remove unused libraries (#1161)
This commit is contained in:
Shuchang Zheng
@@ -1,7 +0,0 @@
|
||||
|
||||
#!/bin/bash
|
||||
|
||||
set -e
|
||||
|
||||
# Run the command and pass in all three arguments
|
||||
streamlit run streamlit_app/visualizer/streamlit.py --server.port 8501
|
||||
1561
poetry.lock
generated
1561
poetry.lock
generated
File diff suppressed because it is too large
Load Diff
@@ -10,20 +10,14 @@ packages = [{ include = "skyvern" }]
|
||||
python = "^3.11,<3.12"
|
||||
python-dotenv = "^1.0.0"
|
||||
openai = "<2.0"
|
||||
tenacity = "^8.2.2"
|
||||
sqlalchemy = {extras = ["mypy"], version = "^2.0.29"}
|
||||
aiohttp = "^3.8.5"
|
||||
colorlog = "^6.7.0"
|
||||
chromadb = "^0.4.10"
|
||||
python-multipart = "^0.0.6"
|
||||
toml = "^0.10.2"
|
||||
jinja2 = "^3.1.2"
|
||||
uvicorn = {extras = ["standard"], version = "^0.24.0.post1"}
|
||||
litellm = "1.42.0"
|
||||
duckduckgo-search = "^3.8.0"
|
||||
selenium = "^4.13.0"
|
||||
bs4 = "^0.0.1"
|
||||
webdriver-manager = "^4.0.1"
|
||||
playwright = "1.46.0"
|
||||
pre-commit = "^3.5.0"
|
||||
pillow = "^10.1.0"
|
||||
@@ -41,25 +35,21 @@ commentjson = "^0.9.0"
|
||||
asyncache = "^0.3.1"
|
||||
orjson = "^3.9.10"
|
||||
structlog = "^23.2.0"
|
||||
plotly = "^5.18.0"
|
||||
clipboard = "^0.0.4"
|
||||
curlify = "^2.2.1"
|
||||
typer = "^0.9.0"
|
||||
types-toml = "^0.10.8.7"
|
||||
httpx = "^0.27.0"
|
||||
filetype = "^1.2.0"
|
||||
redis = "^5.0.3"
|
||||
onnxruntime = "<1.17"
|
||||
supabase = "^2.4.3"
|
||||
aioredlock = "^0.7.3"
|
||||
fpdf = "^1.7.2"
|
||||
pypdf = "^4.2.0"
|
||||
stripe = "^9.7.0"
|
||||
tldextract = "^5.1.2"
|
||||
websockets = "^12.0"
|
||||
email-validator = "^2.2.0"
|
||||
temporalio = "^1.6.0"
|
||||
requests-toolbelt = "^1.0.0"
|
||||
posthog = "^3.7.0"
|
||||
aiofiles = "^24.1.0"
|
||||
|
||||
[tool.poetry.group.dev.dependencies]
|
||||
isort = "^5.13.2"
|
||||
@@ -74,14 +64,14 @@ mock = "^5.1.0"
|
||||
autoflake = "^2.2.0"
|
||||
pydevd-pycharm = "^233.6745.319"
|
||||
ipython = "^8.17.2"
|
||||
streamlit = "^1.28.1"
|
||||
typer = "^0.9.0"
|
||||
ipykernel = "^6.26.0"
|
||||
notebook = "^7.0.6"
|
||||
freezegun = "^1.2.2"
|
||||
snoop = "^0.4.3"
|
||||
rich = {extras = ["jupyter"], version = "^13.7.0"}
|
||||
clipboard = "^0.0.4"
|
||||
fpdf = "^1.7.2"
|
||||
pypdf = "^5.0.1"
|
||||
|
||||
|
||||
[build-system]
|
||||
|
||||
@@ -2,7 +2,6 @@ from enum import StrEnum
|
||||
from typing import Annotated, Any, Dict, Type, TypeVar
|
||||
|
||||
import structlog
|
||||
from deprecation import deprecated
|
||||
from litellm import ConfigDict
|
||||
from pydantic import BaseModel, Field, ValidationError
|
||||
|
||||
@@ -177,7 +176,7 @@ class UploadFileAction(WebAction):
|
||||
return f"UploadFileAction(element_id={self.element_id}, file={self.file_url}, is_upload_file_tag={self.is_upload_file_tag})"
|
||||
|
||||
|
||||
@deprecated("This action is not used in the current implementation. Click actions are used instead.")
|
||||
# this is a deprecated action type
|
||||
class DownloadFileAction(WebAction):
|
||||
action_type: ActionType = ActionType.DOWNLOAD_FILE
|
||||
file_name: str
|
||||
|
||||
@@ -8,7 +8,6 @@ from datetime import datetime, timedelta, timezone
|
||||
from typing import Any, Awaitable, Callable, List
|
||||
|
||||
import structlog
|
||||
from deprecation import deprecated
|
||||
from playwright.async_api import FileChooser, Frame, Locator, Page, TimeoutError
|
||||
from pydantic import BaseModel
|
||||
|
||||
@@ -641,7 +640,7 @@ async def handle_upload_file_action(
|
||||
)
|
||||
|
||||
|
||||
@deprecated("This function is deprecated. Downloads are handled by the click action handler now.")
