sitediff 0.0.1 → 1.0.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +5 -5
- data/bin/sitediff +10 -4
- data/lib/sitediff.rb +179 -91
- data/lib/sitediff/cache.rb +106 -0
- data/lib/sitediff/cli.rb +391 -60
- data/lib/sitediff/config.rb +383 -37
- data/lib/sitediff/config/creator.rb +114 -0
- data/lib/sitediff/config/preset.rb +75 -0
- data/lib/sitediff/crawler.rb +131 -0
- data/lib/sitediff/diff.rb +57 -12
- data/lib/sitediff/exception.rb +5 -0
- data/lib/sitediff/fetch.rb +76 -0
- data/lib/sitediff/files/diff.html.erb +20 -2
- data/lib/sitediff/files/jquery.min.js +2 -0
- data/lib/sitediff/files/normalize.css +349 -0
- data/lib/sitediff/files/report.html.erb +144 -0
- data/lib/sitediff/files/sidebyside.html.erb +16 -0
- data/lib/sitediff/files/sitediff.css +236 -29
- data/lib/sitediff/files/sitediff.js +176 -0
- data/lib/sitediff/report.rb +238 -0
- data/lib/sitediff/result.rb +63 -26
- data/lib/sitediff/sanitize.rb +160 -141
- data/lib/sitediff/sanitize/dom_transform.rb +130 -0
- data/lib/sitediff/sanitize/regexp.rb +82 -0
- data/lib/sitediff/uriwrapper.rb +114 -35
- data/lib/sitediff/webserver.rb +94 -0
- data/lib/sitediff/webserver/resultserver.rb +134 -0
- metadata +103 -43
- data/lib/sitediff/files/html_report.html.erb +0 -47
- data/lib/sitediff/util/cache.rb +0 -32
- data/lib/sitediff/util/webserver.rb +0 -77
data/lib/sitediff/config.rb
CHANGED
@@ -1,13 +1,66 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require 'sitediff/config/preset'
|
4
|
+
require 'sitediff/exception'
|
5
|
+
require 'sitediff/sanitize'
|
6
|
+
require 'pathname'
|
1
7
|
require 'yaml'
|
2
8
|
|
3
9
|
class SiteDiff
|
10
|
+
# SiteDiff Configuration.
|
4
11
|
class Config
|
12
|
+
# Default config file.
|
13
|
+
DEFAULT_FILENAME = 'sitediff.yaml'
|
14
|
+
|
15
|
+
# Default paths file.
|
16
|
+
DEFAULT_PATHS_FILENAME = 'paths.txt'
|
17
|
+
|
18
|
+
# Default SiteDiff config.
|
19
|
+
DEFAULT_CONFIG = {
|
20
|
+
'settings' => {
|
21
|
+
'depth' => 3,
|
22
|
+
'interval' => 0,
|
23
|
+
'whitelist' => '',
|
24
|
+
'blacklist' => '',
|
25
|
+
'concurrency' => 3,
|
26
|
+
'preset' => nil
|
27
|
+
},
|
28
|
+
'before' => {},
|
29
|
+
'after' => {},
|
30
|
+
'paths' => []
|
31
|
+
}.freeze
|
32
|
+
|
33
|
+
# Keys allowed in config files.
|
34
|
+
# TODO: Deprecate repeated params before_url and after_url.
|
35
|
+
# TODO: Create a method self.supports
|
36
|
+
# TODO: Deprecate in favor of self.supports key, subkey, subkey...
|
37
|
+
ALLOWED_CONFIG_KEYS = Sanitizer::TOOLS.values.flatten(1) + %w[
|
38
|
+
includes
|
39
|
+
settings
|
40
|
+
before
|
41
|
+
after
|
42
|
+
before_url
|
43
|
+
after_url
|
44
|
+
ignore_whitespace
|
45
|
+
export
|
46
|
+
]
