feedtools 0.2.18 → 0.2.19
Sign up to get free protection for your applications and to get access to all the features.
- data/CHANGELOG +28 -0
- data/lib/feed_tools.rb +328 -63
- data/lib/feed_tools/feed.rb +767 -764
- data/lib/feed_tools/feed_item.rb +684 -625
- data/lib/feed_tools/helpers/debug_helper.rb +37 -0
- data/lib/feed_tools/helpers/feed_tools_helper.rb +45 -41
- data/lib/feed_tools/helpers/generic_helper.rb +164 -0
- data/lib/feed_tools/helpers/retrieval_helper.rb +36 -0
- data/rakefile +298 -2
- data/test/unit/amp_test.rb +70 -69
- data/test/unit/atom_test.rb +91 -9
- data/test/unit/cache_test.rb +30 -11
- data/test/unit/cdf_test.rb +6 -4
- data/test/unit/encoding_test.rb +99 -0
- data/test/unit/generation_test.rb +3 -40
- data/test/unit/helper_test.rb +66 -6
- data/test/unit/interface_test.rb +34 -0
- data/test/unit/itunes_test.rb +19 -0
- data/test/unit/nonstandard_test.rb +22 -4
- data/test/unit/rdf_test.rb +19 -0
- data/test/unit/rss_test.rb +137 -43
- metadata +18 -8
- data/lib/feed_tools/vendor/builder.rb +0 -15
- data/lib/feed_tools/vendor/builder/blankslate.rb +0 -55
- data/lib/feed_tools/vendor/builder/xmlbase.rb +0 -144
- data/lib/feed_tools/vendor/builder/xmlevents.rb +0 -65
- data/lib/feed_tools/vendor/builder/xmlmarkup.rb +0 -299
data/lib/feed_tools/feed_item.rb
CHANGED
@@ -165,14 +165,18 @@ module FeedTools
|
|
165
165
|
# Warning, this method may be slow if you have a
|
166
166
|
# large number of FeedTools::Feed objects. Can't
|
167
167
|
# use a direct reference to the parent because it plays
|
168
|
-
# havoc with the garbage collector.
|
168
|
+
# havoc with the garbage collector. Could've used
|
169
|
+
# a WeakRef object, but really, if there are multiple
|
170
|
+
# parent feeds, something is going to go wrong, and the
|
171
|
+
# programmer needs to be notified. A WeakRef
|
172
|
+
# implementation can't detect this condition.
|
169
173
|
def feed
|
170
174
|
parent_feed = nil
|
171
175
|
ObjectSpace.each_object(FeedTools::Feed) do |feed|
|
172
|
-
if feed.instance_variable_get("@
|
176
|
+
if feed.instance_variable_get("@entries").nil?
|
173
177
|
feed.items
|
174
178
|
end
|
175
|
-
unsorted_items = feed.instance_variable_get("@
|
179
|
+
unsorted_items = feed.instance_variable_get("@entries")
|
176
180
|
for item in unsorted_items
|
177
181
|
if item.object_id == self.object_id
|
178
182
|
if parent_feed.nil?
|
@@ -224,13 +228,21 @@ module FeedTools
|
|
224
228
|
end
|
225
229
|
|
226
230
|
# Returns the first node within the root_node that matches the xpath query.
|
227
|
-
def find_node(xpath)
|
228
|
-
|
231
|
+
def find_node(xpath, select_result_value=false)
|
232
|
+
if feed.feed_data_type != :xml
|
233
|
+
raise "The feed data type is not xml."
|
234
|
+
end
|
235
|
+
return try_xpaths(self.root_node, [xpath],
|
236
|
+
:select_result_value => select_result_value)
|
229
237
|
end
|
230
238
|
|
231
239
|
# Returns all nodes within the root_node that match the xpath query.
|
232
|
-
def find_all_nodes(xpath)
|
233
|
-
|
240
|
+
def find_all_nodes(xpath, select_result_value=false)
|
241
|
+
if feed.feed_data_type != :xml
|
242
|
+
raise "The feed data type is not xml."
|
243
|
+
end
|
244
|
+
return try_xpaths_all(self.root_node, [xpath],
|
245
|
+
:select_result_value => select_result_value)
|
234
246
|
end
|
235
247
|
|
236
248
|
# Returns the root node of the feed item.
|
@@ -247,13 +259,13 @@ module FeedTools
|
|
247
259
|
# Returns the feed items's unique id
|
248
260
|
def id
|
249
261
|
if @id.nil?
|
250
|
-
|
251
|
-
|
252
|
-
|
253
|
-
|
254
|
-
|
255
|
-
|
256
|
-
|
262
|
+
@id = try_xpaths(self.root_node, [
|
263
|
+
"atom10:id/text()",
|
264
|
+
"atom03:id/text()",
|
265
|
+
"atom:id/text()",
|
266
|
+
"id/text()",
|
267
|
+
"guid/text()"
|
268
|
+
], :select_result_value => true)
|
257
269
|
end
|
258
270
|
return @id
|
259
271
|
end
|
@@ -266,37 +278,23 @@ module FeedTools
|
|
266
278
|
# Returns the feed item title
|
267
279
|
def title
|
268
280
|
if @title.nil?
|
269
|
-
|
270
|
-
|
271
|
-
|
272
|
-
|
273
|
-
|
274
|
-
|
275
|
-
|
276
|
-
|
277
|
-
title_node = XPath.first(root_node, "atom03:title",
|
278
|
-
FEED_TOOLS_NAMESPACES)
|
279
|
-
end
|
280
|
-
if title_node.nil?
|
281
|
-
title_node = XPath.first(root_node, "atom:title")
|
282
|
-
end
|
283
|
-
if title_node.nil?
|
284
|
-
title_node = XPath.first(root_node, "dc:title",
|
285
|
-
FEED_TOOLS_NAMESPACES)
|
286
|
-
end
|
287
|
-
if title_node.nil?
|
288
|
-
title_node = XPath.first(root_node, "dc:title")
|
289
|
-
end
|
290
|
-
if title_node.nil?
|
291
|
-
title_node = XPath.first(root_node, "TITLE")
|
292
|
-
end
|
293
|
-
end
|
281
|
+
repair_entities = false
|
282
|
+
title_node = try_xpaths(self.root_node, [
|
283
|
+
"atom10:title",
|
284
|
+
"atom03:title",
|
285
|
+
"atom:title",
|
286
|
+
"title",
|
287
|
+
"dc:title"
|
288
|
+
])
|
294
289
|
if title_node.nil?
|
295
290
|
return nil
|
296
291
|
end
|
297
|
-
title_type =
|
298
|
-
|
299
|
-
|
292
|
+
title_type = try_xpaths(title_node, "@type",
|
293
|
+
:select_result_value => true)
|
294
|
+
title_mode = try_xpaths(title_node, "@mode",
|
295
|
+
:select_result_value => true)
|
296
|
+
title_encoding = try_xpaths(title_node, "@encoding",
|
297
|
+
:select_result_value => true)
|
300
298
|
|
301
299
|
# Note that we're checking for misuse of type, mode and encoding here
|
302
300
|
if title_type == "base64" || title_mode == "base64" ||
|
@@ -318,7 +316,7 @@ module FeedTools
|
|
318
316
|
@title = FeedTools.unescape_entities(@title) if repair_entities
|
319
317
|
@title = FeedTools.tidy_html(@title) unless repair_entities
|
320
318
|
end
|
321
|
-
if
|
319
|
+
if !@title.blank? && FeedTools.configurations[:strip_comment_count]
|
322
320
|
# Some blogging tools include the number of comments in a post
|
323
321
|
# in the title... this is supremely ugly, and breaks any
|
324
322
|
# applications which expect the title to be static, so we're
|
@@ -331,7 +329,7 @@ module FeedTools
|
|
331
329
|
@title.gsub!(/>\n</, "><")
|
332
330
|
@title.gsub!(/\n/, " ")
|
333
331
|
@title.strip!
|
334
|
-
@title = nil if @title
|
332
|
+
@title = nil if @title.blank?
|
335
333
|
end
|
336
334
|
return @title
|
337
335
|
end
|
@@ -341,123 +339,90 @@ module FeedTools
|
|
341
339
|
@title = new_title
|
342
340
|
end
|
343
341
|
|
344
|
-
# Returns the feed item
|
345
|
-
def
|
346
|
-
if @
|
347
|
-
|
348
|
-
|
349
|
-
|
350
|
-
|
351
|
-
|
352
|
-
|
353
|
-
|
354
|
-
|
355
|
-
|
356
|
-
|
357
|
-
|
358
|
-
|
359
|
-
|
360
|
-
|
361
|
-
|
362
|
-
|
363
|
-
|
364
|
-
description_node = XPath.first(root_node, "xhtml:body")
|
365
|
-
end
|
366
|
-
if description_node.nil?
|
367
|
-
description_node = XPath.first(root_node, "xhtml:body",
|
368
|
-
FEED_TOOLS_NAMESPACES)
|
369
|
-
end
|
370
|
-
if description_node.nil?
|
371
|
-
description_node = XPath.first(root_node, "body")
|
372
|
-
end
|
373
|
-
if description_node.nil?
|
374
|
-
description_node = XPath.first(root_node, "description")
|
375
|
-
end
|
376
|
-
if description_node.nil?
|
377
|
-
description_node = XPath.first(root_node, "tagline")
|
378
|
-
end
|
379
|
-
if description_node.nil?
|
380
|
-
description_node = XPath.first(root_node, "subtitle")
|
381
|
-
end
|
382
|
-
if description_node.nil?
|
383
|
-
description_node = XPath.first(root_node, "summary")
|
384
|
-
end
|
385
|
-
if description_node.nil?
|
386
|
-
description_node = XPath.first(root_node, "abstract")
|
387
|
-
end
|
388
|
-
if description_node.nil?
|
389
|
-
description_node = XPath.first(root_node, "ABSTRACT")
|
390
|
-
end
|
391
|
-
if description_node.nil?
|
392
|
-
description_node = XPath.first(root_node, "blurb")
|
393
|
-
end
|
394
|
-
if description_node.nil?
|
395
|
-
description_node = XPath.first(root_node, "info")
|
396
|
-
end
|
397
|
-
end
|
398
|
-
if description_node.nil?
|
342
|
+
# Returns the feed item content
|
343
|
+
def content
|
344
|
+
if @content.nil?
|
345
|
+
repair_entities = false
|
346
|
+
content_node = try_xpaths(self.root_node, [
|
347
|
+
"content:encoded",
|
348
|
+
"content",
|
349
|
+
"fullitem",
|
350
|
+
"xhtml:body",
|
351
|
+
"body",
|
352
|
+
"encoded",
|
353
|
+
"description",
|
354
|
+
"tagline",
|
355
|
+
"subtitle",
|
356
|
+
"summary",
|
357
|
+
"abstract",
|
358
|
+
"blurb",
|
359
|
+
"info"
|
360
|
+
])
|
361
|
+
if content_node.nil?
