Class: CorpPdf::ObjectResolver

Inherits:
Object
  • Object
show all
Defined in:
lib/corp_pdf/object_resolver.rb

Overview

Parses xref (tables and streams) and exposes object bodies uniformly, including objects embedded in /ObjStm. Also gives you the trailer and /Root.

Defined Under Namespace

Classes: Entry

Instance Method Summary collapse

Constructor Details

#initialize(bytes) ⇒ ObjectResolver

Returns a new instance of ObjectResolver.



11
12
13
14
15
16
 
# File 'lib/corp_pdf/object_resolver.rb', line 11

def initialize(bytes)
  @bytes = bytes
  @entries = {}
  @objstm_cache = {}
  parse_cross_reference
end

Instance Method Details

#apply_png_predictor(data, columns) ⇒ Object 



353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
 
# File 'lib/corp_pdf/object_resolver.rb', line 353

def apply_png_predictor(data, columns)
  # PNG predictor: each row starts with a filter byte, followed by 'columns' data bytes
  row_size = columns + 1  # 1 byte for predictor + columns bytes of data
  num_rows = data.bytesize / row_size
  result = []
  prev_row = [0] * columns

  num_rows.times do |i|
    row_start = i * row_size
    filter_type = data.getbyte(row_start)
    row_bytes = (1..columns).map { |j| data.getbyte(row_start + j) }

    decoded_row = case filter_type
                  when 0  # None
                    row_bytes
                  when 1  # Sub
                    out = []
                    columns.times do |j|
                      left = j.positive? ? out[j - 1] : 0
                      out << ((row_bytes[j] + left) & 0xFF)
                    end
                    out
                  when 2  # Up
                    row_bytes.map.with_index { |b, j| (b + prev_row[j]) & 0xFF }
                  when 3  # Average
                    out = []
                    columns.times do |j|
                      left = j.positive? ? out[j - 1] : 0
                      up = prev_row[j]
                      out << ((row_bytes[j] + ((left + up) / 2)) & 0xFF)
                    end
                    out
                  when 4  # Paeth
                    out = []
                    columns.times do |j|
                      left = j.positive? ? out[j - 1] : 0
                      up = prev_row[j]
                      up_left = j.positive? ? prev_row[j - 1] : 0
                      out << ((row_bytes[j] + paeth_predictor(left, up, up_left)) & 0xFF)
                    end
                    out
                  else
                    row_bytes # Unknown filter, pass through
                  end

    result.concat(decoded_row)
    prev_row = decoded_row
  end

  result.pack("C*")
end

#balanced_from(str, start_idx) ⇒ Object 



420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
 
# File 'lib/corp_pdf/object_resolver.rb', line 420

def balanced_from(str, start_idx)
  depth = 0
  j = start_idx
  while j < str.length
    if str[j, 2] == "<<"
      depth += 1
      j += 2
    elsif str[j, 2] == ">>"
      depth -= 1
      j += 2
      return j if depth.zero?
    else
      j += 1
    end
  end
  raise "unterminated dict"
end

#clear_cache ⇒ Object

Clear the object stream cache to free memory



143
144
145
 
# File 'lib/corp_pdf/object_resolver.rb', line 143

def clear_cache
  @objstm_cache.clear
end

#decode_stream_data(dict_src, stream_chunk) ⇒ Object 



329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
 
# File 'lib/corp_pdf/object_resolver.rb', line 329

def decode_stream_data(dict_src, stream_chunk)
  s_match = /\bstream\r?\n/.match(stream_chunk) or raise "stream keyword missing"
  body = stream_chunk[s_match.end(0)..]
  body = body.sub(/\bendstream\b.*/m, "")

  # Decompress if FlateDecode (handle both "/Filter /FlateDecode" and "/Filter/FlateDecode")
  data = if dict_src =~ %r{/Filter\s*/FlateDecode}
           Zlib::Inflate.inflate(body)
         else
           body
         end