|
5
47
|
|
6
|
-
|
7
|
-
|
8
|
-
|
48
|
+
##
|
49
|
+
# Keys allowed in the "settings" key.
|
50
|
+
# TODO: Create a method self.supports
|
51
|
+
# TODO: Deprecate in favor of self.supports key, subkey, subkey...
|
52
|
+
ALLOWED_SETTINGS_KEYS = %w[
|
53
|
+
preset
|
54
|
+
depth
|
55
|
+
whitelist
|
56
|
+
blacklist
|
57
|
+
concurrency
|
58
|
+
interval
|
59
|
+
curl_opts
|
60
|
+
].freeze
|
9
61
|
|
10
|
-
class InvalidConfig <
|
62
|
+
class InvalidConfig < SiteDiffException; end
|
63
|
+
class ConfigNotFound < SiteDiffException; end
|
11
64
|
|
12
65
|
# Takes a Hash and normalizes it to the following form by merging globals
|
13
66
|
# into before and after. A normalized config Hash looks like this:
|
@@ -27,22 +80,24 @@ class SiteDiff
|
|
27
80
|
# selector: body
|
28
81
|
#
|
29
82
|
def self.normalize(conf)
|
30
|
-
tools =
|
83
|
+
tools = Sanitizer::TOOLS
|
31
84
|
|
32
|
-
#
|
85
|
+
# Merge globals
|
33
86
|
%w[before after].each do |pos|
|
34
87
|
conf[pos] ||= {}
|
35
88
|
tools[:array].each do |key|
|
36
89
|
conf[pos][key] ||= []
|
37
90
|
conf[pos][key] += conf[key] if conf[key]
|
38
91
|
end
|
39
|
-
tools[:scalar].each {|key| conf[pos][key] ||= conf[key]}
|
92
|
+
tools[:scalar].each { |key| conf[pos][key] ||= conf[key] }
|
40
93
|
conf[pos]['url'] ||= conf[pos + '_url']
|
94
|
+
conf[pos]['curl_opts'] = conf['curl_opts']
|
41
95
|
end
|
42
|
-
# normalize paths
|
43
|
-
conf['paths'] = Config::normalize_paths(conf['paths'])
|
44
96
|
|
45
|
-
|
97
|
+
# Normalize paths.
|
98
|
+
conf['paths'] = Config.normalize_paths(conf['paths'])
|
99
|
+
|
100
|
+
conf.select { |k, _v| ALLOWED_CONFIG_KEYS.include? k }
|
46
101
|
end
|
47
102
|
|
48
103
|
# Merges two normalized Hashes according to the following rules:
|
@@ -59,77 +114,369 @@ class SiteDiff
|
|
59
114
|
# (h2) before: {selector: bar, sanitization: [pattern: bar]}
|
60
115
|
# (h3) before: {selector: foo, sanitization: [pattern: foo, pattern: bar]}
|
61
116
|
def self.merge(first, second)
|
62
|
-
result = {
|
63
|
-
|
117
|
+
result = {
|
118
|
+
'before' => {},
|
119
|
+
'after' => {},
|
120
|
+
'settings' => {}
|
121
|
+
}
|
122
|
+
|
123
|
+
# Merge sanitization rules.
|
124
|
+
Sanitizer::TOOLS.values.flatten(1).each do |key|
|
125
|
+
result[key] = second[key] || first[key]
|
126
|
+
result.delete(key) unless result[key]
|
127
|
+
end
|
128
|
+
|
129
|
+
# Rule 1.
|
64
130
|
%w[before after].each do |pos|
|
131
|
+
first[pos] ||= {}
|
132
|
+
second[pos] ||= {}
|
133
|
+
|
134
|
+
# If only the second hash has the value.
|
65
135
|
unless first[pos]
|
66
136
|
result[pos] = second[pos] || {}
|
67
137
|
next
|
68
138
|
end
|
139
|
+
|
69
140
|
result[pos] = first[pos].merge!(second[pos]) do |key, a, b|
|
70
|
-
|
71
|
-
|
72
|
-
|
73
|
-
|
74
|
-
|
141
|
+
# Rule 2a.
|
142
|
+
result[pos][key] = if Sanitizer::TOOLS[:array].include? key
|
143
|
+
(a || []) + (b || [])
|
144
|
+
elsif key == 'settings'
|
145
|
+
b
|
146
|
+
else
|
147
|
+
a || b # Rule 2b.
