Automate browser based workflows with AI
Python aiapiautomationbrowserbrowser-automationcomputergptllmplaywrightpowerautomatepuppeteerpythonrpaseleniumvisionworkflow
Jinja interpolation support for natural-language branch criteria (#4325)
Celal Zamanoglub8dc832eea4d
diff --git a/skyvern/forge/sdk/workflow/models/block.py b/skyvern/forge/sdk/workflow/models/block.pyindex 75cda342f..a78a41577 100644--- a/skyvern/forge/sdk/workflow/models/block.py+++ b/skyvern/forge/sdk/workflow/models/block.py@@ -4510,9 +4510,41 @@ class ConditionalBlock(Block):raise ValueError("organization_id is required to evaluate natural language branches")workflow_run_context = evaluation_context.workflow_run_context+ template_data = evaluation_context.build_template_data()++ rendered_branch_criteria: list[dict[str, Any]] = []+ for idx, branch in enumerate(branches):+ expression = branch.criteria.expression if branch.criteria else ""+ rendered_expression = expression++ # Allow Jinja templating inside natural language branch expressions so users can+ # mix free text with dynamic values (e.g., "If response: {{ foo.bar }}").+ if "{{" in expression and "}}" in expression:+ try:+ template = jinja_sandbox_env.from_string(expression)+ except Exception as exc:+ raise FailedToFormatJinjaStyleParameter(+ template=expression,+ msg=str(exc),+ ) from exc++ if settings.WORKFLOW_TEMPLATING_STRICTNESS == "strict":+ if missing := get_missing_variables(expression, template_data):+ raise MissingJinjaVariables(template=expression, variables=missing)++ try:+ rendered_expression = template.render(template_data)+ except Exception as exc:+ raise FailedToFormatJinjaStyleParameter(+ template=expression,+ msg=str(exc),+ ) from exc++ rendered_branch_criteria.append({"index": idx, "expression": rendered_expression})+branch_criteria_payload = [- {"index": idx, "expression": branch.criteria.expression if branch.criteria else ""}- for idx, branch in enumerate(branches)+ {"index": criterion["index"], "expression": criterion["expression"]}+ for criterion in rendered_branch_criteria]extraction_goal = prompt_engine.load_prompt(