flpdf 0.1.5

A pure-Rust PDF toolkit modeled on qpdf — read, rewrite, linearize, and inspect PDFs.
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
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
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
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
200
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
247
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
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
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
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
use crate::ascii85;
use crate::ascii_hex;
use crate::run_length;
use crate::security::standard::{decrypt_cipher_bytes, StringCipher};
use crate::{Dictionary, Error, Object, Result};
use flate2::read::ZlibDecoder;
use flate2::write::ZlibEncoder;
use flate2::Compression;
use std::io::{Read, Write};

pub fn decode_stream_data(dict: &Dictionary, stream_data: &[u8]) -> Result<Vec<u8>> {
    decode_stream_data_with_filters(dict.get("Filter"), dict.get("DecodeParms"), stream_data)
}

// Wired into encrypted document reads by later flpdf-9hc.3 layers.
#[allow(dead_code)]
pub(crate) fn decode_stream_data_with_decryption(
    dict: &Dictionary,
    stream_data: &[u8],
    cipher: StringCipher<'_>,
) -> Result<Vec<u8>> {
    let mut decrypted = stream_data.to_vec();
    decrypt_cipher_bytes(&mut decrypted, cipher)?;
    decode_stream_data(dict, &decrypted)
}

pub(crate) fn decode_stream_data_with_crypt_filter<F>(
    dict: &Dictionary,
    stream_data: &[u8],
    mut decrypt_crypt: F,
) -> Result<Vec<u8>>
where
    F: FnMut(Option<&Object>, &[u8]) -> Result<Vec<u8>>,
{
    decode_stream_data_with_filters_and_crypt(
        dict.get("Filter"),
        dict.get("DecodeParms"),
        stream_data,
        &mut decrypt_crypt,
    )
}

pub fn encode_stream_data(dict: &Dictionary, stream_data: &[u8]) -> Result<Vec<u8>> {
    encode_stream_data_with_filters(dict.get("Filter"), dict.get("DecodeParms"), stream_data)
}

fn decode_stream_data_with_filters(
    filter: Option<&Object>,
    decode_params: Option<&Object>,
    stream_data: &[u8],
) -> Result<Vec<u8>> {
    decode_stream_data_with_filters_and_crypt(filter, decode_params, stream_data, &mut |_, _| {
        Err(Error::Unsupported(
            "unsupported stream filter: Crypt".to_string(),
        ))
    })
}

fn decode_stream_data_with_filters_and_crypt<F>(
    filter: Option<&Object>,
    decode_params: Option<&Object>,
    stream_data: &[u8],
    decrypt_crypt: &mut F,
) -> Result<Vec<u8>>
where
    F: FnMut(Option<&Object>, &[u8]) -> Result<Vec<u8>>,
{
    match filter {
        None => Ok(stream_data.to_vec()),
        Some(filter) => {
            if let Some(filter_name) = filter.as_name() {
                if filter_name == b"Crypt" {
                    return decrypt_crypt(get_decode_params(decode_params, 0), stream_data);
                }
                let params = get_decode_params(decode_params, 0);
                let decoded = apply_single_filter_decode(filter_name, stream_data, params)
                    .map_err(Error::Unsupported)?;
                return apply_decode_params(params, &decoded);
            }

            if let Some(filters) = filter.as_array() {
                let mut decoded = stream_data.to_vec();
                for (index, filter) in filters.iter().enumerate() {
                    let Some(filter_name) = filter.as_name() else {
                        return Err(Error::Unsupported(
                            "unsupported stream filter type: expected name".to_string(),
                        ));
                    };
                    if filter_name == b"Crypt" {
                        decoded = decrypt_crypt(get_decode_params(decode_params, index), &decoded)?;
                    } else {
                        let params = get_decode_params(decode_params, index);
                        decoded = apply_single_filter_decode(filter_name, &decoded, params)
                            .map_err(Error::Unsupported)?;
                        decoded = apply_decode_params(params, &decoded)?;
                    }
                }
                return Ok(decoded);
            }

            Err(Error::Unsupported(format!(
                "unsupported stream filter syntax: {}",
                object_debug_repr(filter)
            )))
        }
    }
}

fn encode_stream_data_with_filters(
    filter: Option<&Object>,
    decode_params: Option<&Object>,
    stream_data: &[u8],
) -> Result<Vec<u8>> {
    match filter {
        None => Ok(stream_data.to_vec()),
        Some(filter) => {
            if let Some(filter_name) = filter.as_name() {
                let params = get_decode_params(decode_params, 0);
                let after_predictor = apply_encode_params(params, stream_data)?;
                return apply_single_filter_encode(filter_name, &after_predictor)
                    .map_err(Error::Unsupported);
            }

            if let Some(filters) = filter.as_array() {
                // ISO 32000-1 §7.4.2: the /Filter array names filters in *decode*
                // order, so encoding must apply them in reverse for round-tripping.
                let mut encoded = stream_data.to_vec();
                for (index, filter) in filters.iter().enumerate().rev() {
                    let Some(filter_name) = filter.as_name() else {
                        return Err(Error::Unsupported(
                            "unsupported stream filter type: expected name".to_string(),
                        ));
                    };
                    let params = get_decode_params(decode_params, index);
                    encoded = apply_encode_params(params, &encoded)?;
                    encoded = apply_single_filter_encode(filter_name, &encoded)
                        .map_err(Error::Unsupported)?;
                }
                return Ok(encoded);
            }

