curlyq: writer.py annotate

annotate writer.py @ 21:35f29952b51e

Remove deadwood.

author	David Barts <n5jrn@me.com>
date	Sat, 28 Dec 2019 06:32:53 -0800
parents	091c03f1b2e8
children

rev	line source
0 984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	1 #!/usr/bin/env python3
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	2 # -- coding: utf-8 --
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	3
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	4 # A simple HTML writer, so we can process HTML in a streamwise fashion
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	5 # via callbacks, which compared to a document tree tends to be all of:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	6 # easier to program, uses less memory, and uses less processor time.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	7
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	8 # I m p o r t s
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	9
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	10 import os, sys
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	11 import codecs
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	12 import io
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	13 import html
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	14
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	15 # V a r i a b l e s
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	16
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	17 # We only support ASCII, ISO-8859-1, and anything capable of encoding
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	18 # the full Unicode set. Anything else is too sticky a wicket to want
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	19 # to mess with.
3 091c03f1b2e8 Getting it working... David Barts <n5jrn@me.com> parents: 0 diff changeset	20 CODECS_TO_NAME = {}
0 984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	21 for i in [ "US-ASCII", "ISO-8859-1", "UTF-8", "UTF-16", "UTF-16LE", "UTF-16BE", "UTF-32", "UTF-32LE", "UTF-32BE" ]:
3 091c03f1b2e8 Getting it working... David Barts <n5jrn@me.com> parents: 0 diff changeset	22 CODECS_TO_NAME[codecs.lookup(i)] = i
0 984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	23 del i
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	24 _MAXCHAR = { "US-ASCII": 127, "ISO-8859-1": 255 }
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	25
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	26 # There are WAY more HTML entities than this, but we're pessimistic about
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	27 # what browsers "in the wild" support, so we stick to what XML supports.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	28 _QUOTE_ENTITIES = {
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	29 "\"": "quot",
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	30 "'": "apos"
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	31 }
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	32 _OTHER_ENTITIES = {
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	33 "&": "amp",
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	34 "<": "lt",
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	35 ">": "gt"
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	36 }
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	37
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	38 # C l a s s e s
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	39
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	40 class HtmlStreamWriter(object):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	41 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	42 A simple HTML writer, intended to be used in a streamwise fashion.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	43 This class takes REASONABLE precautions against writing garbage, but
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	44 does not check every last thing. It will happily write tags like
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	45 "<garb<<age>>>" etc. if you feed it the right garbage in.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	46 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	47 def __init__(self, stream, encoding):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	48 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	49 Initialize this writer. An encoding is mandatory, even though we
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	50 produce character output, because the encoding governs which
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	51 characters we can send on for I/O without entity-escaping them.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	52 The supplied stream should be buffered or performance will suffer.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	53 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	54 # Stream we're using is available to the caller as .stream
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	55 self.stream = stream
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	56 try:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	57 # A codec to use is available to the caller as .codec
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	58 self.codec = codecs.lookup(encoding)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	59 # Normalized encoding name is available to the caller as .encoding
3 091c03f1b2e8 Getting it working... David Barts <n5jrn@me.com> parents: 0 diff changeset	60 self.encoding = CODECS_TO_NAME[self.codec]
0 984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	61 except (KeyError, LookupError) as e:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	62 raise ValueError("invalid encoding {0!r}".format(encoding))
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	63 self._maxchar = _MAXCHAR.get(self.encoding, 0x7fffffff)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	64
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	65 # html.escape drops the ball, badly. It is too optimistic about what
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	66 # entity names are likely to be understood, and is too stupid to know
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	67 # that ASCII streams need lots of things escaped.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	68 def _escape(self, string, quote=False):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	69 for ch in string:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	70 entity = None
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	71 if quote and ch in _QUOTE_ENTITIES:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	72 entity = _QUOTE_ENTITIES[ch]
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	73 elif ch in _OTHER_ENTITIES:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	74 entity = _OTHER_ENTITIES[ch]
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	75 if entity:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	76 self.stream.write("&")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	77 self.stream.write(entity)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	78 self.stream.write(";")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	79 continue
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	80 ordch = ord(ch)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	81 if ordch > self._maxchar:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	82 self.stream.write("&#")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	83 self.stream.write(str(ordch))
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	84 self.stream.write(";")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	85 continue
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	86 self.stream.write(ch)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	87
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	88 def write_starttag(self, tag, attrs):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	89 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	90 Write a start tag.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	91 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	92 self.stream.write("<")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	93 self.stream.write(tag)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	94 self._writeattrs(attrs)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	95 self.stream.write(">")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	96
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	97 def _writeattrs(self, attrs):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	98 for k, v in attrs:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	99 self.stream.write(" ")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	100 self.stream.write(k)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	101 self.stream.write("=\"")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	102 self._escape(v, quote=True)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	103 self.stream.write("\"")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	104
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	105 def write_endtag(self, tag):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	106 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	107 Write an end tag.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	108 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	109 self.stream.write("</")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	110 self.stream.write(tag)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	111 self.stream.write(">")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	112
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	113 def write_startendtag(self, tag, attrs):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	114 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	115 Write a "start-end" (i.e. empty) tag.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	116 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	117 self.stream.write("<")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	118 self.stream.write(tag)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	119 self._writeattrs(attrs)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	120 self.stream.write("/>")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	121
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	122 def write_data(self, data):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	123 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	124 Write text data.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	125 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	126 self._escape(data)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	127
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	128 def write_raw_data(self, data):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	129 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	130 Write raw data (e.g. style sheets, scripts, etc.)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	131 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	132 self.stream.write(data)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	133
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	134 def write_charref(self, name):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	135 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	136 Write character reference (normally not needed).
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	137 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	138 is_number = False
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	139 try:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	140 junk = int(name)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	141 is_number = True
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	142 except ValueError:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	143 pass
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	144 if name.startswith("x"):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	145 try:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	146 junk = int(name[1:], 16)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	147 is_number = True
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	148 except ValueError:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	149 pass
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	150 self.stream.write("&")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	151 if is_number:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	152 self.stream.write("#")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	153 self.stream.write(name)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	154 else:
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	155 self.stream.write(name)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	156 self.stream.write(";")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	157
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	158 def write_comment(self, data):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	159 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	160 Write a comment.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	161 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	162 self.stream.write("<!--")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	163 self.stream.write(data)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	164 self.stream.write("-->")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	165
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	166 def write_decl(self, decl):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	167 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	168 Write a declarationm.
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	169 """
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	170 self.stream.write("<!")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	171 self.stream.write(decl)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	172 self.stream.write(">")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	173
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	174 def write_pi(self, data):
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	175 self.stream.write("<?")
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	176 self.stream.write(decl)
984876b6a095 Initial commit of first two classes. David Barts <n5jrn@me.com> parents: diff changeset	177 self.stream.write(">")

Mercurial > cgi-bin > hgweb.cgi > curlyq

annotate writer.py @ 21:35f29952b51e