|
399
362
|
return nil
|
400
363
|
end
|
401
|
-
|
402
|
-
|
403
|
-
|
364
|
+
content_type = try_xpaths(content_node, "@type",
|
365
|
+
:select_result_value => true)
|
366
|
+
content_mode = try_xpaths(content_node, "@mode",
|
367
|
+
:select_result_value => true)
|
368
|
+
content_encoding = try_xpaths(content_node, "@encoding",
|
369
|
+
:select_result_value => true)
|
404
370
|
|
405
371
|
# Note that we're checking for misuse of type, mode and encoding here
|
406
|
-
if
|
407
|
-
@
|
372
|
+
if !content_encoding.blank?
|
373
|
+
@content =
|
408
374
|
"[Embedded data objects are not currently supported.]"
|
409
|
-
elsif
|
410
|
-
@
|
411
|
-
elsif
|
412
|
-
|
413
|
-
@
|
414
|
-
elsif
|
415
|
-
|
416
|
-
|
417
|
-
@
|
418
|
-
elsif
|
419
|
-
@
|
420
|
-
|
375
|
+
elsif content_node.cdatas.size > 0
|
376
|
+
@content = content_node.cdatas.first.value
|
377
|
+
elsif content_type == "base64" || content_mode == "base64" ||
|
378
|
+
content_encoding == "base64"
|
379
|
+
@content = Base64.decode64(content_node.inner_xml.strip)
|
380
|
+
elsif content_type == "xhtml" || content_mode == "xhtml" ||
|
381
|
+
content_type == "xml" || content_mode == "xml" ||
|
382
|
+
content_type == "application/xhtml+xml"
|
383
|
+
@content = content_node.inner_xml
|
384
|
+
elsif content_type == "escaped" || content_mode == "escaped"
|
385
|
+
@content = FeedTools.unescape_entities(
|
386
|
+
content_node.inner_xml)
|
421
387
|
else
|
422
|
-
@
|
388
|
+
@content = content_node.inner_xml
|
423
389
|
repair_entities = true
|
424
390
|
end
|
425
|
-
if @
|
426
|
-
@
|
427
|
-
@description = "" if @description.nil?
|
391
|
+
if @content.blank?
|
392
|
+
@content = self.itunes_summary
|
428
393
|
end
|
429
|
-
if @
|
430
|
-
@
|
431
|
-
@description = "" if @description.nil?
|
394
|
+
if @content.blank?
|
395
|
+
@content = self.itunes_subtitle
|
432
396
|
end
|
433
397
|
|
434
|
-
unless @
|
435
|
-
@
|
436
|
-
@
|
437
|
-
@
|
398
|
+
unless @content.blank?
|
399
|
+
@content = FeedTools.sanitize_html(@content, :strip)
|
400
|
+
@content = FeedTools.unescape_entities(@content) if repair_entities
|
401
|
+
@content = FeedTools.tidy_html(@content)
|
438
402
|
end
|
439
403
|
|
440
|
-
@
|
441
|
-
@
|
404
|
+
@content = @content.strip unless @content.nil?
|
405
|
+
@content = nil if @content.blank?
|
442
406
|
end
|
443
|
-
return @
|
407
|
+
return @content
|
444
408
|
end
|
445
409
|
|
446
|
-
# Sets the feed item
|
447
|
-
def
|
448
|
-
@
|
410
|
+
# Sets the feed item content
|
411
|
+
def content=(new_content)
|
412
|
+
@content = new_content
|
449
413
|
end
|
450
414
|
|
451
415
|
# Returns the contents of the itunes:summary element
|
452
416
|
def itunes_summary
|
453
417
|
if @itunes_summary.nil?
|
454
|
-
@itunes_summary =
|
455
|
-
"itunes:summary/text()"
|
456
|
-
|
457
|
-
|
458
|
-
|
459
|
-
unless @itunes_summary.nil?
|
418
|
+
@itunes_summary = try_xpaths(self.root_node, [
|
419
|
+
"itunes:summary/text()"
|
420
|
+
])
|
421
|
+
unless @itunes_summary.blank?
|
422
|
+
@itunes_summary = FeedTools.unescape_entities(@itunes_summary)
|
460
423
|
@itunes_summary = FeedTools.sanitize_html(@itunes_summary)
|
424
|
+
else
|
425
|
+
@itunes_summary = nil
|
461
426
|
end
|
462
427
|
end
|
463
428
|
return @itunes_summary
|
@@ -471,13 +436,14 @@ module FeedTools
|
|
471
436
|
# Returns the contents of the itunes:subtitle element
|
472
437
|
def itunes_subtitle
|
473
438
|
if @itunes_subtitle.nil?
|
474
|
-
@itunes_subtitle =
|
475
|
-
"itunes:subtitle/text()"
|
476
|
-
|
477
|
-
|
478
|
-
|
479
|
-
unless @itunes_subtitle.nil?
|
439
|
+
@itunes_subtitle = try_xpaths(self.root_node, [
|
440
|
+
"itunes:subtitle/text()"
|
441
|
+
])
|
442
|
+
unless @itunes_subtitle.blank?
|
443
|
+
@itunes_subtitle = FeedTools.unescape_entities(@itunes_subtitle)
|
480
444
|
@itunes_subtitle = FeedTools.sanitize_html(@itunes_subtitle)
|
445
|
+
else
|
446
|
+
@itunes_subtitle = nil
|
481
447
|
end
|
482
448
|
end
|
483
449
|
return @itunes_subtitle
|
@@ -511,39 +477,31 @@ module FeedTools
|
|
511
477
|
# Returns the feed item link
|
512
478
|
def link
|
513
479
|
if @link.nil?
|
514
|
-
|
515
|
-
|
516
|
-
|
517
|
-
|
518
|
-
|
519
|
-
|
520
|
-
|
521
|
-
|
522
|
-
|
523
|
-
|
524
|
-
|
525
|
-
|
526
|
-
|
527
|
-
|
528
|
-
|
529
|
-
|
530
|
-
|
531
|
-
|
532
|
-
|
533
|
-
|
534
|
-
if @link == ""
|
535
|
-
@link = XPath.first(root_node, "@HREF").to_s
|
536
|
-
end
|
537
|
-
if @link == ""
|
538
|
-
@link = XPath.first(root_node, "A/@HREF").to_s
|
539
|
-
end
|
540
|
-
end
|
541
|
-
if @link == "" || @link.nil?
|
480
|
+
@link = try_xpaths(self.root_node, [
|
481
|
+
"atom10:link[@type='application/xhtml+xml']/@href",
|
482
|
+
"atom10:link[@type='text/html']/@href",
|
483
|
+
"atom10:link[@rel='alternate']/@href",
|
484
|
+
"atom03:link[@type='application/xhtml+xml']/@href",
|
485
|
+
"atom03:link[@type='text/html']/@href",
|
486
|
+
"atom03:link[@rel='alternate']/@href",
|
487
|
+
"atom:link[@type='application/xhtml+xml']/@href",
|
488
|
+
"atom:link[@type='text/html']/@href",
|
489
|
+
"atom:link[@rel='alternate']/@href",
|
490
|
+
"link[@type='application/xhtml+xml']/@href",
|
491
|
+
"link[@type='text/html']/@href",
|
492
|
+
"link[@rel='alternate']/@href",
|
493
|
+
"link/text()",
|
494
|
+
"@rdf:about",
|
495
|
+
"guid[@isPermaLink='true']/text()",
|
496
|
+
"@href",
|
497
|
+
"a/@href"
|
498
|
+
], :select_result_value => true)
|
499
|
+
if @link.blank?
|
542
500
|
if FeedTools.is_uri? self.guid
|
543
501
|
@link = self.guid
|
544
502
|
end
|
545
503
|
end
|
546
|
-
if
|
504
|
+
if !@link.blank?
|
547
505
|
@link = FeedTools.unescape_entities(@link)
|
548
506
|
end
|
549
507
|
# TODO: Actually implement proper relative url resolving instead of this crap
|
@@ -556,7 +514,50 @@ module FeedTools
|
|
556
514
|
# # prepend the base to the link since they seem to have used a relative path
|
557
515
|
# @link = feed.base + @link
|
558
516
|
# end
|
559
|
-
@link
|
517
|
+
if @link.blank?
|
518
|
+
link_node = try_xpaths(self.root_node, [
|
519
|
+
"atom10:link",
|
520
|
+
"atom03:link",
|
521
|
+
"atom:link",
|
522
|
+
"link"
|
523
|
+
])
|
524
|
+
if link_node != nil
|
525
|
+
if link_node.attributes['type'].to_s =~ /^image/ ||
|
526
|
+
link_node.attributes['type'].to_s =~ /^application/ ||
|
527
|
+
link_node.attributes['type'].to_s =~ /xml/ ||
|
528
|
+
link_node.attributes['rel'].to_s =~ /self/
|
529
|
+
for child in self.root_node
|
530
|
+
if child.class == REXML::Element
|
531
|
+
if child.name.downcase == "link"
|
532
|
+
if child.attributes['type'].to_s =~ /^image/ ||
|
533
|
+
child.attributes['type'].to_s =~ /^application/ ||
|
534
|
+
child.attributes['type'].to_s =~ /xml/ ||
|
535
|
+
child.attributes['rel'].to_s =~ /self/
|
536
|
+
@link = nil
|
537
|
+
next
|
538
|
+
else
|
539
|
+
@link = child.attributes['href'].to_s
|
540
|
+
if @link.blank?
|
541
|
+
@link = child.inner_xml
|
542
|
+
end
|
543
|
+
if @link.blank?
|
544
|
+
next
|
545
|
+
end
|
546
|
+
break
|
547
|
+
end
|
548
|
+
end
|
549
|
+
end
|
550
|
+
end
|
551
|
+
else
|
552
|
+
@link = link_node.attributes['href'].to_s
|
553
|
+
end
|
554
|
+
end
|
555
|
+
end
|
556
|
+
@link = self.comments if @link.blank?
|
557
|
+
@link = nil if @link.blank?
|
558
|
+
if FeedTools.configurations[:url_normalization_enabled]
|
559
|
+
@link = FeedTools.normalize_url(@link)
|
560
|
+
end
|
560
561
|
end
|
561
562
|
return @link
|
562
563
|
end
|
@@ -570,30 +571,24 @@ module FeedTools
|
|
570
571
|
def categories
|
571
572
|
if @categories.nil?
|
572
573
|
@categories = []
|
573
|
-
category_nodes =
|
574
|
-
|
575
|
-
|
576
|
-
|
577
|
-
|
578
|
-
|
579
|
-
|
580
|
-
|
581
|
-
|
582
|
-
|
583
|
-
|
584
|
-
|
585
|
-
|
586
|
-
|
587
|
-
|
588
|
-
|
589
|
-
|
590
|
-
|
591
|
-
category.scheme = XPath.first(category_node, "@domain").to_s
|
592
|
-
end
|
593
|
-
category.scheme.strip! unless category.scheme.nil?