|
148
|
+
end
|
149
|
+
end
|
150
|
+
end
|
151
|
+
|
152
|
+
# Merge settings.
|
153
|
+
result['settings'] = merge_deep(
|
154
|
+
first['settings'] || {},
|
155
|
+
second['settings'] || {}
|
156
|
+
)
|
157
|
+
|
158
|
+
result
|
159
|
+
end
|
160
|
+
|
161
|
+
##
|
162
|
+
# Merges 2 iterable objects deeply.
|
163
|
+
def self.merge_deep(first, second)
|
164
|
+
first.merge(second) do |_key, val1, val2|
|
165
|
+
if val1.is_a? Hash
|
166
|
+
self.class.merge_deep(val1, val2 || {})
|
167
|
+
elsif val1.is_a? Array
|
168
|
+
val1 + (val2 || [])
|
169
|
+
else
|
170
|
+
val2
|
75
171
|
end
|
76
172
|
end
|
173
|
+
end
|
174
|
+
|
175
|
+
##
|
176
|
+
# Gets all loaded configuration except defaults.
|
177
|
+
#
|
178
|
+
# @return [Hash]
|
179
|
+
# Config data.
|
180
|
+
def all
|
181
|
+
result = Marshal.load(Marshal.dump(@config))
|
182
|
+
self.class.remove_defaults(result)
|
183
|
+
end
|
184
|
+
|
185
|
+
##
|
186
|
+
# Removes default parameters from a config hash.
|
187
|
+
#
|
188
|
+
# I know this is weird, but it'll be fixed. The config management needs to
|
189
|
+
# be streamlined further.
|
190
|
+
def self.remove_defaults(data)
|
191
|
+
# Create a deep copy of the config data.
|
192
|
+
result = data
|
193
|
+
|
194
|
+
# Exclude default settings.
|
195
|
+
result['settings'].delete_if do |key, value|
|
196
|
+
value == DEFAULT_CONFIG['settings'][key] || !value
|
197
|
+
end
|
198
|
+
|
199
|
+
# Exclude default curl opts.
|
200
|
+
result['settings']['curl_opts'] ||= {}
|
201
|
+
result['settings']['curl_opts'].delete_if do |key, value|
|
202
|
+
value == UriWrapper::DEFAULT_CURL_OPTS[key.to_sym]
|
203
|
+
end
|
204
|
+
|
205
|
+
# Delete curl opts if empty.
|
206
|
+
unless result['settings']['curl_opts'].length.positive?
|
207
|
+
result['settings'].delete('curl_opts')
|
208
|
+
end
|
209
|
+
|
77
210
|
result
|
78
211
|
end
|
79
212
|
|
80
|
-
|
81
|
-
|
82
|
-
|
83
|
-
|
213
|
+
# Creates a SiteDiff Config object.
|
214
|
+
def initialize(file, directory)
|
215
|
+
# Fallback to default config filename, if none is specified.
|
216
|
+
file = File.join(directory, DEFAULT_FILENAME) if file.nil?
|
217
|
+
unless File.exist?(file)
|
218
|
+
path = File.expand_path(file)
|
219
|
+
raise InvalidConfig, "Missing config file #{path}."
