1698
|
1 " Vim indent file
|
11062
|
2 " Language: DTD (Document Type Definition for XML)
|
|
3 " Previous Maintainer: Nikolai Weibull <now@bitwi.se>
|
|
4 " Latest Revision: 2011-07-08
|
1698
|
5
|
|
6 let s:cpo_save = &cpo
|
|
7 set cpo&vim
|
|
8
|
|
9 setlocal indentexpr=GetDTDIndent()
|
|
10 setlocal indentkeys=!^F,o,O,>
|
|
11 setlocal nosmartindent
|
|
12
|
|
13 if exists("*GetDTDIndent")
|
|
14 finish
|
|
15 endif
|
|
16
|
|
17 " TODO: Needs to be adjusted to stop at [, <, and ].
|
|
18 let s:token_pattern = '^[^[:space:]]\+'
|
|
19
|
|
20 function s:lex1(input, start, ...)
|
|
21 let pattern = a:0 > 0 ? a:1 : s:token_pattern
|
|
22 let start = matchend(a:input, '^\_s*', a:start)
|
|
23 if start == -1
|
|
24 return ["", a:start]
|
|
25 endif
|
|
26 let end = matchend(a:input, pattern, start)
|
|
27 if end == -1
|
|
28 return ["", a:start]
|
|
29 endif
|
|
30 let token = strpart(a:input, start, end - start)
|
|
31 return [token, end]
|
|
32 endfunction
|
|
33
|
|
34 function s:lex(input, start, ...)
|
|
35 let pattern = a:0 > 0 ? a:1 : s:token_pattern
|
|
36 let info = s:lex1(a:input, a:start, pattern)
|
|
37 while info[0] == '--'
|
|
38 let info = s:lex1(a:input, info[1], pattern)
|
|
39 while info[0] != "" && info[0] != '--'
|
|
40 let info = s:lex1(a:input, info[1], pattern)
|
|
41 endwhile
|
|
42 if info[0] == ""
|
|
43 return info
|
|
44 endif
|
|
45 let info = s:lex1(a:input, info[1], pattern)
|
|
46 endwhile
|
|
47 return info
|
|
48 endfunction
|
|
49
|
|
50 function s:indent_to_innermost_parentheses(line, end)
|
|
51 let token = '('
|
|
52 let end = a:end
|
|
53 let parentheses = [end - 1]
|
|
54 while token != ""
|
2965
|
55 let [token, end] = s:lex(a:line, end, '^\%([(),|]\|[A-Za-z0-9_-]\+\|#P\=CDATA\|%[A-Za-z0-9_-]\+;\)[?*+]\=')
|
1698
|
56 if token[0] == '('
|
|
57 call add(parentheses, end - 1)
|
|
58 elseif token[0] == ')'
|
|
59 if len(parentheses) == 1
|
|
60 return [-1, end]
|
|
61 endif
|
|
62 call remove(parentheses, -1)
|
|
63 endif
|
|
64 endwhile
|
|
65 return [parentheses[-1] - strridx(a:line, "\n", parentheses[-1]), end]
|
|
66 endfunction
|
|
67
|
|
68 " TODO: Line and end could be script global (think OO members).
|
|
69 function GetDTDIndent()
|
|
70 if v:lnum == 1
|
|
71 return 0
|
|
72 endif
|
|
73
|
|
74 " Begin by searching back for a <! that isn’t inside a comment.
|
|
75 " From here, depending on what follows immediately after, parse to
|
|
76 " where we’re at to determine what to do.
|
|
77 if search('<!', 'bceW') == 0
|
|
78 return indent(v:lnum - 1)
|
|
79 endif
|
|
80 let lnum = line('.')
|
|
81 let col = col('.')
|
|
82 let indent = indent('.')
|
2587
|
83 let line = lnum == v:lnum ? getline(lnum) : join(getline(lnum, v:lnum - 1), "\n")
|
1698
|
84
|
|
85 let [declaration, end] = s:lex1(line, col)
|
|
86 if declaration == ""
|
11160
|
87 return indent + shiftwidth()
|
1698
|
88 elseif declaration == '--'
|
|
89 " We’re looking at a comment. Now, simply determine if the comment is
|
|
90 " terminated or not. If it isn’t, let Vim take care of that using
|
|
91 " 'comments' and 'autoindent'. Otherwise, indent to the first lines level.