|
594
|
-
category.scheme = nil if category.scheme == ""
|
595
|
-
@categories << category
|
596
|
-
end
|
574
|
+
category_nodes = try_xpaths_all(self.root_node, [
|
575
|
+
"category",
|
576
|
+
"dc:subject"
|
577
|
+
])
|
578
|
+
for category_node in category_nodes
|
579
|
+
category = FeedTools::Feed::Category.new
|
580
|
+
category.term = try_xpaths(category_node, ["@term", "text()"],
|
581
|
+
:select_result_value => true)
|
582
|
+
category.term.strip! unless category.term.nil?
|
583
|
+
category.label = try_xpaths(category_node, ["@label"],
|
584
|
+
:select_result_value => true)
|
585
|
+
category.label.strip! unless category.label.nil?
|
586
|
+
category.scheme = try_xpaths(category_node, [
|
587
|
+
"@scheme",
|
588
|
+
"@domain"
|
589
|
+
], :select_result_value => true)
|
590
|
+
category.scheme.strip! unless category.scheme.nil?
|
591
|
+
@categories << category
|
597
592
|
end
|
598
593
|
end
|
599
594
|
return @categories
|
@@ -603,53 +598,61 @@ module FeedTools
|
|
603
598
|
def images
|
604
599
|
if @images.nil?
|
605
600
|
@images = []
|
606
|
-
image_nodes =
|
607
|
-
|
608
|
-
|
609
|
-
|
610
|
-
|
611
|
-
|
612
|
-
|
613
|
-
|
614
|
-
|
615
|
-
end
|
616
|
-
unless image_nodes.nil?
|
601
|
+
image_nodes = try_xpaths_all(self.root_node, [
|
602
|
+
"image",
|
603
|
+
"logo",
|
604
|
+
"atom10:link",
|
605
|
+
"atom03:link",
|
606
|
+
"atom:link",
|
607
|
+
"link"
|
608
|
+
])
|
609
|
+
unless image_nodes.blank?
|
617
610
|
for image_node in image_nodes
|
618
611
|
image = FeedTools::Feed::Image.new
|
619
|
-
image.url =
|
620
|
-
|
621
|
-
|
622
|
-
|
623
|
-
if image.url
|
624
|
-
(image_node.attributes['type'] =~ /^image/) == 0)
|
625
|
-
image.url =
|
612
|
+
image.url = try_xpaths(image_node, [
|
613
|
+
"url/text()",
|
614
|
+
"@rdf:resource"
|
615
|
+
], :select_result_value => true)
|
616
|
+
if image.url.blank? && (image_node.name == "logo" ||
|
617
|
+
(image_node.attributes['type'].to_s =~ /^image/) == 0)
|
618
|
+
image.url = try_xpaths(image_node, [
|
619
|
+
"@atom10:href",
|
620
|
+
"@atom03:href",
|
621
|
+
"@atom:href",
|
622
|
+
"@href"
|
623
|
+
], :select_result_value => true)
|
624
|
+
if image.url == self.link && image.url != nil
|
625
|
+
image.url = nil
|
626
|
+
end
|
626
627
|
end
|
627
|
-
if image.url
|
628
|
-
image.url =
|
628
|
+
if image.url.blank? && image_node.name == "LOGO"
|
629
|
+
image.url = try_xpaths(image_node, [
|
630
|
+
"@href"
|
631
|
+
], :select_result_value => true)
|
629
632
|
end
|
630
633
|
image.url.strip! unless image.url.nil?
|
631
|
-
image.
|
632
|
-
|
634
|
+
image.title = try_xpaths(image_node,
|
635
|
+
["title/text()"], :select_result_value => true)
|
633
636
|
image.title.strip! unless image.title.nil?
|
634
|
-
image.
|
635
|
-
|
636
|
-
XPath.first(image_node, "description/text()").to_s
|
637
|
+
image.description = try_xpaths(image_node,
|
638
|
+
["description/text()"], :select_result_value => true)
|
637
639
|
image.description.strip! unless image.description.nil?
|
638
|
-
image.
|
639
|
-
|
640
|
+
image.link = try_xpaths(image_node,
|
641
|
+
["link/text()"], :select_result_value => true)
|
640
642
|
image.link.strip! unless image.link.nil?
|
641
|
-
image.
|
642
|
-
|
643
|
+
image.height = try_xpaths(image_node,
|
644
|
+
["height/text()"], :select_result_value => true).to_i
|
643
645
|
image.height = nil if image.height <= 0
|
644
|
-
image.width =
|
646
|
+
image.width = try_xpaths(image_node,
|
647
|
+
["width/text()"], :select_result_value => true).to_i
|
645
648
|
image.width = nil if image.width <= 0
|
646
|
-
image.style =
|
647
|
-
|
648
|
-
|
649
|
-
|
649
|
+
image.style = try_xpaths(image_node, [
|
650
|
+
"style/text()",
|
651
|
+
"@style"
|
652
|
+
], :select_result_value => true)
|
650
653
|
image.style.strip! unless image.style.nil?
|
651
|
-
image.style
|
652
|
-
@images << image
|
654
|
+
image.style.downcase! unless image.style.nil?
|
655
|
+
@images << image unless image.url.nil?
|
653
656
|
end
|
654
657
|
end
|
655
658
|
end
|
@@ -657,19 +660,15 @@ module FeedTools
|
|
657
660
|
end
|
658
661
|
|
659
662
|
# Returns the feed item itunes image link
|
660
|
-
#
|
661
|
-
# If it's not present, falls back to the normal image link.
|
662
|
-
# Technically, the itunes spec says that the image needs to be
|
663
|
-
# square and larger than 300x300, but hey, if there's an image
|
664
|
-
# to be had, it's better than none at all.
|
665
663
|
def itunes_image_link
|
666
664
|
if @itunes_image_link.nil?
|
667
|
-
|
668
|
-
|
669
|
-
|
670
|
-
|
665
|
+
@itunes_image_link = try_xpaths(self.root_node, [
|
666
|
+
"itunes:image/@href",
|
667
|
+
"itunes:link[@rel='image']/@href"
|
668
|
+
], :select_result_value => true)
|
669
|
+
if FeedTools.configurations[:url_normalization_enabled]
|
670
|
+
@itunes_image_link = FeedTools.normalize_url(@itunes_image_link)
|
671
671
|
end
|
672
|
-
@itunes_image_link = FeedTools.normalize_url(@itunes_image_link)
|
673
672
|
end
|
674
673
|
return @itunes_image_link
|
675
674
|
end
|
@@ -680,13 +679,14 @@ module FeedTools
|
|
680
679
|
end
|
681
680
|
|
682
681
|
# Returns the feed item media thumbnail link
|
683
|
-
#
|
684
|
-
# If it's not present, falls back to the normal image link.
|
685
682
|
def media_thumbnail_link
|
686
683
|
if @media_thumbnail_link.nil?
|
687
|
-
|
688
|
-
|
689
|
-
|
684
|
+
@media_thumbnail_link = try_xpaths(self.root_node, [
|
685
|
+
"media:thumbnail/@url"
|
686
|
+
], :select_result_value => true)
|
687
|
+
if FeedTools.configurations[:url_normalization_enabled]
|
688
|
+
@media_thumbnail_link = FeedTools.normalize_url(@media_thumbnail_link)
|
689
|
+
end
|
690
690
|
end
|
691
691
|
return @media_thumbnail_link
|
692
692
|
end
|
@@ -699,44 +699,28 @@ module FeedTools
|
|
699
699
|
# Returns the feed item's copyright information
|
700
700
|
def copyright
|
701
701
|
if @copyright.nil?
|
702
|
-
|
703
|
-
|
704
|
-
|
705
|
-
|
706
|
-
|
707
|
-
|
708
|
-
|
709
|
-
|
710
|
-
|
711
|
-
|
712
|
-
FEED_TOOLS_NAMESPACES)
|
713
|
-
end
|
714
|
-
if copyright_node.nil?
|
715
|
-
copyright_node = XPath.first(root_node, "copyright",
|
716
|
-
FEED_TOOLS_NAMESPACES)
|
717
|
-
end
|
718
|
-
if copyright_node.nil?
|
719
|
-
copyright_node = XPath.first(root_node, "atom03:copyright",
|
720
|
-
FEED_TOOLS_NAMESPACES)
|
721
|
-
end
|
722
|
-
if copyright_node.nil?
|
723
|
-
copyright_node = XPath.first(root_node, "atom10:copyright",
|
724
|
-
FEED_TOOLS_NAMESPACES)
|
725
|
-
end
|
726
|
-
if copyright_node.nil?
|
727
|
-
copyright_node = XPath.first(root_node, "copyrights",
|
728
|
-
FEED_TOOLS_NAMESPACES)
|
729
|
-
end
|
730
|
-
end
|
702
|
+
repair_entities = false
|
703
|
+
copyright_node = try_xpaths(self.root_node, [
|
704
|
+
"atom10:copyright",
|
705
|
+
"atom03:copyright",
|
706
|
+
"atom:copyright",
|
707
|
+
"copyright",
|
708
|
+
"copyrights",
|
709
|
+
"dc:rights",
|
710
|
+
"rights"
|
711
|
+
])
|
731
712
|
if copyright_node.nil?
|
732
713
|
return nil
|
733
714
|
end
|
734
|
-
copyright_type =
|
735
|
-
|
736
|
-
|
715
|
+
copyright_type = try_xpaths(copyright_node, "@type",
|
716
|
+
:select_result_value => true)
|
717
|
+
copyright_mode = try_xpaths(copyright_node, "@mode",
|
718
|
+
:select_result_value => true)
|
719
|
+
copyright_encoding = try_xpaths(copyright_node, "@encoding",
|
720
|
+
:select_result_value => true)
|
737
721
|
|
738
722
|
# Note that we're checking for misuse of type, mode and encoding here
|
739
|
-
if copyright_encoding
|
723
|
+
if !copyright_encoding.blank?
|
740
724
|
@copyright =
|
741
725
|
"[Embedded data objects are not currently supported.]"
|
742
726
|
elsif copyright_node.cdatas.size > 0
|
@@ -763,7 +747,7 @@ module FeedTools
|
|
763
747
|
end
|
764
748
|
|
765
749
|
@copyright = @copyright.strip unless @copyright.nil?
|
766
|
-
@copyright = nil if @copyright
|
750
|
+
@copyright = nil if @copyright.blank?
