vikarti.anatra/Skyvern
1
0
Fork 0
mirror of https://github.com/Skyvern-AI/skyvern.git synced 2026-04-26 10:41:14 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 3
Skyvern/skyvern/core/script_generations/generate_script.py

3386 lines
127 KiB
Python
Raw Blame History

# skyvern_codegen_cst.py
"""
Generate a runnable Skyvern workflow script.
"""
from __future__ import annotations
import asyncio
import hashlib
import keyword
import re
from dataclasses import dataclass
from typing import Any
import libcst as cst
import structlog
from libcst import Attribute, Call, Dict, DictElement, FunctionDef, Name, Param
from skyvern.config import settings
from skyvern.core.script_generations.constants import SCRIPT_TASK_BLOCKS, SCRIPT_TASK_BLOCKS_WITH_COMPLETE_ACTION
from skyvern.core.script_generations.generate_workflow_parameters import (
CUSTOM_FIELD_ACTIONS,
generate_workflow_parameters_schema,
hydrate_input_text_actions_with_field_names,
)
from skyvern.forge import app
from skyvern.schemas.workflows import FileStorageType
from skyvern.utils.strings import sanitize_identifier
from skyvern.webeye.actions.action_types import ActionType
LOG = structlog.get_logger(__name__)
GENERATE_CODE_AI_MODE_PROACTIVE = "proactive"
GENERATE_CODE_AI_MODE_FALLBACK = "fallback"
@dataclass
class ScriptBlockSource:
label: str
code: str
run_signature: str | None
workflow_run_id: str | None
workflow_run_block_id: str | None
input_fields: list[str] | None
requires_agent: bool | None = None
def _build_existing_field_assignments(
blocks: list[dict[str, Any]],
actions_by_task: dict[str, list[dict[str, Any]]],
cached_blocks: dict[str, ScriptBlockSource],
updated_block_labels: set[str],
) -> dict[int, str]:
"""
Build a mapping of action index (1-based) to existing field names for unchanged blocks.
This is used to tell the LLM which field names must be preserved when regenerating
the workflow parameters schema, preventing schema mismatches with cached block code.
Args:
blocks: List of block dictionaries from the workflow
actions_by_task: Dictionary mapping task IDs to lists of action dictionaries
cached_blocks: Dictionary mapping block labels to their cached ScriptBlockSource
updated_block_labels: Set of block labels that have been updated (should not preserve)
Returns:
Dictionary mapping action index (1-based) to the existing field name that must be preserved
"""
# Build mapping of block label -> task_id
block_label_to_task_id: dict[str, str] = {}
for idx, block in enumerate(blocks):
if block.get("block_type") not in SCRIPT_TASK_BLOCKS:
continue
label = block.get("label") or block.get("title") or block.get("task_id") or f"task_{idx}"
task_id = block.get("task_id")
if task_id:
block_label_to_task_id[label] = task_id
# Build mapping of task_id -> list of existing field names (for unchanged blocks)
task_id_to_existing_fields: dict[str, list[str]] = {}
for label, cached_source in cached_blocks.items():
# Skip blocks that have been updated - they need new field names
if label in updated_block_labels:
continue
# Skip blocks without input_fields
if not cached_source.input_fields:
continue
# Find the task_id for this block
task_id = block_label_to_task_id.get(label)
if task_id:
task_id_to_existing_fields[task_id] = list(cached_source.input_fields)
# Now iterate through actions in the same order as generate_workflow_parameters_schema
# to build the action index -> field name mapping
existing_field_assignments: dict[int, str] = {}
action_counter = 1
# Track position within each task's field list
task_field_position: dict[str, int] = {}
for task_id, actions in actions_by_task.items():
for action in actions:
action_type = action.get("action_type", "")
if action_type not in CUSTOM_FIELD_ACTIONS:
continue
# Check if this task has existing field names to preserve
if task_id in task_id_to_existing_fields:
existing_fields = task_id_to_existing_fields[task_id]
position = task_field_position.get(task_id, 0)
if position < len(existing_fields):
existing_field_assignments[action_counter] = existing_fields[position]
task_field_position[task_id] = position + 1
action_counter += 1
return existing_field_assignments
# --------------------------------------------------------------------- #
# 1. helpers #
# --------------------------------------------------------------------- #
def sanitize_variable_name(name: str) -> str:
"""
Sanitize a string to be a valid Python variable name.
- Converts to snake_case
- Removes invalid characters (via shared sanitize_identifier)
- Ensures it doesn't start with a number
- Handles Python keywords by appending underscore
- Converts to lowercase
"""
# Convert to snake_case: handle camelCase and PascalCase
name = re.sub(r"([a-z0-9])([A-Z])", r"\1_\2", name)
# Convert to lowercase before sanitizing
name = name.lower()
# Use shared sanitize_identifier for core cleanup (uses "_" prefix for digit-leading names)
name = sanitize_identifier(name, default="param")
# For script variable names, use "param_" prefix instead of bare "_" for digit-leading names
if name.startswith("_") and len(name) > 1 and name[1].isdigit():
name = f"param{name}"
# Handle Python keywords
if keyword.iskeyword(name):
name = f"{name}_"
return name
ACTION_MAP = {
"click": "click",
"hover": "hover",
"input_text": "fill",
"upload_file": "upload_file",
"select_option": "select_option",
"goto": "goto",
"scroll": "scroll",
"keypress": "keypress",
"type": "type",
"move": "move",
"drag": "drag",
"solve_captcha": "solve_captcha",
"verification_code": "verification_code",
"wait": "wait",
"extract": "extract",
"complete": "complete",
"download_file": "download_file",
}
ACTIONS_WITH_XPATH = [
"click",
"hover",
"input_text",
"type",
"fill",
"upload_file",
"select_option",
]
def _build_semantic_selector(act: dict[str, Any]) -> str | None:
"""Build a semantic CSS selector from element data.
Priority order:
1. aria-label (most reliable semantic identifier)
2. placeholder (for inputs)
3. name attribute (common form field identifier)
4. text content (for buttons, links)
5. Fall back to None (use ai='fallback' with only prompt=)
"""
element_data = act.get("skyvern_element_data") or {}
attrs = element_data.get("attributes", {})
tag = element_data.get("tagName", "")
text = (element_data.get("text") or "").strip()
# Try aria-label first
aria_label = attrs.get("aria-label", "")
if aria_label:
escaped = aria_label.replace('"', '\\"')
return f'{tag}[aria-label="{escaped}"]' if tag else f'[aria-label="{escaped}"]'
# Try placeholder (for input/textarea)
placeholder = attrs.get("placeholder", "")
if placeholder and tag in ("input", "textarea"):
escaped = placeholder.replace('"', '\\"')
return f'{tag}[placeholder="{escaped}"]'
# Try name attribute
name = attrs.get("name", "")
if name and tag in ("input", "textarea", "select"):
escaped = name.replace('"', '\\"')
return f'{tag}[name="{escaped}"]'
# Try text content (for buttons, links, labels)
if text and tag in ("button", "a"):
short_text = text[:50].replace('"', '\\"')
return f'{tag}:has-text("{short_text}")'
# No good semantic selector — return None (caller will use ai-only mode)
return None
ACTIONS_OPT_OUT_INTENTION_FOR_PROMPT = ["extract"]
INDENT = " " * 4
DOUBLE_INDENT = " " * 8
# Minimum length for a parameter value to be eligible for substitution in click prompts.
# Short values (e.g. "1", "No", "CA") cause too many false-positive replacements.
MIN_PARAM_VALUE_LENGTH_FOR_PROMPT_SUB = 4
def _build_value_to_param_lookup(
actions_by_task: dict[str, list[dict[str, Any]]],
) -> dict[str, str]:
"""Build a mapping from literal action values to their parameter field names.
After ``hydrate_input_text_actions_with_field_names`` runs, custom-field actions
carry both the original literal value (``text``, ``option``, or ``file_url``) and
the ``field_name`` assigned by the LLM. This function collects those pairs so
that click-prompt generation can replace matching literals with
``context.parameters['field_name']`` f-string references.
The returned dict is sorted by *descending value length* so that callers who
iterate in order will replace longer matches first, avoiding partial collisions.
"""
raw: dict[str, str] = {}
for _task_id, actions in actions_by_task.items():
for action in actions:
field_name = action.get("field_name")
if not field_name:
continue
action_type = action.get("action_type", "")
if action_type == ActionType.INPUT_TEXT:
value = action.get("text", "")
elif action_type == ActionType.UPLOAD_FILE:
value = action.get("file_url", "")
elif action_type == ActionType.SELECT_OPTION:
value = action.get("option", "")
else:
continue
if value and len(value) >= MIN_PARAM_VALUE_LENGTH_FOR_PROMPT_SUB:
# First writer wins — the field-level name is more specific than a
# workflow-level parameter that may happen to share the same value.
if value not in raw:
raw[value] = field_name
# Sort by descending value length so longer matches are attempted first.
return dict(sorted(raw.items(), key=lambda kv: len(kv[0]), reverse=True))
def _escape_for_fstring_text(text: str) -> str:
"""Escape ``{`` and ``}`` so they survive inside a ``FormattedStringText`` node.
Jinja2 templates (e.g. ``{{param}}``) would otherwise be interpreted as
f-string expressions. Doubling the braces turns them into literal braces
in the rendered f-string.
"""
return text.replace("{", "{{").replace("}", "}}")
def _build_parameterized_prompt_cst(
intention: str,
value_to_param: dict[str, str],
) -> cst.BaseExpression | None:
"""If *intention* contains any literal parameter values, return a ``FormattedString``
CST node (an f-string) with those values replaced by
``context.parameters['field_name']`` expressions.
Returns ``None`` when no substitution is needed (the caller should fall back to
emitting a plain string literal).
"""
# Identify all non-overlapping matches, preferring longer values (dict is
# already sorted by descending length).
# Each match is (start, end, field_name).
matches: list[tuple[int, int, str]] = []
for value, field_name in value_to_param.items():
start = 0
while True:
idx = intention.find(value, start)
if idx == -1:
break
end = idx + len(value)
# Check overlap with already-accepted matches.
overlaps = any(not (end <= ms or idx >= me) for ms, me, _ in matches)
if not overlaps:
matches.append((idx, end, field_name))
start = end
if not matches:
return None
# Sort matches by position so we can build the f-string left-to-right.
matches.sort(key=lambda m: m[0])
parts: list[cst.BaseFormattedStringContent] = []
cursor = 0
for start, end, field_name in matches:
# Text segment before this match.
if start > cursor:
parts.append(cst.FormattedStringText(_escape_for_fstring_text(intention[cursor:start])))
# The {context.parameters['field_name']} expression.
parts.append(
cst.FormattedStringExpression(
expression=cst.Subscript(
value=cst.Attribute(
value=cst.Name("context"),
attr=cst.Name("parameters"),
),
slice=[
cst.SubscriptElement(
slice=cst.Index(value=_value(field_name)),
)
],
)
)
)
cursor = end
# Trailing text after last match.
if cursor < len(intention):
parts.append(cst.FormattedStringText(_escape_for_fstring_text(intention[cursor:])))
# Use triple-quote f-string when the content contains newlines or quotes
# (run_task prompts always have newlines from the appended navigation_payload).
raw_text = intention
if "\n" in raw_text or '"' in raw_text or "'" in raw_text:
quote = '"""'
else:
quote = '"'
return cst.FormattedString(parts=parts, start=f"f{quote}", end=quote)
def _requires_mini_agent(act: dict[str, Any]) -> bool:
"""
Determine whether an input/select action should be forced into proactive mode.