|
||||
# This function is deprecated. Downloads are handled by the click action handler now.
|
||||
async def handle_download_file_action(
|
||||
action: actions.DownloadFileAction,
|
||||
page: Page,
|
||||
|
||||
@@ -1,100 +0,0 @@
|
||||
import json
|
||||
from typing import Any
|
||||
|
||||
import curlify
|
||||
import requests
|
||||
import structlog
|
||||
from requests import PreparedRequest
|
||||
|
||||
from skyvern.forge.sdk.schemas.tasks import TaskRequest
|
||||
|
||||
LOG = structlog.get_logger()
|
||||
|
||||
|
||||
class SkyvernClient:
|
||||
def __init__(self, base_url: str, credentials: str):
|
||||
self.base_url = base_url
|
||||
self.credentials = credentials
|
||||
|
||||
def generate_curl_params(self, task_request_body: TaskRequest, max_steps: int | None = None) -> PreparedRequest:
|
||||
url = f"{self.base_url}/tasks"
|
||||
payload = task_request_body.model_dump()
|
||||
headers = {
|
||||
"Content-Type": "application/json",
|
||||
"x-api-key": self.credentials,
|
||||
}
|
||||
if max_steps is not None:
|
||||
headers["x-max-steps-override"] = str(max_steps)
|
||||
|
||||
return url, payload, headers
|
||||
|
||||
def create_task(self, task_request_body: TaskRequest, max_steps: int | None = None) -> str | None:
|
||||
url, payload, headers = self.generate_curl_params(task_request_body, max_steps=max_steps)
|
||||
|
||||
response = requests.post(url, headers=headers, data=json.dumps(payload))
|
||||
if "task_id" not in response.json():
|
||||
LOG.info(f"Failed to create task: {response.text}")
|
||||
return None
|
||||
return response.json()["task_id"]
|
||||
|
||||
def copy_curl(self, task_request_body: TaskRequest) -> str:
|
||||
url, payload, headers = self.generate_curl_params(task_request_body)
|
||||
|
||||
req = requests.Request("POST", url, headers=headers, data=json.dumps(payload, indent=4))
|
||||
|
||||
return curlify.to_curl(req.prepare())
|
||||
|
||||
def get_task(self, task_id: str) -> dict[str, Any] | None:
|
||||
"""Get a task by id."""
|
||||
url = f"{self.base_url}/internal/tasks/{task_id}"
|
||||
headers = {"x-api-key": self.credentials}
|
||||
response = requests.get(url, headers=headers)
|
||||
if response.status_code != 200:
|
||||
return None
|
||||
return response.json()
|
||||
|
||||
def get_agent_tasks(self, page: int = 1, page_size: int = 15) -> dict[str, Any]:
|
||||
"""Get all tasks with pagination."""
|
||||
url = f"{self.base_url}/internal/tasks"
|
||||
params = {"page": page, "page_size": page_size}
|
||||
headers = {"x-api-key": self.credentials}
|
||||
response = requests.get(url, params=params, headers=headers)
|
||||
return response.json()
|
||||
|
||||
def get_agent_task_steps(self, task_id: str, page: int = 1, page_size: int = 15) -> list[dict[str, Any]]:
|
||||
"""Get all steps for a task with pagination."""
|
||||
url = f"{self.base_url}/tasks/{task_id}/steps"
|
||||
params = {"page": page, "page_size": page_size}
|
||||
headers = {"x-api-key": self.credentials}
|
||||
response = requests.get(url, params=params, headers=headers)
|
||||
steps = response.json()
|
||||
for step in steps:
|
||||
if (
|
||||
"output" in step
|
||||
and step["output"] is not None
|
||||
and "actions_and_results" in step["output"]
|
||||
and step["output"]["actions_and_results"] is not None
|
||||
):
|
||||
step["output"]["actions_and_results"] = json.dumps(step["output"]["actions_and_results"])
|
||||
return steps
|
||||
|
||||
def get_agent_task_video_artifact(self, task_id: str) -> dict[str, Any] | None:
|
||||
"""Get the video artifact from the first step artifact of the task."""
|
||||
steps = self.get_agent_task_steps(task_id)
|
||||
if not steps:
|
||||
return None
|
||||
|
||||
first_step_id = steps[0]["step_id"]
|
||||
artifacts = self.get_agent_artifacts(task_id, first_step_id)
|
||||
for artifact in artifacts:
|
||||
if artifact["artifact_type"] == "recording":
|
||||
return artifact
|
||||
|
||||
return None
|
||||
|
||||
def get_agent_artifacts(self, task_id: str, step_id: str) -> list[dict[str, Any]]:
|
||||
"""Get all artifacts for a list of steps."""
|
||||
url = f"{self.base_url}/tasks/{task_id}/steps/{step_id}/artifacts"
|
||||
headers = {"x-api-key": self.credentials}
|
||||
response = requests.get(url, headers=headers)
|
||||
return response.json()
|
||||
@@ -1,81 +0,0 @@
|
||||
import asyncio
|
||||
import random
|
||||
import string
|
||||
import typing
|
||||
from typing import Any, Callable
|
||||
|
||||
from PIL import Image
|
||||
|
||||
from skyvern.forge.sdk.api.aws import AsyncAWSClient
|
||||
|
||||
async_s3_client = AsyncAWSClient()
|
||||
|
||||
|
||||
def read_artifact(uri: str, is_image: bool = False, is_webm: bool = False) -> Image.Image | str | bytes:
|
||||
"""Load and display an artifact based on its URI."""
|
||||
if uri.startswith("s3://"):
|
||||
downloaded_bytes = asyncio.run(async_s3_client.download_file(uri))
|
||||
if is_image:
|
||||
return downloaded_bytes
|
||||
elif is_webm:
|
||||
return downloaded_bytes
|
||||
else:
|
||||
return downloaded_bytes.decode("utf-8")
|
||||
elif uri.startswith("file://"):
|
||||
# Remove file:// prefix
|
||||
uri = uri[7:]
|
||||
# Means it's a local file
|
||||
if is_image:
|
||||
with open(uri, "rb") as f:
|
||||
image = Image.open(f)
|
||||
image.load()
|
||||
return image
|
||||
elif is_webm:
|
||||
with open(uri, "rb") as f:
|
||||
return f.read()
|
||||
else:
|
||||
with open(uri, "r") as f:
|
||||
return f.read()
|
||||
else:
|
||||
raise ValueError(f"Unsupported URI: {uri}")
|
||||
|
||||
|
||||
def read_artifact_safe(uri: str, is_image: bool = False, is_webm: bool = False) -> Image.Image | str | bytes:
|
||||
"""Load and display an artifact based on its URI."""
