1 | require "forwardable"
|
---|
2 | require "open-uri"
|
---|
3 |
|
---|
4 | require "rss/rss"
|
---|
5 |
|
---|
6 | module RSS
|
---|
7 |
|
---|
8 | class NotWellFormedError < Error
|
---|
9 | attr_reader :line, :element
|
---|
10 |
|
---|
11 | # Create a new NotWellFormedError for an error at +line+
|
---|
12 | # in +element+. If a block is given the return value of
|
---|
13 | # the block ends up in the error message.
|
---|
14 | def initialize(line=nil, element=nil)
|
---|
15 | message = "This is not well formed XML"
|
---|
16 | if element or line
|
---|
17 | message << "\nerror occurred"
|
---|
18 | message << " in #{element}" if element
|
---|
19 | message << " at about #{line} line" if line
|
---|
20 | end
|
---|
21 | message << "\n#{yield}" if block_given?
|
---|
22 | super(message)
|
---|
23 | end
|
---|
24 | end
|
---|
25 |
|
---|
26 | class XMLParserNotFound < Error
|
---|
27 | def initialize
|
---|
28 | super("available XML parser was not found in " <<
|
---|
29 | "#{AVAILABLE_PARSER_LIBRARIES.inspect}.")
|
---|
30 | end
|
---|
31 | end
|
---|
32 |
|
---|
33 | class NotValidXMLParser < Error
|
---|
34 | def initialize(parser)
|
---|
35 | super("#{parser} is not an available XML parser. " <<
|
---|
36 | "Available XML parser"<<
|
---|
37 | (AVAILABLE_PARSERS.size > 1 ? "s are ": " is ") <<
|
---|
38 | "#{AVAILABLE_PARSERS.inspect}.")
|
---|
39 | end
|
---|
40 | end
|
---|
41 |
|
---|
42 | class NSError < InvalidRSSError
|
---|
43 | attr_reader :tag, :prefix, :uri
|
---|
44 | def initialize(tag, prefix, require_uri)
|
---|
45 | @tag, @prefix, @uri = tag, prefix, require_uri
|
---|
46 | super("prefix <#{prefix}> doesn't associate uri " <<
|
---|
47 | "<#{require_uri}> in tag <#{tag}>")
|
---|
48 | end
|
---|
49 | end
|
---|
50 |
|
---|
51 | class Parser
|
---|
52 |
|
---|
53 | extend Forwardable
|
---|
54 |
|
---|
55 | class << self
|
---|
56 |
|
---|
57 | @@default_parser = nil
|
---|
58 |
|
---|
59 | def default_parser
|
---|
60 | @@default_parser || AVAILABLE_PARSERS.first
|
---|
61 | end
|
---|
62 |
|
---|
63 | # Set @@default_parser to new_value if it is one of the
|
---|
64 | # available parsers. Else raise NotValidXMLParser error.
|
---|
65 | def default_parser=(new_value)
|
---|
66 | if AVAILABLE_PARSERS.include?(new_value)
|
---|
67 | @@default_parser = new_value
|
---|
68 | else
|
---|
69 | raise NotValidXMLParser.new(new_value)
|
---|
70 | end
|
---|
71 | end
|
---|
72 |
|
---|
73 | def parse(rss, do_validate=true, ignore_unknown_element=true,
|
---|
74 | parser_class=default_parser)
|
---|
75 | parser = new(rss, parser_class)
|
---|
76 | parser.do_validate = do_validate
|
---|
77 | parser.ignore_unknown_element = ignore_unknown_element
|
---|
78 | parser.parse
|
---|
79 | end
|
---|
80 | end
|
---|
81 |
|
---|
82 | def_delegators(:@parser, :parse, :rss,
|
---|
83 | :ignore_unknown_element,
|
---|
84 | :ignore_unknown_element=, :do_validate,
|
---|
85 | :do_validate=)
|
---|
86 |
|
---|
87 | def initialize(rss, parser_class=self.class.default_parser)
|
---|
88 | @parser = parser_class.new(normalize_rss(rss))
|
---|
89 | end
|
---|
90 |
|
---|
91 | private
|
---|
92 |
|
---|
93 | # Try to get the XML associated with +rss+.