|
84
220
|
end
|
221
|
+
@config = Config.merge(DEFAULT_CONFIG, Config.load_conf(file))
|
222
|
+
@file = file
|
223
|
+
@directory = directory
|
224
|
+
|
225
|
+
# Validate configurations.
|
226
|
+
validate
|
85
227
|
end
|
86
228
|
|
87
|
-
|
88
|
-
|
229
|
+
# Get "before" site configuration.
|
230
|
+
def before(apply_preset = false)
|
231
|
+
section :before, apply_preset
|
89
232
|
end
|
90
|
-
|
91
|
-
|
233
|
+
|
234
|
+
# Get "before" site URL.
|
235
|
+
def before_url
|
236
|
+
result = before
|
237
|
+
result['url'] if result
|
92
238
|
end
|
93
239
|
|
240
|
+
# Get "after" site configuration.
|
241
|
+
def after(apply_preset = false)
|
242
|
+
section :after, apply_preset
|
243
|
+
end
|
244
|
+
|
245
|
+
# Get "after" site URL.
|
246
|
+
def after_url
|
247
|
+
result = after
|
248
|
+
result['url'] if result
|
249
|
+
end
|
250
|
+
|
251
|
+
# Get paths.
|
94
252
|
def paths
|
95
253
|
@config['paths']
|
96
254
|
end
|
255
|
+
|
256
|
+
# Set paths.
|
97
257
|
def paths=(paths)
|
98
|
-
|
258
|
+
raise 'Paths must be an Array' unless paths.is_a? Array
|
259
|
+
|
260
|
+
@config['paths'] = Config.normalize_paths(paths)
|
261
|
+
end
|
262
|
+
|
263
|
+
# Get ignore_whitespace option
|
264
|
+
def ignore_whitespace
|
265
|
+
@config['ignore_whitespace']
|
266
|
+
end
|
267
|
+
|
268
|
+
# Set ignore_whitespace option
|
269
|
+
def ignore_whitespace=(ignore_whitespace)
|
270
|
+
@config['ignore_whitespace'] = ignore_whitespace
|
271
|
+
end
|
272
|
+
|
273
|
+
# Get export option
|
274
|
+
def export
|
275
|
+
@config['export']
|
276
|
+
end
|
277
|
+
|
278
|
+
# Set export option
|
279
|
+
def export=(export)
|
280
|
+
@config['export'] = export
|
281
|
+
end
|
282
|
+
|
283
|
+
##
|
284
|
+
# Writes an array of paths to a file.
|
285
|
+
#
|
286
|
+
# @param [Array] paths
|
287
|
+
# An array of paths.
|
288
|
+
# @param [String] file
|
289
|
+
# Optional path to a file.
|
290
|
+
def paths_file_write(paths, file = nil)
|
291
|
+
unless paths.is_a?(Array) && paths.length.positive?
|
292
|
+
raise SiteDiffException, 'Write failed. Invalid paths.'
|
293
|
+
end
|
294
|
+
|
295
|
+
file ||= File.join(@directory, DEFAULT_PATHS_FILENAME)
|
296
|
+
File.open(file, 'w+') { |f| f.puts(paths) }
|
297
|
+
end
|
298
|
+
|
299
|
+
##
|
300
|
+
# Reads a collection of paths from a file.
|
301
|
+
#
|
302
|
+
# @param [String] file
|
303
|
+
# A file containing one path per line.
|
304
|
+
#
|
305
|
+
# @return [Integer]
|
306
|
+
# Number of paths read.
|
307
|
+
def paths_file_read(file = nil)
|
308
|
+
file ||= File.join(@directory, DEFAULT_PATHS_FILENAME)
|
309
|
+
|
310
|
+
unless File.exist? file
|
311
|
+
raise Config::InvalidConfig, "File not found: #{file}"
|
312
|
+
end
|
313
|
+
|
314
|
+
self.paths = File.readlines(file)