Mirrors runtime logic that treats some inputs as mini-agent flows or TOTP-sensitive.
NOTE: Multi-field TOTP sequences do NOT require proactive mode because we use
get_totp_digit() to provide the exact digit value. Using proactive mode would
cause the AI to override our value with its own generated one.
"""
if act.get("has_mini_agent", False):
return True
# context = act.get("input_or_select_context") or {}
# if isinstance(context, dict) and any(
# context.get(flag) for flag in ("is_location_input", "is_date_related", "date_format")
# ):
# return True
# Multi-field TOTP sequences should NOT use proactive mode - we provide the
# exact digit via get_totp_digit() and want that value used directly
# if act.get("totp_timing_info") and act.get("totp_timing_info", {}).get("is_totp_sequence"):
# return True
return False
def _annotate_multi_field_totp_sequence(actions: list[dict[str, Any]]) -> list[dict[str, Any]]:
"""
Detect and annotate multi-field TOTP sequences in the action list.
Multi-field TOTP is when a 6-digit code needs to be split across 6 individual input fields.
This function identifies such sequences and adds totp_timing_info with the action_index
so that each field gets the correct digit (e.g., totp_code[0], totp_code[1], etc.).
Args:
actions: List of actions to analyze and annotate
Returns:
The same actions list with totp_timing_info added to multi-field TOTP actions
"""
if len(actions) < 4:
return actions
# Identify consecutive runs of single-digit TOTP inputs
# A multi-field TOTP sequence is 4+ consecutive INPUT_TEXT actions with single-digit text
# and the same field_name (typically 'totp_code')
consecutive_start = None
consecutive_count = 0
totp_field_name = None
for idx, act in enumerate(actions):
is_single_digit_totp = (
act.get("action_type") == ActionType.INPUT_TEXT
and act.get("field_name")
and act.get("text")
and len(str(act.get("text", ""))) == 1
and str(act.get("text", "")).isdigit()
)
if is_single_digit_totp:
current_field = act.get("field_name")
if consecutive_start is None:
# Start a new sequence
consecutive_start = idx
totp_field_name = current_field
consecutive_count = 1
elif current_field == totp_field_name:
# Same field, continue the sequence
consecutive_count += 1
else:
# Different field - finalize current sequence if valid, then start new one
if consecutive_count >= 4:
for seq_idx in range(consecutive_count):
actions[consecutive_start + seq_idx]["totp_timing_info"] = {
"is_totp_sequence": True,
"action_index": seq_idx,
"total_digits": consecutive_count,
"field_name": totp_field_name,
}
LOG.debug(
"Annotated multi-field TOTP sequence (field change)",
start_idx=consecutive_start,
count=consecutive_count,
field_name=totp_field_name,
)
# Start new sequence with different field
consecutive_start = idx
totp_field_name = current_field
consecutive_count = 1
else:
# End of consecutive sequence - check if it was a multi-field TOTP
if consecutive_count >= 4 and consecutive_start is not None:
# Annotate all actions in this sequence
for seq_idx in range(consecutive_count):
actions[consecutive_start + seq_idx]["totp_timing_info"] = {
"is_totp_sequence": True,
"action_index": seq_idx,
"total_digits": consecutive_count,
"field_name": totp_field_name,
}
LOG.debug(
"Annotated multi-field TOTP sequence for script generation",
start_idx=consecutive_start,
count=consecutive_count,
field_name=totp_field_name,
)
consecutive_start = None
consecutive_count = 0
totp_field_name = None
# Handle sequence at end of actions list
if consecutive_count >= 4 and consecutive_start is not None:
for seq_idx in range(consecutive_count):
actions[consecutive_start + seq_idx]["totp_timing_info"] = {
"is_totp_sequence": True,
"action_index": seq_idx,
"total_digits": consecutive_count,
"field_name": totp_field_name,
}
LOG.debug(
"Annotated multi-field TOTP sequence for script generation (at end)",
start_idx=consecutive_start,
count=consecutive_count,
field_name=totp_field_name,
)
return actions
def safe_name(label: str) -> str:
s = "".join(c if c.isalnum() else "_" for c in label).lower()
if not s or s[0].isdigit() or keyword.iskeyword(s):
s = f"_{s}"
while "__" in s:
s = s.replace("__", "_")
return s
def _value(value: Any) -> cst.BaseExpression:
"""Convert simple Python objects to CST expressions."""
if isinstance(value, str):
if "\n" in value:
# For multi-line strings, use repr() which handles all escaping properly
# This will use triple quotes when appropriate and escape them when needed
return cst.SimpleString(repr(value))
return cst.SimpleString(repr(value))
if isinstance(value, (int, float, bool)) or value is None:
return cst.parse_expression(repr(value))
if isinstance(value, dict):
return Dict(
[
DictElement(
key=_value(k),
value=_value(v),
)
for k, v in value.items()
]
)
if isinstance(value, (list, tuple)):
elts = [cst.Element(_value(v)) for v in value]
return cst.List(elts) if isinstance(value, list) else cst.Tuple(elts)
# fallback
return cst.SimpleString(repr(str(value)))
def _render_value(
prompt_text: str | None = None,
data_variable_name: str | None = None,
render_func_name: str = "render_template",
) -> cst.BaseExpression:
"""Create a prompt value with template rendering logic if needed."""
if not prompt_text:
return cst.SimpleString("")
if "{{" in prompt_text and "}}" in prompt_text:
args = [cst.Arg(value=_value(prompt_text))]
if data_variable_name:
args.append(
cst.Arg(
keyword=cst.Name("data"),
value=cst.Name(data_variable_name),
)
)
return cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name(render_func_name)),
args=args,
)
else:
# Return the prompt as a simple string value
return _value(prompt_text)
# --------------------------------------------------------------------- #
# 2. utility builders #
# --------------------------------------------------------------------- #
def _workflow_decorator(wf_req: dict[str, Any]) -> cst.Decorator:
"""
Build @skyvern.workflow(
title="...", totp_url=..., totp_identifier=..., webhook_callback_url=..., max_steps=...
)
"""
# helper that skips “None” so the output is concise
def kw(key: str, value: Any) -> cst.Arg | None:
if value is None:
return None
return cst.Arg(keyword=cst.Name(key), value=_value(value))
args: list = list(
filter(
None,
[
kw("title", wf_req.get("title", "")),
kw("totp_url", wf_req.get("totp_url")),
kw("totp_identifier", wf_req.get("totp_identifier")),
kw("webhook_url", wf_req.get("webhook_url")),
kw("max_steps", wf_req.get("max_steps")),
],
)
)
return cst.Decorator(
decorator=cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("workflow")),
args=args,
)
)
def make_decorator(block_label: str, block: dict[str, Any]) -> cst.Decorator:
kwargs = [
cst.Arg(
keyword=cst.Name("cache_key"),
value=_value(block_label),
)
]
return cst.Decorator(
decorator=Call(
func=Attribute(value=cst.Name("skyvern"), attr=cst.Name("cached")),
args=kwargs,
)
)
def _action_to_stmt(
act: dict[str, Any],
task: dict[str, Any],
assign_to_output: bool = False,
value_to_param: dict[str, str] | None = None,
use_semantic_selectors: bool = False,
) -> cst.BaseStatement:
"""
Turn one Action dict into:
await page.<method>(selector=..., prompt=..., data=context.parameters)
Or if assign_to_output is True for extract actions:
output = await page.extract(...)
When *value_to_param* is provided, click prompt strings that contain literal
parameter values will be emitted as f-strings referencing
``context.parameters['field_name']`` instead of hardcoded text.
"""
method = ACTION_MAP[act["action_type"]]
args: list[cst.Arg] = []
if method in ACTIONS_WITH_XPATH:
if use_semantic_selectors:
semantic = _build_semantic_selector(act)
if semantic:
args.append(
cst.Arg(
keyword=cst.Name("selector"),
value=_value(semantic),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
# If no semantic selector, skip selector arg — ai with prompt= handles it
else:
args.append(
cst.Arg(
keyword=cst.Name("selector"),
value=_value(f"xpath={act['xpath']}"),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if method == "click":
if use_semantic_selectors:
# With semantic selectors, try selector first, AI only if miss
ai_mode = GENERATE_CODE_AI_MODE_FALLBACK
else:
ai_mode = GENERATE_CODE_AI_MODE_PROACTIVE
click_context = act.get("click_context")
if click_context and isinstance(click_context, dict) and click_context.get("single_option_click"):
ai_mode = GENERATE_CODE_AI_MODE_FALLBACK
args.append(
cst.Arg(
keyword=cst.Name("ai"),
value=_value(ai_mode),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
elif method == "hover":
hold_seconds = act.get("hold_seconds")
if hold_seconds and hold_seconds > 0:
args.append(
cst.Arg(
keyword=cst.Name("hold_seconds"),
value=_value(hold_seconds),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
elif method in ["type", "fill"]:
# Use context.parameters if field_name is available, otherwise fallback to direct value
if act.get("field_name"):
# Check if this is a multi-field TOTP sequence that needs digit indexing
totp_info = act.get("totp_timing_info") or {}
if totp_info.get("is_totp_sequence") and "action_index" in totp_info:
# Generate: await page.get_totp_digit(context, 'field_name', digit_index)
# This method properly resolves the TOTP code from credentials and returns the specific digit
text_value = cst.Await(
expression=cst.Call(
func=cst.Attribute(
value=cst.Name("page"),
attr=cst.Name("get_totp_digit"),
),
args=[
cst.Arg(value=cst.Name("context")),
cst.Arg(value=_value(act["field_name"])),
cst.Arg(value=_value(totp_info["action_index"])),
],
)
)
else:
text_value = cst.Subscript(
value=cst.Attribute(
value=cst.Name("context"),
attr=cst.Name("parameters"),
),
slice=[cst.SubscriptElement(slice=cst.Index(value=_value(act["field_name"])))],
)
else:
text_value = _value(act["text"])
ai_mode = GENERATE_CODE_AI_MODE_FALLBACK
if _requires_mini_agent(act):
ai_mode = GENERATE_CODE_AI_MODE_PROACTIVE
args.append(
cst.Arg(
keyword=cst.Name("value"),
value=text_value,
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
args.append(
cst.Arg(
keyword=cst.Name("ai"),
value=_value(ai_mode),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if act.get("totp_code_required"):
if task.get("totp_identifier"):
args.append(
cst.Arg(
keyword=cst.Name("totp_identifier"),
value=_value(task.get("totp_identifier")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if task.get("totp_url"):
args.append(
cst.Arg(
keyword=cst.Name("totp_url"),
value=_value(task.get("totp_verification_url")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
elif method == "select_option":
option = act.get("option", {})
value = option.get("value")
label = option.get("label")
value = value or label
if value:
# TODO: consider supporting fallback mode for select_option actions
# ai_mode = GENERATE_CODE_AI_MODE_FALLBACK
# if _requires_mini_agent(act):
ai_mode = GENERATE_CODE_AI_MODE_PROACTIVE
if act.get("field_name"):
option_value = cst.Subscript(
value=cst.Attribute(
value=cst.Name("context"),
attr=cst.Name("parameters"),
),
slice=[cst.SubscriptElement(slice=cst.Index(value=_value(act["field_name"])))],
)
else:
option_value = _value(value)
args.append(
cst.Arg(
keyword=cst.Name("value"),
value=option_value,
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
)
args.append(
cst.Arg(
keyword=cst.Name("ai"),
value=_value(ai_mode),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
elif method == "upload_file":
if act.get("field_name"):
file_url_value = cst.Subscript(
value=cst.Attribute(
value=cst.Name("context"),
attr=cst.Name("parameters"),
),
slice=[cst.SubscriptElement(slice=cst.Index(value=_value(act["field_name"])))],
)
else:
file_url_value = _value(act["file_url"])
args.append(
cst.Arg(
keyword=cst.Name("files"),
value=file_url_value,
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
args.append(
cst.Arg(
keyword=cst.Name("ai"),
value=_value(GENERATE_CODE_AI_MODE_PROACTIVE),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
elif method == "keypress":
args.append(
cst.Arg(
keyword=cst.Name("keys"),
value=_value(act.get("keys", ["Enter"])),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if act.get("hold"):
args.append(
cst.Arg(
keyword=cst.Name("hold"),
value=_value(act["hold"]),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if act.get("duration"):
args.append(
cst.Arg(
keyword=cst.Name("duration"),
value=_value(act["duration"]),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
elif method == "wait":
args.append(
cst.Arg(
keyword=cst.Name("seconds"),
value=_value(act["seconds"]),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
elif method == "download_file":
args.append(
cst.Arg(
keyword=cst.Name("file_name"),
value=_value(act.get("file_name", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if act.get("download_url"):
args.append(
cst.Arg(
keyword=cst.Name("download_url"),
value=_value(act["download_url"]),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
elif method == "extract":
args.append(
cst.Arg(
keyword=cst.Name("prompt"),
value=_value(act["data_extraction_goal"]),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if act.get("data_extraction_schema"):
args.append(
cst.Arg(
keyword=cst.Name("schema"),
value=_value(act["data_extraction_schema"]),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
comma=cst.Comma(),
)
)
action_context = act.get("input_or_select_context")
if action_context and action_context.get("date_format") and method in ["type", "fill", "select_option"]:
date_format_value = action_context.get("date_format")
data = {"date_format": date_format_value}
args.append(
cst.Arg(
keyword=cst.Name("data"),
value=_value(data),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
intention = act.get("intention") or act.get("reasoning") or ""
if intention and method not in ACTIONS_OPT_OUT_INTENTION_FOR_PROMPT:
# Try to parameterize the prompt for click actions so cached scripts
# don't embed recording-specific values (e.g. a patient ID).
prompt_value: cst.BaseExpression | None = None
if value_to_param:
prompt_value = _build_parameterized_prompt_cst(intention, value_to_param)
if prompt_value is None:
prompt_value = _value(intention)
args.extend(
[
cst.Arg(
keyword=cst.Name("prompt"),
value=prompt_value,
whitespace_after_arg=cst.ParenthesizedWhitespace(indent=True),
comma=cst.Comma(),
),
]
)
_mark_last_arg_as_comma(args)
# Only use indented parentheses if we have arguments
if args:
call = cst.Call(
func=cst.Attribute(value=cst.Name("page"), attr=cst.Name(method)),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
else:
call = cst.Call(
func=cst.Attribute(value=cst.Name("page"), attr=cst.Name(method)),
args=args,
)
# await page.method(...)
await_expr = cst.Await(call)
# If this is an extract action and we want to assign to output
if assign_to_output and method == "extract":
# output = await page.extract(...)
assign = cst.Assign(
targets=[cst.AssignTarget(cst.Name("output"))],
value=await_expr,
)
return cst.SimpleStatementLine([assign])
else:
# Wrap in a statement line: await ...
return cst.SimpleStatementLine([cst.Expr(await_expr)])
def _collect_block_input_fields(
block: dict[str, Any],
actions_by_task: dict[str, list[dict[str, Any]]],
) -> list[str]:
"""
Gather the sequence of workflow parameter field names referenced by custom field actions within a block.
"""
task_id = block.get("task_id")
if not task_id:
return []
all_fields: list[str] = []
for action in actions_by_task.get(task_id, []):
action_type = action.get("action_type")
# Keep in sync with CUSTOM_FIELD_ACTIONS used for schema generation
if action_type not in CUSTOM_FIELD_ACTIONS:
continue
field_name = action.get("field_name")
if not field_name or not isinstance(field_name, str):
continue
all_fields.append(field_name)
return all_fields
def _is_form_filling_block(
actions: list[dict[str, Any]],
) -> bool:
"""Check if a block's actions indicate a form-filling pattern (4+ input/select actions)."""
form_action_count = sum(
1 for a in actions if a.get("action_type") in (ActionType.INPUT_TEXT, ActionType.SELECT_OPTION)
)
return form_action_count >= 4
def _build_form_filling_block_fn(
block: dict[str, Any],
actions: list[dict[str, Any]],
value_to_param: dict[str, str] | None = None,
) -> FunctionDef:
"""Generate a form-filling block function that uses page.fill_form().
Produces a simple function that calls page.fill_form(context.parameters)
which internally uses LLM-based field mapping (extract → map → fill).
"""
name = safe_name(block.get("label") or block.get("title") or f"block_{block.get('workflow_run_block_id')}")
cache_key = block.get("label") or block.get("title") or f"block_{block.get('workflow_run_block_id')}"
navigation_goal = block.get("navigation_goal") or "Fill out the form"
# Include page.goto(url) if the block has a URL, just like _build_block_fn does
goto_line = ""
if block.get("url"):
goto_line = f" await page.goto({repr(block['url'])})\n"
func_code = (
f"async def {name}(page: SkyvernPage, context: RunContext):\n"
f"{goto_line}"
f" await page.fill_form(context.parameters, prompt={repr(navigation_goal)})\n"
)
# Parse and add decorator
parsed = cst.parse_module(func_code)
func_def = parsed.body[0]
assert isinstance(func_def, FunctionDef)
# Add the @skyvern.cached decorator
decorated = func_def.with_changes(decorators=[make_decorator(cache_key, block)])
return decorated
def _detect_block_ats_platform(block: dict[str, Any], all_blocks: list[dict[str, Any]] | None = None) -> str | None:
"""Check if a block belongs to a known platform with optimized scripts.
Checks the block's own URL first. If the block has no URL (common for
continuation blocks that share the previous block's page), checks all
other blocks in the workflow for a URL match.
Delegates to app.AGENT_FUNCTION which is overridden in cloud builds.
Returns None in OSS (no detection).
"""
# Check block URL first
result = app.AGENT_FUNCTION.detect_ats_platform(block.get("url") or "")
if result:
return result
# Fall back: check any sibling block's URL (all blocks in the same
# workflow share the same site context)
if all_blocks:
for sibling in all_blocks:
sibling_url = sibling.get("url") or ""
if sibling_url:
result = app.AGENT_FUNCTION.detect_ats_platform(sibling_url)
if result:
return result
return None
def _build_block_fn(
block: dict[str, Any],
actions: list[dict[str, Any]],
value_to_param: dict[str, str] | None = None,
use_semantic_selectors: bool = False,
is_in_for_loop: bool = False,
all_blocks: list[dict[str, Any]] | None = None,
) -> FunctionDef:
# Check for platform-specific pipeline (cloud-only; returns None in OSS)
if use_semantic_selectors:
ats_platform = _detect_block_ats_platform(block, all_blocks=all_blocks)
if ats_platform:
pipeline_fn = app.AGENT_FUNCTION.build_ats_pipeline_block_fn(block, ats_platform)
if pipeline_fn:
LOG.info(
"Code 2.0: platform detected, generating optimized pipeline",
ats_platform=ats_platform,
block_label=block.get("label"),
block_url=block.get("url"),
)
return pipeline_fn
# Check if this block is form-filling (4+ input_text/select actions)
if use_semantic_selectors and _is_form_filling_block(actions):
return _build_form_filling_block_fn(block, actions, value_to_param)
name = safe_name(block.get("label") or block.get("title") or f"block_{block.get('workflow_run_block_id')}")
cache_key = block.get("label") or block.get("title") or f"block_{block.get('workflow_run_block_id')}"
body_stmts: list[cst.BaseStatement] = []
# Detect and annotate multi-field TOTP sequences so each fill gets the correct digit index
actions = _annotate_multi_field_totp_sequence(actions)
if block.get("url"):
body_stmts.append(cst.parse_statement(f"await page.goto({repr(block['url'])})"))
# For file_download blocks inside for-loops, generate a dynamic click that uses
# per-iteration context instead of hardcoded xpath/prompt from iteration 0.
block_type = block.get("block_type")
if is_in_for_loop and block_type == "file_download":
body_stmts.append(
cst.parse_statement(
'await page.click(selector=context.download_selector(), prompt=context.prompt, ai="fallback")'
)
)
else:
for act in actions:
if act["action_type"] in [
ActionType.COMPLETE,
ActionType.TERMINATE,
ActionType.NULL_ACTION,
]:
continue
# For extraction blocks, assign extract action results to output variable
assign_to_output = act["action_type"] == "extract"
body_stmts.append(
_action_to_stmt(
act,
block,
assign_to_output=assign_to_output,
value_to_param=value_to_param,
use_semantic_selectors=use_semantic_selectors,
)
)
# add complete action
if block_type in SCRIPT_TASK_BLOCKS_WITH_COMPLETE_ACTION:
complete_action = {"action_type": "complete"}
body_stmts.append(_action_to_stmt(complete_action, block))
# For extraction blocks, add return output statement if we have actions
if any(
act["action_type"] == "extract"
for act in actions
if act["action_type"] not in [ActionType.COMPLETE, ActionType.TERMINATE, ActionType.NULL_ACTION]
):
body_stmts.append(cst.parse_statement("return output"))
elif not body_stmts:
body_stmts.append(cst.parse_statement("return None"))
return FunctionDef(
name=Name(name),
params=cst.Parameters(
params=[
Param(name=Name("page"), annotation=cst.Annotation(cst.Name("SkyvernPage"))),
Param(name=Name("context"), annotation=cst.Annotation(cst.Name("RunContext"))),
]
),
decorators=[make_decorator(cache_key, block)],
body=cst.IndentedBlock(body_stmts),
returns=None,
asynchronous=cst.Asynchronous(),
)
def _build_task_v2_block_fn(block: dict[str, Any], child_blocks: list[dict[str, Any]]) -> FunctionDef:
"""Build a cached function for task_v2 blocks that calls child workflow sub-tasks."""