|
767
751
|
end
|
768
752
|
return @copyright
|
769
753
|
end
|
@@ -779,16 +763,27 @@ module FeedTools
|
|
779
763
|
@enclosures = []
|
780
764
|
|
781
765
|
# First, load up all the different possible sources of enclosures
|
782
|
-
rss_enclosures =
|
783
|
-
|
784
|
-
|
785
|
-
|
786
|
-
|
787
|
-
|
788
|
-
|
766
|
+
rss_enclosures =
|
767
|
+
try_xpaths_all(self.root_node, ["enclosure"])
|
768
|
+
atom_enclosures =
|
769
|
+
try_xpaths_all(self.root_node, [
|
770
|
+
"atom10:link[@rel='enclosure']",
|
771
|
+
"atom03:link[@rel='enclosure']",
|
772
|
+
"atom:link[@rel='enclosure']",
|
773
|
+
"link[@rel='enclosure']"
|
774
|
+
])
|
775
|
+
media_content_enclosures =
|
776
|
+
try_xpaths_all(self.root_node, ["media:content"])
|
777
|
+
media_group_enclosures =
|
778
|
+
try_xpaths_all(self.root_node, ["media:group"])
|
779
|
+
|
780
|
+
# Parse RSS-type enclosures. Thanks to a few buggy enclosures
|
781
|
+
# implementations, sometimes these also manage to show up in atom
|
782
|
+
# files.
|
789
783
|
for enclosure_node in rss_enclosures
|
790
784
|
enclosure = Enclosure.new
|
791
|
-
enclosure.url = FeedTools.unescape_entities(
|
785
|
+
enclosure.url = FeedTools.unescape_entities(
|
786
|
+
enclosure_node.attributes["url"].to_s)
|
792
787
|
enclosure.type = enclosure_node.attributes["type"].to_s
|
793
788
|
enclosure.file_size = enclosure_node.attributes["length"].to_i
|
794
789
|
enclosure.credits = []
|
@@ -796,10 +791,11 @@ module FeedTools
|
|
796
791
|
@enclosures << enclosure
|
797
792
|
end
|
798
793
|
|
799
|
-
# Parse atom-type enclosures. If there are repeats of the same
|
800
|
-
# we merge the two together.
|
794
|
+
# Parse atom-type enclosures. If there are repeats of the same
|
795
|
+
# enclosure object, we merge the two together.
|
801
796
|
for enclosure_node in atom_enclosures
|
802
|
-
enclosure_url = FeedTools.unescape_entities(
|
797
|
+
enclosure_url = FeedTools.unescape_entities(
|
798
|
+
enclosure_node.attributes["href"].to_s)
|
803
799
|
enclosure = nil
|
804
800
|
new_enclosure = false
|
805
801
|
for existing_enclosure in @enclosures
|
@@ -822,13 +818,15 @@ module FeedTools
|
|
822
818
|
end
|
823
819
|
end
|
824
820
|
|
825
|
-
# Creates an anonymous method to parse content objects from the media
|
826
|
-
# do this to avoid excessive duplication of code since we
|
827
|
-
# processing for content objects within group
|
821
|
+
# Creates an anonymous method to parse content objects from the media
|
822
|
+
# module. We do this to avoid excessive duplication of code since we
|
823
|
+
# have to do identical processing for content objects within group
|
824
|
+
# objects.
|
828
825
|
parse_media_content = lambda do |media_content_nodes|
|
829
826
|
affected_enclosures = []
|
830
827
|
for enclosure_node in media_content_nodes
|
831
|
-
enclosure_url = FeedTools.unescape_entities(
|
828
|
+
enclosure_url = FeedTools.unescape_entities(
|
829
|
+
enclosure_node.attributes["url"].to_s)
|
832
830
|
enclosure = nil
|
833
831
|
new_enclosure = false
|
834
832
|
for existing_enclosure in @enclosures
|
@@ -849,73 +847,84 @@ module FeedTools
|
|
849
847
|
enclosure.width = enclosure_node.attributes["width"].to_i
|
850
848
|
enclosure.bitrate = enclosure_node.attributes["bitrate"].to_i
|
851
849
|
enclosure.framerate = enclosure_node.attributes["framerate"].to_i
|
852
|
-
enclosure.expression =
|
850
|
+
enclosure.expression =
|
851
|
+
enclosure_node.attributes["expression"].to_s
|
853
852
|
enclosure.is_default =
|
854
853
|
(enclosure_node.attributes["isDefault"].to_s.downcase == "true")
|
855
|
-
|
854
|
+
enclosure_thumbnail_url = try_xpaths(enclosure_node,
|
855
|
+
["media:thumbnail/@url"], :select_result_value => true)
|
856
|
+
if !enclosure_thumbnail_url.blank?
|
856
857
|
enclosure.thumbnail = EnclosureThumbnail.new(
|
857
|
-
FeedTools.unescape_entities(
|
858
|
-
FeedTools.unescape_entities(
|
859
|
-
|
858
|
+
FeedTools.unescape_entities(enclosure_thumbnail_url),
|
859
|
+
FeedTools.unescape_entities(
|
860
|
+
try_xpaths(enclosure_node, ["media:thumbnail/@height"],
|
861
|
+
:select_result_value => true)),
|
862
|
+
FeedTools.unescape_entities(
|
863
|
+
try_xpaths(enclosure_node, ["media:thumbnail/@width"],
|
864
|
+
:select_result_value => true))
|
860
865
|
)
|
861
|
-
if enclosure.thumbnail.height == ""
|
862
|
-
enclosure.thumbnail.height = nil
|
863
|
-
end
|
864
|
-
if enclosure.thumbnail.width == ""
|
865
|
-
enclosure.thumbnail.width = nil
|
866
|
-
end
|
867
866
|
end
|
868
867
|
enclosure.categories = []
|
869
|
-
for category in
|
868
|
+
for category in try_xpaths_all(enclosure_node, ["media:category"])
|
870
869
|
enclosure.categories << FeedTools::Feed::Category.new
|
871
870
|
enclosure.categories.last.term =
|
872
|
-
FeedTools.unescape_entities(category.
|
871
|
+
FeedTools.unescape_entities(category.inner_xml)
|
873
872
|
enclosure.categories.last.scheme =
|
874
|
-
FeedTools.unescape_entities(
|
873
|
+
FeedTools.unescape_entities(
|
874
|
+
category.attributes["scheme"].to_s)
|
875
875
|
enclosure.categories.last.label =
|
876
|
-
FeedTools.unescape_entities(
|
877
|
-
|
876
|
+
FeedTools.unescape_entities(
|
877
|
+
category.attributes["label"].to_s)
|
878
|
+
if enclosure.categories.last.scheme.blank?
|
878
879
|
enclosure.categories.last.scheme = nil
|
879
880
|
end
|
880
|
-
if enclosure.categories.last.label
|
881
|
+
if enclosure.categories.last.label.blank?
|
881
882
|
enclosure.categories.last.label = nil
|
882
883
|
end
|
883
884
|
end
|
884
|
-
|
885
|
+
enclosure_media_hash = try_xpaths(enclosure_node,
|
886
|
+
["media:hash/text()"], :select_result_value => true)
|
887
|
+
if !enclosure_media_hash.nil?
|
885
888
|
enclosure.hash = EnclosureHash.new(
|
886
|
-
FeedTools.sanitize_html(FeedTools.unescape_entities(
|
887
|
-
|
889
|
+
FeedTools.sanitize_html(FeedTools.unescape_entities(
|
890
|
+
enclosure_media_hash), :strip),
|
888
891
|
"md5"
|
889
892
|
)
|
890
893
|
end
|
891
|
-
|
894
|
+
enclosure_media_player_url = try_xpaths(enclosure_node,
|
895
|
+
["media:player/@url"], :select_result_value => true)
|
896
|
+
if !enclosure_media_player_url.blank?
|
892
897
|
enclosure.player = EnclosurePlayer.new(
|
893
|
-
FeedTools.unescape_entities(
|
894
|
-
FeedTools.unescape_entities(
|
895
|
-
|
898
|
+
FeedTools.unescape_entities(enclosure_media_player_url),
|
899
|
+
FeedTools.unescape_entities(
|
900
|
+
try_xpaths(enclosure_node,
|
901
|
+
["media:player/@height"], :select_result_value => true)),
|
902
|
+
FeedTools.unescape_entities(
|
903
|
+
try_xpaths(enclosure_node,
|
904
|
+
["media:player/@width"], :select_result_value => true))
|
896
905
|
)
|
897
|
-
if enclosure.player.height == ""
|
898
|
-
enclosure.player.height = nil
|
899
|
-
end
|
900
|
-
if enclosure.player.width == ""
|
901
|
-
enclosure.player.width = nil
|
902
|
-
end
|
903
906
|
end
|
904
907
|
enclosure.credits = []
|
905
|
-
for credit in
|
908
|
+
for credit in try_xpaths_all(enclosure_node, ["media:credit"])
|
906
909
|
enclosure.credits << EnclosureCredit.new(
|
907
|
-
FeedTools.unescape_entities(credit.
|
908
|
-
FeedTools.unescape_entities(
|
910
|
+
FeedTools.unescape_entities(credit.inner_xml.to_s.strip),
|
911
|
+
FeedTools.unescape_entities(
|
912
|
+
credit.attributes["role"].to_s.downcase)
|
909
913
|
)
|
910
|
-
if enclosure.credits.last.
|
914
|
+
if enclosure.credits.last.name.blank?
|
915
|
+
enclosure.credits.last.name = nil
|
916
|
+
end
|
917
|
+
if enclosure.credits.last.role.blank?
|
911
918
|
enclosure.credits.last.role = nil
|
912
919
|
end
|
913
920
|
end
|
914
|
-
enclosure.explicit = (
|
915
|
-
"media:adult/text()").to_s.downcase == "true")
|
916
|
-
|
917
|
-
|
918
|
-
|
921
|
+
enclosure.explicit = (try_xpaths(enclosure_node,
|
922
|
+
["media:adult/text()"]).to_s.downcase == "true")
|
923
|
+
enclosure_media_text =
|
924
|
+
try_xpaths(enclosure_node, ["media:text/text()"])
|
925
|
+
if !enclosure_media_text.blank?
|
926
|
+
enclosure.text = FeedTools.unescape_entities(
|
927
|
+
enclosure_media_text)
|
919
928
|
end
|
920
929
|
affected_enclosures << enclosure
|
921
930
|
if new_enclosure
|
@@ -933,7 +942,7 @@ module FeedTools
|
|
933
942
|
# Parse the group objects.
|
934
943
|
for media_group in media_group_enclosures
|
935
944
|
group_media_content_enclosures =
|
936
|
-
|
945
|
+
try_xpaths_all(media_group, ["media:content"])
|
937
946
|
|
938
947
|
# Parse the content objects within the group objects.
|
939
948
|
affected_enclosures =
|
@@ -942,82 +951,87 @@ module FeedTools
|
|
942
951
|
# Now make sure that content objects inherit certain properties from
|
943
952
|
# the group objects.
|
944
953
|
for enclosure in affected_enclosures
|
945
|
-
|
946
|
-
|
954
|
+
media_group_thumbnail = try_xpaths(media_group,
|
955
|
+
["media:thumbnail/@url"], :select_result_value => true)
|
956
|
+
if enclosure.thumbnail.nil? && !media_group_thumbnail.blank?
|
947
957
|
enclosure.thumbnail = EnclosureThumbnail.new(
|
948
958
|
FeedTools.unescape_entities(
|
949
|
-
|
959
|
+
media_group_thumbnail),
|
950
960
|
FeedTools.unescape_entities(
|
951
|
-
|
961
|
+
try_xpaths(media_group, ["media:thumbnail/@height"],
|
962
|
+
:select_result_value => true)),
|
952
963
|
FeedTools.unescape_entities(
|
953
|
-
|
964
|
+
try_xpaths(media_group, ["media:thumbnail/@width"],
|
965
|
+
:select_result_value => true))
|
954
966
|
)
|
955
|
-
if enclosure.thumbnail.height == ""
|
956
|
-
enclosure.thumbnail.height = nil
|
957
|
-
end
|
958
|
-
if enclosure.thumbnail.width == ""
|
959
|
-
enclosure.thumbnail.width = nil
|
960
|
-
end
|
961
967
|
end
|
962
|
-
if (enclosure.categories.