  # Apply PNG predictor if present
  if dict_src =~ %r{/DecodeParms\s*<<[^>]*/Predictor\s+(\d+)}
    predictor = ::Regexp.last_match(1).to_i
    if predictor.between?(10, 15) # PNG predictors
      columns = dict_src =~ %r{/Columns\s+(\d+)} ? ::Regexp.last_match(1).to_i : 1
      data = apply_png_predictor(data, columns)
    end
  end

  data
end

#each_object ⇒ Object 



136
137
138
139
140
 
# File 'lib/corp_pdf/object_resolver.rb', line 136

def each_object
  @entries.each_key do |ref|
    yield(ref, object_body(ref))
  end
end

#extract_trailer_from_classic_xref(start) ⇒ Object 



65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
 
# File 'lib/corp_pdf/object_resolver.rb', line 65

def extract_trailer_from_classic_xref(start)
  pos = @bytes.rindex("xref", start) or return nil
  i = pos + 4

  # Skip xref entries
  loop do
    m = /\s*(\d+)\s+(\d+)/m.match(@bytes, i) or break
    first = m[1].to_i
    count = m[2].to_i
    i = m.end(0)

    count.times do |_k|
      # Skip whitespace/newlines before the 20-byte record
      i += 1 while (ch = @bytes.getbyte(i)) && [0x0A, 0x0D, 0x20].include?(ch)
      i += 20
      # consume line ending(s)
      i += 1 while (ch = @bytes.getbyte(i)) && [0x0A, 0x0D].include?(ch)
    end

    break if @bytes[i, 7] == "trailer"
  end

  tpos = @bytes.index("trailer", i)
  return nil unless tpos

  dpos = @bytes.index("<<", tpos)
  return nil unless dpos

  dend = balanced_from(@bytes, dpos)
  @bytes[dpos...dend]
end

#extract_trailer_from_xref_stream(header_ofs) ⇒ Object 



97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
 
# File 'lib/corp_pdf/object_resolver.rb', line 97

def extract_trailer_from_xref_stream(header_ofs)
  # Expect "<num> <gen> obj" at header_ofs
  m = /\A(\d+)\s+(\d+)\s+obj\b/m.match(@bytes[header_ofs, 50])
  unless m
    # Sometimes header_ofs might land on whitespace; search forward a bit
    win = @bytes[header_ofs, 256]
    m2 = /(\d+)\s+(\d+)\s+obj\b/m.match(win) or return nil
    header_ofs += m2.begin(0)
    m = m2
  end

  dpos = @bytes.index("<<", header_ofs + m[0].length) or return nil
  dend = balanced_from(@bytes, dpos)
  @bytes[dpos...dend]
end

#find_startxref(bytes) ⇒ Object 



438
439
440
441
442
443
444
445
446
447
448
449
450
 
# File 'lib/corp_pdf/object_resolver.rb', line 438

def find_startxref(bytes)
  return nil if bytes.nil? || bytes.empty?

  if bytes =~ /startxref\s+(\d+)\s*%%EOF\s*\z/m
    return Integer(::Regexp.last_match(1))
  end

  m = bytes.rindex("startxref")
  return nil unless m

  tail = bytes[m, bytes.length - m]
  tail[/startxref\s+(\d+)/m, 1]&.to_i
end

#get_trailer_at_offset(offset) ⇒ Object 



55
56
57
58
59
60
61
62
63
 
# File 'lib/corp_pdf/object_resolver.rb', line 55

def get_trailer_at_offset(offset)
  if @bytes[offset, 4] == "xref"
    # Extract trailer from classic xref without modifying state
    extract_trailer_from_classic_xref(offset)
  else
    # Xref stream case - extract dictionary without modifying state
    extract_trailer_from_xref_stream(offset)
  end
end