|
||||
try:
|
||||
return read_artifact(uri, is_image, is_webm)
|
||||
except Exception as e:
|
||||
return f"Failed to load artifact: {e}"
|
||||
|
||||
|
||||
def streamlit_content_safe(st_obj: Any, f: Callable, content: bytes, message: str, **kwargs: dict[str, Any]) -> None:
|
||||
try:
|
||||
if content:
|
||||
f(content, **kwargs)
|
||||
else:
|
||||
st_obj.write(message)
|
||||
except Exception:
|
||||
st_obj.write(message)
|
||||
|
||||
|
||||
@typing.no_type_check
|
||||
def streamlit_show_recording(st_obj: Any, uri: str) -> None:
|
||||
# ignoring type because is_webm will return bytes
|
||||
content = read_artifact_safe(uri, is_webm=True) # type: ignore
|
||||
if content:
|
||||
random_key = "".join(random.choices(string.ascii_uppercase + string.digits, k=6))
|
||||
st_obj.download_button(
|
||||
"Download recording",
|
||||
content,
|
||||
f"recording{uri.split('/')[-1]}.webm",
|
||||
key=random_key,
|
||||
)
|
||||
|
||||
streamlit_content_safe(
|
||||
st_obj,
|
||||
st_obj.video,
|
||||
content,
|
||||
"No recording available.",
|
||||
format="video/webm",
|
||||
start_time=0,
|
||||
)
|
||||
@@ -1,30 +0,0 @@
|
||||
from typing import Any, Optional
|
||||
|
||||
from streamlit_app.visualizer.api import SkyvernClient
|
||||
|
||||
|
||||
class TaskRepository:
|
||||
def __init__(self, client: SkyvernClient):
|
||||
self.client = client
|
||||
|
||||
def get_task(self, task_id: str) -> dict[str, Any] | None:
|
||||
return self.client.get_task(task_id)
|
||||
|
||||
def get_tasks(self, page: int = 1, page_size: int = 15) -> dict[str, Any]:
|
||||
"""Get tasks with pagination."""
|
||||
return self.client.get_agent_tasks(page=page, page_size=page_size)
|
||||
|
||||
def get_task_steps(self, task_id: str) -> list[dict[str, Any]]:
|
||||
"""Get steps for a specific task with pagination."""
|
||||
return self.client.get_agent_task_steps(task_id)
|
||||
|
||||
def get_artifacts(self, task_id: str, step_id: str) -> list[dict[str, Any]]:
|
||||
"""Get artifacts for a specific task and steps."""
|
||||
return self.client.get_agent_artifacts(task_id, step_id)
|
||||
|
||||
def get_task_recording_uri(self, task: dict[str, Any]) -> Optional[str]:
|
||||
"""Get the recording URI for a task."""
|
||||
video_artifact = self.client.get_agent_task_video_artifact(task["task_id"])
|
||||
if video_artifact is None:
|
||||
return None
|
||||
return video_artifact["uri"]
|
||||
@@ -1,242 +0,0 @@
|
||||
from skyvern.forge.sdk.schemas.tasks import TaskRequest
|
||||
|
||||
|
||||
class SampleTaskRequest(TaskRequest):
|
||||
name: str
|
||||
|
||||
|
||||
bci_seguros_sample_data = SampleTaskRequest(
|
||||
name="bci_seguros",
|
||||
url="https://www.bciseguros.cl/nuestros_seguros/personas/seguro-automotriz/",
|
||||
navigation_goal="Generate an auto insurance quote. A quote has been generated when there's a table of coverages shown on the website.",
|
||||
data_extraction_goal="Extract ALL quote information in JSON format, with one entry per plan visible on the page. The output should include: the selected UF coverage value (3), auto plan name, the online price",
|
||||
navigation_payload={
|
||||
"Rut": "7.250.199-3",
|
||||
"Sexo": "Masculino",
|
||||
"Fecha de Nacimiento": "03-02-2000",
|
||||
"Telefono": "96908116",
|
||||
"Comuna": "Lo Barnachea",
|
||||
"e-mail": "notarealemail@gmail.com",
|
||||
"estado": "Usado",
|
||||
"patente": "HZZV68",
|
||||
"marca": "Subaru",
|
||||
"modelo": "XV",
|
||||
"ano": "2016",
|
||||
"tipo de combustible": "Bencina",
|
||||
"km approx a recorrer": "28,000",
|
||||
},
|
||||
)
|
||||
|
||||
|
||||
california_edd_sample_data = SampleTaskRequest(
|
||||
name="California_EDD",
|
||||
url="https://eddservices.edd.ca.gov/acctservices/AccountManagement/AccountServlet?Command=NEW_SIGN_UP",
|
||||
navigation_goal="Navigate through the employer services online enrollment form. Terminate when the form is completed",
|
||||
navigation_payload={
|
||||
"username": "isthisreal1",
|
||||
"password": "Password123!",
|
||||
"first_name": "John",
|
||||
"last_name": "Doe",
|
||||
"pin": "1234",
|
||||
"email": "isthisreal1@gmail.com",
|
||||
"phone_number": "412-444-1234",
|
||||
},
|
||||
)
|
||||
|
||||
finditparts_sample_data = SampleTaskRequest(
|
||||
name="Finditparts",
|
||||
url="https://www.finditparts.com",
|
||||
navigation_goal="Search for the specified product id, add it to cart and then navigate to the cart page",
|
||||
data_extraction_goal="Extract all product quantity information from the cart page",
|
||||
navigation_payload={
|
||||
"product_id": "W01-377-8537",
|
||||
},
|
||||
)
|
||||
|
||||
|
||||
geico_sample_data = SampleTaskRequest(
|
||||
name="Geico",
|
||||
url="https://www.geico.com",
|
||||
navigation_goal="Navigate through the website until you generate an auto insurance quote. Do not generate a home insurance quote. If you're on a page showing an auto insurance quote (with premium amounts), your goal is COMPLETE.",
|
||||
data_extraction_goal="Extract all quote information in JSON format including the premium amount, the timeframe for the quote.",
|
||||
navigation_payload={
|
||||
"licensed_at_age": 19,
|
||||
"education_level": "HIGH_SCHOOL",
|
||||
"phone_number": "8042221111",
|
||||
"full_name": "Chris P. Bacon",
|
||||
"past_claim": [],
|
||||
"has_claims": False,
|
||||
"spouse_occupation": "Florist",
|
||||
