rclchm
344 lines (295 with data), 12.4 kB
#!/usr/bin/env python2
"""Extract Html files from a Microsoft Compiled Html Help file (.chm)
Needs at least python 2.2 for HTMLParser (chmlib needs 2.2 too)"""
from __future__ import print_function
# Note: this is not converted to python3, libchm does not have a
# python3 wrapper at this point (2015-11)
# Do we return individual chapters as html pages or concatenate everything?
rclchm_catenate = 0
# Use special html type to allow for mimeconf/mimeview Open magic,
# Or go the regular html way with text/html
#rclchm_html_mtype = "text/x-chm-html"
rclchm_html_mtype = "text/html"
import sys
import os
import re
import posixpath
import urlparse
import urllib
if rclchm_catenate:
import subprocess
import rclexecm
try:
from chm import chm,chmlib
except:
print("RECFILTERROR HELPERNOTFOUND python:chm")
sys.exit(1);
try:
from HTMLParser import HTMLParser
except:
print("RECFILTERROR HELPERNOTFOUND python:HTMLParser")
sys.exit(1);
# Small helper routines
def getfile(chmfile, path):
"""Extract internal file text from chm object, given path"""
res, ui = chmfile.ResolveObject(path)
if res != chmlib.CHM_RESOLVE_SUCCESS:
#print("ResolveObject failed: %s" % path, file=sys.stderr)
return ""
res, doc = chmfile.RetrieveObject(ui)
if not res:
print("RetrieveObject failed: %s" % path, file=sys.stderr)
return ""
return doc
def peekfile(chmfile, path):
"""Check that path resolves in chm object"""
res, ui = chmfile.ResolveObject(path)
if res != chmlib.CHM_RESOLVE_SUCCESS:
return False
return True
# CHM Topics tree handler
class ChmTopicsParser(HTMLParser):
"""Parse the chm's Topic file which is basically
a listing of internal nodes (html files mostly). Build a list of
all nodes (parent.contents), which will then be used to walk and index
the chm.
Most nodes in the Topic file look like the following:
<LI> <OBJECT type="text/sitemap">
<param name="Name" value="Global Module Index">
<param name="Local" value="modindex.html">
</OBJECT>
Maybe we should filter out non "text/sitemap" Objects, and maybe there are
things of interest whose name is not Local, but for now, we just take
all values for parameters named "Local" (with some filtering/massaging),
until proven wrong
"""
def __init__(self, rclchm):
HTMLParser.__init__(self)
self.em = rclchm.em
self.rclchm = rclchm
def handle_starttag(self, tag, attrs):
#self.em.rclog("Beginning of a %s tag" % tag)
# If this is a param tag with name Local, we're interested in
# the value which lists a file ref. Discard those with #
# in them (references inside files)
# Sometimes it seems that refs are like Vendor:filename::path,
# we only keep the path, and only if the file matches
if tag != 'param':
return
name = ''
value = ''
for (nm,val) in attrs:
if nm == 'name':
name = val
if nm == 'value':
value = val
#self.em.rclog("Name [%s] value [%s]" %(name, value))
if name != 'Local' or value == '':
return
# value may be url-encoded. Decode it. If there are no % in there, will
# do nothing
value = urllib.unquote(value)
localpath = ""
ll = value.split(":")
if len(ll) == 1:
localpath = value
elif len(ll) == 4 and ll[-1] and ll[-3]:
#self.em.rclog("File: [%s] sfn [%s]" % ((ll[-3]), self.rclchm.sfn))
# We used to test against the simple file name, but this does
# not work if the file is renamed. Just check that the internal
# path resolves. Old: if ll[-3] == self.rclchm.sfn:
localpath = ll[-1]
if not peekfile(self.rclchm.chm, localpath):
#self.em.rclog("SKIPPING %s" % ll[-3])
localpath = ""
if len(localpath) != 0 and localpath.find("#") == -1:
if localpath[0] != '/':
localpath = "/" + localpath
self.rclchm.contents.append(localpath)
# Used when there is no Topics node. Walk the links tree
class ChmWalker(HTMLParser):
"""Links tree walker. This recursively follows all internal links
found in the tree from the top node given as input, and augments
the contents list."""