|
968
|
+
if (enclosure.categories.blank?)
|
963
969
|
enclosure.categories = []
|
964
|
-
for category in
|
970
|
+
for category in try_xpaths_all(media_group, ["media:category"])
|
965
971
|
enclosure.categories << FeedTools::Feed::Category.new
|
966
972
|
enclosure.categories.last.term =
|
967
|
-
FeedTools.unescape_entities(category.
|
973
|
+
FeedTools.unescape_entities(category.inner_xml)
|
968
974
|
enclosure.categories.last.scheme =
|
969
|
-
FeedTools.unescape_entities(
|
975
|
+
FeedTools.unescape_entities(
|
976
|
+
category.attributes["scheme"].to_s)
|
970
977
|
enclosure.categories.last.label =
|
971
|
-
FeedTools.unescape_entities(
|
972
|
-
|
978
|
+
FeedTools.unescape_entities(
|
979
|
+
category.attributes["label"].to_s)
|
980
|
+
if enclosure.categories.last.scheme.blank?
|
973
981
|
enclosure.categories.last.scheme = nil
|
974
982
|
end
|
975
|
-
if enclosure.categories.last.label
|
983
|
+
if enclosure.categories.last.label.blank?
|
976
984
|
enclosure.categories.last.label = nil
|
977
985
|
end
|
978
986
|
end
|
979
987
|
end
|
980
|
-
|
981
|
-
|
988
|
+
enclosure_media_group_hash = try_xpaths(enclosure_node,
|
989
|
+
["media:hash/text()"], :select_result_value => true)
|
990
|
+
if enclosure.hash.nil? && !enclosure_media_group_hash.blank?
|
982
991
|
enclosure.hash = EnclosureHash.new(
|
983
|
-
FeedTools.
|
992
|
+
FeedTools.sanitize_html(FeedTools.unescape_entities(
|
993
|
+
enclosure_media_group_hash), :strip),
|
984
994
|
"md5"
|
985
995
|
)
|
986
996
|
end
|
987
|
-
|
988
|
-
|
997
|
+
enclosure_media_group_url = try_xpaths(media_group,
|
998
|
+
"media:player/@url", :select_result_value => true)
|
999
|
+
if enclosure.player.nil? && !enclosure_media_group_url.blank?
|
989
1000
|
enclosure.player = EnclosurePlayer.new(
|
990
|
-
FeedTools.unescape_entities(
|
991
|
-
FeedTools.unescape_entities(
|
992
|
-
|
1001
|
+
FeedTools.unescape_entities(enclosure_media_group_url),
|
1002
|
+
FeedTools.unescape_entities(
|
1003
|
+
try_xpaths(media_group, ["media:player/@height"],
|
1004
|
+
:select_result_value => true)),
|
1005
|
+
FeedTools.unescape_entities(
|
1006
|
+
try_xpaths(media_group, ["media:player/@width"],
|
1007
|
+
:select_result_value => true))
|
993
1008
|
)
|
994
|
-
if enclosure.player.height == ""
|
995
|
-
enclosure.player.height = nil
|
996
|
-
end
|
997
|
-
if enclosure.player.width == ""
|
998
|
-
enclosure.player.width = nil
|
999
|
-
end
|
1000
1009
|
end
|
1001
1010
|
if enclosure.credits.nil? || enclosure.credits.size == 0
|
1002
1011
|
enclosure.credits = []
|
1003
|
-
for credit in
|
1012
|
+
for credit in try_xpaths_all(media_group, ["media:credit"])
|
1004
1013
|
enclosure.credits << EnclosureCredit.new(
|
1005
|
-
FeedTools.unescape_entities(credit.
|
1006
|
-
FeedTools.unescape_entities(
|
1014
|
+
FeedTools.unescape_entities(credit.inner_xml),
|
1015
|
+
FeedTools.unescape_entities(
|
1016
|
+
credit.attributes["role"].to_s.downcase)
|
1007
1017
|
)
|
1008
|
-
if enclosure.credits.last.role
|
1018
|
+
if enclosure.credits.last.role.blank?
|
1009
1019
|
enclosure.credits.last.role = nil
|
1010
1020
|
end
|
1011
1021
|
end
|
1012
1022
|
end
|
1013
1023
|
if enclosure.explicit?.nil?
|
1014
|
-
enclosure.explicit = (
|
1015
|
-
"media:adult/text()"
|
1024
|
+
enclosure.explicit = ((try_xpaths(media_group, [
|
1025
|
+
"media:adult/text()"
|
1026
|
+
], :select_result_value => true).downcase == "true") ?
|
1027
|
+
true : false)
|
1016
1028
|
end
|
1017
|
-
|
1018
|
-
|
1019
|
-
|
1020
|
-
|
1029
|
+
enclosure_media_group_text = try_xpaths(media_group,
|
1030
|
+
["media:text/text()"], :select_result_value => true)
|
1031
|
+
if enclosure.text.nil? && !enclosure_media_group_text.blank?
|
1032
|
+
enclosure.text = FeedTools.sanitize_html(
|
1033
|
+
FeedTools.unescape_entities(
|
1034
|
+
enclosure_media_group_text), :strip)
|
1021
1035
|
end
|
1022
1036
|
end
|
1023
1037
|
|
@@ -1033,10 +1047,14 @@ module FeedTools
|
|
1033
1047
|
end
|
1034
1048
|
|
1035
1049
|
# Add all the itunes categories
|
1036
|
-
|
1050
|
+
itunes_categories =
|
1051
|
+
try_xpaths_all(self.root_node, ["itunes:category"])
|
1052
|
+
for itunes_category in itunes_categories
|
1037
1053
|
genre = "Podcasts"
|
1038
1054
|
category = itunes_category.attributes["text"].to_s
|
1039
|
-
subcategory =
|
1055
|
+
subcategory =
|
1056
|
+
try_xpaths(itunes_category, ["itunes:category/@text"],
|
1057
|
+
:select_result_value => true)
|
1040
1058
|
category_path = genre
|
1041
1059
|
if category != ""
|
1042
1060
|
category_path << "/" + category
|
@@ -1061,7 +1079,7 @@ module FeedTools
|
|
1061
1079
|
for enclosure in @enclosures
|
1062
1080
|
# Clean up any of those attributes that incorrectly have ""
|
1063
1081
|
# or 0 as their values
|
1064
|
-
if enclosure.type
|
1082
|
+
if enclosure.type.blank?
|
1065
1083
|
enclosure.type = nil
|
1066
1084
|
end
|
1067
1085
|
if enclosure.file_size == 0
|
@@ -1082,12 +1100,13 @@ module FeedTools
|
|
1082
1100
|
if enclosure.framerate == 0
|
1083
1101
|
enclosure.framerate = nil
|
1084
1102
|
end
|
1085
|
-
if enclosure.expression
|
1103
|
+
if enclosure.expression.blank?
|
1086
1104
|
enclosure.expression = "full"
|
1087
1105
|
end
|
1088
1106
|
|
1089
|
-
# If an enclosure is missing the text field, fall back on the
|
1090
|
-
|
1107
|
+
# If an enclosure is missing the text field, fall back on the
|
1108
|
+
# itunes:summary field
|
1109
|
+
if enclosure.text.blank?
|
1091
1110
|
enclosure.text = self.itunes_summary
|
1092
1111
|
end
|
1093
1112
|
|
@@ -1128,8 +1147,8 @@ module FeedTools
|
|
1128
1147
|
end
|
1129
1148
|
end
|
1130
1149
|
|
1131
|
-
# If we have a single enclosure, it's safe to inherit the
|
1132
|
-
# if it's missing.
|
1150
|
+
# If we have a single enclosure, it's safe to inherit the
|
1151
|
+
# itunes:duration field if it's missing.
|
1133
1152
|
if @enclosures.size == 1
|
1134
1153
|
if @enclosures.first.duration.nil? || @enclosures.first.duration == 0
|
1135
1154
|
@enclosures.first.duration = self.itunes_duration
|
@@ -1147,48 +1166,29 @@ module FeedTools
|
|
1147
1166
|
def author
|
1148
1167
|
if @author.nil?
|
1149
1168
|
@author = FeedTools::Feed::Author.new
|
1150
|
-
|
1151
|
-
|
1152
|
-
|
1153
|
-
|
1154
|
-
|
1155
|
-
|
1156
|
-
|
1157
|
-
|
1158
|
-
|
1159
|
-
|
1160
|
-
if author_node.nil?
|
1161
|
-
author_node = XPath.first(root_node, "author")
|
1162
|
-
end
|
1163
|
-
if author_node.nil?
|
1164
|
-
author_node = XPath.first(root_node, "managingEditor")
|
1165
|
-
end
|
1166
|
-
if author_node.nil?
|
1167
|
-
author_node = XPath.first(root_node, "dc:author",
|
1168
|
-
FEED_TOOLS_NAMESPACES)
|
1169
|
-
end
|
1170
|
-
if author_node.nil?
|
1171
|
-
author_node = XPath.first(root_node, "dc:author")
|
1172
|
-
end
|
1173
|
-
if author_node.nil?
|
1174
|
-
author_node = XPath.first(root_node, "dc:creator",
|
1175
|
-
FEED_TOOLS_NAMESPACES)
|
1176
|
-
end
|
1177
|
-
if author_node.nil?
|
1178
|
-
author_node = XPath.first(root_node, "dc:creator")
|
1179
|
-
end
|
1180
|
-
end
|
1169
|
+
author_node = try_xpaths(self.root_node, [
|
1170
|
+
"atom10:author",
|
1171
|
+
"atom03:author",
|
1172
|
+
"atom:author",
|
1173
|
+
"author",
|
1174
|
+
"managingEditor",
|
1175
|
+
"dc:author",
|
1176
|
+
"dc:creator",
|
1177
|
+
"creator"
|
1178
|
+
])
|
1181
1179
|
unless author_node.nil?
|
1182
1180
|
@author.raw = FeedTools.unescape_entities(
|
1183
|
-
XPath.first(author_node, "text()").to_s)
|
1184
|
-
@author.raw = nil if @author.raw
|
1181
|
+
XPath.first(author_node, "text()").to_s).strip
|
1182
|
+
@author.raw = nil if @author.raw.blank?
|
1185
1183
|
unless @author.raw.nil?
