Commit | Line | Data |
---|---|---|
62551769 | 1 | from . import cons |
ff79cdbf FT |
2 | |
3 | def findnsnames(el): | |
4 | names = {} | |
5 | nid = [1] | |
6 | def proc(el): | |
7 | if isinstance(el, cons.element): | |
8 | if el.ns not in names: | |
62551769 | 9 | names[el.ns] = "n" + str(nid[0]) |
ff79cdbf FT |
10 | nid[:] = [nid[0] + 1] |
11 | for ch in el.children: | |
12 | proc(ch) | |
13 | proc(el) | |
14 | if None in names: | |
15 | names[None] = None | |
16 | else: | |
17 | names[el.ns] = None | |
18 | return names | |
19 | ||
20 | class formatter(object): | |
21 | def __init__(self, out, root, nsnames=None, charset="utf-8", doctype=None): | |
22 | self.root = root | |
23 | if nsnames is None: | |
24 | nsnames = findnsnames(root) | |
25 | self.nsnames = nsnames | |
26 | self.out = out | |
27 | self.charset = charset | |
28 | self.doctype = doctype | |
29 | ||
30 | def write(self, text): | |
31 | self.out.write(text.encode(self.charset)) | |
32 | ||
33 | def quotewrite(self, buf): | |
34 | for ch in buf: | |
62551769 FT |
35 | if ch == '&': |
36 | self.write("&") | |
37 | elif ch == '<': | |
38 | self.write("<") | |
39 | elif ch == '>': | |
40 | self.write(">") | |
ff79cdbf FT |
41 | else: |
42 | self.write(ch) | |
43 | ||
44 | def text(self, el): | |
45 | self.quotewrite(el) | |
46 | ||
47 | def attrval(self, buf): | |
62551769 | 48 | qc, qt = ("'", "'") if '"' in buf else ('"', """) |
ff79cdbf FT |
49 | self.write(qc) |
50 | for ch in buf: | |
62551769 FT |
51 | if ch == '&': |
52 | self.write("&") | |
53 | elif ch == '<': | |
54 | self.write("<") | |
55 | elif ch == '>': | |
56 | self.write(">") | |
ff79cdbf FT |
57 | elif ch == qc: |
58 | self.write(qt) | |
59 | else: | |
60 | self.write(ch) | |
61 | self.write(qc) | |
62 | ||
63 | def attr(self, k, v): | |
64 | self.write(k) | |
62551769 | 65 | self.write('=') |
ff79cdbf FT |
66 | self.attrval(v) |
67 | ||
68 | def shorttag(self, el, **extra): | |
62551769 FT |
69 | self.write('<' + self.elname(el)) |
70 | for k, v in el.attrs.items(): | |
71 | self.write(' ') | |
ff79cdbf | 72 | self.attr(k, v) |
62551769 FT |
73 | for k, v in extra.items(): |
74 | self.write(' ') | |
ff79cdbf | 75 | self.attr(k, v) |
62551769 | 76 | self.write(" />") |
ff79cdbf FT |
77 | |
78 | def elname(self, el): | |
79 | ns = self.nsnames[el.ns] | |
80 | if ns is None: | |
81 | return el.name | |
82 | else: | |
62551769 | 83 | return ns + ':' + el.name |
ff79cdbf FT |
84 | |
85 | def starttag(self, el, **extra): | |
62551769 FT |
86 | self.write('<' + self.elname(el)) |
87 | for k, v in el.attrs.items(): | |
88 | self.write(' ') | |
ff79cdbf | 89 | self.attr(k, v) |
62551769 FT |
90 | for k, v in extra.items(): |
91 | self.write(' ') | |
ff79cdbf | 92 | self.attr(k, v) |
62551769 | 93 | self.write('>') |
ff79cdbf FT |
94 | |
95 | def endtag(self, el): | |
62551769 | 96 | self.write('</' + self.elname(el) + '>') |
ff79cdbf FT |
97 | |
98 | def longtag(self, el): | |
99 | self.starttag(el, **extra) | |
100 | for ch in el.children: | |
101 | self.node(ch) | |
102 | self.endtag(el) | |
103 | ||
104 | def element(self, el, **extra): | |
105 | if len(el.children) == 0: | |
106 | self.shorttag(el, **extra) | |
107 | else: | |
