Automate browser based workflows with AI
Python aiapiautomationbrowserbrowser-automationcomputergptllmplaywrightpowerautomatepuppeteerpythonrpaseleniumvisionworkflow
revert pdf (#2328)
Shuchang Zheng04ed3c8dab84
diff --git a/poetry.lock b/poetry.lockindex a8c0a129f..c6dc972f2 100644--- a/poetry.lock+++ b/poetry.lock@@ -5081,24 +5081,6 @@ files = [[package.extras]windows-terminal = ["colorama (>=0.4.6)"]-[[package]]-name = "pymupdf"-version = "1.25.5"-description = "A high performance Python library for data extraction, analysis, conversion & manipulation of PDF (and other) documents."-optional = false-python-versions = ">=3.9"-groups = ["main"]-files = [- {file = "pymupdf-1.25.5-cp39-abi3-macosx_10_9_x86_64.whl", hash = "sha256:cde4e1c9cfb09c0e1e9c2b7f4b787dd6bb34a32cfe141a4675e24af7c0c25dd3"},- {file = "pymupdf-1.25.5-cp39-abi3-macosx_11_0_arm64.whl", hash = "sha256:5a35e2725fae0ab57f058dff77615c15eb5961eac50ba04f41ebc792cd8facad"},- {file = "pymupdf-1.25.5-cp39-abi3-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:d94b800e9501929c42283d39bc241001dd87fdeea297b5cb40d5b5714534452f"},- {file = "pymupdf-1.25.5-cp39-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:ee22155d3a634642d76553204867d862ae1bdd9f7cf70c0797d8127ebee6bed5"},- {file = "pymupdf-1.25.5-cp39-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:6ed7fc25271004d6d3279c20a80cb2bb4cda3efa9f9088dcc07cd790eca0bc63"},- {file = "pymupdf-1.25.5-cp39-abi3-win32.whl", hash = "sha256:65e18ddb37fe8ec4edcdbebe9be3a8486b6a2f42609d0a142677e42f3a0614f8"},- {file = "pymupdf-1.25.5-cp39-abi3-win_amd64.whl", hash = "sha256:7f44bc3d03ea45b2f68c96464f96105e8c7908896f2fb5e8c04f1fb8dae7981e"},- {file = "pymupdf-1.25.5.tar.gz", hash = "sha256:5f96311cacd13254c905f6654a004a0a2025b71cabc04fda667f5472f72c15a0"},-]-[[package]]name = "pyotp"version = "2.9.0"@@ -7388,4 +7370,4 @@ type = ["pytest-mypy"][metadata]lock-version = "2.1"python-versions = "^3.11,<3.12"-content-hash = "762f5581ccfcd1c18787d26dbabb9744ee58dd0ecc9aaaf97e81d5e66d36448c"+content-hash = "81fb59e4ecd1124ab664c84c633ad015acdfc059fda8495a7fe4d40cad33435f"diff --git a/pyproject.toml b/pyproject.tomlindex ec1291bc6..e2f09860b 100644--- a/pyproject.toml+++ b/pyproject.toml@@ -57,7 +57,6 @@ psutil = ">=7.0.0"tiktoken = ">=0.9.0"anthropic = "^0.50.0"google-cloud-aiplatform = "^1.90.0"-pymupdf = "^1.25.5"[tool.poetry.group.dev.dependencies]isort = "^5.13.2"diff --git a/skyvern/webeye/utils/page.py b/skyvern/webeye/utils/page.pyindex f82b3aef7..69c31d7f0 100644--- a/skyvern/webeye/utils/page.py+++ b/skyvern/webeye/utils/page.py@@ -2,12 +2,9 @@ from __future__ import annotationsimport asyncioimport time-from io import BytesIOfrom typing import Any, Dict, List-import pymupdfimport structlog-from PIL import Imagefrom playwright._impl._errors import TimeoutErrorfrom playwright.async_api import ElementHandle, Frame, Page@@ -37,6 +34,7 @@ JS_FUNCTION_DEFS = load_js_script()async def _current_viewpoint_screenshot_helper(page: Page,file_path: str | None = None,+ full_page: bool = False,timeout: float = settings.BROWSER_SCREENSHOT_TIMEOUT_MS,) -> bytes:if page.is_closed():@@ -50,11 +48,13 @@ async def _current_viewpoint_screenshot_helper(screenshot = await page.screenshot(path=file_path,timeout=timeout,+ full_page=full_page,animations="disabled",)else:screenshot = await page.screenshot(timeout=timeout,+ full_page=full_page,animations="disabled",)end_time = time.time()@@ -150,49 +150,9 @@ class SkyvernFrame:file_path: str | None = None,timeout: float = settings.BROWSER_SCREENSHOT_TIMEOUT_MS,) -> bytes:- if not full_page:- return await _current_viewpoint_screenshot_helper(page=page, file_path=file_path, timeout=timeout)-- LOG.debug("Page is fully loaded, agent is about to generate the full page screenshot")- start_time = time.time()- async with asyncio.timeout(timeout):- pdf_bytes = await page.pdf(- print_background=True, width=f"{settings.BROWSER_WIDTH}px", height=f"{settings.BROWSER_HEIGHT}px"- )-- with pymupdf.open(stream=pdf_bytes, filetype="pdf") as doc:- images = []- for pdf_page in doc:- pix = pdf_page.get_pixmap()- img = Image.frombytes("RGB", [pix.width, pix.height], pix.samples)- images.append(img)-- total_height = sum(img.height for img in images)- max_width = max(img.width for img in images)-- merged_img = Image.new("RGB", (max_width, total_height), color=(255, 255, 255))-- current_y = 0- for img in images:- merged_img.paste(img, (0, current_y))- current_y += img.height-- buffer = BytesIO()- merged_img.save(buffer, format="PNG")- buffer.seek(0)-- img_data = buffer.read()- if file_path is not None:- with open(file_path, "wb") as f:- f.write(img_data)-- end_time = time.time()- LOG.debug(- "Full page screenshot taking time",- screenshot_time=end_time - start_time,- file_path=file_path,- )- return img_data+ return await _current_viewpoint_screenshot_helper(+ page=page, file_path=file_path, full_page=full_page, timeout=timeout+ )@staticmethodasync def take_split_screenshots(