|
1186
1184
|
raw_scan = @author.raw.scan(
|
1187
1185
|
/(.*)\((\b[A-Z0-9._%-\+]+@[A-Z0-9._%-]+\.[A-Z]{2,4}\b)\)/i)
|
1188
1186
|
if raw_scan.nil? || raw_scan.size == 0
|
1189
1187
|
raw_scan = @author.raw.scan(
|
1190
1188
|
/(\b[A-Z0-9._%-\+]+@[A-Z0-9._%-]+\.[A-Z]{2,4}\b)\s*\((.*)\)/i)
|
1191
|
-
|
1189
|
+
unless raw_scan.size == 0
|
1190
|
+
author_raw_pair = raw_scan.first.reverse
|
1191
|
+
end
|
1192
1192
|
else
|
1193
1193
|
author_raw_pair = raw_scan.first
|
1194
1194
|
end
|
@@ -1205,42 +1205,44 @@ module FeedTools
|
|
1205
1205
|
else
|
1206
1206
|
unless @author.raw.include?("@")
|
1207
1207
|
# We can be reasonably sure we are looking at something
|
1208
|
-
# that the creator didn't intend to contain an email address
|
1209
|
-
# it got through the preceeding regexes and it doesn't
|
1208
|
+
# that the creator didn't intend to contain an email address
|
1209
|
+
# if it got through the preceeding regexes and it doesn't
|
1210
1210
|
# contain the tell-tale '@' symbol.
|
1211
1211
|
@author.name = @author.raw
|
1212
1212
|
end
|
1213
1213
|
end
|
1214
1214
|
end
|
1215
|
-
|
1216
|
-
if @author.name == ""
|
1215
|
+
if @author.name.blank?
|
1217
1216
|
@author.name = FeedTools.unescape_entities(
|
1218
|
-
|
1217
|
+
try_xpaths(author_node, [
|
1218
|
+
"name/text()",
|
1219
|
+
"@name"
|
1220
|
+
], :select_result_value => true)
|
1221
|
+
)
|
1219
1222
|
end
|
1220
|
-
if @author.
|
1221
|
-
@author.name = FeedTools.unescape_entities(
|
1222
|
-
XPath.first(author_node, "@name").to_s)
|
1223
|
-
end
|
1224
|
-
if @author.email == ""
|
1225
|
-
@author.email = FeedTools.unescape_entities(
|
1226
|
-
XPath.first(author_node, "email/text()").to_s)
|
1227
|
-
end
|
1228
|
-
if @author.email == ""
|
1223
|
+
if @author.email.blank?
|
1229
1224
|
@author.email = FeedTools.unescape_entities(
|
1230
|
-
|
1231
|
-
|
1232
|
-
|
1233
|
-
|
1234
|
-
|
1225
|
+
try_xpaths(author_node, [
|
1226
|
+
"email/text()",
|
1227
|
+
"@email"
|
1228
|
+
], :select_result_value => true)
|
1229
|
+
)
|
1235
1230
|
end
|
1236
|
-
if @author.url
|
1231
|
+
if @author.url.blank?
|
1237
1232
|
@author.url = FeedTools.unescape_entities(
|
1238
|
-
|
1239
|
-
|
1240
|
-
|
1241
|
-
|
1242
|
-
|
1243
|
-
|
1233
|
+
try_xpaths(author_node, [
|
1234
|
+
"url/text()",
|
1235
|
+
"uri/text()",
|
1236
|
+
"@url",
|
1237
|
+
"@uri",
|
1238
|
+
"@href"
|
1239
|
+
], :select_result_value => true)
|
1240
|
+
)
|
1241
|
+
end
|
1242
|
+
@author.name = nil if @author.name.blank?
|
1243
|
+
@author.raw = nil if @author.raw.blank?
|
1244
|
+
@author.email = nil if @author.email.blank?
|
1245
|
+
@author.url = nil if @author.url.blank?
|
1244
1246
|
end
|
1245
1247
|
# Fallback on the itunes module if we didn't find an author name
|
1246
1248
|
begin
|
@@ -1276,12 +1278,11 @@ module FeedTools
|
|
1276
1278
|
|
1277
1279
|
# Set the author name
|
1278
1280
|
@publisher.raw = FeedTools.unescape_entities(
|
1279
|
-
|
1280
|
-
|
1281
|
-
|
1282
|
-
|
1283
|
-
|
1284
|
-
unless @publisher.raw == ""
|
1281
|
+
try_xpaths(self.root_node, [
|
1282
|
+
"dc:publisher/text()",
|
1283
|
+
"webMaster/text()"
|
1284
|
+
], :select_result_value => true))
|
1285
|
+
unless @publisher.raw.blank?
|
1285
1286
|
raw_scan = @publisher.raw.scan(
|
1286
1287
|
/(.*)\((\b[A-Z0-9._%-\+]+@[A-Z0-9._%-]+\.[A-Z]{2,4}\b)\)/i)
|
1287
1288
|
if raw_scan.nil? || raw_scan.size == 0
|
@@ -1314,10 +1315,10 @@ module FeedTools
|
|
1314
1315
|
end
|
1315
1316
|
end
|
1316
1317
|
|
1317
|
-
@publisher.name = nil if @publisher.name
|
1318
|
-
@publisher.raw = nil if @publisher.raw
|
1319
|
-
@publisher.email = nil if @publisher.email
|
1320
|
-
@publisher.url = nil if @publisher.url
|
1318
|
+
@publisher.name = nil if @publisher.name.blank?
|
1319
|
+
@publisher.raw = nil if @publisher.raw.blank?
|
1320
|
+
@publisher.email = nil if @publisher.email.blank?
|
1321
|
+
@publisher.url = nil if @publisher.url.blank?
|
1321
1322
|
end
|
1322
1323
|
return @publisher
|
1323
1324
|
end
|
@@ -1345,10 +1346,10 @@ module FeedTools
|
|
1345
1346
|
# elements. They're actually amazingly common. People don't read specs.
|
1346
1347
|
def itunes_author
|
1347
1348
|
if @itunes_author.nil?
|
1348
|
-
@itunes_author = FeedTools.unescape_entities(
|
1349
|
-
|
1350
|
-
|
1351
|
-
@itunes_author =
|
1349
|
+
@itunes_author = FeedTools.unescape_entities(
|
1350
|
+
try_xpaths(self.root_node,
|
1351
|
+
["itunes:author/text()"], :select_result_value => true))
|
1352
|
+
@itunes_author = feed.itunes_author if @itunes_author.blank?
|
1352
1353
|
end
|
1353
1354
|
return @itunes_author
|
1354
1355
|
end
|
@@ -1361,14 +1362,15 @@ module FeedTools
|
|
1361
1362
|
# Returns the number of seconds that the associated media runs for
|
1362
1363
|
def itunes_duration
|
1363
1364
|
if @itunes_duration.nil?
|
1364
|
-
raw_duration = FeedTools.unescape_entities(
|
1365
|
-
|
1366
|
-
|
1365
|
+
raw_duration = FeedTools.unescape_entities(
|
1366
|
+
try_xpaths(self.root_node,
|
1367
|
+
["itunes:duration/text()"], :select_result_value => true))
|
1368
|
+
if !raw_duration.blank?
|
1367
1369
|
hms = raw_duration.split(":").map { |x| x.to_i }
|
1368
1370
|
if hms.size == 3
|
1369
|
-
@itunes_duration = hms[0].
|
1371
|
+
@itunes_duration = hms[0].hours + hms[1].minutes + hms[2]
|
1370
1372
|
elsif hms.size == 2
|
1371
|
-
@itunes_duration = hms[0].
|
1373
|
+
@itunes_duration = hms[0].minutes + hms[1]
|
1372
1374
|
elsif hms.size == 1
|
1373
1375
|
@itunes_duration = hms[0]
|
1374
1376
|
end
|
@@ -1388,39 +1390,51 @@ module FeedTools
|
|
1388
1390
|
options.keys)
|
1389
1391
|
options = { :estimate_timestamp => true }.merge(options)
|
1390
1392
|
if @time.nil?
|
1391
|
-
|
1392
|
-
|
1393
|
-
|
1394
|
-
|
1395
|
-
|
1396
|
-
|
1397
|
-
|
1398
|
-
|
1399
|
-
|
1400
|
-
|
1401
|
-
|
1402
|
-
|
1403
|
-
|
1404
|
-
|
1405
|
-
|
1393
|
+
time_string = try_xpaths(self.root_node, [
|
1394
|
+
"atom10:updated/text()",
|
1395
|
+
"atom03:updated/text()",
|
1396
|
+
"atom:updated/text()",
|
1397
|
+
"updated/text()",
|
1398
|
+
"atom10:modified/text()",
|
1399
|
+
"atom03:modified/text()",
|
1400
|
+
"atom:modified/text()",
|
1401
|
+
"modified/text()",
|
1402
|
+
"time/text()",
|
1403
|
+
"atom10:issued/text()",
|
1404
|
+
"atom03:issued/text()",
|
1405
|
+
"atom:issued/text()",
|
1406
|
+
"issued/text()",
|
1407
|
+
"atom10:published/text()",
|
1408
|
+
"atom03:published/text()",
|
1409
|
+
"atom:published/text()",
|
1410
|
+
"published/text()",
|
1411
|
+
"pubDate/text()",
|
1412
|
+
"dc:date/text()",
|
1413
|
+
"date/text()"
|
1414
|
+
], :select_result_value => true)
|
1406
1415
|
begin
|
1407
|
-
|
1408
|
-
if time_string != ""
|
1416
|
+
if !time_string.blank?
|
1409
1417
|
@time = Time.parse(time_string).gmtime
|
1418
|
+
elsif FeedTools.configurations[:timestamp_estimation_enabled] &&
|
1419
|
+
!self.title.nil? &&
|
1420
|
+
(Time.parse(self.title) - Time.now).abs > 100
|
1421
|
+
@time = Time.parse(self.title).gmtime
|
1410
1422
|
end
|
1411
1423
|
rescue
|
1412
1424
|
end
|
1413
|
-
if
|
1414
|
-
if
|
1415
|
-
|
1416
|
-
|
1425
|
+
if FeedTools.configurations[:timestamp_estimation_enabled]
|
1426
|
+
if options[:estimate_timestamp]
|
1427
|
+
if @time.nil?
|
1428
|
+
begin
|
1429
|
+
@time = succ_time
|
1430
|
+
if @time.nil?
|
1431
|
+
@time = prev_time
|
1432
|
+
end
|
1433
|
+
rescue
|
1434
|
+
end
|
1417
1435
|
if @time.nil?
|
1418
|
-
@time =
|
1436
|
+
@time = Time.now.gmtime
|
1419
1437
|
end
|
1420
|
-
rescue
|
1421
|
-
end
|
1422
|
-
if @time.nil?
|
1423
|
-
@time = Time.now.gmtime
|
1424
1438
|
end
|
1425
1439
|
end
|
1426
1440
|
end
|
@@ -1440,10 +1454,10 @@ module FeedTools
|
|
1440
1454
|
if parent_feed.nil?
|
1441
1455
|
return nil
|
1442
1456
|
end
|
1443
|
-
if parent_feed.instance_variable_get("@
|
1457
|
+
if parent_feed.instance_variable_get("@entries").nil?
|
1444
1458
|
parent_feed.items
|
1445
1459
|
end
|
1446
|
-
unsorted_items = parent_feed.instance_variable_get("@
|
1460
|
+
unsorted_items = parent_feed.instance_variable_get("@entries")
|
1447
1461
|
item_index = unsorted_items.index(self)
|
1448
1462
|
if item_index.nil?