108 | self.longtag(el, **extra) | |
109 | ||
110 | def node(self, el): | |
111 | if isinstance(el, cons.element): | |
112 | self.element(el) | |
113 | elif isinstance(el, cons.text): | |
114 | self.text(el) | |
115 | else: | |
116 | raise Exception("Unknown object in element tree: " + el) | |
117 | ||
118 | def start(self): | |
62551769 | 119 | self.write('<?xml version="1.0" encoding="' + self.charset + '" ?>\n') |
ff79cdbf | 120 | if self.doctype: |
62551769 FT |
121 | self.write('<!DOCTYPE %s PUBLIC "%s" "%s">\n' % (self.root.name, |
122 | self.doctype[0], | |
123 | self.doctype[1])) | |
ff79cdbf | 124 | extra = {} |
62551769 | 125 | for uri, nm in self.nsnames.items(): |
ff79cdbf FT |
126 | if uri is None: |
127 | continue | |
128 | if nm is None: | |
62551769 | 129 | extra["xmlns"] = uri |
ff79cdbf | 130 | else: |
62551769 | 131 | extra["xmlns:" + nm] = uri |
ff79cdbf FT |
132 | self.element(self.root, **extra) |
133 | ||
134 | @classmethod | |
135 | def output(cls, out, el, *args, **kw): | |
136 | cls(out=out, root=el, *args, **kw).start() | |
137 | ||
138 | def update(self, **ch): | |
139 | ret = type(self).__new__(type(self)) | |
140 | ret.__dict__.update(self.__dict__) | |
141 | ret.__dict__.update(ch) | |
142 | return ret | |
143 | ||
144 | class iwriter(object): | |
145 | def __init__(self, out): | |
146 | self.out = out | |
147 | self.atbol = True | |
148 | self.col = 0 | |
149 | ||
150 | def write(self, buf): | |
62551769 FT |
151 | for i in range(len(buf)): |
152 | c = buf[i:i + 1] | |
153 | if c == b'\n': | |
ff79cdbf FT |
154 | self.col = 0 |
155 | else: | |
156 | self.col += 1 | |
157 | self.out.write(c) | |
158 | self.atbol = False | |
159 | ||
160 | def indent(self, indent): | |
161 | if self.atbol: | |
162 | return | |
163 | if self.col != 0: | |
62551769 | 164 | self.write(b'\n') |
ff79cdbf FT |
165 | self.write(indent) |
166 | self.atbol = True | |
167 | ||
168 | class indenter(formatter): | |
62551769 | 169 | def __init__(self, indent=" ", *args, **kw): |
ff79cdbf FT |
170 | super(indenter, self).__init__(*args, **kw) |
171 | self.out = iwriter(self.out) | |
172 | self.indent = indent | |
62551769 | 173 | self.curind = "" |
ff79cdbf FT |
174 | |
175 | def simple(self, el): | |
176 | for ch in el.children: | |
177 | if not isinstance(ch, cons.text): | |
178 | return False | |
179 | return True | |
180 | ||
181 | def longtag(self, el, **extra): | |
182 | self.starttag(el, **extra) | |
183 | sub = self | |
184 | reind = False | |
185 | if not self.simple(el): | |
186 | sub = self.update(curind=self.curind + self.indent) | |
62551769 | 187 | sub.reindent() |
ff79cdbf FT |
188 | reind = True |
189 | for ch in el.children: | |
190 | sub.node(ch) | |
191 | if reind: | |
62551769 | 192 | self.reindent() |
ff79cdbf FT |
193 | self.endtag(el) |
194 | ||
195 | def element(self, el, **extra): | |
196 | super(indenter, self).element(el, **extra) | |
197 | if self.out.col > 80 and self.simple(el): | |
62551769 FT |
198 | self.reindent() |
199 | ||
200 | def reindent(self): | |
201 | self.out.indent(self.curind.encode(self.charset)) | |
ff79cdbf FT |
202 | |
203 | def start(self): | |
204 | super(indenter, self).start() | |
205 | self.write('\n') |