cache_key = block.get("label") or block.get("title") or f"block_{block.get('workflow_run_block_id')}"
name = safe_name(block.get("label") or block.get("title") or f"block_{block.get('workflow_run_block_id')}")
body_stmts: list[cst.BaseStatement] = []
# Add calls to child workflow sub-tasks
has_extract_block = False
for child_block in child_blocks:
is_extract_block = child_block.get("block_type") == "extraction"
if is_extract_block:
has_extract_block = True
stmt = _build_block_statement(child_block, assign_output=is_extract_block)
body_stmts.append(stmt)
if not body_stmts:
body_stmts.append(cst.parse_statement("return None"))
elif has_extract_block:
body_stmts.append(cst.parse_statement("return output"))
return FunctionDef(
name=Name(name),
params=cst.Parameters(
params=[
Param(name=Name("page"), annotation=cst.Annotation(cst.Name("SkyvernPage"))),
Param(name=Name("context"), annotation=cst.Annotation(cst.Name("RunContext"))),
]
),
decorators=[make_decorator(cache_key, block)],
body=cst.IndentedBlock(body_stmts),
returns=None,
asynchronous=cst.Asynchronous(),
)
def _build_model(workflow: dict[str, Any]) -> cst.ClassDef:
"""
class WorkflowParameters(BaseModel):
param1: str
param2: str
...
"""
ann_lines: list[cst.BaseStatement] = []
for parameter in workflow["workflow_definition"]["parameters"]:
if parameter["parameter_type"] != "workflow":
continue
ann = cst.AnnAssign(
target=cst.Name(sanitize_variable_name(parameter["key"])),
annotation=cst.Annotation(cst.Name("str")),
value=None,
)
ann_lines.append(cst.SimpleStatementLine([ann]))
if not ann_lines: # no parameters
ann_lines.append(cst.SimpleStatementLine([cst.Pass()]))
return cst.ClassDef(
name=cst.Name("WorkflowParameters"),
bases=[cst.Arg(cst.Name("BaseModel"))],
body=cst.IndentedBlock(ann_lines), # ← wrap in block
)
def _build_generated_model_from_schema(schema_code: str) -> cst.ClassDef | None:
"""
Parse the generated schema code and return a ClassDef, or None if parsing fails.
"""
try:
# Parse the schema code and extract just the class definition
parsed_module = cst.parse_module(schema_code)
# Find the GeneratedWorkflowParameters class in the parsed module
for node in parsed_module.body:
if isinstance(node, cst.ClassDef) and node.name.value == "GeneratedWorkflowParameters":
return node
# If no class found, return None
return None
except Exception as e:
LOG.warning("Failed to parse generated schema code", error=str(e))
return None
# --------------------------------------------------------------------- #
# 3. statement builders #
# --------------------------------------------------------------------- #
def _build_run_task_statement(
block_title: str,
block: dict[str, Any],
data_variable_name: str | None = None,
value_to_param: dict[str, str] | None = None,
) -> cst.SimpleStatementLine:
"""Build a skyvern.run_task statement."""
args = __build_base_task_statement(block_title, block, data_variable_name, value_to_param=value_to_param)
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("run_task")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_download_statement(
block_title: str,
block: dict[str, Any],
data_variable_name: str | None = None,
value_to_param: dict[str, str] | None = None,
) -> cst.SimpleStatementLine:
"""Build a skyvern.download statement."""
args = __build_base_task_statement(block_title, block, data_variable_name, value_to_param=value_to_param)
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("download")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_action_statement(
block_title: str, block: dict[str, Any], data_variable_name: str | None = None
) -> cst.SimpleStatementLine:
"""Build a skyvern.action statement."""
args = [
cst.Arg(
keyword=cst.Name("prompt"),
value=_value(block.get("navigation_goal", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
]
if block.get("model"):
args.append(
cst.Arg(
keyword=cst.Name("model"),
value=_value(block.get("model")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("label"):
args.append(
cst.Arg(
keyword=cst.Name("label"),
value=_value(block.get("label")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
),
comma=cst.Comma(),
)
)
_mark_last_arg_as_comma(args)
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("action")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_login_statement(
block_title: str,
block: dict[str, Any],
data_variable_name: str | None = None,
value_to_param: dict[str, str] | None = None,
) -> cst.SimpleStatementLine:
"""Build a skyvern.login statement."""
args = __build_base_task_statement(block_title, block, data_variable_name, value_to_param=value_to_param)
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("login")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_extract_statement(
block_title: str,
block: dict[str, Any],
data_variable_name: str | None = None,
assign_output: bool = True,
) -> cst.SimpleStatementLine:
"""Build a skyvern.extract statement."""
args = [
cst.Arg(
keyword=cst.Name("prompt"),
value=_value(block.get("data_extraction_goal", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
cst.Arg(
keyword=cst.Name("schema"),
value=_value(block.get("data_schema", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
]
if block.get("model"):
args.append(
cst.Arg(
keyword=cst.Name("model"),
value=_value(block.get("model")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("label"):
args.append(
cst.Arg(
keyword=cst.Name("label"),
value=_value(block_title),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
),
comma=cst.Comma(),
)
)
_mark_last_arg_as_comma(args)
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("extract")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
if assign_output:
return cst.SimpleStatementLine(
[
cst.Assign(
targets=[cst.AssignTarget(target=cst.Name("output"))],
value=cst.Await(call),
)
]
)
else:
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_navigate_statement(
block_title: str,
block: dict[str, Any],
data_variable_name: str | None = None,
value_to_param: dict[str, str] | None = None,
) -> cst.SimpleStatementLine:
"""Build a skyvern.navigate statement."""
args = __build_base_task_statement(block_title, block, data_variable_name, value_to_param=value_to_param)
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("run_task")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_send_email_statement(block: dict[str, Any]) -> cst.SimpleStatementLine:
"""Build a skyvern.send_email statement."""
args = [
cst.Arg(
keyword=cst.Name("sender"),
value=_value(block.get("sender", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
cst.Arg(
keyword=cst.Name("recipients"),
value=_value(block.get("recipients", [])),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
cst.Arg(
keyword=cst.Name("subject"),
value=_value(block.get("subject", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
cst.Arg(
keyword=cst.Name("body"),
value=_value(block.get("body", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
cst.Arg(
keyword=cst.Name("file_attachments"),
value=_value(block.get("file_attachments", [])),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
cst.Arg(
keyword=cst.Name("label"),
value=_value(block.get("label", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
),
comma=cst.Comma(),
),
]
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("send_email")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_validate_statement(
block_title: str, block: dict[str, Any], data_variable_name: str | None = None
) -> cst.SimpleStatementLine:
"""Build a skyvern.validate statement."""
args = []
# Add complete_criterion if it exists
if block.get("complete_criterion") is not None:
args.append(
cst.Arg(
keyword=cst.Name("complete_criterion"),
value=_value(block.get("complete_criterion")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
# Add terminate_criterion if it exists
if block.get("terminate_criterion") is not None:
args.append(
cst.Arg(
keyword=cst.Name("terminate_criterion"),
value=_value(block.get("terminate_criterion")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
# Add error_code_mapping if it exists
if block.get("error_code_mapping") is not None:
args.append(
cst.Arg(
keyword=cst.Name("error_code_mapping"),
value=_value(block.get("error_code_mapping")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("model"):
args.append(
cst.Arg(
keyword=cst.Name("model"),
value=_value(block.get("model")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
# Add label if it exists
if block.get("label") is not None:
args.append(
cst.Arg(
keyword=cst.Name("label"),
value=_value(block.get("label")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
),
)
)
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("validate")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_human_interaction_statement(
block: dict[str, Any],
) -> cst.SimpleStatementLine:
LOG.warning("Human interaction code generation is not yet implemented.", block=block)
return cst.SimpleStatementLine([cst.Expr(cst.Comment("# TODO: Implement human interaction logic"))])
def _build_wait_statement(block: dict[str, Any]) -> cst.SimpleStatementLine:
"""Build a skyvern.wait statement."""
args = [
cst.Arg(
keyword=cst.Name("seconds"),
value=_value(block.get("wait_sec", 1)),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
cst.Arg(
keyword=cst.Name("label"),
value=_value(block.get("label")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
),
comma=cst.Comma(),
),
]
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("wait")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_goto_statement(block: dict[str, Any], data_variable_name: str | None = None) -> cst.SimpleStatementLine:
"""Build a skyvern.goto statement."""
args = [
cst.Arg(
keyword=cst.Name("url"),
value=_value(block.get("url", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
cst.Arg(
keyword=cst.Name("label"),
value=_value(block.get("label") or block.get("title") or f"block_{block.get('workflow_run_block_id')}"),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
),
comma=cst.Comma(),
),
]
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("goto")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_code_statement(block: dict[str, Any]) -> cst.SimpleStatementLine:
"""Build a skyvern.run_code statement."""
parameters = block.get("parameters", [])
parameter_list = [parameter["key"] for parameter in parameters]
args = [
cst.Arg(
keyword=cst.Name("code"),
value=_value(block.get("code", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
cst.Arg(
keyword=cst.Name("label"),
value=_value(block.get("label") or block.get("title") or f"block_{block.get('workflow_run_block_id')}"),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
cst.Arg(
keyword=cst.Name("parameters"),
value=_value(parameter_list),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
),
comma=cst.Comma(),
),
]
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("run_code")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_file_upload_statement(block: dict[str, Any]) -> cst.SimpleStatementLine:
"""Build a skyvern.upload_file statement."""
parameters = block.get("parameters", [])
parameter_list = [parameter["key"] for parameter in parameters]
args = [
cst.Arg(
keyword=cst.Name("label"),
value=_value(block.get("label", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
cst.Arg(
keyword=cst.Name("parameters"),
value=_value(parameter_list),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
cst.Arg(
keyword=cst.Name("storage_type"),
value=_value(str(block.get("storage_type", FileStorageType.S3))),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
]
for key in [
"s3_bucket",
"aws_access_key_id",
"aws_secret_access_key",
"region_name",
"azure_storage_account_name",
"azure_storage_account_key",
"azure_blob_container_name",
"path",
]:
if block.get(key) is not None:
args.append(
cst.Arg(
keyword=cst.Name(key),
value=_value(block.get(key, "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
_mark_last_arg_as_comma(args)
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("upload_file")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_pdf_parser_statement(block: dict[str, Any]) -> cst.SimpleStatementLine:
"""Build a skyvern.parse_pdf statement."""