"auto_current_carrier": "None",
|
||||
"home_commercial_uses": None,
|
||||
"spouse_full_name": "Amy Stake",
|
||||
"auto_commercial_uses": None,
|
||||
"requires_sr22": False,
|
||||
"previous_address_move_date": None,
|
||||
"line_of_work": None,
|
||||
"spouse_age": "1987-12-12",
|
||||
"auto_insurance_deadline": None,
|
||||
"email": "chris.p.bacon@abc.com",
|
||||
"net_worth_numeric": 1000000,
|
||||
"spouse_gender": "F",
|
||||
"marital_status": "married",
|
||||
"spouse_licensed_at_age": 20,
|
||||
"license_number": "AAAAAAA090AA",
|
||||
"spouse_license_number": "AAAAAAA080AA",
|
||||
"how_much_can_you_lose": 25000,
|
||||
"vehicles": [
|
||||
{
|
||||
"annual_mileage": 10000,
|
||||
"commute_mileage": 4000,
|
||||
"existing_coverages": None,
|
||||
"ideal_coverages": {
|
||||
"bodily_injury_per_incident_limit": 50000,
|
||||
"bodily_injury_per_person_limit": 25000,
|
||||
"collision_deductible": 1000,
|
||||
"comprehensive_deductible": 1000,
|
||||
"personal_injury_protection": None,
|
||||
"property_damage_per_incident_limit": None,
|
||||
"property_damage_per_person_limit": 25000,
|
||||
"rental_reimbursement_per_incident_limit": None,
|
||||
"rental_reimbursement_per_person_limit": None,
|
||||
"roadside_assistance_limit": None,
|
||||
"underinsured_motorist_bodily_injury_per_incident_limit": 50000,
|
||||
"underinsured_motorist_bodily_injury_per_person_limit": 25000,
|
||||
"underinsured_motorist_property_limit": None,
|
||||
},
|
||||
"ownership": "Owned",
|
||||
"parked": "Garage",
|
||||
"purpose": "commute",
|
||||
"vehicle": {
|
||||
"style": "AWD 3.0 quattro TDI 4dr Sedan",
|
||||
"model": "A8 L",
|
||||
"price_estimate": 29084,
|
||||
"year": 2015,
|
||||
"make": "Audi",
|
||||
},
|
||||
"vehicle_id": None,
|
||||
"vin": None,
|
||||
}
|
||||
],
|
||||
"additional_drivers": [],
|
||||
"home": [
|
||||
{
|
||||
"home_ownership": "owned",
|
||||
}
|
||||
],
|
||||
"spouse_line_of_work": "Agriculture, Forestry and Fishing",
|
||||
"occupation": "Customer Service Representative",
|
||||
"id": None,
|
||||
"gender": "M",
|
||||
"credit_check_authorized": False,
|
||||
"age": "1987-11-11",
|
||||
"license_state": "Washington",
|
||||
"cash_on_hand": "$10000–14999",
|
||||
"address": {
|
||||
"city": "HOUSTON",
|
||||
"country": "US",
|
||||
"state": "TX",
|
||||
"street": "9625 GARFIELD AVE.",
|
||||
"zip": "77082",
|
||||
},
|
||||
"spouse_education_level": "MASTERS",
|
||||
"spouse_email": "amy.stake@abc.com",
|
||||
"spouse_added_to_auto_policy": True,
|
||||
},
|
||||
extracted_information_schema={
|
||||
"additionalProperties": False,
|
||||
"properties": {
|
||||
"quotes": {
|
||||
"items": {
|
||||
"additionalProperties": False,
|
||||
"properties": {
|
||||
"coverages": {
|
||||
"items": {
|
||||
"additionalProperties": False,
|
||||
"properties": {
|
||||
"amount": {
|
||||
"description": "The coverage amount in USD, which can be a single value or a range (e.g., '$300,000' or '$300,000/$300,000').",
|
||||
"type": "string",
|
||||
},
|
||||
"included": {
|
||||
"description": "Indicates whether the coverage is included in the policy (true or False).",
|
||||
"type": "boolean",
|
||||
},
|
||||
"type": {
|
||||
"description": "The limit of the coverage (e.g., 'bodily_injury_limit', 'property_damage_limit', 'underinsured_motorist_bodily_injury_limit').\nTranslate the english name of the coverage to snake case values in the following list:\n * bodily_injury_limit\n * property_damage_limit\n * underinsured_motorist_bodily_injury_limit\n * personal_injury_protection\n * accidental_death\n * work_loss_exclusion\n",
|
||||
"type": "string",
|
||||
},
|
||||
},
|
||||
"type": "object",
|
||||
},
|
||||
"type": "array",
|
||||
},
|
||||
"premium_amount": {
|
||||
"description": "The total premium amount for the whole quote timeframe in USD, formatted as a string (e.g., '$321.57').",
|
||||
"type": "string",
|
||||
},
|
||||
"quote_number": {
|
||||
"description": "The quote number generated by the carrier that identifies this quote",
|
||||
"type": "string",
|
||||
},
|
||||
"timeframe": {
|
||||
"description": "The duration of the coverage, typically expressed in months or years.",
|
||||
"type": "string",
|
||||
},
|
||||
"vehicle_coverages": {
|
||||
"items": {
|
||||
"additionalProperties": False,
|
||||
"properties": {
|
||||
"collision_deductible": {
|
||||
"description": "The collision deductible amount in USD, which is a single value (e.g., '$500') or null if it is not included",
|
||||
"type": "string",
|
||||
},
|
||||
"comprehensive_deductible": {
|
||||
"description": "The collision deductible amount in USD, which is a single value (e.g., '$500') or null if it is not included",
|
||||
"type": "string",
|
||||
},
|
||||
"for_vehicle": {
|
||||
"additionalProperties": False,
|
||||
"description": "The vehicle that the collision and comprehensive coverage is for",
|
||||
"properties": {
|
||||
"make": {
|
||||
"description": "The make of the vehicle",
|
||||
"type": "string",
|
||||
},
|
||||
"model": {
|
||||
"description": "The model of the vehicle",
|
||||
"type": "string",
|
||||
},
|
||||
"year": {
|
||||
"description": "The year of the vehicle",
|
||||
"type": "string",
|
||||
},
|
||||
},
|
||||
"type": "object",
|
||||
},
|
||||
"underinsured_property_damage": {
|
||||