#JSON_like_array(tok) ⇒ Object 



324
325
326
327
 
# File 'lib/corp_pdf/object_resolver.rb', line 324

def JSON_like_array(tok)
  inner = tok[1..-2]
  inner.split(/\s+/).map { |t| t =~ /\A\d+\z/ ? t.to_i : t }
end

#load_objstm(container_ref) ⇒ Object 



452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
 
# File 'lib/corp_pdf/object_resolver.rb', line 452

def load_objstm(container_ref)
  return if @objstm_cache.key?(container_ref)

  body = object_body(container_ref)
  raise "Object stream #{container_ref.inspect} not found in xref table" unless body

  dict_start = body.index("<<") || 0
  dict_end = balanced_from(body, dict_start)
  dict_src = body[dict_start...dict_end]
  s_pos = body.index(/\bstream\r?\n/m, dict_end) or raise "objstm stream missing"
  e_pos = body.index(/\bendstream\b/m, s_pos) or raise "objstm end missing"
  data = body[s_pos..e_pos]
  raw = decode_stream_data(dict_src, data)
  n = DictScan.value_token_after("/N", dict_src).to_i
  first = DictScan.value_token_after("/First", dict_src).to_i
  parsed = CorpPdf::ObjStm.parse(raw, n: n, first: first)
  @objstm_cache[container_ref] = parsed
end

#object_body(ref) ⇒ Object 



147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
 
# File 'lib/corp_pdf/object_resolver.rb', line 147

def object_body(ref)
  case (e = @entries[ref])&.type
  when :in_file
    i = e.offset
    # Find "obj" start near offset (handle any preceding whitespace)
    hdr = /\bobj\b/m.match(@bytes, i) or return nil
    after = hdr.end(0)
    # Skip optional whitespace and one line break if present
    after += 1 while (ch = @bytes.getbyte(after)) && ch <= 0x20
    j = @bytes.index(/\bendobj\b/m, after) or return nil
    @bytes[after...j]
  when :in_objstm
    load_objstm([e.objstm_num, 0])
    @objstm_cache[[e.objstm_num, 0]][e.objstm_index][:body]
  end
end

#paeth_predictor(a, b, c) ⇒ Object 



405
406
407
408
409
410
411
412
413
414
415
416
417
418
 
# File 'lib/corp_pdf/object_resolver.rb', line 405

def paeth_predictor(a, b, c)
  # a = left, b = up, c = up-left
  p = a + b - c
  pa = (p - a).abs
  pb = (p - b).abs
  pc = (p - c).abs
  if pa <= pb && pa <= pc
    a
  elsif pb <= pc
    b
  else
    c
  end
end

#parse_classic_xref(start) ⇒ Object 



201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
 
# File 'lib/corp_pdf/object_resolver.rb', line 201

def parse_classic_xref(start)
  pos = @bytes.rindex("xref", start) or raise "xref not found"
  i = pos + 4

  loop do
    m = /\s*(\d+)\s+(\d+)/m.match(@bytes, i) or break
    first = m[1].to_i
    count = m[2].to_i
    i = m.end(0)

    count.times do |k|
      # Skip whitespace/newlines before reading the 20-byte record
      i += 1 while (ch = @bytes.getbyte(i)) && [0x0A, 0x0D, 0x20].include?(ch)

      rec = @bytes[i, 20]
      raise "bad xref record" unless rec && rec.bytesize == 20

      off = rec[0, 10].to_i
      gen = rec[11, 5].to_i
      typ = rec[17, 1]
      i += 20
      # consume line ending(s)
      i += 1 while (ch = @bytes.getbyte(i)) && [0x0A, 0x0D].include?(ch)

      ref = [first + k, gen]
      @entries[ref] ||= Entry.new(type: :in_file, offset: off) if typ == "n"
      # (ignore 'f' free entries)
    end

    break if @bytes[i, 7] == "trailer"
  end

  tpos = @bytes.index("trailer", i)
  if tpos
    dpos = @bytes.index("<<", tpos)
    if dpos
      dend = balanced_from(@bytes, dpos)
      @last_xref_stream_dict = nil
      @trailer_explicit = @bytes[dpos...dend]
      return @trailer_explicit
    end
  end