args = [
cst.Arg(
keyword=cst.Name("file_url"),
value=_value(block.get("file_url", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
]
if block.get("json_schema") is not None:
args.append(
cst.Arg(
keyword=cst.Name("schema"),
value=_value(block.get("json_schema")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("model"):
args.append(
cst.Arg(
keyword=cst.Name("model"),
value=_value(block.get("model")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("label") is not None:
args.append(
cst.Arg(
keyword=cst.Name("label"),
value=_value(block.get("label")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
_mark_last_arg_as_comma(args)
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("parse_pdf")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_file_url_parser_statement(block: dict[str, Any]) -> cst.SimpleStatementLine:
"""Build a skyvern.parse_file statement."""
args = [
cst.Arg(
keyword=cst.Name("file_url"),
value=_value(block.get("file_url", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
cst.Arg(
keyword=cst.Name("file_type"),
value=_value(str(block.get("file_type"))),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
]
# Add optional parameters if they exist
if block.get("json_schema") is not None:
args.append(
cst.Arg(
keyword=cst.Name("schema"),
value=_value(block.get("json_schema")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("model"):
args.append(
cst.Arg(
keyword=cst.Name("model"),
value=_value(block.get("model")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("label") is not None:
args.append(
cst.Arg(
keyword=cst.Name("label"),
value=_value(block.get("label")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
_mark_last_arg_as_comma(args)
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("parse_file")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_http_request_statement(block: dict[str, Any]) -> cst.SimpleStatementLine:
"""Build a skyvern.http_request statement."""
args = [
cst.Arg(
keyword=cst.Name("method"),
value=_value(block.get("method", "GET")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
cst.Arg(
keyword=cst.Name("url"),
value=_value(block.get("url", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
]
# Add optional parameters if they exist
if block.get("headers") is not None:
args.append(
cst.Arg(
keyword=cst.Name("headers"),
value=_value(block.get("headers")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("body") is not None:
args.append(
cst.Arg(
keyword=cst.Name("body"),
value=_value(block.get("body")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("timeout") is not None:
args.append(
cst.Arg(
keyword=cst.Name("timeout"),
value=_value(block.get("timeout")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("follow_redirects") is not None:
args.append(
cst.Arg(
keyword=cst.Name("follow_redirects"),
value=_value(block.get("follow_redirects")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("label") is not None:
args.append(
cst.Arg(
keyword=cst.Name("label"),
value=_value(block.get("label")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
_mark_last_arg_as_comma(args)
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("http_request")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_prompt_statement(block: dict[str, Any]) -> cst.SimpleStatementLine:
"""Build a skyvern.prompt statement."""
args = [
cst.Arg(
keyword=cst.Name("prompt"),
value=_value(block.get("prompt", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
]
# Add optional parameters if they exist
if block.get("json_schema") is not None:
args.append(
cst.Arg(
keyword=cst.Name("schema"),
value=_value(block.get("json_schema")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("model"):
args.append(
cst.Arg(
keyword=cst.Name("model"),
value=_value(block.get("model")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("label") is not None:
args.append(
cst.Arg(
keyword=cst.Name("label"),
value=_value(block.get("label")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("parameters"):
parameters = block.get("parameters", [])
parameter_list = [parameter["key"] for parameter in parameters]
args.append(
cst.Arg(
keyword=cst.Name("parameters"),
value=_value(parameter_list),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
_mark_last_arg_as_comma(args)
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("prompt")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_conditional_statement(
block: dict[str, Any],
blocks_by_label: dict[str, dict[str, Any]],
consumed_labels: set[str],
) -> list[cst.BaseStatement]:
"""
Build if/elif/else statements for a conditional block.
Returns a list of statements:
1. result = await skyvern.conditional(label='...')
2. if/elif/else block with branch child blocks nested inside
An example:
```
result = await skyvern.conditional(label='block_2')
if result.get('branch_index') == 0:
await skyvern.prompt(prompt='output: "expression 1 successful."', label='block_3')
else:
await skyvern.prompt(prompt='output: "expression 1 failed"', label='block_4')
```
"""
label = block.get("label", "conditional")
branches = block.get("branch_conditions") or block.get("branches") or block.get("ordered_branches") or []
merge_label = block.get("next_block_label")
# Separate non-default branches from the default (else) branch
non_default_branches: list[tuple[int, dict[str, Any]]] = []
default_branch: tuple[int, dict[str, Any]] | None = None
for i, branch in enumerate(branches):
if branch.get("is_default"):
default_branch = (i, branch)
else:
non_default_branches.append((i, branch))
def _collect_branch_body(start_label: str | None) -> list[cst.BaseStatement]:
"""Follow next_block_label chain to collect all blocks in a branch.
Note: mutates ``consumed_labels`` (from enclosing scope) as a side effect
so the top-level iteration in _build_run_fn skips blocks already nested here.
Limitation: nested conditional blocks inside a branch are rendered via
_build_block_statement which falls through to the comment-string codepath.
Supporting recursive conditionals would require calling
_build_conditional_statement here, which is left for a future change.
"""
stmts: list[cst.BaseStatement] = []
current = start_label
while current and current != merge_label:
b = blocks_by_label.get(current)
if b:
consumed_labels.add(current)
stmts.append(_build_block_statement(b, assign_output=False))
current = b.get("next_block_label")
else:
break
if not stmts:
stmts.append(cst.parse_statement("pass"))
return stmts
# Build the else clause (default branch) — only when there are non-default
# branches to form the if/elif chain. When only a default branch exists,
# we emit the body directly (handled below) and skip the orelse node.
orelse: cst.If | cst.Else | None = None
if default_branch and non_default_branches:
_, d_branch = default_branch
d_body = _collect_branch_body(d_branch.get("next_block_label"))
orelse = cst.Else(
body=cst.IndentedBlock(body=d_body),
)
# Build elif chain from bottom up (last non-default branch first)
for idx in range(len(non_default_branches) - 1, 0, -1):
branch_index, branch = non_default_branches[idx]
branch_body = _collect_branch_body(branch.get("next_block_label"))
orelse = cst.If(
test=cst.parse_expression(f"result.get('branch_index') == {branch_index}"),
body=cst.IndentedBlock(body=branch_body),
orelse=orelse,
leading_lines=[],
whitespace_before_test=cst.SimpleWhitespace(" "),
)
# Build the top-level if statement (first non-default branch)
# Return both the conditional call and the if/else block
conditional_call = cst.parse_statement(f"result = await skyvern.conditional(label='{label}')")
if non_default_branches:
first_index, first_branch = non_default_branches[0]
first_body = _collect_branch_body(first_branch.get("next_block_label"))
if_stmt = cst.If(
test=cst.parse_expression(f"result.get('branch_index') == {first_index}"),
body=cst.IndentedBlock(body=first_body),
orelse=orelse,
leading_lines=[],
whitespace_before_test=cst.SimpleWhitespace(" "),
)
return [conditional_call, if_stmt]
# Only a default branch exists — emit the conditional call and default body
# directly without an if/else wrapper (there's nothing to branch on).
if default_branch:
_, d_branch = default_branch
d_body = _collect_branch_body(d_branch.get("next_block_label"))
return [conditional_call, *d_body]
# No branches at all — just the conditional call
return [conditional_call]
def _build_workflow_trigger_statement(block: dict[str, Any]) -> cst.SimpleStatementLine:
"""Build a skyvern.trigger_workflow statement.
WorkflowTriggerBlock makes zero LLM calls — it's pure orchestration
(template resolution, workflow dispatch, output collection). Executed
as-is during cached script runs.
"""
args = [
cst.Arg(
keyword=cst.Name("workflow_permanent_id"),
value=_value(block.get("workflow_permanent_id", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
]
if block.get("payload") is not None:
args.append(
cst.Arg(
keyword=cst.Name("payload"),
value=_value(block.get("payload")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("wait_for_completion") is not None:
args.append(
cst.Arg(
keyword=cst.Name("wait_for_completion"),
value=_value(block.get("wait_for_completion")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("use_parent_browser_session"):
args.append(
cst.Arg(
keyword=cst.Name("use_parent_browser_session"),
value=_value(True),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("browser_session_id"):
args.append(
cst.Arg(
keyword=cst.Name("browser_session_id"),
value=_value(block.get("browser_session_id")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
args.append(
cst.Arg(
keyword=cst.Name("label"),
value=_value(block.get("label", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
),
),
)
_mark_last_arg_as_comma(args)
call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("trigger_workflow")),
args=args,
whitespace_before_args=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
return cst.SimpleStatementLine([cst.Expr(cst.Await(call))])
def _build_for_loop_statement(block_title: str, block: dict[str, Any]) -> cst.For:
"""
Build a for loop statement.
All the blocks within the for loop block statement will run without cache_key.
An example of a for loop statement:
```
async for current_value in skyvern.loop(context.parameters["urls"]):
await skyvern.goto(
url=current_value,
label="block_4",
)
await skyvern.extract(
prompt="Get a summary of the page",
schema={
"type": "object",
"properties": {
"summary": {
"type": "string",
"description": "A concise summary of the main content or purpose of the page"
}
},
"required": [
"summary"
]
},
label="block_5",
)
```
"""
# Extract loop configuration.
# For loops can reference values in two ways:
# 1. loop_variable_reference — references another block's output (e.g., "extract_rows")
# 2. loop_over — a workflow parameter object with a "key" field (e.g., {"key": "items", ...})
# The script passes this to skyvern.loop(values=...) which resolves it at runtime.
loop_over_parameter_key = block.get("loop_variable_reference") or ""
if not loop_over_parameter_key:
loop_over = block.get("loop_over")
if isinstance(loop_over, dict) and loop_over.get("key"):
loop_over_parameter_key = loop_over["key"]
loop_blocks = block.get("loop_blocks", [])
# Create the for loop target (current_value)
target = cst.Name("current_value")
# Build body statements from loop_blocks
body_statements = []
# Add loop_data assignment as the first statement
for loop_block in loop_blocks:
stmt = _build_block_statement(loop_block)
body_statements.append(stmt)
# create skyvern.loop(loop_over_parameter_key, label=block_title)
loop_call_args = [cst.Arg(keyword=cst.Name("values"), value=_value(loop_over_parameter_key))]
if block.get("complete_if_empty"):
loop_call_args.append(
cst.Arg(keyword=cst.Name("complete_if_empty"), value=_value(block.get("complete_if_empty")))
)
loop_call_args.append(cst.Arg(keyword=cst.Name("label"), value=_value(block_title)))
loop_call = cst.Call(
func=cst.Attribute(value=cst.Name("skyvern"), attr=cst.Name("loop")),
args=loop_call_args,
)
# Create the async for loop
for_loop = cst.For(
target=target,
iter=loop_call,
body=cst.IndentedBlock(body=body_statements),
asynchronous=cst.Asynchronous(),
whitespace_after_for=cst.SimpleWhitespace(" "),
whitespace_before_in=cst.SimpleWhitespace(" "),
whitespace_after_in=cst.SimpleWhitespace(" "),
whitespace_before_colon=cst.SimpleWhitespace(""),
)
return for_loop
def _mark_last_arg_as_comma(args: list[cst.Arg]) -> None:
if not args:
return
last_arg = args.pop()
new_arg = cst.Arg(
keyword=last_arg.keyword,
value=last_arg.value,
comma=cst.Comma(),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
),
)
args.append(new_arg)
def __build_base_task_statement(
block_title: str,
block: dict[str, Any],
data_variable_name: str | None = None,
value_to_param: dict[str, str] | None = None,
) -> list[cst.Arg]:
block_type = block.get("block_type")
prompt = block.get("prompt") if block_type == "task_v2" else block.get("navigation_goal")
# add parameters to prompt
parameters = block.get("parameters", [])
navigation_payload = {}
# make all parameters as jinja2 template parameters in the generated code
for parameter in parameters:
parameter_key = parameter["key"]
navigation_payload[parameter_key] = "{{" + parameter_key + "}}"
if navigation_payload:
prompt = prompt or ""
prompt = f"{prompt}\n{navigation_payload}"
