Class: Meteor::Ml::Html::ParserImpl

Inherits:

Core::Kernel

• Object
• Parser
• Core::Kernel
• Meteor::Ml::Html::ParserImpl

show all

Defined in:

lib/meteor.rb,
lib/meteor.rb

Overview

Constant Summary

KAIGYO_CODE =

KAIGYO_CODE = "r?n|r" KAIGYO_CODE = "rn|n|r"

["\r\n", "\n", "\r"]

NBSP_2 =

' '

NBSP_3 =

'nbsp'

BR_1 =

"\r?\n|\r"

BR_2 =

'<br>'

META =

'META'

META_S =

'meta'

HTTP_EQUIV =

'http-equiv'

CONTENT_TYPE =

'Content-Type'

CONTENT =

'content'

OPTION =

'option'

SELECTED =

'selected'

INPUT =

'input'

CHECKED =

'checked'

RADIO =

'radio'

DISABLE_ELEMENT =

DISABLE_ELEMENT = "input|textarea|select|optgroup"

['input', 'textarea', 'select', 'optgroup']

DISABLED =

'disabled'

READONLY_TYPE =

READONLY_TYPE = "text|password"

['text', 'password']

TEXTAREA =

'textarea'

READONLY =

'readonly'

SELECT =

'select'

MULTIPLE =

'multiple'

SELECTED_M =

@@pattern_option = Regexp.new(OPTION) @@pattern_selected = Regexp.new(SELECTED) @@pattern_input = Regexp.new(INPUT) @@pattern_checked = Regexp.new(CHECKED) @@pattern_radio = Regexp.new(RADIO) @@pattern_disable_element = Regexp.new(DISABLE_ELEMENT) @@pattern_disabled = Regexp.new(DISABLED) @@pattern_readonly_type = Regexp.new(READONLY_TYPE) @@pattern_textarea = Regexp.new(TEXTAREA) @@pattern_readonly = Regexp.new(READONLY) @@pattern_select = Regexp.new(SELECT) @@pattern_multiple = Regexp.new(MULTIPLE)

'\\sselected\\s|\\sselected$|\\sSELECTED\\s|\\sSELECTED$'

SELECTED_R =

SELECTED_M = [' selected ',' selected',' SELECTED ',' SELECTED']

'selected\\s|selected$|SELECTED\\s|SELECTED$'

CHECKED_M =

'\\schecked\\s|\\schecked$|\\sCHECKED\\s|\\sCHECKED$'

CHECKED_R =

CHECKED_M = [' checked ',' checked',' CHECKED ',' CHECKED']

'checked\\s|checked$|CHECKED\\s|CHECKED$'

DISABLED_M =

'\\sdisabled\\s|\\sdisabled$|\\sDISABLED\\s|\\sDISABLED$'

DISABLED_R =

DISABLED_M = [' disabled ',' disiabled',' DISABLED ',' DISABLED']

'disabled\\s|disabled$|DISABLED\\s|DISABLED$'

READONLY_M =

'\\sreadonly\\s|\\sreadonly$|\\sREADONLY\\s|\\sREADONLY$'

READONLY_R =

READONLY_M = [' readonly ',' readonly',' READONLY ',' READONLY']

'readonly\\s|readonly$|READONLY\\s|READONLY$'

MULTIPLE_M =

'\\smultiple\\s|\\smultiple$|\\sMULTIPLE\\s|\\sMULTIPLE$'

MULTIPLE_R =

MULTIPLE_M = [' multiple ',' multiple',' MULTIPLE ',' MULTIPLE']

'multiple\\s|multiple$|MULTIPLE\\s|MULTIPLE$'

TRUE =

'true'

FALSE =

'false'

TYPE_L =

@@pattern_true = Regexp.new(TRUE) @@pattern_false = Regexp.new(FALSE)

'type'

TYPE_U =

'TYPE'

PATTERN_UNESCAPE =

'&(amp|quot|apos|gt|lt|nbsp);'

GET_ATTRS_MAP2 =

'\\s(disabled|readonly|checked|selected|multiple)'

TABLE_FOR_ESCAPE_ =

{
    '&' => '&',
    '"' => '"',
    '\'' => ''',
    '<' => '&lt;',
    '>' => '&gt;',
    ' ' => '&nbsp;',
}

TABLE_FOR_ESCAPE_CONTENT_ =

{
    '&' => '&',
    '"' => '"',
    '\'' => ''',
    '<' => '&lt;',
    '>' => '&gt;',
    ' ' => '&nbsp;',
    "\r\n" => '<br>',
    "\r" => '<br>',
    "\n" => '<br>',
}

