emdee/phantompy
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

add setup.py

Browse source
This commit is contained in:
emdee 2022-11-16 18:33:59 +00:00
parent 1d92e0ec65
commit c6a7d839d9
5 changed files with 84 additions and 81 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

3
__main__.py
View file

@ -13,4 +13,5 @@ try:
vsetup_logging(log_level, logfile='', stream=sys.stderr)
except: pass
iMain(sys.argv[1:], bgui=False)
if __name__ == '__main__':
iMain(sys.argv[1:], bgui=False)

19
lookupdns.py
View file

@ -1,9 +1,14 @@
#!/usr/local/bin/python3.sh
# -*-mode: python; indent-tabs-mode: nil; py-indent-offset: 4; coding: utf-8 -*
# Looks for urls https://dns.google/resolve?
# and parses them to extract a magic field.
# https://dns.google/resolve?name=domain.name&type=TXT&cd=true&do=true
"""
Looks for urls https://dns.google/resolve?
https://dns.google/resolve?name=domain.name&type=TXT&cd=true&do=true
and parses them to extract a magic field.
A good example of how you can parse json embedded in HTML with phantomjs.
"""
import sys
import os
@ -17,7 +22,7 @@ warnings.filterwarnings('ignore')
LOG = logging.getLogger()
class LookFor(Render):
def __init__(self, app, do_print=True, do_save=False):
app.lfps = []
self._app = app
@ -37,7 +42,7 @@ class LookFor(Render):
fp = fp[:i]
# threadsafe?
self._app.lfps.append(fp)
def _html_callback(self, *args):
"""print(self, QPrinter, Callable[[bool], None])"""
if type(args[0]) is str:
@ -72,8 +77,8 @@ class LookFor(Render):
self.we_run_this_tor_relay = False
LOG.warn(f"BAD {self.uri}")
return 2
def _loadFinished(self, result):
LOG.debug(f"phantom.py: Loading finished {self.uri}")
self.toHtml(self._html_callback)

109
phantompy.py
View file

@ -13,8 +13,8 @@ replacement for other bulky headless browser frameworks.
If you have a display attached:
./phantom.py [--pdf_output <pdf-file>] [--js_input <javascript-file>] <url-or-html-file>
./phantom.py [--pdf_output <pdf-file>] [--js_input <javascript-file>] <url-or-html-file>
If you don't have a display attached (i.e. on a remote server), you can use
xvfb-run, or don't add --show_gui - it should work without a display.
@ -64,7 +64,7 @@ CSS @media types, etc.
Installation of dependencies in Debian Stretch is easy:
apt-get install xvfb python3-pyqt5 python3-pyqt5.qtwebkit
Finding the equivalent for other OSes is an exercise that I leave to you.
@ -80,16 +80,16 @@ Given the following file /tmp/test.html
document.getElementById('id1').innerHTML = "bar";
</script>
</html>
... and the following file /tmp/test.js:
document.getElementById('id2').innerHTML = "baz";
console.log("__PHANTOM_PY_DONE__");
... and running this script (without attached display) ...
xvfb-run python3 phantom.py /tmp/test.html /tmp/out.pdf /tmp/test.js
... you will get a PDF file /tmp/out.pdf with the contents "foo bar baz".
Note that the second occurrence of "foo" has been replaced by the web page's own
@ -130,8 +130,6 @@ from PyQt5.QtWidgets import QApplication
from PyQt5.QtPrintSupport import QPrinter
from PyQt5.QtWebEngineWidgets import QWebEnginePage
from support_phantompy import vsetup_logging
global LOG
import logging
import warnings
@ -161,19 +159,19 @@ def prepare(sdir='/tmp'):
</html>
""")
LOG.debug(f"wrote {sfile} ")
class Render(QWebEnginePage):
def __init__(self, app, do_print=False, do_save=True):
app.ldone = []
self._app = app
self.do_print = do_print
self.do_save = do_save
self.percent = 0
self.uri = None
self.jsfile = None
self.htmlfile = None
self.pdffile = None
QWebEnginePage.__init__(self)
app.ldone = []
self._app = app
self.do_print = do_print
self.do_save = do_save
self.percent = 0
self.uri = None
self.jsfile = None
self.htmlfile = None
self.pdffile = None
QWebEnginePage.__init__(self)
def run(self, url, pdffile, htmlfile, jsfile):
self._app.lstart.append(id(self))
@ -184,64 +182,65 @@ class Render(QWebEnginePage):
self.pdffile = pdffile
self.outfile = pdffile or htmlfile
LOG.debug(f"phantom.py: URL={url} OUTFILE={outfile} JSFILE={jsfile}")
qurl = QUrl.fromUserInput(url)
qurl = QUrl.fromUserInput(url)
# The PDF generation only happens when the special string __PHANTOM_PY_DONE__
# is sent to console.log(). The following JS string will be executed by
# default, when no external JavaScript file is specified.
self.js_contents = "setTimeout(function() { console.log('__PHANTOM_PY_DONE__') }, 5000);";
if jsfile:
try:
with open(self.jsfile, 'rt') as f:
self.js_contents = f.read()
except Exception as e:
LOG.exception(f"error reading jsfile {self.jsfile}")
self.loadFinished.connect(self._loadFinished)
self.percent = 20
self.load(qurl)
self.javaScriptConsoleMessage = self._onConsoleMessage
LOG.debug(f"phantom.py: loading 10")
def _onConsoleMessage(self, *args):
if len(args) > 3:
level, txt, lineno, filename = args
else:
level = 1
txt, lineno, filename = args
LOG.debug(f"CONSOLE {lineno} {txt} {filename}")
if "__PHANTOM_PY_DONE__" in txt:
self.percent = 40
# If we get this magic string, it means that the external JS is done
if self.do_save:
self.toHtml(self._html_callback)
return
# drop through
txt = "__PHANTOM_PY_SAVED__"
if "__PHANTOM_PY_SAVED__" in txt:
self.percent = 50
if self.do_print:
self._print()
return
txt = "__PHANTOM_PY_PRINTED__"
if "__PHANTOM_PY_PRINTED__" in txt:
self.percent = 60
self._exit(level)
if len(args) > 3:
level, txt, lineno, filename = args
else:
level = 1
txt, lineno, filename = args
LOG.debug(f"CONSOLE {lineno} {txt} {filename}")
if "__PHANTOM_PY_DONE__" in txt:
self.percent = 40
# If we get this magic string, it means that the external JS is done
if self.do_save:
self.toHtml(self._html_callback)
return
# drop through
txt = "__PHANTOM_PY_SAVED__"
if "__PHANTOM_PY_SAVED__" in txt:
self.percent = 50
if self.do_print:
self._print()
return
txt = "__PHANTOM_PY_PRINTED__"
if "__PHANTOM_PY_PRINTED__" in txt:
self.percent = 60
self._exit(level)
def _loadFinished(self, result):
self.percent = 30
LOG.info(f"phantom.py: _loadFinished {result} {self.percent}")
LOG.debug(f"phantom.py: Evaluating JS from {self.jsfile}")
self.runJavaScript("document.documentElement.contentEditable=true")
self.runJavaScript(self.js_contents)
# RenderProcessTerminationStatus ?
self.percent = 30
LOG.info(f"phantom.py: _loadFinished {result} {self.percent}")
LOG.debug(f"phantom.py: Evaluating JS from {self.jsfile}")
self.runJavaScript("document.documentElement.contentEditable=true")
self.runJavaScript(self.js_contents)
def _html_callback(self, *args):
"""print(self, QPrinter, Callable[[bool], None])"""
if type(args[0]) is str:
self._save(args[0])
self._onConsoleMessage(0, "__PHANTOM_PY_SAVED__", 0 , '')
def _save(self, html):
sfile = self.htmlfile
# CompleteHtmlSaveFormat SingleHtmlSaveFormat MimeHtmlSaveFormat
@ -267,7 +266,7 @@ class Render(QWebEnginePage):
printer.setOutputFileName(sfile)
self.print(printer, self._printer_callback)
LOG.debug("phantom.py: Printed")
def _exit(self, val):
self.percent = 100
LOG.debug(f"phantom.py: Exiting with val {val}")

