Class: Rpdfium::Structure::Element

Inherits:
Object
  • Object
show all
Defined in:
lib/rpdfium/structure/element.rb

Overview

Element of a tagged PDF StructTree.

An Element represents a node of the document's logical structure: Document, P (paragraph), H1..H6 (headings), Table, TR, TH, TD, Figure, Span, Lbl, LI, Caption, etc. See PDF spec §14.8 for the complete taxonomy.

Elements have no independent lifetime: they belong to the Tree that produced them. When the Tree is closed, the elements become invalid. Do not call methods on an element after tree.close.

All methods are read-only: PDFium exposes no API to modify the StructTree (it is a "read-only" structure even in its public C API).

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(tree, handle) ⇒ Element

Returns a new instance of Element.



22
23
24
25
# File 'lib/rpdfium/structure/element.rb', line 22

def initialize(tree, handle)
  @tree = tree
  @handle = handle
end

Instance Attribute Details

#handleObject (readonly)

Returns the value of attribute handle.



20
21
22
# File 'lib/rpdfium/structure/element.rb', line 20

def handle
  @handle
end

#treeObject (readonly)

Returns the value of attribute tree.



20
21
22
# File 'lib/rpdfium/structure/element.rb', line 20

def tree
  @tree
end

Instance Method Details

#actual_textObject

ActualText: override of the "logical" text for the element. Resolves ligatures (the PDF shows but actual_text says "fi"), math symbols ("∫" → "integral"), abbreviations. When present, it takes precedence over the graphical text for accessibility and search.



63
64
65
# File 'lib/rpdfium/structure/element.rb', line 63

def actual_text
  read_utf16_string(:FPDF_StructElement_GetActualText)
end

#alt_textObject

AltText: alternative text for Figure / Formula / images. PDF/UA requires every Figure to have a non-empty alt_text.



69
70
71
# File 'lib/rpdfium/structure/element.rb', line 69

def alt_text
  read_utf16_string(:FPDF_StructElement_GetAltText)
end

#attributesObject

Structural PDF attributes. Returns a Hash { name => value } with all attributes declared on this element (RowSpan, ColSpan, Scope, Headers, BBox, etc.). Values are Ruby-native: Integer, Float, String, true/false, or Array for "Headers" attributes that contain lists of IDs.



169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
# File 'lib/rpdfium/structure/element.rb', line 169

def attributes
  result = {}
  attr_count = Raw.FPDF_StructElement_GetAttributeCount(@handle)
  return result if attr_count <= 0

  (0...attr_count).each do |ai|
    attr = Raw.FPDF_StructElement_GetAttributeAtIndex(@handle, ai)
    next if attr.null?

    key_count = Raw.FPDF_StructElement_Attr_GetCount(attr)
    (0...key_count).each do |ki|
      name = read_attr_name(attr, ki)
      next if name.nil? || name.empty?

      value = read_attr_value(attr, name)
      result[name] = value unless value.nil?
    end
  end
  result
end

#childrenObject

Direct children of the element. Ordered as declared in the PDF (top-to-bottom, left-to-right for reading order).



108
109
110
111
112
113
114
115
116
# File 'lib/rpdfium/structure/element.rb', line 108

def children
  n = Raw.FPDF_StructElement_CountChildren(@handle)
  return [] if n <= 0

  (0...n).filter_map do |i|
    child_handle = Raw.FPDF_StructElement_GetChildAtIndex(@handle, i)
    child_handle.null? ? nil : Element.new(@tree, child_handle)
  end
end

#expansionObject

Expansion text for abbreviations (e.g. an element of type "Span" with content "Dr." and expansion "Doctor"). Used for text-to-speech.



75
76
77
# File 'lib/rpdfium/structure/element.rb', line 75

def expansion
  read_utf16_string(:FPDF_StructElement_GetExpansion)
end

#idObject

Unique ID of the element (if declared in the /ID dictionary of the StructTreeRoot). Enables cross-element references (e.g. the Headers attribute of a TD cell pointing to a TH by id).



49
50
51
# File 'lib/rpdfium/structure/element.rb', line 49

def id
  read_utf16_string(:FPDF_StructElement_GetID)
end

#inspectObject



201
202
203
# File 'lib/rpdfium/structure/element.rb', line 201

def inspect
  "#<Rpdfium::Structure::Element #{self}>"
end

#langObject

Language declared on the element (e.g. "it-IT", "en-US"). Inherited from the parent if not overridden. Useful for language-aware pipelines.



55
56
57
# File 'lib/rpdfium/structure/element.rb', line 55

def lang
  read_utf16_string(:FPDF_StructElement_GetLang)
end

#leavesObject

Leaves of the sub-tree (elements without children). These are the nodes that typically hold the direct MCID.