|
---|
94 | # Return +rss+ if it already looks like XML, or treat it as a URI,
|
---|
95 | # or a file to get the XML,
|
---|
96 | def normalize_rss(rss)
|
---|
97 | return rss if maybe_xml?(rss)
|
---|
98 |
|
---|
99 | uri = to_uri(rss)
|
---|
100 |
|
---|
101 | if uri.respond_to?(:read)
|
---|
102 | uri.read
|
---|
103 | elsif !rss.tainted? and File.readable?(rss)
|
---|
104 | File.open(rss) {|f| f.read}
|
---|
105 | else
|
---|
106 | rss
|
---|
107 | end
|
---|
108 | end
|
---|
109 |
|
---|
110 | # maybe_xml? tests if source is a string that looks like XML.
|
---|
111 | def maybe_xml?(source)
|
---|
112 | source.is_a?(String) and /</ =~ source
|
---|
113 | end
|
---|
114 |
|
---|
115 | # Attempt to convert rss to a URI, but just return it if
|
---|
116 | # there's a ::URI::Error
|
---|
117 | def to_uri(rss)
|
---|
118 | return rss if rss.is_a?(::URI::Generic)
|
---|
119 |
|
---|
120 | begin
|
---|
121 | URI(rss)
|
---|
122 | rescue ::URI::Error
|
---|
123 | rss
|
---|
124 | end
|
---|
125 | end
|
---|
126 | end
|
---|
127 |
|
---|
128 | class BaseParser
|
---|
129 |
|
---|
130 | class << self
|
---|
131 | def raise_for_undefined_entity?
|
---|
132 | listener.raise_for_undefined_entity?
|
---|
133 | end
|
---|
134 | end
|
---|
135 |
|
---|
136 | def initialize(rss)
|
---|
137 | @listener = self.class.listener.new
|
---|
138 | @rss = rss
|
---|
139 | end
|
---|
140 |
|
---|
141 | def rss
|
---|
142 | @listener.rss
|
---|
143 | end
|
---|
144 |
|
---|
145 | def ignore_unknown_element
|
---|
146 | @listener.ignore_unknown_element
|
---|
147 | end
|
---|
148 |
|
---|
149 | def ignore_unknown_element=(new_value)
|
---|
150 | @listener.ignore_unknown_element = new_value
|
---|
151 | end
|
---|
152 |
|
---|
153 | def do_validate
|
---|
154 | @listener.do_validate
|
---|
155 | end
|
---|
156 |
|
---|
157 | def do_validate=(new_value)
|
---|
158 | @listener.do_validate = new_value
|
---|
159 | end
|
---|
160 |
|
---|
161 | def parse
|
---|
162 | if @listener.rss.nil?
|
---|
163 | _parse
|
---|
164 | end
|
---|
165 | @listener.rss
|
---|
166 | end
|
---|
167 |
|
---|
168 | end
|
---|
169 |
|
---|
170 | class BaseListener
|
---|
171 |
|
---|
172 | extend Utils
|
---|
173 |
|
---|
174 | class << self
|
---|
175 |
|
---|
176 | @@setters = {}
|
---|
177 | @@registered_uris = {}
|
---|
178 | @@class_names = {}
|
---|
179 |
|
---|
180 | # return the setter for the uri, tag_name pair, or nil.
|
---|
181 | def setter(uri, tag_name)
|
---|
182 | begin
|
---|
183 | @@setters[uri][tag_name]
|
---|
184 | rescue NameError
|
---|
185 | nil
|
---|
186 | end
|
---|
187 | end
|
---|
188 |
|
---|
189 |
|
---|
190 | # return the tag_names for setters associated with uri
|
---|
191 | def available_tags(uri)
|
---|
192 | begin
|
---|
193 | @@setters[uri].keys
|
---|
194 | rescue NameError
|
---|
195 | []
|
---|
196 | end
|
---|
197 | end
|
---|
198 |
|
---|
199 | # register uri against this name.
