, Paragraph, TextRun, CommentRangeStart, CommentRangeEnd, CommentReference, Comments } = require('docx');
 const JSZip = require('jszip');
-const mongoose = require('mongoose');
-
-function safePkgVersion(name) {
-  try {
-    // Some packages (like docx) don't export package.json, so read from disk.
-    const fs = require('fs');
-    const path = require('path');
-    const p = path.join(process.cwd(), 'node_modules', name, 'package.json');
-    const j = JSON.parse(fs.readFileSync(p, 'utf8'));
-    return j?.version || null;
-  } catch {
-    return null;
-  }
-}
 
 const router = express.Router();
 const RefinityTask = require('../models/RefinityTask');
@@ -70,25 +56,6 @@ const RefinityVersion = require('../models/RefinityVersion');
 const upload = multer({ storage: multer.memoryStorage(), limits: { fileSize: 8 * 1024 * 1024 } });
 const RefinityAnnotation = require('../models/RefinityAnnotation');
 
-// Diagnostic endpoint (helps verify which backend build HF is serving)
-router.get('/test', (req, res) => {
-  res.json({
-    message: 'refinity route is working',
-    compareCommentsImpl: 'tutorial-refinity-copied',
-    build: {
-      signature: 'toolkit-export-origin-slice-remap-v3',
-      node: process.version,
-      docx: safePkgVersion('docx'),
-      diff: safePkgVersion('diff'),
-      db: {
-        name: mongoose?.connection?.name || null,
-        state: mongoose?.connection?.readyState ?? null,
-      },
-    },
-    timestamp: new Date().toISOString(),
-  });
-});
-
 // ---- Helpers ----
 function encodeRFC5987ValueChars(str) {
   return encodeURIComponent(str)
@@ -299,9 +266,6 @@ router.post('/compare-comments-with-corrections', async (req, res) => {
     const annotationVersionId = req.body?.annotationVersionId;
     const authorName = String(req.body?.authorName || 'Refinity');
     const authorInitials = String(req.body?.authorInitials || (authorName.split(/\s+/).map(s=>s[0]||'').join('').slice(0,3).toUpperCase()) || 'RF');
-    const requester = String(req.headers['x-user-name'] || req.headers['x-user-email'] || '').toLowerCase();
-    const roleHdr = String(req.headers['x-user-role'] || req.headers['user-role'] || '').toLowerCase();
-    const isAdmin = roleHdr === 'admin';
 
     if (!annotationVersionId) {
       return res.status(400).json({ error: 'annotationVersionId is required' });
@@ -314,23 +278,16 @@ router.post('/compare-comments-with-corrections', async (req, res) => {
     }
 
     // Find all versions in the same task that come after the older version
-    // (kept for compatibility / potential future use)
     const laterVersions = await RefinityVersion.find({
       taskId: olderVersion.taskId,
       versionNumber: { $gt: olderVersion.versionNumber }
     }).sort({ versionNumber: 1 }).lean();
 
-    // Match Tutorial DR compare-sidebar behavior:
-    // include annotations from the older version AND all later versions in the same task.
-    // This allows compare(v1, v4) to surface annotations added while revising v2/v3 as well.
+    // Get all annotations from the older version and all later versions
     const versionIds = [annotationVersionId, ...laterVersions.map(v => v._id)];
-    const annQuery = { versionId: { $in: versionIds } };
-    // Toolkit DR is shared across users; avoid mixing other users' highlights/corrections in exports.
-    // Non-admin: only export annotations created by the requester.
-    if (requester && !isAdmin) {
-      annQuery.createdBy = requester;
-    }
-    const allAnns = await RefinityAnnotation.find(annQuery).sort({ start: 1, end: 1 }).lean();
+    const allAnns = await RefinityAnnotation.find({
+      versionId: { $in: versionIds }
+    }).sort({ start: 1, end: 1 }).lean();
 
     // Build a map of annotations by their text content for matching
     // Key: normalized text content, Value: annotation with error type, correction, and positions
@@ -342,50 +299,7 @@ router.post('/compare-comments-with-corrections', async (req, res) => {
         : laterVersions.find(v => v._id.toString() === ann.versionId.toString());
       if (!version) continue;
       