|
1449
1463
|
return nil
|
@@ -1457,7 +1471,7 @@ module FeedTools
|
|
1457
1471
|
return nil
|
1458
1472
|
end
|
1459
1473
|
end
|
1460
|
-
|
1474
|
+
private :succ_time
|
1461
1475
|
|
1462
1476
|
# Returns 1 second before the succeeding item's time.
|
1463
1477
|
def prev_time #:nodoc:
|
@@ -1466,10 +1480,10 @@ module FeedTools
|
|
1466
1480
|
if parent_feed.nil?
|
1467
1481
|
return nil
|
1468
1482
|
end
|
1469
|
-
if parent_feed.instance_variable_get("@
|
1483
|
+
if parent_feed.instance_variable_get("@entries").nil?
|
1470
1484
|
parent_feed.items
|
1471
1485
|
end
|
1472
|
-
unsorted_items = parent_feed.instance_variable_get("@
|
1486
|
+
unsorted_items = parent_feed.instance_variable_get("@entries")
|
1473
1487
|
item_index = unsorted_items.index(self)
|
1474
1488
|
if item_index.nil?
|
1475
1489
|
return nil
|
@@ -1483,18 +1497,22 @@ module FeedTools
|
|
1483
1497
|
return nil
|
1484
1498
|
end
|
1485
1499
|
end
|
1486
|
-
|
1487
|
-
|
1500
|
+
private :prev_time
|
1501
|
+
|
1488
1502
|
# Returns the feed item updated time
|
1489
1503
|
def updated
|
1490
1504
|
if @updated.nil?
|
1491
|
-
|
1492
|
-
|
1493
|
-
|
1494
|
-
|
1495
|
-
|
1496
|
-
|
1497
|
-
|
1505
|
+
updated_string = try_xpaths(self.root_node, [
|
1506
|
+
"atom10:updated/text()",
|
1507
|
+
"atom03:updated/text()",
|
1508
|
+
"atom:updated/text()",
|
1509
|
+
"updated/text()",
|
1510
|
+
"atom10:modified/text()",
|
1511
|
+
"atom03:modified/text()",
|
1512
|
+
"atom:modified/text()",
|
1513
|
+
"modified/text()"
|
1514
|
+
], :select_result_value => true)
|
1515
|
+
if !updated_string.blank?
|
1498
1516
|
@updated = Time.parse(updated_string).gmtime rescue nil
|
1499
1517
|
else
|
1500
1518
|
@updated = nil
|
@@ -1508,23 +1526,24 @@ module FeedTools
|
|
1508
1526
|
@updated = new_updated
|
1509
1527
|
end
|
1510
1528
|
|
1511
|
-
# Returns the feed item
|
1512
|
-
def
|
1513
|
-
if @
|
1514
|
-
|
1515
|
-
|
1516
|
-
|
1517
|
-
|
1518
|
-
|
1519
|
-
|
1520
|
-
|
1521
|
-
|
1522
|
-
|
1523
|
-
|
1524
|
-
|
1525
|
-
|
1526
|
-
|
1527
|
-
|
1529
|
+
# Returns the feed item published time
|
1530
|
+
def published
|
1531
|
+
if @published.nil?
|
1532
|
+
published_string = try_xpaths(self.root_node, [
|
1533
|
+
"atom10:issued/text()",
|
1534
|
+
"atom03:issued/text()",
|
1535
|
+
"atom:issued/text()",
|
1536
|
+
"issued/text()",
|
1537
|
+
"atom10:published/text()",
|
1538
|
+
"atom03:published/text()",
|
1539
|
+
"atom:published/text()",
|
1540
|
+
"published/text()",
|
1541
|
+
"pubDate/text()",
|
1542
|
+
"dc:date/text()",
|
1543
|
+
"date/text()"
|
1544
|
+
], :select_result_value => true)
|
1545
|
+
if !published_string.blank?
|
1546
|
+
@issued = Time.parse(published_string).gmtime rescue nil
|
1528
1547
|
else
|
1529
1548
|
@issued = nil
|
1530
1549
|
end
|
@@ -1532,17 +1551,19 @@ module FeedTools
|
|
1532
1551
|
return @issued
|
1533
1552
|
end
|
1534
1553
|
|
1535
|
-
# Sets the feed item
|
1536
|
-
def
|
1537
|
-
@
|
1554
|
+
# Sets the feed item published time
|
1555
|
+
def published=(new_published)
|
1556
|
+
@published = new_published
|
1538
1557
|
end
|
1539
1558
|
|
1540
1559
|
# Returns the url for posting comments
|
1541
1560
|
def comments
|
1542
1561
|
if @comments.nil?
|
1543
|
-
@comments =
|
1544
|
-
|
1545
|
-
|
1562
|
+
@comments = try_xpaths(self.root_node, ["comments/text()"],
|
1563
|
+
:select_result_value => true)
|
1564
|
+
if FeedTools.configurations[:url_normalization_enabled]
|
1565
|
+
@comments = FeedTools.normalize_url(@comments)
|
1566
|
+
end
|
1546
1567
|
end
|
1547
1568
|
return @comments
|
1548
1569
|
end
|
@@ -1556,10 +1577,10 @@ module FeedTools
|
|
1556
1577
|
def source
|
1557
1578
|
if @source.nil?
|
1558
1579
|
@source = FeedTools::Feed::Link.new
|
1559
|
-
@source.url =
|
1560
|
-
|
1561
|
-
@source.value =
|
1562
|
-
|
1580
|
+
@source.url = try_xpaths(self.root_node, ["source/@url"],
|
1581
|
+
:select_result_value => true)
|
1582
|
+
@source.value = try_xpaths(self.root_node, ["source/text()"],
|
1583
|
+
:select_result_value => true)
|
1563
1584
|
end
|
1564
1585
|
return @source
|
1565
1586
|
end
|
@@ -1575,26 +1596,26 @@ module FeedTools
|
|
1575
1596
|
end
|
1576
1597
|
if @tags.nil? || @tags.size == 0
|
1577
1598
|
@tags = []
|
1578
|
-
tag_list =
|
1579
|
-
|
1580
|
-
|
1581
|
-
|
1582
|
-
end
|
1583
|
-
if tag_list != nil && tag_list.size > 1
|
1599
|
+
tag_list = try_xpaths_all(self.root_node,
|
1600
|
+
["dc:subject/rdf:Bag/rdf:li/text()"],
|
1601
|
+
:select_result_value => true)
|
1602
|
+
if tag_list != nil && tag_list.size > 0
|
1584
1603
|
for tag in tag_list
|
1585
|
-
@tags << tag.
|
1604
|
+
@tags << tag.downcase.strip
|
1586
1605
|
end
|
1587
1606
|
end
|
1588
1607
|
end
|
1589
1608
|
if @tags.nil? || @tags.size == 0
|
1590
1609
|
# messy effort to find ourselves some tags, mainly for del.icio.us
|
1591
1610
|
@tags = []
|
1592
|
-
rdf_bag =
|
1611
|
+
rdf_bag = try_xpaths_all(self.root_node,
|
1612
|
+
["taxo:topics/rdf:Bag/rdf:li"])
|
1593
1613
|
if rdf_bag != nil && rdf_bag.size > 0
|
1594
1614
|
for tag_node in rdf_bag
|
1595
1615
|
begin
|
1596
|
-
tag_url =
|
1597
|
-
|
1616
|
+
tag_url = try_xpaths(tag_node, ["@resource"],
|
1617
|
+
:select_result_value => true)
|
1618
|
+
tag_match = tag_url.scan(/\/(tag|tags)\/(\w+)$/)
|
1598
1619
|
if tag_match.size > 0
|
1599
1620
|
@tags << tag_match.first.last.downcase.strip
|
1600
1621
|
end
|
@@ -1605,21 +1626,36 @@ module FeedTools
|
|
1605
1626
|
end
|
1606
1627
|
if @tags.nil? || @tags.size == 0
|
1607
1628
|
@tags = []
|
1608
|
-
tag_list =
|
1629
|
+
tag_list = try_xpaths_all(self.root_node, ["category/text()"],
|
1630
|
+
:select_result_value => true)
|
1609
1631
|
for tag in tag_list
|
1610
1632
|
@tags << tag.to_s.downcase.strip
|
1611
1633
|
end
|
1612
1634
|
end
|
1613
1635
|
if @tags.nil? || @tags.size == 0
|
1614
1636
|
@tags = []
|
1615
|
-
tag_list =
|
1637
|
+
tag_list = try_xpaths_all(self.root_node, ["dc:subject/text()"],
|
1638
|
+
:select_result_value => true)
|
1616
1639
|
for tag in tag_list
|
1617
1640
|
@tags << tag.to_s.downcase.strip
|
1618
1641
|
end
|
1619
1642
|
end
|
1620
|
-
if @tags.
|
1643
|
+
if @tags.blank?
|
1621
1644
|
begin
|
1622
|
-
|
1645
|
+
itunes_keywords_string = try_xpaths(self.root_node, [
|
1646
|
+
"itunes:keywords/text()"
|
1647
|
+
], :select_result_value => true)
|
1648
|
+
unless itunes_keywords_string.blank?
|
1649
|
+
@tags = itunes_keywords_string.downcase.split(",")
|
1650
|
+
if @tags.size == 1
|
1651
|
+
@tags = itunes_keywords_string.downcase.split(" ")
|
1652
|
+
@tags = @tags.map { |tag| tag.chomp(",") }
|
1653
|
+
end
|
1654
|
+
if @tags.size == 1
|
1655
|
+
@tags = itunes_keywords_string.downcase.split(",")
|
1656
|
+
end
|
1657
|
+
@tags = @tags.map { |tag| tag.strip }
|
1658
|
+
end
|
1623
1659
|
rescue
|
1624
1660
|
@tags = []
|
1625
1661
|
end
|
@@ -1642,12 +1678,11 @@ module FeedTools
|
|
1642
1678
|
# isn't explicitly marked as explicit.
|
1643
1679
|
def explicit?
|
1644
1680
|
if @explicit.nil?
|
1645
|
-
|
1646
|
-
|
1647
|
-
|
1648
|
-
|
1649
|
-
|
1650
|
-
"itunes:explicit/text()").to_s.downcase == "true" ||
|
1681
|
+
explicit_string = try_xpaths(self.root_node, [
|
1682
|
+
"media:adult/text()",
|
1683
|
+
"itunes:explicit/text()"
|
1684
|
+
], :select_result_value => true)
|
1685
|
+
if explicit_string == "true" || explicit_string == "yes" ||
|
1651
1686
|
feed.explicit?
|
1652
1687
|
@explicit = true
|
1653
1688
|
else
|
@@ -1669,8 +1704,9 @@ module FeedTools
|
|
1669
1704
|
end
|
1670
1705
|
|
1671
1706
|
# Generates xml based on the content of the feed item
|
1672
|
-
def build_xml(feed_type=(self.feed.feed_type or "
|
1673
|
-
xml_builder=Builder::XmlMarkup.new(
|
1707
|
+
def build_xml(feed_type=(self.feed.feed_type or "atom"), version=nil,
|
1708
|
+
xml_builder=Builder::XmlMarkup.new(
|
1709
|
+
:indent => 2, :escape_attrs => false))
|
1674
1710
|
if feed_type == "rss" && (version == nil || version == 0.0)
|
1675
1711
|
version = 1.0
|
1676
1712
|
elsif feed_type == "atom" && (version == nil || version == 0.0)
|
@@ -1681,7 +1717,8 @@ module FeedTools
|
|
1681
1717
|
if link.nil?