|
---|
200 | def register_uri(uri, name)
|
---|
201 | @@registered_uris[name] ||= {}
|
---|
202 | @@registered_uris[name][uri] = nil
|
---|
203 | end
|
---|
204 |
|
---|
205 | # test if this uri is registered against this name
|
---|
206 | def uri_registered?(uri, name)
|
---|
207 | @@registered_uris[name].has_key?(uri)
|
---|
208 | end
|
---|
209 |
|
---|
210 | # record class_name for the supplied uri and tag_name
|
---|
211 | def install_class_name(uri, tag_name, class_name)
|
---|
212 | @@class_names[uri] ||= {}
|
---|
213 | @@class_names[uri][tag_name] = class_name
|
---|
214 | end
|
---|
215 |
|
---|
216 | # retrieve class_name for the supplied uri and tag_name
|
---|
217 | # If it doesn't exist, capitalize the tag_name
|
---|
218 | def class_name(uri, tag_name)
|
---|
219 | begin
|
---|
220 | @@class_names[uri][tag_name]
|
---|
221 | rescue NameError
|
---|
222 | tag_name[0,1].upcase + tag_name[1..-1]
|
---|
223 | end
|
---|
224 | end
|
---|
225 |
|
---|
226 | def install_get_text_element(uri, name, setter)
|
---|
227 | install_setter(uri, name, setter)
|
---|
228 | def_get_text_element(uri, name, *get_file_and_line_from_caller(1))
|
---|
229 | end
|
---|
230 |
|
---|
231 | def raise_for_undefined_entity?
|
---|
232 | true
|
---|
233 | end
|
---|
234 |
|
---|
235 | private
|
---|
236 | # set the setter for the uri, tag_name pair
|
---|
237 | def install_setter(uri, tag_name, setter)
|
---|
238 | @@setters[uri] ||= {}
|
---|
239 | @@setters[uri][tag_name] = setter
|
---|
240 | end
|
---|
241 |
|
---|
242 | def def_get_text_element(uri, name, file, line)
|
---|
243 | register_uri(uri, name)
|
---|
244 | unless private_instance_methods(false).include?("start_#{name}")
|
---|
245 | module_eval(<<-EOT, file, line)
|
---|
246 | def start_#{name}(name, prefix, attrs, ns)
|
---|
247 | uri = _ns(ns, prefix)
|
---|
248 | if self.class.uri_registered?(uri, #{name.inspect})
|
---|
249 | start_get_text_element(name, prefix, ns, uri)
|
---|
250 | else
|
---|
251 | start_else_element(name, prefix, attrs, ns)
|
---|
252 | end
|
---|
253 | end
|
---|
254 | EOT
|
---|
255 | __send__("private", "start_#{name}")
|
---|
256 | end
|
---|
257 | end
|
---|
258 |
|
---|
259 | end
|
---|
260 |
|
---|
261 | end
|
---|
262 |
|
---|
263 | module ListenerMixin
|
---|
264 |
|
---|
265 | attr_reader :rss
|
---|
266 |
|
---|
267 | attr_accessor :ignore_unknown_element
|
---|
268 | attr_accessor :do_validate
|
---|
269 |
|
---|
270 | def initialize
|
---|
271 | @rss = nil
|
---|
272 | @ignore_unknown_element = true
|
---|
273 | @do_validate = true
|
---|
274 | @ns_stack = [{}]
|
---|
275 | @tag_stack = [[]]
|
---|
276 | @text_stack = ['']
|
---|
277 | @proc_stack = []
|
---|
278 | @last_element = nil
|
---|
279 | @version = @encoding = @standalone = nil
|
---|
280 | @xml_stylesheets = []
|
---|
281 | end
|
---|
282 |
|
---|