PATTERN_ESCAPE =

"[&\"'<> ]"

PATTERN_ESCAPE_CONTENT =

"[&\"'<> \\n]"

@@match_tag =

MATCH_TAG = "br|hr|img|input|meta|base"

['br', 'hr', 'img', 'input', 'meta', 'base']

@@match_tag_2 =

@@match_tag_2 = "textarea|option|pre"

['textarea', 'option', 'pre']

@@match_tag_sng =

Array

入れ子にできない要素

['texarea', 'select', 'option', 'form', 'fieldset']

@@attr_logic =

Array

論理値で指定する属性

['disabled', 'readonly', 'checked', 'selected', 'multiple']

@@pattern_selected_m =

Regexp.new(SELECTED_M)

@@pattern_selected_r =

Regexp.new(SELECTED_R)

@@pattern_checked_m =

Regexp.new(CHECKED_M)

@@pattern_checked_r =

Regexp.new(CHECKED_R)

@@pattern_disabled_m =

Regexp.new(DISABLED_M)

@@pattern_disabled_r =

Regexp.new(DISABLED_R)

@@pattern_readonly_m =

Regexp.new(READONLY_M)

@@pattern_readonly_r =

Regexp.new(READONLY_R)

@@pattern_multiple_m =

Regexp.new(MULTIPLE_M)

@@pattern_multiple_r =

Regexp.new(MULTIPLE_R)

@@pattern_unescape =

Regexp.new(PATTERN_UNESCAPE)

@@pattern_get_attrs_map2 =

Regexp.new(GET_ATTRS_MAP2)

@@pattern_escape =

Regexp.new(PATTERN_ESCAPE)

@@pattern_escape_content =

Regexp.new(PATTERN_ESCAPE_CONTENT)

@@pattern_br_2 =

Regexp.new(BR_2)

@@pattern_and_1 =

Regexp.new(AND_1)

@@pattern_lt_1 =

Regexp.new(LT_1)

@@pattern_gt_1 =

Regexp.new(GT_1)

@@pattern_dq_1 =

Regexp.new(DOUBLE_QUATATION)

@@pattern_space_1 =

Regexp.new(SPACE)

@@pattern_br_1 =

Regexp.new(BR_1)

@@pattern_lt_2 =

Regexp.new(LT_2)

@@pattern_gt_2 =

Regexp.new(GT_2)

@@pattern_dq_2 =

Regexp.new(QO_2)

@@pattern_space_2 =

Regexp.new(NBSP_2)

@@pattern_and_2 =

Regexp.new(AND_2)

Constants inherited from Core::Kernel

AND_1, AND_2, AND_3, AP_1, AP_2, AP_3, ATTR_EQ, CLEAN_1, CLEAN_2, DOUBLE_QUATATION, EMPTY, EN_1, ERASE_ATTR_1, ESCAPE_ENTITY_REF, GET_ATTRS_MAP, GET_ATTR_1, GT_1, GT_2, GT_3, LT_1, LT_2, LT_3, MODE, MODE_AF, MODE_BF, MODE_UTF8, PATTERN_FIND_1, PATTERN_FIND_2_1, PATTERN_FIND_2_2, PATTERN_FIND_2_3, PATTERN_FIND_3, PATTERN_FIND_4, PATTERN_FIND_5, QO_2, QO_3, SEARCH_CX_1, SEARCH_CX_2, SEARCH_CX_3, SEARCH_CX_4, SEARCH_CX_5, SEARCH_CX_6, SET_ATTR_1, SET_CX_1, SET_CX_2, SET_CX_3, SET_CX_4, SET_MONO_1, SPACE, TAG_CLOSE, TAG_CLOSE3, TAG_OPEN, TAG_OPEN3, TAG_SEARCH_1_1, TAG_SEARCH_1_2, TAG_SEARCH_1_3, TAG_SEARCH_1_4, TAG_SEARCH_1_4_2, TAG_SEARCH_2_1, TAG_SEARCH_2_1_2, TAG_SEARCH_2_2, TAG_SEARCH_2_2_2, TAG_SEARCH_2_3, TAG_SEARCH_2_3_2, TAG_SEARCH_2_3_2_2, TAG_SEARCH_2_4, TAG_SEARCH_2_4_2, TAG_SEARCH_2_4_2_2, TAG_SEARCH_2_4_2_3, TAG_SEARCH_2_4_3, TAG_SEARCH_2_4_3_2, TAG_SEARCH_2_4_4, TAG_SEARCH_2_6, TAG_SEARCH_2_7, TAG_SEARCH_3_1, TAG_SEARCH_3_1_2, TAG_SEARCH_3_1_2_2, TAG_SEARCH_3_2, TAG_SEARCH_3_2_2, TAG_SEARCH_3_2_2_2, TAG_SEARCH_4_1, TAG_SEARCH_4_2, TAG_SEARCH_4_3, TAG_SEARCH_4_4, TAG_SEARCH_4_5, TAG_SEARCH_4_6, TAG_SEARCH_4_7, TAG_SEARCH_4_7_2, TAG_SEARCH_NC_1_1, TAG_SEARCH_NC_1_2, TAG_SEARCH_NC_1_3, TAG_SEARCH_NC_1_4, TAG_SEARCH_NC_1_4_2, TAG_SEARCH_NC_2_1, TAG_SEARCH_NC_2_1_2, TAG_SEARCH_NC_2_2, TAG_SEARCH_NC_2_2_2, TAG_SEARCH_NC_2_3, TAG_SEARCH_NC_2_3_2, TAG_SEARCH_NC_2_3_2_2, TAG_SEARCH_NC_2_4, TAG_SEARCH_NC_2_4_2, TAG_SEARCH_NC_2_4_2_2, TAG_SEARCH_NC_2_4_2_3, TAG_SEARCH_NC_2_4_3, TAG_SEARCH_NC_2_4_3_2, TAG_SEARCH_NC_2_4_4, TAG_SEARCH_NC_2_6, TAG_SEARCH_NC_2_7, TAG_SEARCH_NC_3_1, TAG_SEARCH_NC_3_1_2, TAG_SEARCH_NC_3_1_2_2, TAG_SEARCH_NC_3_2, TAG_SEARCH_NC_3_2_2, TAG_SEARCH_NC_3_2_2_2

