def Scrape(urls, outdir, size, pos, timeout=20, **kwargs):
"""Invoke a browser, send it to a series of URLs, and save its output.
Args:
urls: list of URLs to scrape
outdir: directory to place output
size: size of browser window to use
pos: position of browser window
timeout: amount of time to wait for page to load
kwargs: miscellaneous keyword args
Returns:
None if success, else an error string
"""
path = r"c:\program files\internet explorer\iexplore.exe"
if "path" in kwargs and kwargs["path"]: path = kwargs["path"]
(iewnd, ieproc, address_bar, render_pane,
tab_window) = (InvokeBrowser(path))
# Resize and reposition the frame
windowing.MoveAndSizeWindow(iewnd, pos, size, render_pane)
# Visit each URL we're given
if type(urls) in types.StringTypes: urls = [urls]
timedout = False
for url in urls:
# Double-click in the address bar, type the name, and press Enter
mouse.DoubleClickInWindow(address_bar)
keyboard.TypeString(url)
keyboard.TypeString("\n")
# Wait for the page to finish loading
load_time = windowing.WaitForThrobber(tab_window, (6, 8, 22, 24),
timeout)
timedout = load_time < 0
if timedout:
break
# Scrape the page
image = windowing.ScrapeWindow(render_pane)
# Save to disk
if "filename" in kwargs:
if callable(kwargs["filename"]):
filename = kwargs["filename"](url)
else:
filename = kwargs["filename"]
else:
filename = windowing.URLtoFilename(url, outdir, ".bmp")
image.save(filename)
windowing.EndProcess(ieproc)
if timedout:
return "timeout"
def Time(urls, size, timeout, **kwargs):
"""Measure how long it takes to load each of a series of URLs
Args:
urls: list of URLs to time
size: size of browser window to use
timeout: amount of time to wait for page to load
kwargs: miscellaneous keyword args
Returns:
A list of tuples (url, time). "time" can be "crashed" or "timeout"
"""
if "path" in kwargs and kwargs["path"]: path = kwargs["path"]
else: path = DEFAULT_PATH
proc = None
# Visit each URL we're given
if type(urls) in types.StringTypes: urls = [urls]
ret = []
for url in urls:
try:
# Invoke the browser if necessary
if not proc:
(wnd, proc, address_bar, render_pane,
tab_window) = InvokeBrowser(path)
# Resize and reposition the frame
windowing.MoveAndSizeWindow(wnd, (0, 0), size, render_pane)
# Double-click in the address bar, type the name, and press Enter
mouse.DoubleClickInWindow(address_bar)
keyboard.TypeString(url)
keyboard.TypeString("\n")
# Wait for the page to finish loading
load_time = windowing.WaitForThrobber(tab_window, (6, 8, 22, 24),
timeout)
timedout = load_time < 0
if timedout:
load_time = "timeout"
# Send an alt-F4 to make the browser close; if this times out,
# we've probably got a crash
keyboard.TypeString(r"{\4}", use_modifiers=True)
if not windowing.WaitForProcessExit(proc, timeout):
windowing.EndProcess(proc)
load_time = "crashed"
proc = None
except pywintypes.error:
load_time = "crashed"
proc = None
ret.append((url, load_time))
# Send an alt-F4 to make the browser close; if this times out,
# we've probably got a crash
if proc:
keyboard.TypeString(r"{\4}", use_modifiers=True)
if not windowing.WaitForProcessExit(proc, timeout):
windowing.EndProcess(proc)
return ret