|
1682
1718
|
raise "Cannot generate an rdf-based feed item with a nil link field."
|
1683
1719
|
end
|
1684
|
-
return xml_builder.item("rdf:about" =>
|
1720
|
+
return xml_builder.item("rdf:about" =>
|
1721
|
+
FeedTools.escape_entities(link)) do
|
1685
1722
|
unless title.nil? || title == ""
|
1686
1723
|
xml_builder.title(title)
|
1687
1724
|
else
|
@@ -1708,24 +1745,37 @@ module FeedTools
|
|
1708
1745
|
end
|
1709
1746
|
end
|
1710
1747
|
end
|
1711
|
-
|
1748
|
+
if self.feed.podcast?
|
1749
|
+
xml_builder.tag!("itunes:keywords", tags.join(", "))
|
1750
|
+
end
|
1712
1751
|
end
|
1713
1752
|
build_xml_hook(feed_type, version, xml_builder)
|
1714
1753
|
end
|
1715
1754
|
elsif feed_type == "rss"
|
1716
1755
|
# normal rss format
|
1717
1756
|
return xml_builder.item do
|
1718
|
-
unless title.
|
1719
|
-
xml_builder.title(title)
|
1757
|
+
unless self.title.blank?
|
1758
|
+
xml_builder.title(self.title)
|
1720
1759
|
end
|
1721
|
-
unless link.
|
1760
|
+
unless self.link.blank?
|
1722
1761
|
xml_builder.link(link)
|
1723
1762
|
end
|
1724
|
-
unless description.
|
1725
|
-
xml_builder.description(description)
|
1763
|
+
unless self.description.blank?
|
1764
|
+
xml_builder.description(self.description)
|
1726
1765
|
end
|
1727
|
-
unless time.nil?
|
1728
|
-
xml_builder.pubDate(time.rfc822)
|
1766
|
+
unless self.time.nil?
|
1767
|
+
xml_builder.pubDate(self.time.rfc822)
|
1768
|
+
end
|
1769
|
+
unless self.guid.blank?
|
1770
|
+
if FeedTools.is_uri?(self.guid)
|
1771
|
+
xml_builder.guid(self.guid, "isPermaLink" => "true")
|
1772
|
+
else
|
1773
|
+
xml_builder.guid(self.guid, "isPermaLink" => "false")
|
1774
|
+
end
|
1775
|
+
else
|
1776
|
+
unless self.link.blank?
|
1777
|
+
xml_builder.guid(self.link, "isPermaLink" => "true")
|
1778
|
+
end
|
1729
1779
|
end
|
1730
1780
|
unless tags.nil? || tags.size == 0
|
1731
1781
|
xml_builder.tag!("taxo:topics") do
|
@@ -1735,53 +1785,51 @@ module FeedTools
|
|
1735
1785
|
end
|
1736
1786
|
end
|
1737
1787
|
end
|
1738
|
-
|
1739
|
-
|
1740
|
-
build_xml_hook(feed_type, version, xml_builder)
|
1741
|
-
end
|
1742
|
-
elsif feed_type == "atom" && version == 0.3
|
1743
|
-
# normal atom format
|
1744
|
-
return xml_builder.entry("xmlns" =>
|
1745
|
-
FEED_TOOLS_NAMESPACES['atom03']) do
|
1746
|
-
unless title.nil? || title == ""
|
1747
|
-
xml_builder.title(title,
|
1748
|
-
"mode" => "escaped",
|
1749
|
-
"type" => "text/html")
|
1750
|
-
end
|
1751
|
-
xml_builder.author do
|
1752
|
-
unless self.author.nil? || self.author.name.nil?
|
1753
|
-
xml_builder.name(self.author.name)
|
1754
|
-
else
|
1755
|
-
xml_builder.name("n/a")
|
1756
|
-
end
|
1757
|
-
unless self.author.nil? || self.author.email.nil?
|
1758
|
-
xml_builder.email(self.author.email)
|
1759
|
-
end
|
1760
|
-
unless self.author.nil? || self.author.url.nil?
|
1761
|
-
xml_builder.url(self.author.url)
|
1788
|
+
if self.feed.podcast?
|
1789
|
+
xml_builder.tag!("itunes:keywords", tags.join(", "))
|
1762
1790
|
end
|
1763
1791
|
end
|
1764
|
-
unless
|
1765
|
-
|
1766
|
-
|
1767
|
-
|
1768
|
-
|
1769
|
-
|
1770
|
-
|
1771
|
-
|
1772
|
-
|
1773
|
-
|
1774
|
-
|
1775
|
-
|
1776
|
-
|
1777
|
-
|
1778
|
-
|
1779
|
-
|
1780
|
-
|
1792
|
+
unless self.enclosures.blank? || self.enclosures.size == 0
|
1793
|
+
for enclosure in self.enclosures
|
1794
|
+
attribute_hash = {}
|
1795
|
+
next if enclosure.url.blank?
|
1796
|
+
begin
|
1797
|
+
if enclosure.file_size.blank? || enclosure.file_size.to_i == 0
|
1798
|
+
# We can't use this enclosure because it's missing the
|
1799
|
+
# required file size. Check alternate versions for
|
1800
|
+
# file_size.
|
1801
|
+
if !enclosure.versions.blank? && enclosure.versions.size > 0
|
1802
|
+
for alternate in enclosure.versions
|
1803
|
+
if alternate.file_size != nil &&
|
1804
|
+
alternate.file_size.to_i > 0
|
1805
|
+
enclosure = alternate
|
1806
|
+
break
|
1807
|
+
end
|
1808
|
+
end
|
1809
|
+
end
|
1810
|
+
end
|
1811
|
+
rescue
|
1812
|
+
end
|
1813
|
+
attribute_hash["url"] = FeedTools.normalize_url(enclosure.url)
|
1814
|
+
if enclosure.type != nil
|
1815
|
+
attribute_hash["type"] = enclosure.type
|
1816
|
+
end
|
1817
|
+
if enclosure.file_size != nil && enclosure.file_size.to_i > 0
|
1818
|
+
attribute_hash["length"] = enclosure.file_size.to_s
|
1819
|
+
else
|
1820
|
+
# We couldn't find an alternate and the problem is still
|
1821
|
+
# there. Give up and go on.
|
1822
|
+
xml_builder.comment!(
|
1823
|
+
"*** Enclosure failed to include file size. Ignoring. ***")
|
1824
|
+
next
|
1825
|
+
end
|
1826
|
+
xml_builder.enclosure(attribute_hash)
|
1781
1827
|
end
|
1782
1828
|
end
|
1783
1829
|
build_xml_hook(feed_type, version, xml_builder)
|
1784
1830
|
end
|
1831
|
+
elsif feed_type == "atom" && version == 0.3
|
1832
|
+
raise "Atom 0.3 is obsolete."
|
1785
1833
|
elsif feed_type == "atom" && version == 1.0
|
1786
1834
|
# normal atom format
|
1787
1835
|
return xml_builder.entry("xmlns" =>
|
@@ -1800,20 +1848,20 @@ module FeedTools
|
|
1800
1848
|
xml_builder.email(self.author.email)
|
1801
1849
|
end
|
1802
1850
|
unless self.author.nil? || self.author.url.nil?
|
1803
|
-
xml_builder.
|
1851
|
+
xml_builder.uri(self.author.url)
|
1804
1852
|
end
|
1805
1853
|
end
|
1806
1854
|
unless link.nil? || link == ""
|
1807
|
-
xml_builder.link("href" => link,
|
1855
|
+
xml_builder.link("href" => FeedTools.escape_entities(self.link),
|
1808
1856
|
"rel" => "alternate",
|
1809
1857
|
"type" => "text/html",
|
1810
|
-
"title" => title)
|
1858
|
+
"title" => FeedTools.escape_entities(title))
|
1811
1859
|
end
|
1812
|
-
|
1860
|
+
if !description.blank?
|
1813
1861
|
xml_builder.content(description,
|
1814
1862
|
"type" => "html")
|
1815
|
-
|
1816
|
-
xml_builder.content(FeedTools.no_content_string,
|
1863
|
+
elsif !FeedTools.configurations[:no_content_string].blank?
|
1864
|
+
xml_builder.content(FeedTools.configurations[:no_content_string],
|
1817
1865
|
"type" => "html")
|
1818
1866
|
end
|
1819
1867
|
if self.updated != nil
|
@@ -1851,25 +1899,36 @@ module FeedTools
|
|
1851
1899
|
xml_builder.category("term" => tag)
|
1852
1900
|
end
|
1853
1901
|
end
|
1902
|
+
unless self.enclosures.blank? || self.enclosures.size == 0
|
1903
|
+
for enclosure in self.enclosures
|
1904
|
+
attribute_hash = {}
|
1905
|
+
next if enclosure.url.blank?
|
1906
|
+
attribute_hash["rel"] = "enclosure"
|
1907
|
+
attribute_hash["href"] = FeedTools.normalize_url(enclosure.url)
|
1908
|
+
if enclosure.type != nil
|
1909
|
+
attribute_hash["type"] = enclosure.type
|
1910
|
+
end
|
1911
|
+
if enclosure.file_size != nil && enclosure.file_size.to_i > 0
|
1912
|
+
attribute_hash["length"] = enclosure.file_size.to_s
|
1913
|
+
end
|
1914
|
+
xml_builder.link(attribute_hash)
|
1915
|
+
end
|
1916
|
+
end
|
1854
1917
|
build_xml_hook(feed_type, version, xml_builder)
|
1855
1918
|
end
|
1919
|
+
else
|
1920
|
+
raise "Unsupported feed format/version."
|
1856
1921
|
end
|
1857
1922
|
end
|
1858
1923
|
|
1859
|
-
alias_method :
|
1860
|
-
alias_method :
|
1861
|
-
alias_method :
|
1862
|
-
alias_method :
|
1863
|
-
alias_method :
|
1864
|
-
alias_method :
|
1865
|
-
alias_method :abstract, :description
|
1866
|
-
alias_method :abstract=, :description=
|
1867
|
-
alias_method :content, :description
|
1868
|
-
alias_method :content=, :description=
|
1924
|
+
alias_method :summary, :content
|
1925
|
+
alias_method :summary=, :content=
|
1926
|
+
alias_method :abstract, :content
|
1927
|
+
alias_method :abstract=, :content=
|
1928
|
+
alias_method :description, :content
|
1929
|
+
alias_method :description=, :content=
|
1869
1930
|
alias_method :guid, :id
|
1870
1931
|
alias_method :guid=, :id=
|
1871
|
-
alias_method :published, :issued
|
1872
|
-
alias_method :published=, :issued=
|
1873
1932
|
|
1874
1933
|
# Returns a simple representation of the feed item object's state.
|
1875
1934
|
def inspect
|