Fred Drake | 3843bae | 1998-12-01 19:00:58 +0000 | [diff] [blame] | 1 | """Miscellaneous utility functions useful for dealing with ESIS streams.""" |
| 2 | __version__ = '$Revision$' |
| 3 | |
| 4 | import re |
| 5 | import string |
| 6 | import sys |
Fred Drake | 2697694 | 1999-05-07 21:14:28 +0000 | [diff] [blame] | 7 | import xml.dom.core |
| 8 | import xml.dom.esis_builder |
Fred Drake | 3843bae | 1998-12-01 19:00:58 +0000 | [diff] [blame] | 9 | |
| 10 | |
| 11 | _data_rx = re.compile(r"[^\\][^\\]*") |
| 12 | |
| 13 | def decode(s): |
| 14 | r = '' |
| 15 | while s: |
| 16 | m = _data_rx.match(s) |
| 17 | if m: |
| 18 | r = r + m.group() |
| 19 | s = s[len(m.group()):] |
| 20 | elif s[1] == "\\": |
| 21 | r = r + "\\" |
| 22 | s = s[2:] |
| 23 | elif s[1] == "n": |
| 24 | r = r + "\n" |
| 25 | s = s[2:] |
| 26 | else: |
| 27 | raise ValueError, "can't handle " + `s` |
| 28 | return r |
| 29 | |
| 30 | |
| 31 | _charmap = {} |
| 32 | for c in map(chr, range(256)): |
| 33 | _charmap[c] = c |
| 34 | _charmap["\n"] = r"\n" |
| 35 | _charmap["\\"] = r"\\" |
| 36 | del c |
| 37 | |
| 38 | def encode(s): |
| 39 | return string.join(map(_charmap.get, s), '') |
| 40 | |
| 41 | |
Fred Drake | 3843bae | 1998-12-01 19:00:58 +0000 | [diff] [blame] | 42 | class ExtendedEsisBuilder(xml.dom.esis_builder.EsisBuilder): |
| 43 | def __init__(self, *args, **kw): |
| 44 | self.__empties = {} |
| 45 | self.__is_empty = 0 |
| 46 | apply(xml.dom.esis_builder.EsisBuilder.__init__, (self,) + args, kw) |
Fred Drake | ada2de6 | 1999-07-29 22:13:31 +0000 | [diff] [blame] | 47 | self.buildFragment() |
Fred Drake | 3843bae | 1998-12-01 19:00:58 +0000 | [diff] [blame] | 48 | |
| 49 | def feed(self, data): |
| 50 | for line in string.split(data, '\n'): |
| 51 | if not line: |
| 52 | break |
| 53 | event = line[0] |
| 54 | text = line[1:] |
| 55 | if event == '(': |
| 56 | element = self.document.createElement(text, self.attr_store) |
| 57 | self.attr_store = {} |
| 58 | self.push(element) |
| 59 | if self.__is_empty: |
| 60 | self.__empties[text] = text |
| 61 | self.__is_empty = 0 |
| 62 | elif event == ')': |
| 63 | self.pop() |
| 64 | elif event == 'A': |
| 65 | l = re.split(' ', text, 2) |
| 66 | name = l[0] |
| 67 | value = decode(l[2]) |
| 68 | self.attr_store[name] = value |
| 69 | elif event == '-': |
| 70 | text = self.document.createText(decode(text)) |
| 71 | self.push(text) |
| 72 | elif event == 'C': |
| 73 | return |
| 74 | elif event == 'e': |
| 75 | self.__is_empty = 1 |
Fred Drake | abb158f | 1999-08-26 18:04:32 +0000 | [diff] [blame^] | 76 | elif event == '&': |
| 77 | eref = self.document.createEntityReference(text) |
| 78 | self.push(eref) |
Fred Drake | 3843bae | 1998-12-01 19:00:58 +0000 | [diff] [blame] | 79 | else: |
| 80 | sys.stderr.write('Unknown event: %s\n' % line) |
Fred Drake | abb158f | 1999-08-26 18:04:32 +0000 | [diff] [blame^] | 81 | |
Fred Drake | 3843bae | 1998-12-01 19:00:58 +0000 | [diff] [blame] | 82 | def get_empties(self): |
| 83 | return self.__empties.keys() |