  # No trailer found (might be at an intermediate xref in the chain)
  nil
end

#parse_cross_reference ⇒ Object

--- internals -----------------------------------------------------------



166
167
168
169
 
# File 'lib/corp_pdf/object_resolver.rb', line 166

def parse_cross_reference
  start = find_startxref(@bytes) or raise "startxref not found"
  parse_xref_at_offset(start)
end

#parse_xref_at_offset(offset) ⇒ Object 



171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
 
# File 'lib/corp_pdf/object_resolver.rb', line 171

def parse_xref_at_offset(offset)
  # 1) If 'xref' is literally at that offset => classic table
  if @bytes[offset, 4] == "xref"
    tr = parse_classic_xref(offset)

    # 2) Classic trailers may include /XRefStm <offset> to an xref stream with compressed entries
    xrefstm_tok = DictScan.value_token_after("/XRefStm", tr) if tr
    if xrefstm_tok && (ofs = xrefstm_tok.to_i).positive?
      parse_xref_stream_at(ofs) # merge entries from xref stream (type 0/1/2)
    end

    # 3) Follow /Prev pointer if present
    prev_tok = DictScan.value_token_after("/Prev", tr) if tr
    if prev_tok && (prev_ofs = prev_tok.to_i).positive?
      parse_xref_at_offset(prev_ofs)
    end
  else
    # Direct xref stream case (offset points to the xref stream obj header)
    dict_src = parse_xref_stream_at(offset)

    # Follow /Prev in the xref stream's dictionary
    if dict_src
      prev_tok = DictScan.value_token_after("/Prev", dict_src)
      if prev_tok && (prev_ofs = prev_tok.to_i).positive?
        parse_xref_at_offset(prev_ofs)
      end
    end
  end
end

#parse_xref_stream_at(header_ofs) ⇒ Object 



248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
 
# File 'lib/corp_pdf/object_resolver.rb', line 248

def parse_xref_stream_at(header_ofs)
  # Expect "<num> <gen> obj" at header_ofs
  m = /\A(\d+)\s+(\d+)\s+obj\b/m.match(@bytes[header_ofs, 50])
  unless m
    # Sometimes header_ofs might land on whitespace; search forward a bit
    win = @bytes[header_ofs, 256]
    m2 = /(\d+)\s+(\d+)\s+obj\b/m.match(win) or raise "xref stream header not found"
    header_ofs += m2.begin(0)
    m = m2
  end
  obj_ref = [m[1].to_i, m[2].to_i]

  dpos = @bytes.index("<<", header_ofs + m[0].length) or raise "xref stream dict missing"
  dend = balanced_from(@bytes, dpos)
  dict_src = @bytes[dpos...dend]
  @last_xref_stream_dict ||= dict_src # Keep first one for trailer_dict

  spos = @bytes.index(/\bstream\r?\n/m, dend) or raise "xref stream body missing"
  epos = @bytes.index(/\bendstream\b/m, spos) or raise "xref stream end missing"
  data = @bytes[spos..epos]
  raw = decode_stream_data(dict_src, data)

  # W is mandatory in xref streams; if missing, bail (don't crash)
  w_tok = DictScan.value_token_after("/W", dict_src)
  return nil unless w_tok

  w = JSON_like_array(w_tok)
  idx_tok = DictScan.value_token_after("/Index", dict_src)
  index = idx_tok ? JSON_like_array(idx_tok) : [0, DictScan.value_token_after("/Size", dict_src).to_i]

  parse_xref_stream_records(raw, w, index)

