Merge pull request #101 from SentienceAPI/missing_fields2

trace should include snapshot elements; more fixes to seq_num, etc

Author: rcholic

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "sentienceapi"
-version = "0.91.0"
+version = "0.91.1"
 description = "Python SDK for Sentience AI Agent Browser Automation"
 readme = "README.md"
 requires-python = ">=3.11"

sentience/__init__.py

@@ -70,7 +70,7 @@
 )
 from .wait import wait_for
 
-__version__ = "0.91.0"
+__version__ = "0.91.1"
 
 __all__ = [
     # Core SDK

sentience/agent.py

@@ -198,29 +198,16 @@ def act(  # noqa: C901
 
                 # Emit snapshot trace event if tracer is enabled
                 if self.tracer:
-                    # Include element data for live overlay visualization
-                    # Use filtered_elements for overlay (only relevant elements)
-                    elements_data = [
-                        {
-                            "id": el.id,
-                            "bbox": {
-                                "x": el.bbox.x,
-                                "y": el.bbox.y,
-                                "width": el.bbox.width,
-                                "height": el.bbox.height,
-                            },
-                            "role": el.role,
-                            "text": el.text[:50] if el.text else "",  # Truncate for brevity
-                        }
-                        for el in filtered_elements[:50]  # Limit to first 50 for performance
-                    ]
+                    # Include ALL elements with full data for DOM tree display
+                    # Use snap.elements (all elements) not filtered_elements
+                    elements_data = [el.model_dump() for el in snap.elements]
 
                     # Build snapshot event data
                     snapshot_data = {
                         "url": snap.url,
                         "element_count": len(snap.elements),
                         "timestamp": snap.timestamp,
-                        "elements": elements_data,  # Add element data for overlay
+                        "elements": elements_data,  # Full element data for DOM tree
                     }
 
                     # Always include screenshot in trace event for studio viewer compatibility
@@ -983,29 +970,16 @@ async def act(  # noqa: C901
 
                 # Emit snapshot trace event if tracer is enabled
                 if self.tracer:
-                    # Include element data for live overlay visualization
-                    # Use filtered_elements for overlay (only relevant elements)
-                    elements_data = [
-                        {
-                            "id": el.id,
-                            "bbox": {
-                                "x": el.bbox.x,
-                                "y": el.bbox.y,
-                                "width": el.bbox.width,
-                                "height": el.bbox.height,
-                            },
-                            "role": el.role,
-                            "text": el.text[:50] if el.text else "",  # Truncate for brevity
-                        }
-                        for el in filtered_elements[:50]  # Limit to first 50 for performance
-                    ]
+                    # Include ALL elements with full data for DOM tree display
+                    # Use snap.elements (all elements) not filtered_elements
+                    elements_data = [el.model_dump() for el in snap.elements]
 
                     # Build snapshot event data
                     snapshot_data = {
                         "url": snap.url,
                         "element_count": len(snap.elements),
                         "timestamp": snap.timestamp,
-                        "elements": elements_data,  # Add element data for overlay
+                        "elements": elements_data,  # Full element data for DOM tree
                     }
 
                     # Always include screenshot in trace event for studio viewer compatibility

sentience/cloud_tracing.py

@@ -270,7 +270,10 @@ def _generate_index(self) -> None:
         try:
             from .trace_indexing import write_trace_index
 
-            write_trace_index(str(self._path))
+            # Use frontend format to ensure 'step' field is present (1-based)
+            # Frontend derives sequence from step.step - 1, so step must be valid
+            index_path = Path(str(self._path).replace(".jsonl", ".index.json"))
+            write_trace_index(str(self._path), str(index_path), frontend_format=True)
         except Exception as e:
             # Non-fatal: log but don't crash
             print(f"⚠️  Failed to generate trace index: {e}")
@@ -322,10 +325,28 @@ def _upload_index(self) -> None:
                     self.logger.warning("No upload URL in index upload response")
                 return
 
-            # Read and compress index file
-            with open(index_path, "rb") as f:
-                index_data = f.read()
+            # Read index file and update trace_file.path to cloud storage path
+            with open(index_path, encoding="utf-8") as f:
+                index_json = json.load(f)
 
+            # Extract cloud storage path from trace upload URL
+            # upload_url format: https://...digitaloceanspaces.com/traces/{run_id}.jsonl.gz
+            # Extract path: traces/{run_id}.jsonl.gz
+            try:
+                from urllib.parse import urlparse
+
+                parsed_url = urlparse(self.upload_url)
+                # Extract path after domain (e.g., /traces/run-123.jsonl.gz -> traces/run-123.jsonl.gz)
+                cloud_trace_path = parsed_url.path.lstrip("/")
+                # Update trace_file.path in index
+                if "trace_file" in index_json and isinstance(index_json["trace_file"], dict):
+                    index_json["trace_file"]["path"] = cloud_trace_path
+            except Exception as e:
+                if self.logger:
+                    self.logger.warning(f"Failed to extract cloud path from upload URL: {e}")
+
+            # Serialize updated index to JSON
+            index_data = json.dumps(index_json, indent=2).encode("utf-8")
             compressed_index = gzip.compress(index_data)
             index_size = len(compressed_index)
             self.index_file_size_bytes = index_size  # Track index file size

sentience/trace_indexing/indexer.py

@@ -58,13 +58,26 @@ def _compute_snapshot_digest(snapshot_data: dict[str, Any]) -> str:
     # Canonicalize elements
     canonical_elements = []
     for elem in elements:
+        # Extract is_primary and is_clickable from visual_cues if present
+        visual_cues = elem.get("visual_cues", {})
+        is_primary = (
+            visual_cues.get("is_primary", False)
+            if isinstance(visual_cues, dict)
+            else elem.get("is_primary", False)
+        )
+        is_clickable = (
+            visual_cues.get("is_clickable", False)
+            if isinstance(visual_cues, dict)
+            else elem.get("is_clickable", False)
+        )
+
         canonical_elem = {
             "id": elem.get("id"),
             "role": elem.get("role", ""),
             "text_norm": _normalize_text(elem.get("text")),
             "bbox": _round_bbox(elem.get("bbox", {"x": 0, "y": 0, "width": 0, "height": 0})),
-            "is_primary": elem.get("is_primary", False),
-            "is_clickable": elem.get("is_clickable", False),
+            "is_primary": is_primary,
+            "is_clickable": is_clickable,
         }
         canonical_elements.append(canonical_elem)
 

sentience/tracing.py

@@ -230,7 +230,10 @@ def _generate_index(self) -> None:
         try:
             from .trace_indexing import write_trace_index
 
-            write_trace_index(str(self.path))
+            # Use frontend format to ensure 'step' field is present (1-based)
+            # Frontend derives sequence from step.step - 1, so step must be valid
+            index_path = Path(self.path).with_suffix(".index.json")
+            write_trace_index(str(self.path), str(index_path), frontend_format=True)
         except Exception as e:
             # Non-fatal: log but don't crash
             print(f"⚠️  Failed to generate trace index: {e}")