283 | # set instance vars for version, encoding, standalone
|
---|
284 | def xmldecl(version, encoding, standalone)
|
---|
285 | @version, @encoding, @standalone = version, encoding, standalone
|
---|
286 | end
|
---|
287 |
|
---|
288 | def instruction(name, content)
|
---|
289 | if name == "xml-stylesheet"
|
---|
290 | params = parse_pi_content(content)
|
---|
291 | if params.has_key?("href")
|
---|
292 | @xml_stylesheets << XMLStyleSheet.new(*params)
|
---|
293 | end
|
---|
294 | end
|
---|
295 | end
|
---|
296 |
|
---|
297 | def tag_start(name, attributes)
|
---|
298 | @text_stack.push('')
|
---|
299 |
|
---|
300 | ns = @ns_stack.last.dup
|
---|
301 | attrs = {}
|
---|
302 | attributes.each do |n, v|
|
---|
303 | if /\Axmlns(?:\z|:)/ =~ n
|
---|
304 | ns[$POSTMATCH] = v
|
---|
305 | else
|
---|
306 | attrs[n] = v
|
---|
307 | end
|
---|
308 | end
|
---|
309 | @ns_stack.push(ns)
|
---|
310 |
|
---|
311 | prefix, local = split_name(name)
|
---|
312 | @tag_stack.last.push([_ns(ns, prefix), local])
|
---|
313 | @tag_stack.push([])
|
---|
314 | if respond_to?("start_#{local}", true)
|
---|
315 | __send__("start_#{local}", local, prefix, attrs, ns.dup)
|
---|
316 | else
|
---|
317 | start_else_element(local, prefix, attrs, ns.dup)
|
---|
318 | end
|
---|
319 | end
|
---|
320 |
|
---|
321 | def tag_end(name)
|
---|
322 | if DEBUG
|
---|
323 | p "end tag #{name}"
|
---|
324 | p @tag_stack
|
---|
325 | end
|
---|
326 | text = @text_stack.pop
|
---|
327 | tags = @tag_stack.pop
|
---|
328 | pr = @proc_stack.pop
|
---|
329 | pr.call(text, tags) unless pr.nil?
|
---|
330 | @ns_stack.pop
|
---|
331 | end
|
---|
332 |
|
---|
333 | def text(data)
|
---|
334 | @text_stack.last << data
|
---|
335 | end
|
---|
336 |
|
---|
337 | private
|
---|
338 | def _ns(ns, prefix)
|
---|
339 | ns.fetch(prefix, "")
|
---|
340 | end
|
---|
341 |
|
---|
342 | CONTENT_PATTERN = /\s*([^=]+)=(["'])([^\2]+?)\2/
|
---|
343 | # Extract the first name="value" pair from content.
|
---|
344 | # Works with single quotes according to the constant
|
---|
345 | # CONTENT_PATTERN. Return a Hash.
|
---|
346 | def parse_pi_content(content)
|
---|
347 | params = {}
|
---|
348 | content.scan(CONTENT_PATTERN) do |name, quote, value|
|
---|
349 | params[name] = value
|
---|
350 | end
|
---|
351 | params
|
---|
352 | end
|
---|
353 |
|
---|
354 | def start_else_element(local, prefix, attrs, ns)
|
---|
355 | class_name = self.class.class_name(_ns(ns, prefix), local)
|
---|
356 | current_class = @last_element.class
|
---|
357 | if current_class.constants.include?(class_name)
|
---|
358 | next_class = current_class.const_get(class_name)
|
---|
359 | start_have_something_element(local, prefix, attrs, ns, next_class)
|
---|
360 | else
|
---|
361 | if !@do_validate or @ignore_unknown_element
|
---|
362 | @proc_stack.push(nil)
|
---|
363 | else
|
---|
364 | parent = "ROOT ELEMENT???"