|
|
92 while declaration != ""
|
|
93 let [declaration, end] = s:lex(line, end)
|
|
94 if declaration == "-->"
|
|
95 return indent
|
|
96 endif
|
|
97 endwhile
|
|
98 return -1
|
|
99 elseif declaration == 'ELEMENT'
|
|
100 " Check for element name. If none exists, indent one level.
|
|
101 let [name, end] = s:lex(line, end)
|
|
102 if name == ""
|
11160
|
103 return indent + shiftwidth()
|
1698
|
104 endif
|
|
105
|
|
106 " Check for token following element name. This can be a specification of
|
|
107 " whether the start or end tag may be omitted. If nothing is found, indent
|
|
108 " one level.
|
2587
|
109 let [token, end] = s:lex(line, end, '^\%([-O(]\|ANY\|EMPTY\)')
|
1698
|
110 let n = 0
|
|
111 while token =~ '[-O]' && n < 2
|
|
112 let [token, end] = s:lex(line, end, '^\%([-O(]\|ANY\|EMPTY\)')
|
|
113 let n += 1
|
|
114 endwhile
|
|
115 if token == ""
|
11160
|
116 return indent + shiftwidth()
|
1698
|
117 endif
|
|
118
|
|
119 " Next comes the content model. If the token we’ve found isn’t a
|
|
120 " parenthesis it must be either ANY, EMPTY or some random junk. Either
|
|
121 " way, we’re done indenting this element, so set it to that of the first
|
|
122 " line so that the terminating “>” winds up having the same indention.
|
|
123 if token != '('
|
|
124 return indent
|
|
125 endif
|
|
126
|
|
127 " Now go through the content model. We need to keep track of the nesting
|
|
128 " of parentheses. As soon as we hit 0 we’re done. If that happens we must
|
|
129 " have a complete content model. Thus set indention to be the same as that
|
|
130 " of the first line so that the terminating “>” winds up having the same
|
|
131 " indention. Otherwise, we’ll indent to the innermost parentheses not yet
|
|
132 " matched.
|
|
133 let [indent_of_innermost, end] = s:indent_to_innermost_parentheses(line, end)
|
|
134 if indent_of_innermost != -1
|
|
135 return indent_of_innermost
|
|
136 endif
|
|
137
|
|
138 " Finally, look for any additions and/or exceptions to the content model.
|
|
139 " This is defined by a “+” or “-” followed by another content model
|
|
140 " declaration.
|
|
141 " TODO: Can the “-” be separated by whitespace from the “(”?
|
|
142 let seen = { '+(': 0, '-(': 0 }
|
|
143 while 1
|
|
144 let [additions_exceptions, end] = s:lex(line, end, '^[+-](')
|
|
145 if additions_exceptions != '+(' && additions_exceptions != '-('
|
|
146 let [token, end] = s:lex(line, end)
|
|
147 if token == '>'
|
|
148 return indent
|
|
149 endif
|
|
150 " TODO: Should use s:lex here on getline(v:lnum) and check for >.
|
11160
|
151 return getline(v:lnum) =~ '^\s*>' || count(values(seen), 0) == 0 ? indent : (indent + shiftwidth())
|
1698
|
152 endif
|
|
153
|
|
154 " If we’ve seen an addition or exception already and this is of the same
|
|
155 " kind, the user is writing a broken DTD. Time to bail.
|
|
156 if seen[additions_exceptions]
|
|
157 return indent
|
|
158 endif
|
|
159 let seen[additions_exceptions] = 1
|
|
160
|
|
161 let [indent_of_innermost, end] = s:indent_to_innermost_parentheses(line, end)
|
|
162 if indent_of_innermost != -1
|
|
163 return indent_of_innermost
|
|
164 endif
|
|
165 endwhile
|
|
166 elseif declaration == 'ATTLIST'
|
|
167 " Check for element name. If none exists, indent one level.