-      // Important safety + backward compatibility for toolkit exports:
-      // Annotation offsets are relative to the version they were created on. Applying those
-      // offsets directly to `prev` can cause wrong edits / duplicated body text.
-      //
-      // If the origin slice at [start,end) doesn't match `prev` at the same offsets, try to
-      // remap by searching the origin slice text inside `prev` and anchoring to the closest
-      // occurrence. If we can't remap, skip (better than inventing edits).
-      const rawStart = Number(ann.start || 0);
-      const rawEnd = Number(ann.end || 0);
-      const originContent = String(version.content || '');
-      const originSlice = originContent.slice(Math.max(0, rawStart), Math.max(0, rawEnd));
-      let mappedStart = rawStart;
-      let mappedEnd = rawEnd;
-      let annText = prev.slice(Math.max(0, rawStart), Math.max(0, rawEnd));
-
-      // If `annText` is empty (out-of-range) OR mismatched, try to remap by text search.
-      if (originSlice && originSlice !== annText) {
-        // Collect all occurrences (not just first) so we can pick the closest to rawStart.
-        const occurrences = [];
-        let fromIdx = 0;
-        while (fromIdx <= prev.length) {
-          const idx = prev.indexOf(originSlice, fromIdx);
-          if (idx === -1) break;
-          occurrences.push(idx);
-          // Advance by 1 to allow overlapping matches (rare, but safer for short CJK spans).
-          fromIdx = idx + 1;
-        }
-        if (!occurrences.length) {
-          continue;
-        }
-        let bestIdx = occurrences[0];
-        let bestDist = Math.abs(bestIdx - rawStart);
-        for (const idx of occurrences) {
-          const d = Math.abs(idx - rawStart);
-          if (d < bestDist) {
-            bestDist = d;
-            bestIdx = idx;
-          }
-        }
-        mappedStart = bestIdx;
-        mappedEnd = bestIdx + originSlice.length;
-        annText = prev.slice(mappedStart, mappedEnd);
-      }
-
+      const annText = prev.slice(ann.start || 0, ann.end || 0);
       if (annText) {
         const normalized = annText.trim().replace(/\s+/g, ' ');
         // Store the annotation with its error type and correction
@@ -394,16 +308,16 @@ router.post('/compare-comments-with-corrections', async (req, res) => {
         }
         annByText.get(normalized).push({
           category: ann.category || 'other',
-          start: mappedStart,
-          end: mappedEnd,
+          start: ann.start,
+          end: ann.end,
           correction: ann.correction || '',
           originalAnn: ann, // Store full annotation for reference
         });
         
         // Also store by position for range matching
         annByPosition.push({
-          start: mappedStart,
-          end: mappedEnd,
+          start: ann.start || 0,
+          end: ann.end || 0,
           category: ann.category || 'other',
           correction: ann.correction || '',
           selectedText: annText,
@@ -434,50 +348,6 @@ router.post('/compare-comments-with-corrections', async (req, res) => {
       return union.size > 0 ? intersection.size / union.size : 0;
     }
     
-    // ANNOTATION-FIRST APPROACH (exactly like Tutorial DR):
-    // Only process annotations from the database. Skip all diff-based processing.
-    const USE_ANNOTATION_FIRST = true;
-    let mergedAnnotations = [];
-    if (USE_ANNOTATION_FIRST) {
-      const annotationBasedItems = [];
-      for (const ann of annByPosition) {
-        const annSelectedText = ann.selectedText;
-        const annCorrection = ann.correction || '';
-        const annCategory = ann.category || 'other';
-        const isDeletion = !annCorrection || annCorrection.trim() === '' || annCorrection.trim() === annSelectedText.trim();
-        const isInsertion = annSelectedText.trim().length <= 2 && annCorrection.trim().length > annSelectedText.trim().length;
-        annotationBasedItems.push({
-          start: ann.start,
-          end: ann.end,
-          removedText: annSelectedText,
-          replacementText: annCorrection,
-          category: annCategory,
-          isDeleted: isDeletion,
-          isInsertion: isInsertion,
-          isMove: false,
-        });
-      }
-      annotationBasedItems.sort((a, b) => a.start - b.start || a.end - b.end);
-      mergedAnnotations = (() => {
-        const out = [];
-        const seen = new Set();
-        for (const it of annotationBasedItems) {
-          const key = [
-            it.start ?? '',
-            it.end ?? '',
-            String(it.category || ''),
-            String(it.replacementText || ''),
-            it.isDeleted ? '1' : '0',
-            it.isInsertion ? '1' : '0',
-            it.isMove ? '1' : '0',
-          ].join('|');
-          if (seen.has(key)) continue;
-          seen.add(key);
-          out.push(it);
-        }
-        return out;
-      })();
-    } else {
     // 1) Run diff to find ALL changes (like Show Diff does)
     const diffParts = safeDiffTokens(prev, current);
     