# Try to parameterize PII in the prompt with f-string context.parameters refs.
prompt_value: cst.BaseExpression | None = None
if value_to_param and prompt:
prompt_value = _build_parameterized_prompt_cst(prompt, value_to_param)
if prompt_value is None:
prompt_value = _value(prompt)
args = [
cst.Arg(
keyword=cst.Name("prompt"),
value=prompt_value,
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
),
]
if block.get("url"):
args.append(
cst.Arg(
keyword=cst.Name("url"),
value=_value(block.get("url", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
max_steps = block.get("max_steps") if block_type == "task_v2" else block.get("max_steps_per_run")
if max_steps:
args.append(
cst.Arg(
keyword=cst.Name("max_steps"),
value=_value(max_steps or settings.MAX_STEPS_PER_RUN),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("download_suffix"):
args.append(
cst.Arg(
keyword=cst.Name("download_suffix"),
value=_value(block.get("download_suffix")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("totp_identifier"):
args.append(
cst.Arg(
keyword=cst.Name("totp_identifier"),
value=_value(block.get("totp_identifier", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("totp_verification_url"):
args.append(
cst.Arg(
keyword=cst.Name("totp_url"),
value=_value(block.get("totp_verification_url", "")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("model"):
args.append(
cst.Arg(
keyword=cst.Name("model"),
value=_value(block.get("model")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
# Add error_code_mapping if it exists
if block.get("error_code_mapping") is not None:
args.append(
cst.Arg(
keyword=cst.Name("error_code_mapping"),
value=_value(block.get("error_code_mapping")),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
if block.get("block_type") == "task_v2":
args.append(
cst.Arg(
keyword=cst.Name("engine"),
value=_value("skyvern-2.0"),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
)
args.append(
cst.Arg(
keyword=cst.Name("label"),
value=_value(block_title),
whitespace_after_arg=cst.ParenthesizedWhitespace(
indent=True,
),
comma=cst.Comma(),
)
)
return args
# --------------------------------------------------------------------- #
# 4. function builders #
# --------------------------------------------------------------------- #
def _build_block_statement(
block: dict[str, Any],
data_variable_name: str | None = None,
assign_output: bool = False,
value_to_param: dict[str, str] | None = None,
) -> cst.SimpleStatementLine:
"""Build a block statement."""
block_type = block.get("block_type")
block_title = block.get("label") or block.get("title") or f"block_{block.get('workflow_run_block_id')}"
if block_type in SCRIPT_TASK_BLOCKS:
# For task blocks, call the custom function with cache_key
if block_type == "task":
stmt = _build_run_task_statement(block_title, block, data_variable_name, value_to_param=value_to_param)
elif block_type == "file_download":
stmt = _build_download_statement(block_title, block, data_variable_name, value_to_param=value_to_param)
elif block_type == "action":
stmt = _build_action_statement(block_title, block, data_variable_name)
elif block_type == "login":
stmt = _build_login_statement(block_title, block, data_variable_name, value_to_param=value_to_param)
elif block_type == "extraction":
stmt = _build_extract_statement(block_title, block, data_variable_name, assign_output)
elif block_type == "navigation":
stmt = _build_navigate_statement(block_title, block, data_variable_name, value_to_param=value_to_param)
elif block_type == "validation":
stmt = _build_validate_statement(block_title, block, data_variable_name)
elif block_type == "human_interaction":
stmt = _build_human_interaction_statement(block)
elif block_type == "task_v2":
stmt = _build_run_task_statement(block_title, block, data_variable_name, value_to_param=value_to_param)
elif block_type == "send_email":
stmt = _build_send_email_statement(block)
elif block_type == "text_prompt":
stmt = _build_prompt_statement(block)
elif block_type == "wait":
stmt = _build_wait_statement(block)
elif block_type == "for_loop":
stmt = _build_for_loop_statement(block_title, block)
elif block_type == "goto_url":
stmt = _build_goto_statement(block, data_variable_name)
elif block_type == "code":
stmt = _build_code_statement(block)
elif block_type == "file_upload":
stmt = _build_file_upload_statement(block)
elif block_type == "file_url_parser":
stmt = _build_file_url_parser_statement(block)
elif block_type == "http_request":
stmt = _build_http_request_statement(block)
elif block_type == "pdf_parser":
stmt = _build_pdf_parser_statement(block)
elif block_type == "workflow_trigger":
stmt = _build_workflow_trigger_statement(block)
elif block_type == "conditional":
# Conditional blocks are evaluated at runtime by the workflow engine.
# Generate a descriptive comment showing this is a runtime branch point.
# The blocks inside conditional branches are processed separately when executed.
branches = block.get("branches") or block.get("ordered_branches") or []
branch_info_lines = []
for i, branch in enumerate(branches):
next_label = branch.get("next_block_label", "?")
condition = branch.get("condition", "")
# Truncate long conditions for readability
if len(condition) > 50:
condition = condition[:47] + "..."
branch_info_lines.append(f"# Branch {i + 1}: {condition!r} → {next_label}")
if branch_info_lines:
branch_info = "\n".join(branch_info_lines)
comment_text = f"# === CONDITIONAL: {block_title} ===\n# Evaluated at runtime by workflow engine. One branch executes:\n{branch_info}"
else:
comment_text = f"# === CONDITIONAL: {block_title} ===\n# Evaluated at runtime by workflow engine."
stmt = cst.SimpleStatementLine([cst.Expr(cst.SimpleString(repr(comment_text)))])
else:
# Default case for unknown block types - use quoted string literal to avoid libcst validation error
stmt = cst.SimpleStatementLine([cst.Expr(cst.SimpleString(f"'# Unknown block type: {block_type}'"))])
return stmt
def _build_run_fn(blocks: list[dict[str, Any]], wf_req: dict[str, Any]) -> FunctionDef:
body = [
cst.parse_statement(
"parameters = parameters.model_dump() if isinstance(parameters, WorkflowParameters) else parameters"
),
cst.parse_statement("page, context = await skyvern.setup(parameters, GeneratedWorkflowParameters)"),
]
# Build lookup for conditional branch resolution
blocks_by_label: dict[str, dict[str, Any]] = {label: b for b in blocks if (label := b.get("label")) is not None}
consumed_labels: set[str] = set()
for block in blocks:
label = block.get("label")
if label and label in consumed_labels:
continue
if block.get("block_type") == "conditional":
stmts = _build_conditional_statement(block, blocks_by_label, consumed_labels)
body.extend(stmts)
else:
stmt = _build_block_statement(block, assign_output=False)
body.append(stmt)
params = cst.Parameters(
params=[
Param(
name=cst.Name("parameters"),
annotation=cst.Annotation(
cst.BinaryOperation(
left=cst.Name("WorkflowParameters"),
operator=cst.BitOr(
whitespace_before=cst.SimpleWhitespace(" "),
whitespace_after=cst.SimpleWhitespace(" "),
),
right=cst.Subscript(
value=cst.Name("dict"),
slice=[
cst.SubscriptElement(
slice=cst.Index(value=cst.Name("str")),
comma=cst.Comma(whitespace_after=cst.SimpleWhitespace(" ")),
),
cst.SubscriptElement(
slice=cst.Index(value=cst.Name("Any")),
),
],
),
)
),
whitespace_after_param=cst.ParenthesizedWhitespace(),
comma=cst.Comma(),
),
]
)
return FunctionDef(
name=cst.Name("run_workflow"),
asynchronous=cst.Asynchronous(),
decorators=[_workflow_decorator(wf_req)],
params=params,
body=cst.IndentedBlock(body),
whitespace_before_params=cst.ParenthesizedWhitespace(
indent=True,
last_line=cst.SimpleWhitespace(INDENT),
),
)
# --------------------------------------------------------------------- #
# 5. entrypoint #
# --------------------------------------------------------------------- #
async def generate_workflow_script_python_code(
*,
file_name: str,
workflow_run_request: dict[str, Any],
workflow: dict[str, Any],
blocks: list[dict[str, Any]],
actions_by_task: dict[str, list[dict[str, Any]]],
task_v2_child_blocks: dict[str, list[dict[str, Any]]] | None = None,
organization_id: str | None = None,
run_id: str | None = None,
script_id: str | None = None,
script_revision_id: str | None = None,
pending: bool = False,
cached_blocks: dict[str, ScriptBlockSource] | None = None,
updated_block_labels: set[str] | None = None,
use_semantic_selectors: bool = False,
adaptive_caching: bool = False,
) -> str:
"""
Build a LibCST Module and emit .code (PEP-8-formatted source).
Cached script blocks can be reused by providing them via `cached_blocks`. Any labels present in
`updated_block_labels` will be regenerated from the latest workflow run execution data.
