Fix/mecat trainer/doc markup break

.github/workflows/medcat-trainer_ci.yml

@@ -285,13 +285,15 @@ jobs:
           tags: cogstacksystems/medcat-trainer:test
           labels: ${{ steps.meta.outputs.labels }}
           cache-from: type=registry,ref=cogstacksystems/medcat-trainer:buildcache
-          cache-to: type=registry,ref=cogstacksystems/medcat-trainer:buildcache,mode=max
 
       - name: Run Django Tests
         run: |
           # run tests using the built image
           docker run --rm cogstacksystems/medcat-trainer:test /home/.venv/bin/python manage.py test
 
+      - name: Prune stale Docker build cache
+        run: docker system prune --all --force --volumes
+
       - name: Push Docker MedCATtrainer image
         id: docker_build_push
         uses: docker/build-push-action@v7

medcat-trainer/webapp/frontend/src/components/common/ClinicalText.vue

@@ -74,8 +74,10 @@ export default {
     formattedText () {
       if (this.loading || !this.text || !this.ents) { return '' }
       if (this.ents.length === 0) {
-        let text = this.text.replace('&', '&amp').replace('<', '&gt').replace('>', '&gt')
-        text = text === 'nan' ? '' : text
+        // Must escape for vue3-runtime-template: raw '<' / '>' / '&' are compiled as Vue
+        // template HTML and throw (e.g. compiler-24 missing end tag), which can leave the
+        // annotator unable to render subsequent documents.
+        const text = this.text === 'nan' ? '' : _.escape(this.text)
         return this.addAnnos ? `<div @contextmenu.prevent.stop="showCtxMenu($event)">${text}</div>` : `<div>${text}</div>`
       }
 

medcat-trainer/webapp/frontend/src/tests/components/ClinicalText.spec.ts

@@ -387,5 +387,54 @@ describe('ClinicalText.vue', () => {
         const formattedText = vm.formattedText
         expect(formattedText).toContain('@contextmenu.prevent.stop')
     })
+
+    it('escapes HTML-like clinical markup when there are no annotations', async () => {
+        const { compile } = await import('@vue/compiler-dom')
+        const text = '18 >[ No Service Recorded ] [ LLAMS MDT ] status <unknown> a & b'
+        const wrapper = mount(ClinicalText, {
+            props: {
+                ...defaultProps,
+                text,
+                ents: []
+            },
+            global: {
+                stubs: ['v-overlay', 'v-progress-circular', 'v-runtime-template', 'vue-simple-context-menu']
+            }
+        })
+
+        const formattedText = (wrapper.vm as any).formattedText
+        expect(formattedText).toContain('&gt;')
+        expect(formattedText).toContain('&lt;unknown&gt;')
+        expect(formattedText).toContain('&amp;')
+        expect(formattedText).not.toContain('<unknown>')
+        expect(() => compile(formattedText)).not.toThrow()
+    })
+
+    it('escapes HTML-like markup in unannotated segments when annotations exist', async () => {
+        const { compile } = await import('@vue/compiler-dom')
+        const text = 'before <tag> Sample after >[ note ]'
+        const ents = [{
+            id: 1,
+            start_ind: 13,
+            end_ind: 19,
+            assignedValues: { 'Concept Anno': 'Correct' },
+            manually_created: false
+        }]
+        const wrapper = mount(ClinicalText, {
+            props: {
+                ...defaultProps,
+                text,
+                ents
+            },
+            global: {
+                stubs: ['v-overlay', 'v-progress-circular', 'v-runtime-template', 'vue-simple-context-menu']
+            }
+        })
+
+        const formattedText = (wrapper.vm as any).formattedText
+        expect(formattedText).toContain('&lt;tag&gt;')
+        expect(formattedText).toContain('&gt;')
+        expect(() => compile(formattedText)).not.toThrow()
+    })
 })
 

medcat-trainer/webapp/frontend/src/tests/views/TrainAnnotations.spec.ts

@@ -52,7 +52,7 @@ describe('TrainAnnotations.vue fetchData', () => {
       if (url.startsWith('/api/project-annotate-entities/')) {
         return Promise.reject({ response: { status: 500, data: { message: 'Database unavailable' } } })
       }
-      return new Promise(() => {})
+      return new Promise(() => { })
     })
 
     wrapper.vm.fetchData()
@@ -68,7 +68,7 @@ describe('TrainAnnotations.vue fetchData', () => {
       if (url.startsWith('/api/project-annotate-entities/')) {
         return Promise.reject({ response: { status: 401, data: { detail: 'Invalid token.' } } })
       }
-      return new Promise(() => {})
+      return new Promise(() => { })
     })
 