"description": "The underinsured property damage limit for this vehicle, which is a limit and a deductible (e.g., '$25,000/$250 deductible') or null if it is not included",
|
||||
"type": "string",
|
||||
},
|
||||
},
|
||||
"type": "object",
|
||||
},
|
||||
"type": "array",
|
||||
},
|
||||
},
|
||||
"type": "object",
|
||||
},
|
||||
"type": "array",
|
||||
}
|
||||
},
|
||||
"type": "object",
|
||||
},
|
||||
)
|
||||
|
||||
|
||||
supported_examples = [
|
||||
geico_sample_data,
|
||||
finditparts_sample_data,
|
||||
california_edd_sample_data,
|
||||
bci_seguros_sample_data,
|
||||
]
|
||||
@@ -1,497 +0,0 @@
|
||||
import json
|
||||
import sys
|
||||
|
||||
import clipboard
|
||||
import pandas as pd
|
||||
import streamlit as st
|
||||
|
||||
from skyvern import analytics
|
||||
from skyvern.forge.sdk.schemas.tasks import ProxyLocation, TaskRequest
|
||||
from streamlit_app.visualizer import styles
|
||||
from streamlit_app.visualizer.api import SkyvernClient
|
||||
from streamlit_app.visualizer.artifact_loader import (
|
||||
read_artifact_safe,
|
||||
streamlit_content_safe,
|
||||
streamlit_show_recording,
|
||||
)
|
||||
from streamlit_app.visualizer.repository import TaskRepository
|
||||
from streamlit_app.visualizer.sample_data import supported_examples
|
||||
|
||||
analytics.capture("skyvern-oss-run-ui")
|
||||
|
||||
# Streamlit UI Configuration
|
||||
st.set_page_config(layout="wide")
|
||||
|
||||
# Apply styles
|
||||
st.markdown(styles.page_font_style, unsafe_allow_html=True)
|
||||
st.markdown(styles.button_style, unsafe_allow_html=True)
|
||||
|
||||
tab_name = sys.argv[1] if len(sys.argv) > 1 else ""
|
||||
|
||||
|
||||
# Configuration
|
||||
def reset_session_state() -> None:
|
||||
# Delete all the items in Session state when env or org is changed
|
||||
for key in st.session_state.keys():
|
||||
del st.session_state[key]
|
||||
|
||||
|
||||
CONFIGS_DICT = st.secrets["skyvern"]["configs"]
|
||||
if not CONFIGS_DICT:
|
||||
raise Exception("No configuration found. Copy the values from 1P and restart the app.")
|
||||
SETTINGS = {}
|
||||
for config in CONFIGS_DICT:
|
||||
env = config["env"]
|
||||
host = config["host"]
|
||||
orgs = config["orgs"]
|
||||
org_dict = {org["name"]: org["cred"] for org in orgs}
|
||||
SETTINGS[env] = {"host": host, "orgs": org_dict}
|
||||
|
||||
st.sidebar.markdown("#### **Settings**")
|
||||
select_env = st.sidebar.selectbox("Environment", list(SETTINGS.keys()), on_change=reset_session_state)
|
||||
select_org = st.sidebar.selectbox(
|
||||
"Organization",
|
||||
list(SETTINGS[select_env]["orgs"].keys()),
|
||||
on_change=reset_session_state,
|
||||
)
|
||||
|
||||
# Hack the sidebar size to be a little bit smaller
|
||||
st.markdown(
|
||||
"""
|
||||
<style>
|
||||
.sidebar .sidebar-content {
|
||||
width: 375px;
|
||||
}
|
||||
</style>
|
||||
""",
|
||||
unsafe_allow_html=True,
|
||||
)
|
||||
|
||||
# Initialize session state
|
||||
if "client" not in st.session_state:
|
||||
st.session_state.client = SkyvernClient(
|
||||
base_url=SETTINGS[select_env]["host"],
|
||||
credentials=SETTINGS[select_env]["orgs"][select_org],
|
||||
)
|
||||
if "repository" not in st.session_state:
|
||||
st.session_state.repository = TaskRepository(st.session_state.client)
|
||||
if "task_page_number" not in st.session_state:
|
||||
st.session_state.task_page_number = 1
|
||||
if "selected_task" not in st.session_state:
|
||||
st.session_state.selected_task = None
|
||||
st.session_state.selected_task_recording_uri = None
|
||||
st.session_state.task_steps = None
|
||||
if "selected_step" not in st.session_state:
|
||||
st.session_state.selected_step = None
|
||||
st.session_state.selected_step_index = None
|
||||
|
||||
client = st.session_state.client
|
||||
repository = st.session_state.repository
|
||||
task_page_number = st.session_state.task_page_number
|
||||
selected_task = st.session_state.selected_task
|
||||
selected_task_recording_uri = st.session_state.selected_task_recording_uri
|
||||
task_steps = st.session_state.task_steps
|
||||
selected_step = st.session_state.selected_step
|
||||
selected_step_index = st.session_state.selected_step_index
|
||||
|
||||
|
||||
# Onclick handlers
|
||||
def select_task(task: dict) -> None:
|
||||
st.session_state.selected_task = task
|
||||
st.session_state.selected_task_recording_uri = repository.get_task_recording_uri(task)
|
||||
# reset step selection
|
||||
st.session_state.selected_step = None
|
||||
# save task's steps in session state
|
||||
st.session_state.task_steps = repository.get_task_steps(task["task_id"])
|
||||
if st.session_state.task_steps:
|
||||
st.session_state.selected_step = st.session_state.task_steps[0]
|
||||
st.session_state.selected_step_index = 0
|
||||
|
||||
|
||||
def go_to_previous_step() -> None:
|
||||
new_step_index = max(0, selected_step_index - 1)
|
||||
select_step(task_steps[new_step_index])
|
||||
|
||||
|
||||
def go_to_next_step() -> None:
|
||||
new_step_index = min(len(task_steps) - 1, selected_step_index + 1)
|
||||
select_step(task_steps[new_step_index])
|
||||
|
||||
|
||||
def select_step(step: dict) -> None:
|
||||
st.session_state.selected_step = step
|
||||
st.session_state.selected_step_index = task_steps.index(step)
|
||||
|
||||
|
||||
# Streamlit UI Logic
|
||||
st.markdown("# **:dragon: Skyvern :dragon:**")
|
||||
st.markdown(f"### **{select_env} - {select_org}**")
|
||||