|
315
|
+
|
316
|
+
# Return the number of paths.
|
317
|
+
paths.length
|
318
|
+
end
|
319
|
+
|
320
|
+
##
|
321
|
+
# Get roots.
|
322
|
+
#
|
323
|
+
# Example: If the config has a "before" and "after" sections, then roots
|
324
|
+
# will be ["before", "after"].
|
325
|
+
def roots
|
326
|
+
@roots = { 'after' => after_url }
|
327
|
+
@roots['before'] = before_url if before
|
328
|
+
@roots
|
329
|
+
end
|
330
|
+
|
331
|
+
##
|
332
|
+
# Gets a setting.
|
333
|
+
#
|
334
|
+
# @param [String] key
|
335
|
+
# A key.
|
336
|
+
#
|
337
|
+
# @return [*]
|
338
|
+
# A value, if exists.
|
339
|
+
def setting(key)
|
340
|
+
key = key.to_s if key.is_a?(Symbol)
|
341
|
+
return @config['settings'][key] if @config['settings'].key?(key)
|
342
|
+
end
|
343
|
+
|
344
|
+
##
|
345
|
+
# Gets all settings.
|
346
|
+
#
|
347
|
+
# TODO: Make sure the settings are not writable.
|
348
|
+
#
|
349
|
+
# @return [Hash]
|
350
|
+
# All settings.
|
351
|
+
def settings
|
352
|
+
@config['settings']
|
99
353
|
end
|
100
354
|
|
101
355
|
# Checks if the configuration is usable for diff-ing.
|
102
|
-
|
103
|
-
|
356
|
+
# TODO: Do we actually need the opts argument?
|
357
|
+
def validate(opts = {})
|
358
|
+
opts = { need_before: true }.merge(opts)
|
359
|
+
|
360
|
+
if opts[:need_before] && !before['url']
|
361
|
+
raise InvalidConfig, "Undefined 'before' base URL."
|
362
|
+
end
|
363
|
+
|
104
364
|
raise InvalidConfig, "Undefined 'after' base URL." unless after['url']
|
105
|
-
|
365
|
+
|
366
|
+
# Validate interval and concurrency.
|
367
|
+
interval = setting(:interval)
|
368
|
+
concurrency = setting(:concurrency)
|
369
|
+
if interval.to_i != 0 && concurrency != 1
|
370
|
+
raise InvalidConfig, 'Concurrency must be 1 when an interval is set.'
|
371
|
+
end
|
372
|
+
|
373
|
+
# Validate preset.
|
374
|
+
Preset.exist? setting(:preset), true if setting(:preset)
|
375
|
+
end
|
376
|
+
|
377
|
+
##
|
378
|
+
# Returns object clone with stringified keys.
|
379
|
+
# TODO: Make this method available globally, if required.
|
380
|
+
def self.stringify_keys(object)
|
381
|
+
# Do nothing if it is not an object.
|
382
|
+
return object unless object.respond_to?('each_key')
|
383
|
+
|
384
|
+
# Convert symbol indices to strings.
|
385
|
+
output = {}
|
386
|
+
object.each_key do |old_k|
|
387
|
+
new_k = old_k.is_a?(Symbol) ? old_k.to_s : old_k
|
388
|
+
output[new_k] = stringify_keys object[old_k]
|
389
|
+
end
|
390
|
+
|
391
|
+
# Return the new hash with string indices.
|
392
|
+
output
|
393
|
+
end
|
394
|
+
|
395
|
+
##
|
396
|
+
# Creates a RegExp from a string.
|
397
|
+
def self.create_regexp(string_param)
|
398
|
+
begin
|
399
|
+
@return_value = string_param == '' ? nil : Regexp.new(string_param)
|
400
|
+
rescue SiteDiffException => e
|
401
|
+
@return_value = nil
|
402
|
+
SiteDiff.log 'Invalid RegExp: ' + string_param, :error
|
403
|
+
SiteDiff.log e.message, :error
|
404
|
+
# TODO: Use SiteDiff.log type :debug
|
405
|
+
# SiteDiff.log e.backtrace, :error if options[:verbose]
|
406
|
+
end
|
407
|
+
@return_value
|
106
408
|
end
|
107
409
|
|
108
410
|
private
|
109
411
|
|
412
|
+
##
|
413
|
+
# Returns one of the "before" or "after" sections.
|
414
|
+
#
|
415
|
+
# @param [String|Symbol]
|
416
|
+
# Section name. Example: before, after.
|
417
|
+
# @param [Boolean] with_preset
|
418
|
+
# Whether to merge with preset config (if any).
|
419
|
+
#
|
420
|
+
# @return [Hash|Nil]
|
421
|
+
# Section data or Nil.
|
422
|
+
def section(name, with_preset = false)
|
423
|
+
name = name.to_s if name.is_a? Symbol
|
424
|
+
|
425
|
+
# Validate section.
|
426
|
+
unless %w[before after].include? name
|
427
|
+
raise SiteDiffException, '"name" must be one of "before" or "after".'