"""
cached_blocks = cached_blocks or {}
updated_block_labels = set(updated_block_labels or [])
# Drop cached entries that do not have usable source
cached_blocks = {label: source for label, source in cached_blocks.items() if source.code}
# Always regenerate the orchestrator block so it stays aligned with the workflow definition
cached_blocks.pop(settings.WORKFLOW_START_BLOCK_LABEL, None)
if task_v2_child_blocks is None:
task_v2_child_blocks = {}
# --- imports --------------------------------------------------------
imports: list[cst.BaseStatement] = [
cst.SimpleStatementLine([cst.Import(names=[cst.ImportAlias(cst.Name("asyncio"))])]),
cst.SimpleStatementLine([cst.Import(names=[cst.ImportAlias(cst.Name("pydantic"))])]),
cst.SimpleStatementLine(
[
cst.ImportFrom(
module=cst.Name("typing"),
names=[
cst.ImportAlias(cst.Name("Any")),
],
)
]
),
cst.SimpleStatementLine(
[
cst.ImportFrom(
module=cst.Name("pydantic"),
names=[
cst.ImportAlias(cst.Name("BaseModel")),
cst.ImportAlias(cst.Name("Field")),
],
)
]
),
cst.SimpleStatementLine([cst.Import(names=[cst.ImportAlias(cst.Name("skyvern"))])]),
cst.SimpleStatementLine(
[
cst.ImportFrom(
module=cst.Name("skyvern"),
names=[
cst.ImportAlias(cst.Name("RunContext")),
cst.ImportAlias(cst.Name("SkyvernPage")),
],
)
]
),
]
# --- generate schema and hydrate actions ---------------------------
# Build existing field assignments from cached blocks to preserve field names
# for unchanged blocks, preventing schema mismatches with cached code
existing_field_assignments = _build_existing_field_assignments(
blocks=blocks,
actions_by_task=actions_by_task,
cached_blocks=cached_blocks,
updated_block_labels=updated_block_labels,
)
generated_schema, field_mappings = await generate_workflow_parameters_schema(
actions_by_task, existing_field_assignments
)
actions_by_task = hydrate_input_text_actions_with_field_names(actions_by_task, field_mappings)
# Build a lookup from literal parameter values to field names so that click
# prompt strings can be parameterized (e.g. patient ID → context.parameters[...]).
value_to_param = _build_value_to_param_lookup(actions_by_task)
# --- class + cached params -----------------------------------------
model_cls = _build_model(workflow)
generated_model_cls = _build_generated_model_from_schema(generated_schema)
# --- blocks ---------------------------------------------------------
block_fns: list[cst.CSTNode] = []
task_v1_blocks = [block for block in blocks if block["block_type"] in SCRIPT_TASK_BLOCKS]
task_v2_blocks = [block for block in blocks if block["block_type"] == "task_v2"]
def append_block_code(block_code: str) -> None:
nonlocal block_fns
parsed = cst.parse_module(block_code)
if block_fns:
block_fns.append(cst.EmptyLine())
block_fns.append(cst.EmptyLine())
block_fns.extend(parsed.body)
# Handle task v1 blocks (excluding child blocks of task_v2)
for idx, task in enumerate(task_v1_blocks):
if task.get("parent_task_v2_label"):
continue
block_name = task.get("label") or task.get("title") or task.get("task_id") or f"task_{idx}"
cached_source = cached_blocks.get(block_name)
use_cached = cached_source is not None and block_name not in updated_block_labels
input_fields = _collect_block_input_fields(task, actions_by_task)
if not input_fields and cached_source and cached_source.input_fields:
input_fields = cached_source.input_fields
if use_cached:
assert cached_source is not None
block_code = cached_source.code
run_signature = cached_source.run_signature
block_workflow_run_id = cached_source.workflow_run_id
block_workflow_run_block_id = cached_source.workflow_run_block_id
else:
task_id = task.get("task_id", "")
block_actions = actions_by_task.get(task_id, [])
# Skip blocks that have no actions AND no task_id — they haven't executed yet.
# Creating script_block entries for actionless blocks causes a permanent
# stuck state where generate_script_if_needed thinks they're cached but
# the Python file has no code for them. (SKY-8443)
# Note: a block WITH task_id but zero actions is valid — it means the block
# executed but completed immediately (e.g., page.complete() with no interaction).
# _build_block_fn handles this correctly by generating a minimal function.
if not block_actions and not task_id:
LOG.debug(
"Skipping block with no actions and no task_id — not yet executed",
block_label=block_name,
script_id=script_id,
)
continue
block_fn_def = _build_block_fn(
task,
block_actions,
value_to_param=value_to_param,
use_semantic_selectors=use_semantic_selectors,
all_blocks=task_v1_blocks,
)
temp_module = cst.Module(body=[block_fn_def])
block_code = temp_module.code
# run_signature is executed in a scope without `context`, so don't
# parameterize the prompt — use literal strings instead.
block_stmt = _build_block_statement(task, value_to_param=None)
run_signature_module = cst.Module(body=[block_stmt])
run_signature = run_signature_module.code.strip()
block_workflow_run_id = task.get("workflow_run_id") or run_id
block_workflow_run_block_id = task.get("workflow_run_block_id")
if script_id and script_revision_id and organization_id:
try:
await create_or_update_script_block(
block_code=block_code,
script_revision_id=script_revision_id,
script_id=script_id,
organization_id=organization_id,
block_label=block_name,
update=pending,
run_signature=run_signature,
workflow_run_id=block_workflow_run_id,
workflow_run_block_id=block_workflow_run_block_id,
input_fields=input_fields,
)
except Exception as e:
LOG.error("Failed to create script block", error=str(e), exc_info=True)
append_block_code(block_code)
# Handle task_v2 blocks
for task_v2 in task_v2_blocks:
task_v2_label = task_v2.get("label") or f"task_v2_{task_v2.get('workflow_run_block_id')}"
child_blocks = task_v2_child_blocks.get(task_v2_label, [])
cached_source = cached_blocks.get(task_v2_label)
use_cached = cached_source is not None and task_v2_label not in updated_block_labels
input_fields = _collect_block_input_fields(task_v2, actions_by_task)
if not input_fields and cached_source and cached_source.input_fields:
input_fields = cached_source.input_fields
block_code = ""
run_signature = None
block_workflow_run_id = task_v2.get("workflow_run_id") or run_id
block_workflow_run_block_id = task_v2.get("workflow_run_block_id")
if use_cached:
assert cached_source is not None
block_code = cached_source.code
run_signature = cached_source.run_signature
block_workflow_run_id = cached_source.workflow_run_id
block_workflow_run_block_id = cached_source.workflow_run_block_id
else:
# Skip task_v2 blocks that haven't executed (no child workflow run).
# Same rationale as task_v1 guard — prevents phantom script_block entries. (SKY-8443)
if not child_blocks and not task_v2.get("block_workflow_run_id"):
LOG.debug(
"Skipping task_v2 block with no child blocks — not yet executed",
block_label=task_v2_label,
script_id=script_id,
)
continue
task_v2_fn_def = _build_task_v2_block_fn(task_v2, child_blocks)
task_v2_block_body: list[cst.CSTNode] = [task_v2_fn_def]
for child_block in child_blocks:
if child_block.get("block_type") in SCRIPT_TASK_BLOCKS and child_block.get("block_type") != "task_v2":
child_fn_def = _build_block_fn(
child_block,
actions_by_task.get(child_block.get("task_id", ""), []),
value_to_param=value_to_param,
use_semantic_selectors=use_semantic_selectors,
)
task_v2_block_body.append(cst.EmptyLine())
task_v2_block_body.append(cst.EmptyLine())
task_v2_block_body.append(child_fn_def)
temp_module = cst.Module(body=task_v2_block_body)
block_code = temp_module.code
task_v2_stmt = _build_block_statement(task_v2, value_to_param=None)
run_signature = cst.Module(body=[task_v2_stmt]).code.strip()
if script_id and script_revision_id and organization_id:
try:
await create_or_update_script_block(
block_code=block_code,
script_revision_id=script_revision_id,
script_id=script_id,
organization_id=organization_id,
block_label=task_v2_label,
update=pending,
run_signature=run_signature,
workflow_run_id=block_workflow_run_id,
workflow_run_block_id=block_workflow_run_block_id,
input_fields=input_fields,
)
except Exception as e:
LOG.error("Failed to create task_v2 script block", error=str(e), exc_info=True)
append_block_code(block_code)
# Handle for_loop blocks
# ForLoop blocks need script_block entries with run_signature so they can be executed via cached scripts
for_loop_blocks = [block for block in blocks if block["block_type"] == "for_loop"]
for for_loop_block in for_loop_blocks:
for_loop_label = for_loop_block.get("label") or f"for_loop_{for_loop_block.get('workflow_run_block_id')}"
cached_source = cached_blocks.get(for_loop_label)
use_cached = cached_source is not None and for_loop_label not in updated_block_labels
block_workflow_run_id = for_loop_block.get("workflow_run_id") or run_id
block_workflow_run_block_id = for_loop_block.get("workflow_run_block_id")
if use_cached:
assert cached_source is not None
block_code = cached_source.code
run_signature = cached_source.run_signature
block_workflow_run_id = cached_source.workflow_run_id
block_workflow_run_block_id = cached_source.workflow_run_block_id
else:
# Build the for loop statement
for_loop_stmt = _build_for_loop_statement(for_loop_label, for_loop_block)
temp_module = cst.Module(body=[for_loop_stmt])
block_code = temp_module.code
run_signature = block_code.strip()
if script_id and script_revision_id and organization_id:
try:
await create_or_update_script_block(
block_code=block_code,
script_revision_id=script_revision_id,
script_id=script_id,
organization_id=organization_id,
block_label=for_loop_label,
update=pending,
run_signature=run_signature,
workflow_run_id=block_workflow_run_id,
workflow_run_block_id=block_workflow_run_block_id,
input_fields=None,
)
except Exception as e:
LOG.error("Failed to create for_loop script block", error=str(e), exc_info=True)
# NOTE: Do NOT call append_block_code() for for_loop blocks.
# Unlike task blocks (which produce function definitions valid at module level),
# for_loop blocks produce bare `async for` statements that cause SyntaxError
# at module level ("async for outside async function"). The for-loop code is
# already correctly inlined inside run_workflow() via _build_block_statement().
# Generate cached function bodies for for_loop inner blocks.
# Inner blocks (e.g. extraction inside a loop) are nested in loop_blocks and
# are NOT in the top-level blocks list, so they need separate processing here.
# This follows the same pattern as task_v2 child block handling (lines 2704-2714).
for loop_block in for_loop_block.get("loop_blocks", []):
if loop_block.get("block_type") not in SCRIPT_TASK_BLOCKS:
continue
inner_label = (
loop_block.get("label") or loop_block.get("title") or f"block_{loop_block.get('workflow_run_block_id')}"
)
# Check if already cached (for progressive caching)
cached_inner = cached_blocks.get(inner_label)
use_inner_cached = cached_inner is not None and inner_label not in updated_block_labels
if use_inner_cached:
assert cached_inner is not None
inner_block_code = cached_inner.code
inner_run_signature = cached_inner.run_signature
inner_wrbi = cached_inner.workflow_run_block_id
inner_wri = cached_inner.workflow_run_id
else:
inner_actions = actions_by_task.get(loop_block.get("task_id", ""), [])
if not inner_actions:
continue # No actions from agent run = can't generate cached function
inner_fn_def = _build_block_fn(
loop_block,
inner_actions,
value_to_param=value_to_param,
use_semantic_selectors=use_semantic_selectors,
is_in_for_loop=True,
)
inner_block_code = cst.Module(body=[inner_fn_def]).code
inner_stmt = _build_block_statement(loop_block, value_to_param=None)
inner_run_signature = cst.Module(body=[inner_stmt]).code.strip()
inner_wrbi = loop_block.get("workflow_run_block_id")
inner_wri = loop_block.get("workflow_run_id") or run_id
# Create script_block entry for preservation across regenerations
if script_id and script_revision_id and organization_id:
try:
inner_input_fields = _collect_block_input_fields(loop_block, actions_by_task)
if not inner_input_fields and cached_inner and cached_inner.input_fields:
inner_input_fields = cached_inner.input_fields
await create_or_update_script_block(
block_code=inner_block_code,
script_revision_id=script_revision_id,
script_id=script_id,
organization_id=organization_id,
block_label=inner_label,
update=pending,
run_signature=inner_run_signature,
workflow_run_id=inner_wri,
workflow_run_block_id=inner_wrbi,
input_fields=inner_input_fields,
)
except Exception as e:
LOG.error(
"Failed to create for_loop inner block script block",
error=str(e),
block_label=inner_label,
exc_info=True,
)
append_block_code(inner_block_code)