execute_tab, visualizer_tab = st.tabs(["Execute", "Visualizer"])
|
||||
|
||||
|
||||
def copy_curl_to_clipboard(task_request_body: TaskRequest) -> None:
|
||||
clipboard.copy(client.copy_curl(task_request_body=task_request_body))
|
||||
|
||||
|
||||
with execute_tab:
|
||||
# Streamlit doesn't support "focusing" on a tab, so this is a workaround to make the requested tab be the "first" tab
|
||||
sorted_supported_examples = sorted(
|
||||
supported_examples,
|
||||
key=lambda x: (-1 if x.name.lower() == tab_name.lower() else 0),
|
||||
)
|
||||
example_tabs = st.tabs([supported_example.name for supported_example in sorted_supported_examples])
|
||||
|
||||
for i, example_tab in enumerate(example_tabs):
|
||||
with example_tab:
|
||||
create_column, explanation_column = st.columns([2, 3])
|
||||
with create_column:
|
||||
run_task, copy_curl = st.columns([3, 1])
|
||||
task_request_body = sorted_supported_examples[i]
|
||||
|
||||
unique_key = f"{task_request_body.name}"
|
||||
copy_curl.button(
|
||||
"Copy cURL",
|
||||
key=f"copy_curl_{unique_key}",
|
||||
on_click=lambda: copy_curl_to_clipboard(task_request_body=task_request_body),
|
||||
)
|
||||
with st.form(f"task_form_{unique_key}"):
|
||||
run_task.markdown("## Run a task")
|
||||
|
||||
example = sorted_supported_examples[i]
|
||||
# Create all the fields to create a TaskRequest object
|
||||
st_url = st.text_input("URL*", value=example.url, key=f"url_{unique_key}")
|
||||
st_webhook_callback_url = st.text_input(
|
||||
"Webhook Callback URL",
|
||||
key=f"webhook_{unique_key}",
|
||||
placeholder="Optional",
|
||||
)
|
||||
st_navigation_goal = st.text_area(
|
||||
"Navigation Goal",
|
||||
key=f"nav_goal_{unique_key}",
|
||||
placeholder="Describe the navigation goal",
|
||||
value=example.navigation_goal,
|
||||
height=120,
|
||||
)
|
||||
st_data_extraction_goal = st.text_area(
|
||||
"Data Extraction Goal",
|
||||
key=f"data_goal_{unique_key}",
|
||||
placeholder="Describe the data extraction goal",
|
||||
value=example.data_extraction_goal,
|
||||
height=120,
|
||||
)
|
||||
st_navigation_payload = st.text_area(
|
||||
"Navigation Payload JSON",
|
||||
key=f"nav_payload_{unique_key}",
|
||||
placeholder='{"name": "John Doe", "email": "abc@123.com"}',
|
||||
value=json.dumps(example.navigation_payload, indent=2),
|
||||
height=200,
|
||||
)
|
||||
st_extracted_information_schema = st.text_area(
|
||||
"Extracted Information Schema",
|
||||
key=f"extracted_info_schema_{unique_key}",
|
||||
placeholder='{"quote_price": "float"}',
|
||||
value=example.extracted_information_schema,
|
||||
)
|
||||
# Create a TaskRequest object from the form fields
|
||||
task_request_body = TaskRequest(
|
||||
url=st_url,
|
||||
webhook_callback_url=st_webhook_callback_url,
|
||||
navigation_goal=st_navigation_goal,
|
||||
data_extraction_goal=st_data_extraction_goal,
|
||||
proxy_location=ProxyLocation.NONE,
|
||||
navigation_payload=st_navigation_payload,
|
||||
extracted_information_schema=st_extracted_information_schema,
|
||||
)
|
||||
# Submit the form
|
||||
if st.form_submit_button("Execute Task", use_container_width=True):
|
||||
# Call the API to create a task
|
||||
task_id = client.create_task(task_request_body)
|
||||
if not task_id:
|
||||
st.error("Failed to create task!")
|
||||
else:
|
||||
st.success("Task created successfully, task_id: " + task_id)
|
||||
|
||||
with explanation_column:
|
||||
st.markdown("### **Task Request**")
|
||||
st.markdown("\n")
|
||||
st.markdown("#### **URL**")
|
||||
st.markdown("The starting URL for the task.")
|
||||
st.markdown("\n")
|
||||
st.markdown("#### **Webhook Callback URL**")
|
||||
st.markdown("The URL to call with the results when the task is completed.")
|
||||
st.markdown("\n")
|
||||
st.markdown("#### **Navigation Goal**")
|
||||
st.markdown("The user's goal for the task. Nullable if the task is only for data extraction.")
|
||||
st.markdown("\n")
|
||||
st.markdown("\n")
|
||||
st.markdown("#### **Data Extraction Goal**")
|
||||
st.markdown("The user's goal for data extraction. Nullable if the task is only for navigation.")
|
||||
st.markdown("\n")
|
||||
st.markdown("\n")
|
||||
st.markdown("#### **Navigation Payload**")
|
||||
st.markdown(
|
||||
"The user's details needed to achieve the task. This is an unstructured field, and information can be passed in in any format you desire. Skyvern will map this information to the questions on the screen in real-time"
|
||||
)
|
||||
st.markdown("\n")
|
||||
st.markdown("\n")
|
||||
st.markdown("\n")
|
||||
st.markdown("#### **Extracted Information Schema**")
|
||||
st.markdown(
|
||||
"(Optional) The requested schema of the extracted information for data extraction goal. This is a JSON object with keys as the field names and values as the data types. The data types can be any of the following: string, number, boolean, date, datetime, time, float, integer, object, array, null. If the schema is not provided, Skyvern will infer the schema from the extracted data."
|
||||
)
|
||||
|
||||
|
||||
with visualizer_tab:
|
||||
task_id_input = st.text_input("task_id", value="")
|
||||
|
||||
def search_task() -> None:
|
||||
if not task_id_input:
|
||||
return
|
||||
|
||||
task = repository.get_task(task_id_input)
|
||||
if task:
|
||||
select_task(task)
|
||||
else:
|
||||
st.error(f"Task with id {task_id_input} not found.")