@@ -1192,31 +1062,7 @@ router.post('/compare-comments-with-corrections', async (req, res) => {
       return finalAnnotations;
     };
 
-    mergedAnnotations = mergeDeletionInsertionPairs(diffBasedAnnotations);
-
-    // De-dupe identical change items so sidebar-comment export doesn't duplicate spans/comments.
-    // This mirrors the tutorial DR fix and keeps behavior minimal/safe.
-    {
-      const out = [];
-      const seen = new Set();
-      for (const it of mergedAnnotations) {
-        const key = [
-          it.start ?? '',
-          it.end ?? '',
-          String(it.category || ''),
-          String(it.removedText || ''),
-          String(it.replacementText || ''),
-          it.isDeleted ? '1' : '0',
-          it.isInsertion ? '1' : '0',
-          it.isMove ? '1' : '0',
-        ].join('|');
-        if (seen.has(key)) continue;
-        seen.add(key);
-        out.push(it);
-      }
-      mergedAnnotations = out;
-    }
-    } // end diff-first fallback
+    const mergedAnnotations = mergeDeletionInsertionPairs(diffBasedAnnotations);
 
     // 3) Split the older text into logical lines with their global offsets
     const lines = [];
@@ -1423,14 +1269,17 @@ router.post('/compare-comments-with-corrections', async (req, res) => {
       
       let commentChildren;
       if (item.isDeleted) {
+        // Deletion: show error type + original text with strikethrough
         commentChildren = [
           new TextRun({ text: `[${categoryLabel}] ` }),
           new TextRun({ text: original, strike: true }),
         ];
       } else if (item.isInsertion) {
+        // Insertion: show error type + → inserted text
         const commentText = `[${categoryLabel}] → ${newer}`;
         commentChildren = [ new TextRun({ text: commentText }) ];
       } else {
+        // Change: show error type + original → newer
         const commentText = `[${categoryLabel}] ${original} → ${newer}`;
         commentChildren = [ new TextRun({ text: commentText }) ];
       }

routes/tutorial-refinity.js

@@ -49,18 +49,6 @@ const TutorialRefinityVersion = require('../models/TutorialRefinityVersion');
 const TutorialRefinityAnnotation = require('../models/TutorialRefinityAnnotation');
 const upload = multer({ storage: multer.memoryStorage(), limits: { fileSize: 8 * 1024 * 1024 } });
 
-function safePkgVersion(name) {
-  try {
-    const fs = require('fs');
-    const path = require('path');
-    const p = path.join(process.cwd(), 'node_modules', name, 'package.json');
-    const j = JSON.parse(fs.readFileSync(p, 'utf8'));
-    return j?.version || null;
-  } catch {
-    return null;
-  }
-}
-
 // Helper to set download headers
 function setDownloadHeaders(res, filename) {
   res.setHeader('Content-Type', 'application/vnd.openxmlformats-officedocument.wordprocessingml.document');
@@ -271,29 +259,8 @@ router.post('/compare-comments-with-corrections', async (req, res) => {
     
     // Sort by position
     annotationBasedItems.sort((a, b) => a.start - b.start || a.end - b.end);
-
-    // De-dupe identical annotations (same span/category/correction/flags).
-    // Without this, duplicated DB rows (or re-annotated identical spans) can cause the
-    // generator to literally duplicate the span text in the document body.
-    const mergedAnnotations = (() => {
-      const out = [];
-      const seen = new Set();
-      for (const it of annotationBasedItems) {
-        const key = [
-          it.start ?? '',
-          it.end ?? '',
-          String(it.category || ''),
-          String(it.replacementText || ''),
-          it.isDeleted ? '1' : '0',
-          it.isInsertion ? '1' : '0',
-          it.isMove ? '1' : '0',
-        ].join('|');
-        if (seen.has(key)) continue;
-        seen.add(key);
-        out.push(it);
-      }
-      return out;
-    })();
+    
+    const mergedAnnotations = annotationBasedItems;
 
     const lines = [];
     let pos = 0;
@@ -588,16 +555,7 @@ router.post('/export-plain', async (req, res) => {
 
 // ----- Task APIs -----
 router.get('/test', (req, res) => {
-  res.json({
-    message: 'tutorial-refinity route is working',
-    build: {
-      signature: 'tutorial-refinity-export-v1',
-      node: process.version,
-      docx: safePkgVersion('docx'),
-      diff: safePkgVersion('diff'),
-    },
-    timestamp: new Date().toISOString(),
-  });
+  res.json({ message: 'tutorial-refinity route is working', timestamp: new Date().toISOString() });
 });
 
 // Diagnostic endpoint to list all tasks (for debugging)