From 76a256ffaaf99e3933ccbd93c367b68643de0885 Mon Sep 17 00:00:00 2001
From: Shuchang Zheng <wintonzheng0325@gmail.com>
Date: Fri, 20 Sep 2024 19:21:00 -0700
Subject: [PATCH] skip large size svg (#869)

Co-authored-by: LawyZheng <lawyzheng1106@gmail.com>
---
 skyvern/config.py                 |  2 ++
 skyvern/forge/agent_functions.py  | 15 +++++++++++++++
 skyvern/webeye/scraper/scraper.py |  6 ++++++
 3 files changed, 23 insertions(+)

diff --git a/skyvern/config.py b/skyvern/config.py
index ce2eb7da..869a3fdd 100644
--- a/skyvern/config.py
+++ b/skyvern/config.py
@@ -122,6 +122,8 @@ class Settings(BaseSettings):
     VERIFICATION_CODE_INITIAL_WAIT_TIME_SECS: int = 40
     VERIFICATION_CODE_POLLING_TIMEOUT_MINS: int = 5
 
+    SVG_MAX_LENGTH: int = 100000
+
     def is_cloud_environment(self) -> bool:
         """
         :return: True if env is not local, else False
diff --git a/skyvern/forge/agent_functions.py b/skyvern/forge/agent_functions.py
index f5d29c91..a75362e5 100644
--- a/skyvern/forge/agent_functions.py
+++ b/skyvern/forge/agent_functions.py
@@ -6,6 +6,7 @@ from typing import Dict, List
 import structlog
 from playwright.async_api import Page
 
+from skyvern.config import settings
 from skyvern.constants import SKYVERN_ID_ATTR
 from skyvern.exceptions import StepUnableToExecuteError, SVGConversionFailed
 from skyvern.forge import app
@@ -62,6 +63,9 @@ async def _convert_svg_to_string(task: Task, step: Step, organization: Organizat
     if element.get("tagName") != "svg":
         return
 
+    if element.get("isDropped", False):
+        return
+
     element_id = element.get("id", "")
     svg_element = _remove_skyvern_attributes(element)
     svg_html = json_to_html(svg_element)
@@ -83,6 +87,17 @@ async def _convert_svg_to_string(task: Task, step: Step, organization: Organizat
     if svg_shape:
         LOG.debug("SVG loaded from cache", element_id=element_id, shape=svg_shape)
     else:
+        if len(svg_html) > settings.SVG_MAX_LENGTH:
+            # TODO: implement a fallback solution for "too large" case, maybe convert by screenshot
+            LOG.warning(
+                "SVG element is too large to convert, going to drop the svg element.",
+                element_id=element_id,
+                length=len(svg_html),
+            )
+            del element["children"]
+            element["isDropped"] = True
+            return
+
         LOG.debug("call LLM to convert SVG to string shape", element_id=element_id)
         svg_convert_prompt = prompt_engine.load_prompt("svg-convert", svg_element=svg_html)
 
diff --git a/skyvern/webeye/scraper/scraper.py b/skyvern/webeye/scraper/scraper.py
index 790ba816..96d3ec63 100644
--- a/skyvern/webeye/scraper/scraper.py
+++ b/skyvern/webeye/scraper/scraper.py
@@ -86,6 +86,12 @@ def build_attribute(key: str, value: Any) -> str:
 
 
 def json_to_html(element: dict, need_skyvern_attrs: bool = True) -> str:
+    """
+    if element is flagged as dropped, the html format is empty
+    """
+    if element.get("isDropped", False):
+        return ""
+
     attributes: dict[str, Any] = copy.deepcopy(element.get("attributes", {}))
 
     if need_skyvern_attrs: