Hubcapp/session2html.py

## session2html.py
#!/usr/bin/python

################################################################################
# usage: session2html.py [-h] -s SESSION_FILE [-o OUTPUT_FILE]
#
# Convert Firefox sessionstore.js file to html.
#
# optional arguments:
#   -h, --help       show this help message and exit
#   -s SESSION_FILE  session file to be converted
#   -o OUTPUT_FILE   write html output to OUTPUT_FILE; if not defined write html
#                    output to stdout
################################################################################


import os, sys, json, time, argparse
from xml.sax.saxutils import escape

html_escape_table = {
    '"': "&quot;",
    "'": "&apos;"
}

# escapes this characters: '&', '<', '>', '"', and '''
def html_escape(text):
    return escape(text, html_escape_table)

# parse script arguments
desc="Convert Firefox sessionstore.js file to html."
parser = argparse.ArgumentParser(description=desc)
parser.add_argument("-s", dest="sessionfile", type=argparse.FileType('r'),
      metavar="SESSION_FILE", required=True, help="session file to be converted")
parser.add_argument("-o", dest="outputfile", type=argparse.FileType('w'),
      default=sys.stdout, metavar="OUTPUT_FILE", help="write html output \
      to OUTPUT_FILE; if not defined write html output to stdout")
args = parser.parse_args()

# read session file and parse it to json structure
ss = json.loads(args.sessionfile.read().decode("utf-8"))


mtimeInfo = "<p>File mtime {0}</p>".format(
      time.ctime(os.stat(args.sessionfile.name).st_mtime))

lastUpdateInfo = ""
if "session" in ss:
    if "lastUpdate" in ss["session"]:
        lastUpdateInfo = "<p>Recorded session.lastUpdate {0}</p>".format(
              time.ctime(ss["session"]["lastUpdate"]/1000.0))


args.outputfile.write("""
<html>
<head>
 <meta http-equiv="Content-type" content="text/html; charset=utf-8">
 <title>Open Tabs from session file</title>
</head>
<body>
<h1>Open Tabs from Session File</h1>
{0}
{1}
<ul style="list-style-type:none">
""".format(mtimeInfo, lastUpdateInfo))

windows = []

#entries need associated window and group. i'll give it a favicon too, why not...
groupedTabs = []
windowCounter = 0
windowGroups = []
for window in ss["windows"]:
    windowGroups.append([])
    groupedTabs.append([])
    for tab in window["tabs"]:
        tabgroup = "none"
        if 'extData' in tab.keys():
            if 'tabview-tab' in tab['extData'].keys():
                if tab['extData']['tabview-tab'] != "null":
                    sss = json.loads(tab['extData']['tabview-tab'].decode("utf-8"))
                    tabgroup = str(sss['groupID'])
                else:
                    tabgroup = "null"

        if tabgroup not in windowGroups[windowCounter]:
            windowGroups[windowCounter].append(tabgroup)

        groupedTabs[windowCounter].append({'favicon': tab['image'], 'entries': tab["entries"], "group" : tabgroup})
    windowCounter += 1

#have now structured data so that we can use tabgroups in the html
counter = 1
for windowNum in range(len(windowGroups)):
    args.outputfile.write('<h2>Window #{0}</h2>'.format(windowNum+1))
    for group in windowGroups[windowNum]:
        args.outputfile.write('<h3>Group #{0}</h3>'.format(group))
        for tab in groupedTabs[windowNum]:
            if tab['group'] == group:
                entries = tab["entries"]
                args.outputfile.write('<li>#{0}<ul style="list-style-type:none">'.format(
                      counter))
                for entry in enumerate(entries):
                    url = entry[1]["url"].encode("utf-8")
                    title = html_escape(entry[1].get("title", url)).encode("utf-8")

                    img = '<img width=32 height=32 src="{0}"/> '.format(tab['favicon'])

                    if entry[0] != len(entries)-1:
                        line = '<li>{0}<a href="{1}">{2}</a> : <tt>{3}</tt></li>\n'.format(
                            img, url, title, url)
                    else:
                        line = '<li>{0}<b><a href="{1}">{2}</a> : <tt>{3}</tt></b></li>\n'.format(
                            img, url, title, url)
                    args.outputfile.write(line)
                args.outputfile.write("</ul>")
                print "tabs processed: {0}".format(counter)
                counter += 1
        counter = 1

args.outputfile.write("""
</ul>
<p></p>
</body>
</html>
"""
)
args.outputfile.close()
	#!/usr/bin/python