  # Ensure the xref stream object itself is registered (type 1 entry usually exists,
  # but if not, add it so object_body can find the stream if needed)
  unless @entries.key?(obj_ref)
    # Approximate offset at header_ofs
    @entries[obj_ref] = Entry.new(type: :in_file, offset: header_ofs)
  end

  dict_src # Return dict for /Prev checking
end

#parse_xref_stream_records(raw, w, index) ⇒ Object 



290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
 
# File 'lib/corp_pdf/object_resolver.rb', line 290

def parse_xref_stream_records(raw, w, index)
  w0, w1, w2 = w
  s = StringScanner.new(raw)
  (0...(index.length / 2)).each do |i|
    obj = index[2 * i].to_i
    count = index[(2 * i) + 1].to_i
    count.times do |k|
      t  = read_int(s, w0)
      f1 = read_int(s, w1)
      f2 = read_int(s, w2)
      ref = [obj + k, 0]
      case t
      when 0 then next # free
      when 1 then @entries[ref] ||= Entry.new(type: :in_file, offset: f1)
      when 2 then @entries[ref] ||= Entry.new(type: :in_objstm, objstm_num: f1, objstm_index: f2)
      end
    end
  end
end

#read_int(scanner, width) ⇒ Object 



310
311
312
313
314
315
316
317
318
319
320
321
322
 
# File 'lib/corp_pdf/object_resolver.rb', line 310

def read_int(scanner, width)
  # Ensure width is an integer
  w = width.is_a?(Integer) ? width : width.to_i
  return 0 if w.zero?

  bytes = scanner.peek(w)
  return 0 unless bytes && bytes.bytesize == w

  scanner.pos += w
  val = 0
  bytes.each_byte { |b| val = (val << 8) | b }
  val
end

#root_ref ⇒ Object 



18
19
20
21
22
23
24
25
26
27
28
 
# File 'lib/corp_pdf/object_resolver.rb', line 18

def root_ref
  # First try the current trailer_dict
  tr = trailer_dict
  if tr && tr =~ %r{/Root\s+(\d+)\s+(\d+)\s+R}
    return [Integer(::Regexp.last_match(1)), Integer(::Regexp.last_match(2))]
  end

  # If not found, search through all trailers by following /Prev chain
  start = find_startxref(@bytes) or return nil
  root_ref_from_trailer_chain(start)
end

#root_ref_from_trailer_chain(offset) ⇒ Object 



30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
 
# File 'lib/corp_pdf/object_resolver.rb', line 30

def root_ref_from_trailer_chain(offset)
  visited = Set.new
  
  loop do
    return nil if visited.include?(offset)
    visited.add(offset)

    # Get trailer at this offset
    tr = get_trailer_at_offset(offset)
    return nil unless tr

    # Check if this trailer has /Root
    if tr =~ %r{/Root\s+(\d+)\s+(\d+)\s+R}
      return [Integer(::Regexp.last_match(1)), Integer(::Regexp.last_match(2))]
    end

    # Follow /Prev pointer if present
    prev_tok = DictScan.value_token_after("/Prev", tr)
    break unless prev_tok && (prev_ofs = prev_tok.to_i).positive?
    offset = prev_ofs
  end

  nil
end

#trailer_dict ⇒ Object 



113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
 
# File 'lib/corp_pdf/object_resolver.rb', line 113

def trailer_dict
  # Priority order:
  # 1. Explicit trailer from classic xref (incremental updates)
  # 2. Xref stream dictionary (original PDFs)
  # 3. Search for trailer (fallback)
  @trailer_dict ||= if @trailer_explicit
                      @trailer_explicit
                    elsif @last_xref_stream_dict
                      @last_xref_stream_dict
                    else
                      # Find last 'trailer << ... >>' before last startxref
                      start = find_startxref(@bytes) || 0
                      head = @bytes[0...start]
                      idx = head.rindex("trailer")
                      raise "trailer not found" unless idx

                      # naive grab following dict
                      m = head.index("<<", idx)
                      n = balanced_from(head, m)
                      head[m...n]
                    end
end