138
139
140
141
142
# File 'lib/rpdfium/structure/element.rb', line 138

def leaves
  return [self] if children.empty?

  children.flat_map(&:leaves)
end

#marked_content_idsObject

Marked Content IDs linked to this element. An element typically has 1 MCID (e.g. a <P> holds all the paragraph text inside a BDC with mcid=N) or 0 (a pure structural element: <Document>, <Table>, <TR> — their MCIDs reside in the leaf children).

To link an MCID to the page text: read the page objects and group by FPDFPageObj_GetMarkedContentID. See Element#text.



86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
# File 'lib/rpdfium/structure/element.rb', line 86

def marked_content_ids
  first = Raw.FPDF_StructElement_GetMarkedContentID(@handle)
  count = Raw.FPDF_StructElement_GetMarkedContentIdCount(@handle)
  # Cases: GetMarkedContentIdCount returns -1 when there are no direct
  # MCIDs (structural element). GetMarkedContentID returns -1 in the
  # same case.
  return [] if count <= 0 && first < 0

  # When a single MCID exists, GetMarkedContentIdCount may return
  # 0 or -1 while GetMarkedContentID provides the value. Coalesce:
  if count <= 0
    first >= 0 ? [first] : []
  else
    (0...count).filter_map do |i|
      mcid = Raw.FPDF_StructElement_GetMarkedContentIdAtIndex(@handle, i)
      mcid >= 0 ? mcid : nil
    end
  end
end

#obj_typeObject

Type of the underlying PDF object: usually "StructElem", but may be "MCR" (Marked Content Reference) or "OBJR" (Object Reference) for specialized nodes. Most users use type.



36
37
38
# File 'lib/rpdfium/structure/element.rb', line 36

def obj_type
  read_utf16_string(:FPDF_StructElement_GetObjType)
end

#parentObject

Parent. Nil for root elements (direct children of the StructTree).



119
120
121
122
123
124
# File 'lib/rpdfium/structure/element.rb', line 119

def parent
  h = Raw.FPDF_StructElement_GetParent(@handle)
  return nil if h.null?

  Element.new(@tree, h)
end

#textObject

Text of the element, reconstructed from the page via MCID. Resolution:

  1. If actual_text is present, use it (handles ligatures/abbreviations).
  2. Otherwise collect all MCIDs of the sub-tree (this element
    • recursively the children) and concatenate the text of the page objects with those MCIDs, in document order.

For pure structural elements (Table, TR) the text is the concatenation of all descendants — useful as a "summary".



152
153
154
155
156
157
158
159
160
161
162
# File 'lib/rpdfium/structure/element.rb', line 152

def text
  return actual_text if actual_text && !actual_text.empty?

  # Collect MCIDs of the entire sub-tree depth-first
  all_mcids = []
  walk { |el| all_mcids.concat(el.marked_content_ids) }
  return "" if all_mcids.empty?

  mcid_map = @tree.send(:mcid_text_map)
  all_mcids.filter_map { |id| mcid_map[id] }.join
end

#titleObject

Title attribute (rare, used in some documents to give the element a descriptive name, e.g. "Capitolo 1").



42
43
44
# File 'lib/rpdfium/structure/element.rb', line 42

def title
  read_utf16_string(:FPDF_StructElement_GetTitle)
end

#to_sObject



190
191
192
193
194
195
196
197
198
199
# File 'lib/rpdfium/structure/element.rb', line 190

def to_s
  parts = ["<#{type || obj_type || '?'}>"]
  mcids = marked_content_ids
  parts << "mcid=#{mcids.first}" if mcids.size == 1
  parts << "mcids=#{mcids.inspect}" if mcids.size > 1
  parts << "lang=#{lang.inspect}" if lang
  parts << "actual_text=#{actual_text.inspect[0, 30]}" if actual_text
  parts << "alt_text=#{alt_text.inspect[0, 30]}" if alt_text
  parts.join(" ")
end

#typeObject

Structural type of the element (e.g. "P", "H1", "Table", "TR", "TD"). Nil if PDFium cannot read it (placeholder element).



29
30
31
# File 'lib/rpdfium/structure/element.rb', line 29

def type
  read_utf16_string(:FPDF_StructElement_GetType)
end

#walk {|_self| ... } ⇒ Object

Depth-first walk of the entire sub-tree starting from this element. Visits self first, then recursively the children. Without a block returns an Enumerator.

Yields:

  • (_self)

Yield Parameters:



129
130
131
132
133
134
# File 'lib/rpdfium/structure/element.rb', line 129

def walk(&block)
  return enum_for(:walk) unless block

  yield self
  children.each { |c| c.walk(&block) }
end