|
|
168 let [name, end] = s:lex(line, end)
|
|
169 if name == ""
|
11160
|
170 return indent + shiftwidth()
|
1698
|
171 endif
|
|
172
|
|
173 " Check for any number of attributes.
|
|
174 while 1
|
|
175 " Check for attribute name. If none exists, indent one level, unless the
|
|
176 " current line is a lone “>”, in which case we indent to the same level
|
|
177 " as the first line. Otherwise, if the attribute name is “>”, we have
|
|
178 " actually hit the end of the attribute list, in which case we indent to
|
|
179 " the same level as the first line.
|
|
180 let [name, end] = s:lex(line, end)
|
|
181 if name == ""
|
|
182 " TODO: Should use s:lex here on getline(v:lnum) and check for >.
|
11160
|
183 return getline(v:lnum) =~ '^\s*>' ? indent : (indent + shiftwidth())
|
1698
|
184 elseif name == ">"
|
|
185 return indent
|
|
186 endif
|
|
187
|
|
188 " Check for attribute value declaration. If none exists, indent two
|
|
189 " levels. Otherwise, if it’s an enumerated value, check for nested
|
|
190 " parentheses and indent to the innermost one if we don’t reach the end
|
|
191 " of the listc. Otherwise, just continue with looking for the default
|
|
192 " attribute value.
|
|
193 " TODO: Do validation of keywords
|
|
194 " (CDATA|NMTOKEN|NMTOKENS|ID|IDREF|IDREFS|ENTITY|ENTITIES)?
|
|
195 let [value, end] = s:lex(line, end, '^\%((\|[^[:space:]]\+\)')
|
|
196 if value == ""
|
11160
|
197 return indent + shiftwidth() * 2
|
1698
|
198 elseif value == 'NOTATION'
|
|
199 " If this is a enumerated value based on notations, read another token
|
|
200 " for the actual value. If it doesn’t exist, indent three levels.
|
|
201 " TODO: If validating according to above, value must be equal to '('.
|
|
202 let [value, end] = s:lex(line, end, '^\%((\|[^[:space:]]\+\)')
|
|
203 if value == ""
|
11160
|
204 return indent + shiftwidth() * 3
|
1698
|
205 endif
|
|
206 endif
|
|
207
|
|
208 if value == '('
|
|
209 let [indent_of_innermost, end] = s:indent_to_innermost_parentheses(line, end)
|
|
210 if indent_of_innermost != -1
|
|
211 return indent_of_innermost
|
|
212 endif
|
|
213 endif
|
|
214
|
|
215 " Finally look for the attribute’s default value. If non exists, indent
|
|
216 " two levels.
|
2587
|
217 let [default, end] = s:lex(line, end, '^\%("\_[^"]*"\|#\(REQUIRED\|IMPLIED\|FIXED\)\)')
|
1698
|
218 if default == ""
|
11160
|
219 return indent + shiftwidth() * 2
|
1698
|
220 elseif default == '#FIXED'
|
|
221 " We need to look for the fixed value. If non exists, indent three
|
|
222 " levels.
|
|
223 let [default, end] = s:lex(line, end, '^"\_[^"]*"')
|
|
224 if default == ""
|
11160
|
225 return indent + shiftwidth() * 3
|
1698
|
226 endif
|
|
227 endif
|
|
228 endwhile
|
|
229 elseif declaration == 'ENTITY'
|
|
230 " Check for entity name. If none exists, indent one level. Otherwise, if
|
|
231 " the name actually turns out to be a percent sign, “%”, this is a
|
|
232 " parameter entity. Read another token to determine the entity name and,
|
|
233 " again, if none exists, indent one level.
|
|
234 let [name, end] = s:lex(line, end)
|
|
235 if name == ""
|
11160
|
236 return indent + shiftwidth()
|
1698
|
237 elseif name == '%'
|
|
238 let [name, end] = s:lex(line, end)
|
|
239 if name == ""
|
11160
|
240 return indent + shiftwidth()
|
1698
|
241 endif
|
|
242 endif
|
|
243
|
|
244 " Now check for the entity value. If none exists, indent one level. If it
|
|
245 " does exist, indent to same level as first line, as we’re now done with
|
|
246 " this entity.