Constants inherited from Parser

HTML, HTML5, XHTML, XHTML5, XML

Instance Attribute Summary

Attributes inherited from Core::Kernel

doc_type, document_hook, element_cache, element_hook

Instance Method Summary (collapse)

• - (String) content_type

  get content type (コンテントタイプを取得する).

• - (Object) escape(content)
• - (Object) escape_content(content, elm)
• - (ParserImpl) initialize(*args) constructor

  initializer (イニシャライザ).

• - (Object) parse(document)

  set document in parser (ドキュメントをパーサにセットする).

• - (Object) read(file_path, encoding)

  read file , set in parser (ファイルを読み込み、パーサにセットする).

Methods inherited from Core::Kernel

#attr, #attr_map, #character_encoding, #character_encoding=, #content, #create_element_pattern, #cxtag, #document, #document=, #element, #execute, #find, #flush, #remove_element, #root_element, #shadow

3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320

# File 'lib/meteor.rb', line 3309

def initialize(*args)
  super()
  @doc_type = Parser::HTML
  case args.length
    when ZERO
      initialize_0
    when ONE
      initialize_1(args[0])
    else
      raise ArgumentError
  end
end

3380
3381
3382

# File 'lib/meteor.rb', line 3380

def content_type
  @root.content_type
end

3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839

# File 'lib/meteor.rb', line 3815

def escape(content)
  #特殊文字の置換
  #「&」->「&」
  if content.include?(AND_1) then
    content.gsub!(@@pattern_and_1, AND_2)
  end
  #「<」->「&lt;」
  if content.include?(LT_1) then
    content.gsub!(@@pattern_lt_1, LT_2)
  end
  #「>」->「&gt;」
  if content.include?(GT_1) then
    content.gsub!(@@pattern_gt_1, GT_2)
  end
  #「"」->「&quotl」
  if content.include?(DOUBLE_QUATATION) then
    content.gsub!(@@pattern_dq_1, QO_2)
  end
  #「 」->「&nbsp;」
  if content.include?(SPACE) then
    content.gsub!(@@pattern_space_1, NBSP_2)
  end

  content
end

3822
3823
3824
3825
3826
3827
3828
3829
3830
3831

# File 'lib/meteor.rb', line 3822

def escape_content(content, elm)
  content = escape(content)

  if elm.cx || !is_match(@@match_tag_2, elm.name) then
    #「¥r?¥n」->「<br>」
    content.gsub!(@@pattern_br_1, BR_2)
  end

  content
end

3347
3348
3349
3350

# File 'lib/meteor.rb', line 3347

def parse(document)
  @root.document = document
  analyze_ml()
end

3357
3358
3359
3360

# File 'lib/meteor.rb', line 3357

def read(file_path, encoding)
  super(file_path, encoding)
  analyze_ml()
end