vodorod/Dorod-Sky
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

add totp_code_required info to the InputTextAction json (#3415)

Browse Source
This commit is contained in:
Shuchang Zheng
2025-09-11 19:05:25 -07:00
committed by GitHub
parent 0e2aecc75d
commit f1aa653b82
8 changed files with 227 additions and 120 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
skyvern/webeye/actions/parse_actions.py
View File

@@ -7,9 +7,11 @@ from openai.types.responses.response import Response as OpenAIResponse
from pydantic import ValidationError
from skyvern.constants import SCROLL_AMOUNT_MULTIPLIER
from skyvern.core.totp import poll_verification_code
from skyvern.exceptions import NoTOTPVerificationCodeFound, UnsupportedActionType
from skyvern.forge import app
from skyvern.forge.prompts import prompt_engine
from skyvern.forge.sdk.core import skyvern_context
from skyvern.forge.sdk.models import Step
from skyvern.forge.sdk.schemas.tasks import Task
from skyvern.utils.image_resizer import Resolution, scale_coordinates
@@ -36,13 +38,17 @@ from skyvern.webeye.actions.actions import (
VerificationCodeAction,
WaitAction,
)
from skyvern.webeye.actions.handler import poll_verification_code
from skyvern.webeye.scraper.scraper import ScrapedPage
LOG = structlog.get_logger()
def parse_action(action: Dict[str, Any], scraped_page: ScrapedPage, data_extraction_goal: str | None = None) -> Action:
def parse_action(
action: Dict[str, Any],
scraped_page: ScrapedPage,
data_extraction_goal: str | None = None,
totp_code_required: bool = False,
) -> Action:
if "id" in action:
element_id = action["id"]
elif "element_id" in action:
@@ -95,7 +101,12 @@ def parse_action(action: Dict[str, Any], scraped_page: ScrapedPage, data_extract
if context_dict and len(context_dict) > 0:
context_dict["intention"] = intention
input_or_select_context = InputOrSelectContext.model_validate(context_dict)
return InputTextAction(**base_action_dict, text=action["text"], input_or_select_context=input_or_select_context)
return InputTextAction(
**base_action_dict,
text=action["text"],
input_or_select_context=input_or_select_context,
totp_code_required=totp_code_required,
)
if action_type == ActionType.UPLOAD_FILE:
# TODO: see if the element is a file input element. if it's not, convert this action into a click action
@@ -162,10 +173,16 @@ def parse_actions(
task: Task, step_id: str, step_order: int, scraped_page: ScrapedPage, json_response: list[Dict[str, Any]]
) -> list[Action]:
actions: list[Action] = []
context = skyvern_context.ensure_context()
totp_code = context.totp_codes.get(task.task_id)
totp_code_required = bool(totp_code)
for idx, action in enumerate(json_response):
try:
action_instance = parse_action(
action=action, scraped_page=scraped_page, data_extraction_goal=task.data_extraction_goal
action=action,
scraped_page=scraped_page,
data_extraction_goal=task.data_extraction_goal,
totp_code_required=totp_code_required,
)
action_instance.organization_id = task.organization_id
action_instance.workflow_run_id = task.workflow_run_id
@@ -793,8 +810,8 @@ async def generate_cua_fallback_actions(
)
try:
verification_code = await poll_verification_code(
task.task_id,
task.organization_id,
organization_id=task.organization_id,
task_id=task.task_id,
workflow_run_id=task.workflow_run_id,
totp_verification_url=task.totp_verification_url,
totp_identifier=task.totp_identifier,