ꦫꦣꦺꦤ꧀ꦄꦤ꧀ꦠꦱꦺꦤ

PATH: usr / lib64 / python2.7 / Tools / scripts

#! /usr/bin/env python
""" Utility for parsing HTML entity definitions available from:

http://www.w3.org/ as e.g.
      http://www.w3.org/TR/REC-html40/HTMLlat1.ent

Input is read from stdin, output is written to stdout in form of a
    Python snippet defining a dictionary "entitydefs" mapping literal
    entity name to character or numeric entity.

Marc-Andre Lemburg, mal@lemburg.com, 1999.
    Use as you like. NO WARRANTIES.

"""
import re,sys
import TextTools

entityRE = re.compile('<!ENTITY +(\w+) +CDATA +"([^"]+)" +-- +((?:.|\n)+?) *-->')

def parse(text,pos=0,endpos=None):

pos = 0
    if endpos is None:
        endpos = len(text)
    d = {}
    while 1:
        m = entityRE.search(text,pos,endpos)
        if not m:
            break
        name,charcode,comment = m.groups()
        d[name] = charcode,comment
        pos = m.end()
    return d

def writefile(f,defs):

f.write("entitydefs = {\n")
    items = defs.items()
    items.sort()
    for name,(charcode,comment) in items:
        if charcode[:2] == '&#':
            code = int(charcode[2:-1])
            if code < 256:
                charcode = "'\%o'" % code
            else:
                charcode = repr(charcode)
        else:
            charcode = repr(charcode)
        comment = TextTools.collapse(comment)
        f.write("    '%s':\t%s,  \t# %s\n" % (name,charcode,comment))
    f.write('\n}\n')

if __name__ == '__main__':
    if len(sys.argv) > 1:
        infile = open(sys.argv[1])
    else:
        infile = sys.stdin
    if len(sys.argv) > 2:
        outfile = open(sys.argv[2],'w')
    else:
        outfile = sys.stdout
    text = infile.read()
    defs = parse(text)
    writefile(outfile,defs)