|
||||
|
||||
st.button("search task", on_click=search_task)
|
||||
|
||||
col_tasks, _, col_steps, _, col_artifacts = st.columns([4, 1, 6, 1, 18])
|
||||
|
||||
col_tasks.markdown("#### Tasks")
|
||||
col_steps.markdown("#### Steps")
|
||||
col_artifacts.markdown("#### Artifacts")
|
||||
tasks_response = repository.get_tasks(task_page_number)
|
||||
if not isinstance(tasks_response, list):
|
||||
st.error("Failed to fetch tasks.")
|
||||
st.error(tasks_response)
|
||||
st.error(
|
||||
"#1 -- Make sure you have both the server (./run_skyvern.sh) and client (./run_ui.sh) running at the same time in different terminals."
|
||||
)
|
||||
st.error(
|
||||
"#2 -- If you're getting a credentials error, Make sure you have the correct organization credentials in .streamlit/secrets.toml."
|
||||
)
|
||||
st.error(
|
||||
"You can validate the credentials against the postgresql credentials by running\n\n"
|
||||
'`psql -U skyvern -h localhost -d skyvern -c "SELECT o.organization_id, o.organization_name, token FROM organizations o JOIN organization_auth_tokens oat ON oat.organization_id = o.organization_id;"`.'
|
||||
"\n\n NOTE: There might be multiple organizations -- each run of ./setup.sh creates a new one. Pick your favourite!"
|
||||
"\n\n If you're running postgres via Docker, please make sure you wrap it in a docker exec command. "
|
||||
"`docker exec postgresql-container psql -U skyvern -h localhost -d skyvern -c 'SELECT o.organization_id, o.organization_name, token FROM organizations o JOIN organization_auth_tokens oat ON oat.organization_id = o.organization_id;'`"
|
||||
)
|
||||
|
||||
else:
|
||||
# Display tasks in sidebar for selection
|
||||
tasks = {task["task_id"]: task for task in tasks_response}
|
||||
task_id_buttons = {
|
||||
task_id: col_tasks.button(
|
||||
f"{task_id}",
|
||||
on_click=select_task,
|
||||
args=(task,),
|
||||
use_container_width=True,
|
||||
type=("primary" if selected_task and task_id == selected_task["task_id"] else "secondary"),
|
||||
)
|
||||
for task_id, task in tasks.items()
|
||||
}
|
||||
|
||||
# Display pagination buttons
|
||||
task_page_prev, _, show_task_page_number, _, task_page_next = col_tasks.columns([1, 1, 1, 1, 1])
|
||||
show_task_page_number.button(str(task_page_number), disabled=True)
|
||||
if task_page_next.button("\>"):
|
||||
st.session_state.task_page_number += 1
|
||||
if task_page_prev.button("\<", disabled=task_page_number == 1):
|
||||
st.session_state.task_page_number = max(1, st.session_state.task_page_number - 1)
|
||||
|
||||
(
|
||||
tab_task,
|
||||
tab_step,
|
||||
tab_recording,
|
||||
tab_screenshot,
|
||||
tab_post_action_screenshot,
|
||||
tab_id_to_xpath,
|
||||
tab_id_to_frame,
|
||||
tab_element_tree,
|
||||
tab_element_tree_trimmed,
|
||||
tab_llm_prompt,
|
||||
tab_llm_request,
|
||||
tab_llm_response_parsed,
|
||||
tab_llm_response_raw,
|
||||
tab_html,
|
||||
) = col_artifacts.tabs(
|
||||
[
|
||||
":green[Task]",
|
||||
":blue[Step]",
|
||||
":violet[Recording]",
|
||||
":rainbow[Screenshot]",
|
||||
":rainbow[Action Screenshots]",
|
||||
":red[ID -> XPath]",
|
||||
":red[ID -> Frame]",
|
||||
":orange[Element Tree]",
|
||||
":blue[Element Tree (Trimmed)]",
|
||||
":yellow[LLM Prompt]",
|
||||
":green[LLM Request]",
|
||||
":blue[LLM Response (Parsed)]",
|
||||
":violet[LLM Response (Raw)]",
|
||||
":rainbow[Html (Raw)]",
|
||||
]
|
||||
)
|
||||
|
||||
tab_task_details, tab_task_steps, tab_task_action_results = tab_task.tabs(
|
||||
["Details", "Steps", "Action Results"]
|
||||
)
|
||||
|
||||
if selected_task:
|
||||
tab_task_details.json(selected_task)
|
||||
if selected_task_recording_uri:
|
||||
streamlit_show_recording(tab_recording, selected_task_recording_uri)
|
||||
|
||||
if task_steps:
|
||||
col_steps_prev, _, col_steps_next = col_steps.columns([3, 1, 3])
|
||||
col_steps_prev.button(
|
||||
"prev",
|
||||
on_click=go_to_previous_step,
|
||||
key="previous_step_button",
|
||||
use_container_width=True,
|
||||
)
|
||||
col_steps_next.button(
|
||||
"next",
|
||||
on_click=go_to_next_step,
|
||||
key="next_step_button",
|
||||
use_container_width=True,
|
||||
)
|
||||
|
||||
step_id_buttons = {
|
||||
step["step_id"]: col_steps.button(
|
||||
f"{step['order']} - {step['retry_index']} - {step['step_id']}",
|
||||
on_click=select_step,
|
||||
args=(step,),
|
||||
use_container_width=True,
|
||||
type=("primary" if selected_step and step["step_id"] == selected_step["step_id"] else "secondary"),
|
||||
)
|
||||
for step in task_steps
|
||||
}
|
||||
|
||||
df = pd.json_normalize(task_steps)
|
||||
tab_task_steps.dataframe(df, use_container_width=True, height=1000)
|
||||
|
||||
task_action_results = []
|
||||
for step in task_steps:
|
||||
output = step.get("output")
|
||||
step_id = step["step_id"]
|
||||
if output:
|
||||
step_action_results = output.get("action_results", [])
|
||||
for action_result in step_action_results:
|
||||
task_action_results.append(
|
||||
{
|
||||
"step_id": step_id,
|
||||
"order": step["order"],