     wrapper.vm.fetchData()
@@ -85,7 +85,7 @@ describe('TrainAnnotations.vue fetchEntities', () => {
         return Promise.reject({ response: { data: { message: 'Invalid token.' } } })
       }
       // Stall created() lifecycle requests so they don't interfere with the test.
-      return new Promise(() => {})
+      return new Promise(() => { })
     })
 
     wrapper.vm.project = project
@@ -134,7 +134,7 @@ describe('TrainAnnotations.vue fetchEntities', () => {
       if (url.startsWith('/api/annotated-entities/')) {
         return Promise.resolve({ data: { results: [], previous: null, next: null } })
       }
-      return new Promise(() => {})
+      return new Promise(() => { })
     })
 
     wrapper.vm.fetchData()
@@ -157,7 +157,7 @@ describe('TrainAnnotations.vue fetchEntities', () => {
           }
         })
       }
-      return new Promise(() => {})
+      return new Promise(() => { })
     })
 
     wrapper.vm.project = project
@@ -173,8 +173,36 @@ describe('TrainAnnotations.vue fetchEntities', () => {
     expect(wrapper.vm.currentEnt.id).toBe(10)
   })
 
+  it('resets annotation pagination state when switching documents', async () => {
+    const { wrapper, mockGet } = mountView((url) => {
+      if (url.startsWith('/api/annotated-entities/')) {
+        return Promise.resolve({
+          data: { results: [], previous: null, next: null }
+        })
+      }
+      return new Promise(() => { })
+    })
+
+    wrapper.vm.project = { ...project, prepared_documents: [123, 456] }
+    wrapper.vm.currentDoc = { id: 123, text: 'first doc' }
+    wrapper.vm.ents = [{ id: 1, start_ind: 0, end_ind: 1, assignedValues: {} }]
+    wrapper.vm.nextEntSetUrl = '/api/annotated-entities/?project=1&document=123&page=2'
+    wrapper.vm.loadingMsg = null
+
+    wrapper.vm.loadDoc({ id: 456, text: 'second doc with <tag> markup' })
+    await flushPromises()
+
+    expect(wrapper.vm.currentDoc.id).toBe(456)
+    expect(wrapper.vm.loadingMsg).toBeNull()
+    // Must request the new document, not continue the previous page URL.
+    const entCalls = mockGet.mock.calls.filter(c => String(c[0]).startsWith('/api/annotated-entities/'))
+    expect(entCalls.some(c => String(c[0]).includes('document=456'))).toBe(true)
+    expect(entCalls.some(c => String(c[0]).includes('page=2'))).toBe(false)
+
+  })
+
   it('does not load a document when the auth cookie is missing', async () => {
-    const { wrapper, mockGet } = mountView(() => new Promise(() => {}), {})
+    const { wrapper, mockGet } = mountView(() => new Promise(() => { }), {})
     mockGet.mockClear()
 
     wrapper.vm.project = project
@@ -183,5 +211,6 @@ describe('TrainAnnotations.vue fetchEntities', () => {
 
     expect(mockGet).not.toHaveBeenCalled()
     expect(wrapper.vm.currentDoc).toBeNull()
+
   })
 })

medcat-trainer/webapp/frontend/src/views/TrainAnnotations.vue

@@ -529,7 +529,12 @@ export default {
           }
         })
       }
+      // Clear prior doc annotation state so we never render the new text with old ents
+      // (or continue paginating the previous document's annotated-entities URL).
       this.currentEnt = null
+      this.ents = null
+      this.nextEntSetUrl = null
+      this.loadingMsg = 'Loading document...'
       this.prepareDoc()
     },
     prepareDoc () {
@@ -620,7 +625,8 @@ export default {
         this.nextEntSetUrl = null
         this.loadingMsg = null
         this.errors.modal = true
-        this.errors.message = 'Failed to load document annotations. Please try again by refreshing the page.'
+        this.errors.message = `Failed to load document annotations for project ID ${this.projectId || ''}${this.currentDoc && this.currentDoc.id ? ', document ID ' + this.currentDoc.id : ''}. Please try again by refreshing the page. If the problem persists, please contact your administrator and provide them with this page's URL: ${window.location.href}`
+
         if (err.response) {
           this.errors.message = err.response.data?.message || this.errors.message
           this.errors.description = err.response.data?.description || ''