	################################################################################
	# usage: session2html.py [-h] -s SESSION_FILE [-o OUTPUT_FILE]
	#
	# Convert Firefox sessionstore.js file to html.
	#
	# optional arguments:
	# -h, --help show this help message and exit
	# -s SESSION_FILE session file to be converted
	# -o OUTPUT_FILE write html output to OUTPUT_FILE; if not defined write html
	# output to stdout
	################################################################################


	import os, sys, json, time, argparse
	from xml.sax.saxutils import escape

	html_escape_table = {
	'"': """,
	"'": "'"
	}

	# escapes this characters: '&', '<', '>', '"', and '''
	def html_escape(text):
	return escape(text, html_escape_table)

	# parse script arguments
	desc="Convert Firefox sessionstore.js file to html."
	parser = argparse.ArgumentParser(description=desc)
	parser.add_argument("-s", dest="sessionfile", type=argparse.FileType('r'),
	metavar="SESSION_FILE", required=True, help="session file to be converted")
	parser.add_argument("-o", dest="outputfile", type=argparse.FileType('w'),
	default=sys.stdout, metavar="OUTPUT_FILE", help="write html output \
	to OUTPUT_FILE; if not defined write html output to stdout")
	args = parser.parse_args()

	# read session file and parse it to json structure
	ss = json.loads(args.sessionfile.read().decode("utf-8"))


	mtimeInfo = "<p>File mtime {0}</p>".format(
	time.ctime(os.stat(args.sessionfile.name).st_mtime))

	lastUpdateInfo = ""
	if "session" in ss:
	if "lastUpdate" in ss["session"]:
	lastUpdateInfo = "<p>Recorded session.lastUpdate {0}</p>".format(
	time.ctime(ss["session"]["lastUpdate"]/1000.0))



	args.outputfile.write("""
	<html>
	<head>
	<meta http-equiv="Content-type" content="text/html; charset=utf-8">
	<title>Open Tabs from session file</title>
	</head>
	<body>
	<h1>Open Tabs from Session File</h1>
	{0}
	{1}
	<ul style="list-style-type:none">
	""".format(mtimeInfo, lastUpdateInfo))

	windows = []

	#entries need associated window and group. i'll give it a favicon too, why not...
	groupedTabs = []
	windowCounter = 0
	windowGroups = []
	for window in ss["windows"]:
	windowGroups.append([])
	groupedTabs.append([])
	for tab in window["tabs"]:
	tabgroup = "none"
	if 'extData' in tab.keys():
	if 'tabview-tab' in tab['extData'].keys():
	if tab['extData']['tabview-tab'] != "null":
	sss = json.loads(tab['extData']['tabview-tab'].decode("utf-8"))
	tabgroup = str(sss['groupID'])
	else:
	tabgroup = "null"

	if tabgroup not in windowGroups[windowCounter]:
	windowGroups[windowCounter].append(tabgroup)

	groupedTabs[windowCounter].append({'favicon': tab['image'], 'entries': tab["entries"], "group" : tabgroup})
	windowCounter += 1

	#have now structured data so that we can use tabgroups in the html
	counter = 1
	for windowNum in range(len(windowGroups)):
	args.outputfile.write('<h2>Window #{0}</h2>'.format(windowNum+1))
	for group in windowGroups[windowNum]:
	args.outputfile.write('<h3>Group #{0}</h3>'.format(group))
	for tab in groupedTabs[windowNum]:
	if tab['group'] == group:
	entries = tab["entries"]
	args.outputfile.write('<li>#{0}<ul style="list-style-type:none">'.format(
	counter))
	for entry in enumerate(entries):
	url = entry[1]["url"].encode("utf-8")
	title = html_escape(entry[1].get("title", url)).encode("utf-8")

	img = '<img width=32 height=32 src="{0}"/> '.format(tab['favicon'])

	if entry[0] != len(entries)-1:
	line = '<li>{0}<a href="{1}">{2}</a> : <tt>{3}</tt></li>\n'.format(
	img, url, title, url)
	else:
	line = '<li>{0}<b><a href="{1}">{2}</a> : <tt>{3}</tt></b></li>\n'.format(
	img, url, title, url)
	args.outputfile.write(line)
	args.outputfile.write("</ul>")
	print "tabs processed: {0}".format(counter)
	counter += 1
	counter = 1

	args.outputfile.write("""
	</ul>
	<p></p>
	</body>
	</html>
	"""
	)
	args.outputfile.close()