|
||||
"retry_index": step["retry_index"],
|
||||
**action_result,
|
||||
}
|
||||
)
|
||||
df = pd.json_normalize(task_action_results)
|
||||
df = df.reindex(sorted(df.columns), axis=1)
|
||||
tab_task_action_results.dataframe(df, use_container_width=True, height=1000)
|
||||
|
||||
if selected_step:
|
||||
tab_step.json(selected_step)
|
||||
|
||||
artifacts_response = repository.get_artifacts(selected_task["task_id"], selected_step["step_id"])
|
||||
split_artifact_uris = [artifact["uri"].split("/") for artifact in artifacts_response]
|
||||
file_name_to_uris = {split_uri[-1]: "/".join(split_uri) for split_uri in split_artifact_uris}
|
||||
|
||||
for file_name, uri in file_name_to_uris.items():
|
||||
file_name = file_name.lower()
|
||||
if file_name.endswith("screenshot_llm.png") or file_name.endswith("screenshot.png"):
|
||||
streamlit_content_safe(
|
||||
tab_screenshot,
|
||||
tab_screenshot.image,
|
||||
read_artifact_safe(uri, is_image=True),
|
||||
"No screenshot available.",
|
||||
use_column_width=True,
|
||||
)
|
||||
elif file_name.endswith("screenshot_action.png"):
|
||||
streamlit_content_safe(
|
||||
tab_post_action_screenshot,
|
||||
tab_post_action_screenshot.image,
|
||||
read_artifact_safe(uri, is_image=True),
|
||||
"No action screenshot available.",
|
||||
use_column_width=True,
|
||||
)
|
||||
elif file_name.endswith("id_xpath_map.json"):
|
||||
streamlit_content_safe(
|
||||
tab_id_to_xpath,
|
||||
tab_id_to_xpath.json,
|
||||
read_artifact_safe(uri),
|
||||
"No ID -> XPath map available.",
|
||||
)
|
||||
elif file_name.endswith("id_frame_map.json"):
|
||||
streamlit_content_safe(
|
||||
tab_id_to_frame,
|
||||
tab_id_to_frame.json,
|
||||
read_artifact_safe(uri),
|
||||
"No ID -> Frame map available.",
|
||||
)
|
||||
elif file_name.endswith("tree.json"):
|
||||
streamlit_content_safe(
|
||||
tab_element_tree,
|
||||
tab_element_tree.json,
|
||||
read_artifact_safe(uri),
|
||||
"No element tree available.",
|
||||
)
|
||||
elif file_name.endswith("tree_trimmed.json"):
|
||||
streamlit_content_safe(
|
||||
tab_element_tree_trimmed,
|
||||
tab_element_tree_trimmed.json,
|
||||
read_artifact_safe(uri),
|
||||
"No element tree trimmed available.",
|
||||
)
|
||||
elif file_name.endswith("llm_prompt.txt"):
|
||||
content = read_artifact_safe(uri)
|
||||
# this is a hacky way to call this generic method to get it working with st.text_area
|
||||
streamlit_content_safe(
|
||||
tab_llm_prompt,
|
||||
tab_llm_prompt.text_area,
|
||||
content,
|
||||
"No LLM prompt available.",
|
||||
value=content,
|
||||
height=1000,
|
||||
label_visibility="collapsed",
|
||||
)
|
||||
# tab_llm_prompt.text_area("collapsed", value=content, label_visibility="collapsed", height=1000)
|
||||
elif file_name.endswith("llm_request.json"):
|
||||
streamlit_content_safe(
|
||||
tab_llm_request,
|
||||
tab_llm_request.json,
|
||||
read_artifact_safe(uri),
|
||||
"No LLM request available.",
|
||||
)
|
||||
elif file_name.endswith("llm_response_parsed.json"):
|
||||
streamlit_content_safe(
|
||||
tab_llm_response_parsed,
|
||||
tab_llm_response_parsed.json,
|
||||
read_artifact_safe(uri),
|
||||
"No parsed LLM response available.",
|
||||
)
|
||||
elif file_name.endswith("llm_response.json"):
|
||||
streamlit_content_safe(
|
||||
tab_llm_response_raw,
|
||||
tab_llm_response_raw.json,
|
||||
read_artifact_safe(uri),
|
||||
"No raw LLM response available.",
|
||||
)
|
||||
elif file_name.endswith("html_scrape.html"):
|
||||
streamlit_content_safe(
|
||||
tab_html,
|
||||
tab_html.text,
|
||||
read_artifact_safe(uri),
|
||||
"No html available.",
|
||||
)
|
||||
elif file_name.endswith("html_action.html"):
|
||||
streamlit_content_safe(
|
||||
tab_html,
|
||||
tab_html.text,
|
||||
read_artifact_safe(uri),
|
||||
"No html available.",
|
||||
)
|
||||
else:
|
||||
st.write(f"Artifact {file_name} not supported.")
|
||||
@@ -1,38 +0,0 @@
|
||||
page_font_style = """
|
||||
<style>
|
||||
@import url('https://fonts.googleapis.com/css2?family=Roboto+Mono:wght@300&display=swap');
|
||||
|
||||
* {
|
||||
font-family: 'Roboto Mono', monospace;
|
||||
}
|
||||
</style>
|
||||
"""
|
||||
|
||||
button_style = """
|
||||
<style>
|
||||
/* Apply the custom styles to Streamlit button */
|
||||
.stButton > button {
|
||||
text-align: center; /* Center button text */
|
||||
font-size: 10px; /* Set font size here */
|
||||
border: none; /* No border */
|
||||
border-radius: 20px; /* Rounded corners */
|
||||
background-color: #67748E;
|
||||
color: ##3C414A;
|
||||
padding: 10px 10px; /* Some padding */
|
||||
box-shadow: 0 4px 8px rgba(0,0,0,0.2); /* Box shadow */
|
||||
}
|
||||
|
||||
.stButton > button[kind="primary"] {
|
||||
border: 3px solid #DCFF94; /* Red border */
|
||||
}
|
||||
|
||||
.stButton > button:disabled {
|
||||
background-color: #636B7D;
|
||||
}
|
||||
|
||||
.stButton > button:hover {
|
||||
background-color: #73678F;
|
||||
color: #B6E359;
|
||||
}
|
||||
</style>
|
||||
"""
|
||||
Reference in New Issue
Block a user