| Fred Drake | 99ff8c4 | 1998-11-18 23:28:24 +0000 | [diff] [blame] | 1 | #! /usr/bin/env python |
| 2 | |
| 3 | """ |
| 4 | """ |
| 5 | __version__ = '$Revision$' |
| 6 | |
| 7 | |
| 8 | import re |
| 9 | import string |
| 10 | |
| 11 | |
| 12 | _data_rx = re.compile(r"[^\\][^\\]*") |
| 13 | |
| 14 | def decode(s): |
| 15 | r = '' |
| 16 | while s: |
| 17 | m = _data_rx.match(s) |
| 18 | if m: |
| 19 | r = r + m.group() |
| 20 | s = s[len(m.group()):] |
| 21 | elif s[1] == "\\": |
| 22 | r = r + "\\" |
| 23 | s = s[2:] |
| 24 | elif s[1] == "n": |
| 25 | r = r + "\n" |
| 26 | s = s[2:] |
| 27 | else: |
| 28 | raise ValueError, "can't handle " + `s` |
| 29 | return r |
| 30 | |
| 31 | |
| 32 | def format_attrs(attrs): |
| 33 | attrs = attrs.items() |
| 34 | attrs.sort() |
| 35 | s = '' |
| 36 | for name, value in attrs: |
| 37 | s = '%s %s="%s"' % (s, name, value) |
| 38 | return s |
| 39 | |
| 40 | |
| 41 | def do_convert(ifp, ofp, knownempties, xml=0): |
| 42 | attrs = {} |
| 43 | lastopened = None |
| 44 | knownempty = 0 |
| 45 | lastempty = 0 |
| 46 | while 1: |
| 47 | line = ifp.readline() |
| 48 | if not line: |
| 49 | break |
| 50 | |
| 51 | type = line[0] |
| 52 | data = line[1:] |
| 53 | if data and data[-1] == "\n": |
| 54 | data = data[:-1] |
| 55 | if type == "-": |
| 56 | data = decode(data) |
| 57 | ofp.write(data) |
| 58 | if "\n" in data: |
| 59 | lastopened = None |
| 60 | knownempty = 0 |
| 61 | lastempty = 0 |
| 62 | elif type == "(": |
| 63 | if knownempty and xml: |
| 64 | ofp.write("<%s%s/>" % (data, format_attrs(attrs))) |
| 65 | else: |
| 66 | ofp.write("<%s%s>" % (data, format_attrs(attrs))) |
| 67 | attrs = {} |
| 68 | lastopened = data |
| 69 | lastempty = knownempty |
| 70 | knownempty = 0 |
| 71 | elif type == ")": |
| 72 | if xml: |
| 73 | if not lastempty: |
| 74 | ofp.write("</%s>" % data) |
| 75 | elif data not in knownempties: |
| 76 | if lastopened == data: |
| 77 | ofp.write("</>") |
| 78 | else: |
| 79 | ofp.write("</%s>" % data) |
| 80 | lastopened = None |
| 81 | lastempty = 0 |
| 82 | elif type == "A": |
| 83 | name, type, value = string.split(data, " ", 2) |
| 84 | attrs[name] = decode(value) |
| 85 | elif type == "e": |
| 86 | knownempty = 1 |
| 87 | |
| 88 | |
| 89 | def sgml_convert(ifp, ofp, knownempties=()): |
| 90 | return do_convert(ifp, ofp, knownempties, xml=0) |
| 91 | |
| 92 | |
| 93 | def xml_convert(ifp, ofp, knownempties=[]): |
| 94 | return do_convert(ifp, ofp, knownempties, xml=1) |
| 95 | |
| 96 | |
| 97 | def main(): |
| 98 | import sys |
| 99 | # |
| 100 | convert = sgml_convert |
| 101 | if sys.argv[1:] and sys.argv[1] in ("-x", "--xml"): |
| 102 | convert = xml_convert |
| 103 | del sys.argv[1] |
| 104 | if len(sys.argv) == 1: |
| 105 | ifp = sys.stdin |
| 106 | ofp = sys.stdout |
| 107 | elif len(sys.argv) == 2: |
| 108 | ifp = open(sys.argv[1]) |
| 109 | ofp = sys.stdout |
| 110 | elif len(sys.argv) == 3: |
| 111 | ifp = open(sys.argv[1]) |
| 112 | ofp = open(sys.argv[2], "w") |
| 113 | else: |
| 114 | usage() |
| 115 | sys.exit(2) |
| 116 | # knownempties is ignored in the XML version |
| 117 | convert(ifp, ofp, knownempties=["rfc", "POSIX", "ASCII", "declaremodule", |
| 118 | "maketitle", "makeindex", "makemodindex", |
| 119 | "localmoduletable", "ABC", "UNIX", "Cpp", |
| 120 | "C", "EOF", "NULL", "manpage", "input", |
| 121 | "label"]) |
| 122 | |
| 123 | |
| 124 | if __name__ == "__main__": |
| 125 | main() |