def __init__(self, rclchm, path, contents):
HTMLParser.__init__(self)
self.rclchm = rclchm
self.chm = rclchm.chm
self.contents = contents
self.path = posixpath.normpath(path)
self.dir = posixpath.dirname(self.path)
contents.append(self.path)
def handle_starttag(self, tag, attrs):
if tag != 'a':
return
href = ''
for (nm,val) in attrs:
if nm == 'href':
href = val
path = ""
res = urlparse.urlparse(href)
if (not res.scheme or res.scheme.lower == "ms-its"):
path = res.path
lpath = path.split(':')
if len(lpath) == 3:
# MS-ITS::somefile.chm:/some/path/file.htm ? As far as I
# know this never happens because there was a runtime error
# in this path
path = lpath[2]
if not peekfile(self.chm, path):
path = ""
elif len(lpath) == 1:
path = lpath[0]
else:
path = ""
if path:
#print "got path", path, "me", self.path, "dir", self.dir
if path[0] == "/":
npath = posixpath.normpath(path)
else:
npath = posixpath.normpath(posixpath.join(self.dir, path))
if not npath in self.contents:
#print("Going into [%s] paths [%s]\n" %
#(npath,str(self.contents)))
text = getfile(self.chm, npath)
if text:
try:
newwalker = ChmWalker(self.rclchm, npath, self.contents)
newwalker.feed(self.rclchm.fixencoding(text))
except:
pass
class rclCHM:
"""RclExecM slave worker for extracting all files from an Msoft chm
file. We first extract the list of internal nodes, and them return them
one by one. The ipath is the node path"""
def __init__(self, em):
self.contents = []
self.chm = chm.CHMFile()
self.em = em
if rclchm_catenate:
self.em.setmimetype("text/plain")
else:
self.em.setmimetype(rclchm_html_mtype)
expr = r'(<meta *http-equiv *= *"content-type".*charset *= *)((us-)?ascii)( *" *>)'
self.asciito1252re = re.compile(expr, re.IGNORECASE)
expr = r'<meta *http-equiv *= *"content-type".*charset *= *([a-z0-9-]+) *" *>'
self.findcharsetre = re.compile(expr, re.IGNORECASE)
def extractone(self, path):
"""Extract one path-named internal file from the chm file"""
#self.em.rclog("extractone: [%s]" % (path,))
iseof = rclexecm.RclExecM.noteof
if self.currentindex >= len(self.contents) -1:
iseof = rclexecm.RclExecM.eofnext
res, ui = self.chm.ResolveObject(path)
#self.em.rclog("extract: ResolveO: %d [%s]" % (res, ui))
if res != chmlib.CHM_RESOLVE_SUCCESS:
return (False, "", path, iseof)
# RetrieveObject() returns len,value
res, doc = self.chm.RetrieveObject(ui)
#self.em.rclog("extract: RetrieveObject: %d [%s]" % (res, doc))
if res > 0:
doc = re.sub('''</[hH][eE][aA][dD]''',
'''<meta name="rclaptg" content="chm"></head>''', doc)
self.em.setmimetype(rclchm_html_mtype)
return (True, doc, path, iseof)
return (False, "", path, iseof)
def dumpall(self):
alltxt=""
for pth in self.contents:
ret,doc,path,iseof = self.extractone(pth)
if not ret:
continue
# Feed doc to lynx
process = subprocess.Popen(["lynx", "-stdin", "-dump", "-nolist",
"-display_charset=utf8",
"-force_html"],
stdin=subprocess.PIPE,
stdout=subprocess.PIPE
)
txt,err = process.communicate(doc)
alltxt += txt
return alltxt
def fixencoding(self, text):
"""Fix encoding for supposedly html document. We do 2 things here:
- Change any 'ASCII' charset decl to windows-1252 because windows
people can't learn and we have to cope.
- Decode the string to unicode if it's originally an str because
that's what Python HTMLParser actually expects even if it does not
really say so. See http://bugs.python.org/issue3932.
"""
# Memo. Charset decl example. Maybe we should also process the
# HTML5 charset tag ?
#<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=US-ASCII">
if isinstance(text, str):
# Fix an ascii charset decl to windows-1252
text = self.asciito1252re.sub(r"\1windows-1252\4", text, 1)
# Convert to unicode according to charset decl
m = self.findcharsetre.search(text)
if m:
text = text.decode(m.group(1))
return text
def openfile(self, params):
"""Open the chm file and build the contents list by extracting and
parsing the Topics object"""
self.currentindex = -1
self.contents = []
filename = params["filename:"]
if not self.chm.LoadCHM(filename):
self.em.rclog("LoadCHM failed")
return False
#self.em.rclog("home [%s] topics [%s] title [%s]" %
# (self.chm.home, self.chm.topics, self.chm.title))
self.topics = self.chm.GetTopicsTree()
if self.topics:
# Parse Topics file and extract list of internal nodes
#self.em.rclog("Got topics");
tp = ChmTopicsParser(self)
tp.feed(self.fixencoding(self.topics))
tp.close()
else:
# No topics. If there is a home, let's try to walk the tree
#self.em.rclog("GetTopicsTree failed")
if not self.chm.home:
self.em.rclog("No topics and no home")
return False
home = self.chm.home
if home[0] != '/':
home = "/" + home
text = getfile(self.chm, home)
if not text:
self.em.rclog("No topics and no home content")
return False
walker = ChmWalker(self, self.chm.home, self.contents)
walker.feed(self.fixencoding(text))
walker.close()
#self.em.rclog("Contents size %d" % len(self.contents))
uniq = set(self.contents)
self.contents = list(uniq)
return True
def getipath(self, params):
return self.extractone(params["ipath:"])
def getnext(self, params):
if rclchm_catenate:
alltxt = self.dumpall()
if alltxt:
return (True, alltxt, "", rclexecm.RclExecM.eofnext)
else:
return (False, "", "", rclexecm.RclExecM.eofnow)
if self.currentindex == -1:
# Return "self" doc
self.currentindex = 0
self.em.setmimetype('text/plain')
if len(self.contents) == 0:
eof = rclexecm.RclExecM.eofnext
else:
eof = rclexecm.RclExecM.noteof
return (True, "", "", eof)
if self.currentindex >= len(self.contents):
return (False, "", "", rclexecm.RclExecM.eofnow)
else:
ret= self.extractone(self.contents[self.currentindex])
self.currentindex += 1
return ret
proto = rclexecm.RclExecM()
extract = rclCHM(proto)
rclexecm.main(proto, extract)