28
qasync_phantompy.py
View file

@ -13,6 +13,7 @@ from PyQt5.QtWidgets import (QProgressBar, QWidget, QVBoxLayout)
from phantompy import Render
# from lookupdns import LookFor as Render
from support_phantompy import vsetup_logging, omain_argparser
global LOG
import logging
@ -35,7 +36,7 @@ class Widget(QtWidgets.QWidget):
i = len(asyncio.all_tasks())
self._label.setText(str(i))
self.progress.setValue(int(text))
class ContextManager:
def __init__(self) -> None:
self._seconds = 0
@ -63,25 +64,22 @@ async def main(widget, app, ilen):
app.exit()
# raise asyncio.CancelledError
return
LOG.debug(f"{app.ldone} {perc} {seconds}")
LOG.debug(f"{app.ldone} {seconds}")
except asyncio.CancelledError as ex:
LOG.debug("Task cancelled")
def iMain(largs):
parser = oMainArgparser()
oargs = parser.parse_args(lArgs)
parser = omain_argparser()
oargs = parser.parse_args(largs)
bgui=oargs.show_gui
try:
from support_phantompy import vsetup_logging
d = int(os.environ.get('DEBUG', 0))
if d > 0:
vsetup_logging(10, stream=sys.stderr)
else:
vsetup_logging(oargs.log_level, stream=sys.stderr)
vsetup_logging(log_level, logfile='', stream=sys.stderr)
oargs.log_level = 10
vsetup_logging(oargs.log_level, logfile='', stream=sys.stderr)
except: pass
app = QtWidgets.QApplication([])
app.lstart = []
if bgui:
@ -90,7 +88,7 @@ def iMain(largs):
widget.show()
else:
widget = None
loop = qasync.QEventLoop(app)
asyncio.set_event_loop(loop)
@ -105,9 +103,9 @@ def iMain(largs):
uri = url.strip()
r.run(uri, pdffile, htmlfile, jsfile)
LOG.debug(f"{r.percent} {app.lstart}")
LOG.info(f"queued {len(app.lstart)} urls")
task = loop.create_task(main(widget, app, 1))
loop.run_forever()
@ -117,6 +115,6 @@ def iMain(largs):
loop.run_until_complete(asyncio.gather(*tasks))
if __name__ == '__main__':
iMain(sys.argv[1:])

6
support_phantompy.py
View file

@ -80,7 +80,7 @@ def vsetup_logging(log_level, logfile='', stream=sys.stdout):
'NOTSET': logging.NOTSET,
}
def omain__argparser(_=None):
def omain_argparser(_=None):
try:
from OpenSSL import SSL
@ -106,9 +106,9 @@ def omain__argparser(_=None):
help="Operate on the HTML file with javascript")
parser.add_argument('--html_output', type=str, default='',
help="Write loaded and javascripted result to a HTML file")
parser.add_argument('--pdf_output', type=str, default=''),
parser.add_argument('--pdf_output', type=str, default='',
help="Write loaded and javascripted result to a PDF file")
parser.add_argument('--show_gui', type=bool, store_action=True),
parser.add_argument('--show_gui', type=bool, default=False, store_action=True),
help="show a progress meter that doesn't work")
parser.add_argument('html_url', type=str, nargs='?',
required=True,