def ScrapeResult(url, proc, wnd, result):
"""Capture and save the scrape."""
if log_file: log_file.write(result)
# Scrape the page
image = windowing.ScrapeWindow(wnd)
filename = windowing.URLtoFilename(url, command["--outdir"], ".bmp")
image.save(filename)
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 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
"""
if "path" in kwargs and kwargs["path"]: path = kwargs["path"]
else: path = DEFAULT_PATH
(wnd, proc, render_pane) = InvokeBrowser(path)
# Resize and reposition the frame
windowing.MoveAndSizeWindow(wnd, pos, size, render_pane)
time.sleep(3)
# Firefox is a bit of a pain: it doesn't use standard edit controls,
# and it doesn't display a throbber when there's no tab. Let's make
# sure there's at least one tab, then select the first one
mouse.ClickInWindow(wnd)
keyboard.TypeString("[t]", True)
mouse.ClickInWindow(wnd, (30, 115))
time.sleep(2)
timedout = False
# Visit each URL we're given
if type(urls) in types.StringTypes: urls = [urls]
for url in urls:
# Use keyboard shortcuts
keyboard.TypeString("{d}", True)
keyboard.TypeString(url)
keyboard.TypeString("\n")
# Wait for the page to finish loading
load_time = windowing.WaitForThrobber(wnd, (10, 96, 26, 112), 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)
# Close all the tabs, cheesily
mouse.ClickInWindow(wnd)
while len(windowing.FindChildWindows(0, "MozillaUIWindowClass")):
keyboard.TypeString("[w]", True)
time.sleep(1)
if timedout:
return "timeout"
def Scrape(urls, outdir, size, pos, timeout, 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
"""
if "path" in kwargs and kwargs["path"]: path = kwargs["path"]
else: path = DEFAULT_PATH
(wnd, proc, address_bar, render_pane) = InvokeBrowser(path)
# Resize and reposition the frame
windowing.MoveAndSizeWindow(wnd, 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.ClickInWindow(address_bar)
keyboard.TypeString(url, 0.1)
keyboard.TypeString("\n")
# Wait for the page to finish loading
load_time = windowing.WaitForThrobber(wnd, (20, 16, 36, 32), 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)
if proc:
windowing.SetForegroundWindow(wnd)
# Send Alt-F4, then wait for process to end
keyboard.TypeString(r"{\4}", use_modifiers=True)
if not windowing.WaitForProcessExit(proc, timeout):
windowing.EndProcess(proc)
return "crashed"
if timedout:
return "timeout"
return None