# --- agent-required blocks (adaptive caching) -----------------------
# Structural blocks (conditional, text_prompt, wait) can't be code-generated
# initially. Create script_block entries with requires_agent=True so the runtime
# knows to execute them via agent even when ai_fallback=False. The script reviewer
# can later upgrade these to code by setting requires_agent=False and providing
# a run_signature.
_AGENT_REQUIRED_BLOCK_TYPES = {"conditional", "text_prompt", "wait"}
if adaptive_caching:
agent_required_blocks = [b for b in blocks if b["block_type"] in _AGENT_REQUIRED_BLOCK_TYPES]
for arb in agent_required_blocks:
arb_label = arb.get("label") or f"{arb['block_type']}_{arb.get('workflow_run_block_id')}"
# Check if the reviewer has already provided code for this block
cached_source = cached_blocks.get(arb_label)
if cached_source and cached_source.run_signature and not cached_source.requires_agent:
# Reviewer upgraded this block to code — preserve it
if script_id and script_revision_id and organization_id:
try:
await create_or_update_script_block(
block_code=cached_source.code,
script_revision_id=script_revision_id,
script_id=script_id,
organization_id=organization_id,
block_label=arb_label,
update=pending,
run_signature=cached_source.run_signature,
workflow_run_id=cached_source.workflow_run_id,
workflow_run_block_id=cached_source.workflow_run_block_id,
requires_agent=False,
)
except Exception as e:
LOG.error("Failed to preserve coded agent block", error=str(e), exc_info=True)
append_block_code(cached_source.code)
else:
# Create a requires_agent entry (no code, no run_signature)
placeholder_code = f"# Block '{arb_label}' ({arb['block_type']}) — executed via agent"
if script_id and script_revision_id and organization_id:
try:
await create_or_update_script_block(
block_code=placeholder_code,
script_revision_id=script_revision_id,
script_id=script_id,
organization_id=organization_id,
block_label=arb_label,
update=pending,
run_signature=None,
requires_agent=True,
)
except Exception as e:
LOG.error("Failed to create agent-required script block", error=str(e), exc_info=True)
# --- preserve cached blocks from unexecuted branches ----------------
# When a workflow has conditional blocks, not all branches execute in a single run.
# transform_workflow_run_to_code_gen_input() only returns blocks that executed,
# so cached blocks from unexecuted branches would be lost during regeneration
# if we don't explicitly preserve them here.
processed_labels: set[str] = set()
for task in task_v1_blocks:
label = task.get("label") or task.get("title") or task.get("task_id")
if label:
processed_labels.add(label)
for task_v2 in task_v2_blocks:
label = task_v2.get("label") or f"task_v2_{task_v2.get('workflow_run_block_id')}"
processed_labels.add(label)
for flb in for_loop_blocks:
label = flb.get("label") or f"for_loop_{flb.get('workflow_run_block_id')}"
processed_labels.add(label)
# Also track inner block labels to prevent duplication in the
# "preserve unexecuted branch" section below
for lb in flb.get("loop_blocks", []):
inner_lbl = lb.get("label") or lb.get("title")
if inner_lbl:
processed_labels.add(inner_lbl)
if adaptive_caching:
for arb in [b for b in blocks if b["block_type"] in _AGENT_REQUIRED_BLOCK_TYPES]:
arb_label = arb.get("label") or f"{arb['block_type']}_{arb.get('workflow_run_block_id')}"
processed_labels.add(arb_label)
preserved_count = 0
for cached_label, cached_source in cached_blocks.items():
if cached_label in processed_labels:
continue # Already processed above
if not cached_source.code or not cached_source.run_signature:
continue # Skip entries without usable code/metadata
if script_id and script_revision_id and organization_id:
try:
await create_or_update_script_block(
block_code=cached_source.code,
script_revision_id=script_revision_id,
script_id=script_id,
organization_id=organization_id,
block_label=cached_label,
update=pending,
run_signature=cached_source.run_signature,
workflow_run_id=cached_source.workflow_run_id,
workflow_run_block_id=cached_source.workflow_run_block_id,
input_fields=cached_source.input_fields,
)
except Exception as e:
LOG.error(
"Failed to preserve cached script block from unexecuted branch",
error=str(e),
block_label=cached_label,
exc_info=True,
)
append_block_code(cached_source.code)
preserved_count += 1
if preserved_count > 0:
LOG.info(
"Preserved cached blocks from unexecuted branches during regeneration",
preserved_count=preserved_count,
preserved_labels=[
label
for label in cached_blocks
if label not in processed_labels and cached_blocks[label].code and cached_blocks[label].run_signature
],
)
# --- runner ---------------------------------------------------------
run_fn = _build_run_fn(blocks, workflow_run_request)
# --- create __start_block__ -----------------------------------------
# Build the __start_block__ content that combines imports, model classes, and run function
start_block_body = [
*imports,
cst.EmptyLine(),
cst.EmptyLine(),
model_cls,
cst.EmptyLine(),
cst.EmptyLine(),
]
# Add generated model class if available
if generated_model_cls:
start_block_body.extend(
[
generated_model_cls,
cst.EmptyLine(),
cst.EmptyLine(),
]
)
# Add run function to start block
start_block_body.extend(
[
run_fn,
cst.EmptyLine(),
cst.EmptyLine(),
]
)
# Create script block for __start_block__ if we have script context
if script_id and script_revision_id and organization_id:
try:
# Create a temporary module to convert the start block content to a function
start_block_module = cst.Module(body=start_block_body)
start_block_code = start_block_module.code
await create_or_update_script_block(
block_code=start_block_code,
script_revision_id=script_revision_id,
script_id=script_id,
organization_id=organization_id,
block_label=settings.WORKFLOW_START_BLOCK_LABEL,
update=pending,
)
except Exception as e:
LOG.error("Failed to create __start_block__", error=str(e), exc_info=True)
# Continue without script block creation if it fails
# Build module body with the start block content and other blocks
module_body = [
*start_block_body,
*block_fns,
]
module = cst.Module(body=module_body)
return module.code
async def create_or_update_script_block(
block_code: str | bytes,
script_revision_id: str,
script_id: str,
organization_id: str,
block_label: str,
update: bool = False,
run_signature: str | None = None,
workflow_run_id: str | None = None,
workflow_run_block_id: str | None = None,
input_fields: list[str] | None = None,
requires_agent: bool | None = None,
) -> None:
"""
Create a script block in the database and save the block code to a script file.
If update is True, the script block will be updated instead of created.
Args:
block_code: The code to save
script_revision_id: The script revision ID
script_id: The script ID
organization_id: The organization ID
block_label: Optional custom name for the block (defaults to function name)
update: Whether to update the script block instead of creating a new one
run_signature: The function call code to execute this block (e.g., "await skyvern.action(...)")
workflow_run_id: The workflow run that generated this cached block
workflow_run_block_id: The workflow run block that generated this cached block
input_fields: Workflow parameter field names referenced by this block's cached actions
requires_agent: Whether this block must be executed via agent (None = don't change on update)
"""
block_code_bytes = block_code if isinstance(block_code, bytes) else block_code.encode("utf-8")
try:
# Step 3: Create script block in database
script_block = await app.DATABASE.scripts.get_script_block_by_label(
organization_id=organization_id,
script_revision_id=script_revision_id,
script_block_label=block_label,
)
if not script_block:
script_block = await app.DATABASE.scripts.create_script_block(
script_revision_id=script_revision_id,
script_id=script_id,
organization_id=organization_id,
script_block_label=block_label,
run_signature=run_signature,
workflow_run_id=workflow_run_id,
workflow_run_block_id=workflow_run_block_id,
input_fields=input_fields,
requires_agent=requires_agent if requires_agent is not None else False,
)
elif any(
value is not None
for value in [run_signature, workflow_run_id, workflow_run_block_id, input_fields, requires_agent]
):
# Update metadata when new values are provided
script_block = await app.DATABASE.scripts.update_script_block(
script_block_id=script_block.script_block_id,
organization_id=organization_id,
run_signature=run_signature,
workflow_run_id=workflow_run_id,
workflow_run_block_id=workflow_run_block_id,
input_fields=input_fields,
requires_agent=requires_agent,
)
# Step 4: Create script file for the block
# Generate a unique filename for the block
file_name = f"{block_label}.skyvern"
file_path = f"blocks/{file_name}"
# Create artifact and upload to S3
artifact_id = None
if update and script_block.script_file_id:
script_file = await app.DATABASE.scripts.get_script_file_by_id(
script_revision_id,
script_block.script_file_id,
organization_id,
)
if script_file and script_file.artifact_id:
artifact = await app.DATABASE.artifacts.get_artifact_by_id(script_file.artifact_id, organization_id)
if artifact:
asyncio.create_task(app.STORAGE.store_artifact(artifact, block_code_bytes))
else:
LOG.error("Script file or artifact not found", script_file_id=script_block.script_file_id)
else:
artifact_id = await app.ARTIFACT_MANAGER.create_script_file_artifact(
organization_id=organization_id,
script_id=script_id,
script_version=1, # Assuming version 1 for now
file_path=file_path,
data=block_code_bytes,
)
# Create script file record
script_file = await app.DATABASE.scripts.create_script_file(
script_revision_id=script_revision_id,
script_id=script_id,
organization_id=organization_id,
file_path=file_path,
file_name=file_name,
file_type="file",
content_hash=f"sha256:{hashlib.sha256(block_code_bytes).hexdigest()}",
file_size=len(block_code_bytes),
mime_type="text/x-python",
artifact_id=artifact_id,
)
# update script block with script file id
await app.DATABASE.scripts.update_script_block(
script_block_id=script_block.script_block_id,
organization_id=organization_id,
script_file_id=script_file.file_id,
workflow_run_id=workflow_run_id,
workflow_run_block_id=workflow_run_block_id,
)
except Exception as e:
# Log error but don't fail the entire generation process
LOG.error("Failed to create script block", error=str(e), exc_info=True)
# For now, just log the error and continue
# In production, you might want to handle this differently
Reference in a new issue View git blame Copy permalink