[+] ..
[-] setup.pyo [edit]
[-] classfix.py [edit]
[-] cvsfiles.pyo [edit]
[-] fixps.pyc [edit]
[-] fixdiv.py [edit]
[-] ptags.pyc [edit]
[-] pysource.pyc [edit]
[-] h2py.py [edit]
[-] svneol.py [edit]
[-] win_add2path.pyc [edit]
[-] suff.py [edit]
[-] untabify.py [edit]
[-] parseentities.py [edit]
[-] objgraph.pyc [edit]
[-] mailerdaemon.pyo [edit]
[-] findlinksto.pyo [edit]
[-] checkappend.py [edit]
[-] fixps.py [edit]
[-] untabify.pyo [edit]
[-] setup.py [edit]
[-] find_recursionlimit.pyc [edit]
[-] nm2def.pyc [edit]
[-] copytime.pyo [edit]
[-] dutree.pyc [edit]
[-] which.py [edit]
[-] db2pickle.pyo [edit]
[-] serve.pyo [edit]
[-] svneol.pyc [edit]
[-] lll.py [edit]
[-] pindent.pyc [edit]
[-] h2py.pyo [edit]
[-] lll.pyo [edit]
[-] analyze_dxp.py [edit]
[-] byteyears.pyc [edit]
[-] reindent-rst.pyo [edit]
[-] serve.py [edit]
[-] mkreal.py [edit]
[-] ndiff.pyo [edit]
[-] combinerefs.pyc [edit]
[-] diff.pyc [edit]
[-] fixnotice.pyo [edit]
[-] texcheck.pyc [edit]
[-] analyze_dxp.pyo [edit]
[-] byext.py [edit]
[-] dutree.py [edit]
[-] which.pyc [edit]
[-] texcheck.pyo [edit]
[-] reindent.pyo [edit]
[-] methfix.pyo [edit]
[-] ndiff.py [edit]
[-] texi2html.pyc [edit]
[-] parseentities.pyc [edit]
[-] checkpyc.pyc [edit]
[-] hotshotmain.py [edit]
[-] ftpmirror.pyo [edit]
[-] fixps.pyo [edit]
[-] redemo.pyo [edit]
[-] fixcid.pyo [edit]
[-] finddiv.pyo [edit]
[-] which.pyo [edit]
[-] redemo.py [edit]
[-] suff.pyc [edit]
[-] cleanfuture.py [edit]
[-] combinerefs.py [edit]
[-] ptags.pyo [edit]
[-] byext.pyc [edit]
[-] db2pickle.py [edit]
[-] cvsfiles.py [edit]
[-] pindent.py [edit]
[-] diff.pyo [edit]
[-] analyze_dxp.pyc [edit]
[-] cleanfuture.pyo [edit]
[-] dutree.pyo [edit]
[-] eptags.pyo [edit]
[-] pysource.py [edit]
[-] reindent.pyc [edit]
[-] google.pyc [edit]
[-] logmerge.py [edit]
[-] lfcr.py [edit]
[-] finddiv.pyc [edit]
[-] classfix.pyo [edit]
[-] lfcr.pyo [edit]
[-] treesync.py [edit]
[-] hotshotmain.pyo [edit]
[-] nm2def.py [edit]
[-] byext.pyo [edit]
[-] ftpmirror.py [edit]
[-] treesync.pyc [edit]
[-] copytime.py [edit]
[-] finddiv.py [edit]
[-] ifdef.py [edit]
[-] patchcheck.pyo [edit]
[-] objgraph.pyo [edit]
[-] ptags.py [edit]
[-] objgraph.py [edit]
[-] cvsfiles.pyc [edit]
[-] pysource.pyo [edit]
[-] pathfix.py [edit]
[-] linktree.pyc [edit]
[-] eptags.py [edit]
[-] patchcheck.py [edit]
[-] mkreal.pyc [edit]
[-] rgrep.pyc [edit]
[-] google.py [edit]
[-] win_add2path.py [edit]
[-] texcheck.py [edit]
[-] pathfix.pyo [edit]
[-] gprof2html.py [edit]
[-] pickle2db.pyc [edit]
[-] findnocoding.pyc [edit]
[-] nm2def.pyo [edit]
[-] checkappend.pyo [edit]
[-] texi2html.py [edit]
[-] crlf.pyc [edit]
[-] treesync.pyo [edit]
[-] findlinksto.py [edit]
[-] pdeps.pyc [edit]
[-] fixheader.pyo [edit]
[-] pdeps.pyo [edit]
[-] findnocoding.py [edit]
[-] setup.pyc [edit]
[-] methfix.pyc [edit]
[-] logmerge.pyo [edit]
[-] reindent.py [edit]
[-] hotshotmain.pyc [edit]
[-] gprof2html.pyc [edit]
[-] redemo.pyc [edit]
[-] fixheader.py [edit]
[-] diff.py [edit]
[-] classfix.pyc [edit]
[-] ftpmirror.pyc [edit]
[-] google.pyo [edit]
[-] db2pickle.pyc [edit]
[-] pathfix.pyc [edit]
[-] ifdef.pyo [edit]
[-] xxci.pyc [edit]
[-] rgrep.pyo [edit]
[-] cleanfuture.pyc [edit]
[-] reindent-rst.pyc [edit]
[-] svneol.pyo [edit]
[-] xxci.pyo [edit]
[-] parseentities.pyo [edit]
[-] md5sum.py [edit]
[-] eptags.pyc [edit]
[-] mailerdaemon.pyc [edit]
[-] pickle2db.pyo [edit]
[-] win_add2path.pyo [edit]
[-] serve.pyc [edit]
[-] byteyears.pyo [edit]
[-] fixdiv.pyo [edit]
[-] logmerge.pyc [edit]
[-] pickle2db.py [edit]
[-] checkpyc.pyo [edit]
[-] fixdiv.pyc [edit]
[-] find_recursionlimit.py [edit]
[-] crlf.pyo [edit]
[-] suff.pyo [edit]
[-] lll.pyc [edit]
[-] rgrep.py [edit]
[-] untabify.pyc [edit]
[-] ifdef.pyc [edit]
[-] fixheader.pyc [edit]
[-] fixcid.pyc [edit]
[-] crlf.py [edit]
[-] checkappend.pyc [edit]
[-] xxci.py [edit]
[-] findnocoding.pyo [edit]
[-] mkreal.pyo [edit]
[-] patchcheck.pyc [edit]
[-] linktree.pyo [edit]
[-] texi2html.pyo [edit]
[-] fixcid.py [edit]
[-] ndiff.pyc [edit]
[-] md5sum.pyo [edit]
[-] findlinksto.pyc [edit]
[-] reindent-rst.py [edit]
[-] linktree.py [edit]
[-] byteyears.py [edit]
[-] md5sum.pyc [edit]
[-] copytime.pyc [edit]
[-] combinerefs.pyo [edit]
[-] fixnotice.py [edit]
[-] pdeps.py [edit]
[-] fixnotice.pyc [edit]
[-] lfcr.pyc [edit]
[-] checkpyc.py [edit]
[-] methfix.py [edit]
[-] pindent.pyo [edit]
[-] mailerdaemon.py [edit]
[-] gprof2html.pyo [edit]
[-] h2py.pyc [edit]
[-] find_recursionlimit.pyo [edit]