|
428
|
+
end
|
429
|
+
|
430
|
+
# Return nil if section is not defined.
|
431
|
+
return nil unless @config[name]
|
432
|
+
|
433
|
+
result = @config[name]
|
434
|
+
|
435
|
+
# Merge preset rules, if required.
|
436
|
+
preset = setting(:preset)
|
437
|
+
if with_preset && !preset.nil?
|
438
|
+
preset_config = Preset.read preset
|
439
|
+
|
440
|
+
# Merge plugins with array values.
|
441
|
+
# TODO: This won't be required after plugin declarations are improved.
|
442
|
+
# See https://rm.ewdev.ca/issues/18301
|
443
|
+
Sanitizer::TOOLS[:array].each do |key|
|
444
|
+
if preset_config[key]
|
445
|
+
result[key] = (result[key] || []) + preset_config[key]
|
446
|
+
end
|
447
|
+
end
|
448
|
+
end
|
449
|
+
|
450
|
+
result
|
451
|
+
end
|
452
|
+
|
110
453
|
def self.normalize_paths(paths)
|
111
454
|
paths ||= []
|
112
|
-
|
455
|
+
paths.map { |p| (p[0] == '/' ? p : "/#{p}").chomp }
|
113
456
|
end
|
114
457
|
|
115
458
|
# reads a YAML file and raises an InvalidConfig if the file is not valid.
|
116
459
|
def self.load_raw_yaml(file)
|
117
|
-
|
460
|
+
# TODO: Only show this in verbose mode.
|
461
|
+
SiteDiff.log "Reading config file: #{Pathname.new(file).expand_path}"
|
118
462
|
conf = YAML.load_file(file) || {}
|
463
|
+
|
119
464
|
unless conf.is_a? Hash
|
120
465
|
raise InvalidConfig, "Invalid configuration file: '#{file}'"
|
121
466
|
end
|
122
|
-
|
123
|
-
|
467
|
+
|
468
|
+
conf.each_key do |k, _v|
|
469
|
+
unless ALLOWED_CONFIG_KEYS.include? k
|
124
470
|
raise InvalidConfig, "Unknown configuration key (#{file}): '#{k}'"
|
125
471
|
end
|
126
472
|
end
|
473
|
+
|
127
474
|
conf
|
128
475
|
end
|
129
476
|
|
130
477
|
# loads a single YAML configuration file, merges all its 'included' files
|
131
478
|
# and returns a normalized Hash.
|
132
|
-
def self.load_conf(file, visited=[])
|
479
|
+
def self.load_conf(file, visited = [])
|
133
480
|
# don't get fooled by a/../a/ or symlinks
|
134
481
|
file = File.realpath(file)
|
135
482
|
if visited.include? file
|
@@ -141,14 +488,13 @@ class SiteDiff
|
|
141
488
|
|
142
489
|
# normalize and merge includes
|
143
490
|
includes = conf['includes'] || []
|
144
|
-
conf = Config
|
491
|
+
conf = Config.normalize(conf)
|
145
492
|
includes.each do |dep|
|
146
493
|
# include paths are relative to the including file.
|
147
494
|
dep = File.join(File.dirname(file), dep)
|
148
|
-
conf = Config
|
495
|
+
conf = Config.merge(conf, load_conf(dep, visited))
|
149
496
|
end
|
150
497
|
conf
|
151
498
|
end
|
152
|
-
|
153
499
|
end
|
154
500
|
end
|