Merge pull request #183 from ufal/annotation-interface-fixes

Annotation interface fixes
ufal · Jan 23, 2025 · d3bde6c · d3bde6c
2 parents 3c11dc3 + 5c3c420
commit d3bde6c
Show file tree

Hide file tree

Showing 6 changed files with 34 additions and 21 deletions.
diff --git a/factgenie/campaign.py b/factgenie/campaign.py
@@ -203,8 +203,8 @@ def get_overview(self):
             example_cnt=pd.NamedAgg(column="example_idx", aggfunc="count"),
             status=pd.NamedAgg(column="status", aggfunc="first"),
             annotator_id=pd.NamedAgg(column="annotator_id", aggfunc="first"),
-            start=pd.NamedAgg(column="start", aggfunc="first"),
-            end=pd.NamedAgg(column="end", aggfunc="first"),
+            start=pd.NamedAgg(column="start", aggfunc="min"),
+            end=pd.NamedAgg(column="end", aggfunc="max"),
         ).reset_index()
 
         for col in ["status", "annotator_id", "start", "end"]:

diff --git a/factgenie/crowdsourcing.py b/factgenie/crowdsourcing.py
@@ -343,7 +343,7 @@ def get_annotator_batch(app, campaign, service_ids, batch_idx=None):
                 # no available batches
                 return []
         else:
-            # preview mode
+            # preview mode with the specific batch
             batch_idx = int(batch_idx)
             annotator_group = 0
 
@@ -352,10 +352,10 @@ def get_annotator_batch(app, campaign, service_ids, batch_idx=None):
         # we do not block the example if we are in preview mode
         if annotator_id != PREVIEW_STUDY_ID:
             db.loc[mask, "status"] = ExampleStatus.ASSIGNED
+            db.loc[mask, "start"] = start
+            db.loc[mask, "annotator_id"] = annotator_id
 
-        db.loc[mask, "start"] = start
-        db.loc[mask, "annotator_id"] = annotator_id
-        campaign.update_db(db)
+            campaign.update_db(db)
 
         annotator_batch = get_examples_for_batch(db, batch_idx, annotator_group)
         logging.info(f"Releasing lock for {annotator_id}")

diff --git a/factgenie/models.py b/factgenie/models.py
@@ -408,18 +408,25 @@ def get_optional_fields(self):
     def postprocess_output(self, output):
         extra_args = self.config.get("extra_args", {})
 
-        if extra_args.get("remove_suffix", ""):
-            suffix = self.config["extra_args"]["remove_suffix"]
-
-            if output.endswith(suffix):
-                output = output[: -len(suffix)]
-
+        # cut model generation at the stopping sequence
         if extra_args.get("stopping_sequence", False):
-            stopping_sequence = self.config["extra_args"]["stopping_sequence"]
+            stopping_sequence = extra_args["stopping_sequence"]
+
+            # re-normalize double backslashes ("\\n" -> "\n")
+            stopping_sequence = stopping_sequence.encode().decode("unicode_escape")
 
             if stopping_sequence in output:
                 output = output[: output.index(stopping_sequence)]
 
+        output = output.strip()
+
+        # strip the suffix from the output
+        if extra_args.get("remove_suffix", ""):
+            suffix = extra_args["remove_suffix"]
+
+            if output.endswith(suffix):
+                output = output[: -len(suffix)]
+
         output = output.strip()
         return output
 

diff --git a/factgenie/static/js/annotate.js b/factgenie/static/js/annotate.js
@@ -138,6 +138,8 @@ function goToAnnotation(example_idx) {
     const options = annotation_set[example_idx].options;
     const textFields = annotation_set[example_idx].textFields;
 
+    annotation_set[example_idx]["lastAccessedTime"] = Math.floor(Date.now() / 1000);
+
     clearExampleLevelFields();
 
     if (flags !== undefined) {
@@ -267,6 +269,7 @@ function saveCurrentAnnotations(example_idx) {
     annotation_set[example_idx]["flags"] = collectFlags();
     annotation_set[example_idx]["options"] = collectOptions();
     annotation_set[example_idx]["textFields"] = collectTextFields();
+    annotation_set[example_idx]["lastSavedTime"] = Math.floor(Date.now() / 1000);
 }
 
 function submitAnnotations(campaign_id) {

diff --git a/factgenie/static/js/span-annotator.js b/factgenie/static/js/span-annotator.js
@@ -125,17 +125,17 @@ class SpanAnnotator {
 
         $element.on('mousedown', (e) => {
             this.isSelecting = true;
-            this.startSpan = this._findClosestSpan(objectId, e.pageX, e.pageY);
+            this.startSpan = this._findClosestSpan(objectId, e.clientX, e.clientY);
         });
 
         $element.on('mousemove', (e) => {
             if (this.isSelecting) {
-                const closestSpan = this.throttledFindClosestSpan(objectId, e.pageX, e.pageY);
+                const closestSpan = this.throttledFindClosestSpan(objectId, e.clientX, e.clientY);
                 if (closestSpan) {
                     this._updateHighlight(objectId, this.startSpan, closestSpan);
                 }
             } else if (this.currentType === -1) {
-                const closestSpan = this._findClosestSpan(objectId, e.pageX, e.pageY);
+                const closestSpan = this._findClosestSpan(objectId, e.clientX, e.clientY);
                 if (closestSpan) {
                     this._previewEraserEffect(objectId, closestSpan);
                 }
@@ -146,7 +146,7 @@ class SpanAnnotator {
         $element.on('mouseup', (e) => {
             if (!this.isSelecting) return;
             this.isSelecting = false;
-            const endSpan = this._findClosestSpan(objectId, e.pageX, e.pageY);
+            const endSpan = this._findClosestSpan(objectId, e.clientX, e.clientY);
 
             if (this.startSpan && endSpan) {
                 if (this.currentType === -1) {
@@ -240,9 +240,9 @@ class SpanAnnotator {
 
         $('.annotatable', doc.element).each((_, span) => {
             const $span = $(span);
-            const offset = $span.offset();
-            const left = offset.left;
-            const top = offset.top;
+            const rect = span.getBoundingClientRect();
+            const left = rect.left;
+            const top = rect.top;
             const right = left + $span.width();
             const bottom = top + $span.height();
 

diff --git a/factgenie/workflows.py b/factgenie/workflows.py
@@ -932,7 +932,10 @@ def save_record(mode, campaign, row, result):
         last_run = campaign.metadata.get("last_run", int(time.time()))
         filename = f"{dataset_id}-{split}-{setup_id}-{last_run}.jsonl"
     elif mode == CampaignMode.CROWDSOURCING:
-        filename = f"{dataset_id}-{split}-{setup_id}-{annotator_id}.jsonl"
+        batch_idx = row["batch_idx"]
+        annotator_group = row.get("annotator_group", 0)
+        batch_end = int(row["end"])
+        filename = f"{batch_idx}-{annotator_group}-{annotator_id}-{batch_end}.jsonl"
     elif mode == CampaignMode.LLM_GEN:
         last_run = campaign.metadata.get("last_run", int(time.time()))
         filename = f"{dataset_id}-{split}-{last_run}.jsonl"