|
---|
365 | if current_class.tag_name
|
---|
366 | parent = current_class.tag_name
|
---|
367 | end
|
---|
368 | raise NotExpectedTagError.new(local, _ns(ns, prefix), parent)
|
---|
369 | end
|
---|
370 | end
|
---|
371 | end
|
---|
372 |
|
---|
373 | NAMESPLIT = /^(?:([\w:][-\w\d.]*):)?([\w:][-\w\d.]*)/
|
---|
374 | def split_name(name)
|
---|
375 | name =~ NAMESPLIT
|
---|
376 | [$1 || '', $2]
|
---|
377 | end
|
---|
378 |
|
---|
379 | def check_ns(tag_name, prefix, ns, require_uri)
|
---|
380 | if @do_validate
|
---|
381 | if _ns(ns, prefix) == require_uri
|
---|
382 | #ns.delete(prefix)
|
---|
383 | else
|
---|
384 | raise NSError.new(tag_name, prefix, require_uri)
|
---|
385 | end
|
---|
386 | end
|
---|
387 | end
|
---|
388 |
|
---|
389 | def start_get_text_element(tag_name, prefix, ns, required_uri)
|
---|
390 | @proc_stack.push Proc.new {|text, tags|
|
---|
391 | setter = self.class.setter(required_uri, tag_name)
|
---|
392 | if @last_element.respond_to?(setter)
|
---|
393 | @last_element.__send__(setter, text.to_s)
|
---|
394 | else
|
---|
395 | if @do_validate and !@ignore_unknown_element
|
---|
396 | raise NotExpectedTagError.new(tag_name, _ns(ns, prefix),
|
---|
397 | @last_element.tag_name)
|
---|
398 | end
|
---|
399 | end
|
---|
400 | }
|
---|
401 | end
|
---|
402 |
|
---|
403 | def start_have_something_element(tag_name, prefix, attrs, ns, klass)
|
---|
404 |
|
---|
405 | check_ns(tag_name, prefix, ns, klass.required_uri)
|
---|
406 |
|
---|
407 | attributes = {}
|
---|
408 | klass.get_attributes.each do |a_name, a_uri, required, element_name|
|
---|
409 |
|
---|
410 | if a_uri.is_a?(String) or !a_uri.respond_to?(:include?)
|
---|
411 | a_uri = [a_uri]
|
---|
412 | end
|
---|
413 | unless a_uri == [""]
|
---|
414 | for prefix, uri in ns
|
---|
415 | if a_uri.include?(uri)
|
---|
416 | val = attrs["#{prefix}:#{a_name}"]
|
---|
417 | break if val
|
---|
418 | end
|
---|
419 | end
|
---|
420 | end
|
---|
421 | if val.nil? and a_uri.include?("")
|
---|
422 | val = attrs[a_name]
|
---|
423 | end
|
---|
424 |
|
---|
425 | if @do_validate and required and val.nil?
|
---|
426 | unless a_uri.include?("")
|
---|
427 | for prefix, uri in ns
|
---|
428 | if a_uri.include?(uri)
|
---|
429 | a_name = "#{prefix}:#{a_name}"
|
---|
430 | end
|
---|
431 | end
|
---|
432 | end
|
---|
433 | raise MissingAttributeError.new(tag_name, a_name)
|
---|
434 | end
|
---|
435 |
|
---|
436 | attributes[a_name] = val
|
---|
437 | end
|
---|
438 |
|
---|
439 | previous = @last_element
|
---|
440 | next_element = klass.new(@do_validate, attributes)
|
---|
441 | previous.instance_eval {set_next_element(tag_name, next_element)}
|
---|
442 | @last_element = next_element
|
---|
443 | @proc_stack.push Proc.new { |text, tags|
|
---|
444 | p(@last_element.class) if DEBUG
|
---|
445 | @last_element.content = text if klass.have_content?
|
---|
446 | if @do_validate
|
---|
447 | @last_element.validate_for_stream(tags, @ignore_unknown_element)
|
---|
448 | end
|
---|
449 | @last_element = previous
|
---|
450 | }
|
---|
451 | end
|
---|
452 |
|
---|
453 | end
|
---|
454 |
|
---|
455 | unless const_defined? :AVAILABLE_PARSER_LIBRARIES
|
---|
456 | AVAILABLE_PARSER_LIBRARIES = [
|
---|
457 | ["rss/xmlparser", :XMLParserParser],
|
---|
458 | ["rss/xmlscanner", :XMLScanParser],
|
---|
459 | ["rss/rexmlparser", :REXMLParser],
|
---|
460 | ]
|
---|
461 | end
|
---|
462 |
|
---|
463 | AVAILABLE_PARSERS = []
|
---|
464 |
|
---|
465 | AVAILABLE_PARSER_LIBRARIES.each do |lib, parser|
|
---|
466 | begin
|
---|
467 | require lib
|
---|
468 | AVAILABLE_PARSERS.push(const_get(parser))
|
---|
469 | rescue LoadError
|
---|
470 | end
|
---|
471 | end
|
---|
472 |
|
---|
473 | if AVAILABLE_PARSERS.empty?
|
---|
474 | raise XMLParserNotFound
|
---|
475 | end
|
---|
476 | end
|
---|