|
|
247 "
|
|
248 " The entity value can be a string in single or double quotes (no escapes
|
|
249 " to worry about, as entities are used instead). However, it can also be
|
|
250 " that this is an external unparsed entity. In that case we have to look
|
|
251 " further for (possibly) a public ID and an URI followed by the NDATA
|
|
252 " keyword and the actual notation name. For the public ID and URI, indent
|
|
253 " two levels, if they don’t exist. If the NDATA keyword doesn’t exist,
|
|
254 " indent one level. Otherwise, if the actual notation name doesn’t exist,
|
|
255 " indent two level. If it does, indent to same level as first line, as
|
|
256 " we’re now done with this entity.
|
|
257 let [value, end] = s:lex(line, end)
|
|
258 if value == ""
|
11160
|
259 return indent + shiftwidth()
|
1698
|
260 elseif value == 'SYSTEM' || value == 'PUBLIC'
|
|
261 let [quoted_string, end] = s:lex(line, end, '\%("[^"]\+"\|''[^'']\+''\)')
|
|
262 if quoted_string == ""
|
11160
|
263 return indent + shiftwidth() * 2
|
1698
|
264 endif
|
|
265
|
|
266 if value == 'PUBLIC'
|
|
267 let [quoted_string, end] = s:lex(line, end, '\%("[^"]\+"\|''[^'']\+''\)')
|
|
268 if quoted_string == ""
|
11160
|
269 return indent + shiftwidth() * 2
|
1698
|
270 endif
|
|
271 endif
|
|
272
|
|
273 let [ndata, end] = s:lex(line, end)
|
|
274 if ndata == ""
|
11160
|
275 return indent + shiftwidth()
|
1698
|
276 endif
|
|
277
|
|
278 let [name, end] = s:lex(line, end)
|
11160
|
279 return name == "" ? (indent + shiftwidth() * 2) : indent
|
1698
|
280 else
|
|
281 return indent
|
|
282 endif
|
|
283 elseif declaration == 'NOTATION'
|
|
284 " Check for notation name. If none exists, indent one level.
|
|
285 let [name, end] = s:lex(line, end)
|
|
286 if name == ""
|
11160
|
287 return indent + shiftwidth()
|
1698
|
288 endif
|
|
289
|
|
290 " Now check for the external ID. If none exists, indent one level.
|
|
291 let [id, end] = s:lex(line, end)
|
|
292 if id == ""
|
11160
|
293 return indent + shiftwidth()
|
1698
|
294 elseif id == 'SYSTEM' || id == 'PUBLIC'
|
|
295 let [quoted_string, end] = s:lex(line, end, '\%("[^"]\+"\|''[^'']\+''\)')
|
|
296 if quoted_string == ""
|
11160
|
297 return indent + shiftwidth() * 2
|
1698
|
298 endif
|
|
299
|
|
300 if id == 'PUBLIC'
|
|
301 let [quoted_string, end] = s:lex(line, end, '\%("[^"]\+"\|''[^'']\+''\|>\)')
|
|
302 if quoted_string == ""
|
|
303 " TODO: Should use s:lex here on getline(v:lnum) and check for >.
|
11160
|
304 return getline(v:lnum) =~ '^\s*>' ? indent : (indent + shiftwidth() * 2)
|
1698
|
305 elseif quoted_string == '>'
|
|
306 return indent
|
|
307 endif
|
|
308 endif
|
|
309 endif
|
|
310
|
|
311 return indent
|
|
312 endif
|
|
313
|
|
314 " TODO: Processing directives could be indented I suppose. But perhaps it’s
|
|
315 " just as well to let the user decide how to indent them (perhaps extending
|
|
316 " this function to include proper support for whatever processing directive
|
|
317 " language they want to use).
|
|
318
|
|
319 " Conditional sections are simply passed along to let Vim decide what to do
|
|
320 " (and hence the user).
|
|
321 return -1
|
|
322 endfunction
|
|
323
|
|
324 let &cpo = s:cpo_save
|
3445
|
325 unlet s:cpo_save
|