            Err(Error::Unsupported(format!(
                "unsupported stream filter syntax: {}",
                object_debug_repr(filter)
            )))
        }
    }
}

fn get_decode_params(params: Option<&Object>, index: usize) -> Option<&Object> {
    let param = params?;
    if param.as_dict().is_some() {
        Some(param)
    } else {
        param.as_array()?.get(index)
    }
}

fn integer_decode_param(params: &Dictionary, key: &str) -> Result<Option<i64>> {
    let Some(value) = params.get(key) else {
        return Ok(None);
    };
    value
        .as_integer()
        .map(Some)
        .ok_or_else(|| Error::Unsupported(format!("/DecodeParms /{key} must be integer")))
}

fn non_negative_usize_param(params: &Dictionary, key: &str) -> Result<Option<usize>> {
    integer_decode_param(params, key)?
        .map(|value| {
            usize::try_from(value).map_err(|_| {
                Error::Unsupported(format!("/DecodeParms /{key} must be non-negative"))
            })
        })
        .transpose()
}

fn non_negative_u8_param(params: &Dictionary, key: &str) -> Result<Option<u8>> {
    integer_decode_param(params, key)?
        .map(|value| {
            u8::try_from(value).map_err(|_| {
                Error::Unsupported(format!("/DecodeParms /{key} must be non-negative"))
            })
        })
        .transpose()
}

/// Extract PNG predictor parameters from a DecodeParms dictionary.
///
/// Returns `Ok(None)` when no predictor is needed (no dict, no Predictor key, or Predictor ≤ 1).
/// Returns `Ok(Some((predictor, row_bytes, bytes_per_pixel)))` for PNG predictors 10..=15.
/// Returns `Err` for Predictor 2 or any other unsupported value.
fn extract_predictor_params(decode_params: Option<&Object>) -> Result<Option<(u8, usize, usize)>> {
    let Some(params) = decode_params.and_then(Object::as_dict) else {
        return Ok(None);
    };

    let Some(predictor) = non_negative_u8_param(params, "Predictor")? else {
        return Ok(None);
    };

    if predictor <= 1 {
        return Ok(None);
    }

    if predictor == 2 {
        return Err(Error::Unsupported(
            "/DecodeParms /Predictor 2 is not supported for this stream type".to_string(),
        ));
    }

    if !(10..=15).contains(&predictor) {
        return Err(Error::Unsupported(format!(
            "unsupported /DecodeParms /Predictor {predictor}"
        )));
    }

    let colors = non_negative_usize_param(params, "Colors")?.unwrap_or(1);
    let bits_per_component = non_negative_usize_param(params, "BitsPerComponent")?.unwrap_or(8);
    let columns = non_negative_usize_param(params, "Columns")?.ok_or_else(|| {
        Error::Unsupported("/DecodeParms /Columns required for PNG predictor".to_string())
    })?;

    let row_bits = columns
        .checked_mul(colors)
        .and_then(|value| value.checked_mul(bits_per_component))
        .ok_or_else(|| Error::Unsupported("/DecodeParms /Predictor overflow".to_string()))?;
    let row_bytes = row_bits.div_ceil(8);
    if row_bytes == 0 {
        return Err(Error::Unsupported(
            "/DecodeParms /Predictor produced zero row width".to_string(),
        ));
    }
    let bits_per_pixel = colors
        .checked_mul(bits_per_component)
        .ok_or_else(|| Error::Unsupported("/DecodeParms /Predictor overflow".to_string()))?;
    let bytes_per_pixel = bits_per_pixel.div_ceil(8).max(1);

    Ok(Some((predictor, row_bytes, bytes_per_pixel)))
}

fn apply_decode_params(decode_params: Option<&Object>, stream_data: &[u8]) -> Result<Vec<u8>> {
    match extract_predictor_params(decode_params)? {
        None => Ok(stream_data.to_vec()),
        Some((_predictor, row_bytes, bytes_per_pixel)) => {
            decode_png_predictor(stream_data, row_bytes, bytes_per_pixel)
        }
    }
}

fn apply_encode_params(decode_params: Option<&Object>, stream_data: &[u8]) -> Result<Vec<u8>> {
    match extract_predictor_params(decode_params)? {
        None => Ok(stream_data.to_vec()),
        Some((predictor, row_bytes, bytes_per_pixel)) => {
            encode_png_predictor(stream_data, row_bytes, bytes_per_pixel, predictor)
        }
    }
}

fn decode_png_predictor(bytes: &[u8], row_bytes: usize, bytes_per_pixel: usize) -> Result<Vec<u8>> {
    let row_size = row_bytes
        .checked_add(1)
        .ok_or_else(|| Error::Unsupported("PNG predictor row size overflow".to_string()))?;
    if !bytes.len().is_multiple_of(row_size) {
        return Err(Error::Unsupported(
            "corrupt PNG predictor stream".to_string(),
        ));
    }

    let row_count = bytes.len() / row_size;
    let mut decoded = Vec::with_capacity(row_count * row_bytes);
    let mut previous_row = vec![0u8; row_bytes];

    for row in bytes.chunks_exact(row_size) {
        let filter = row[0];
        let raw = &row[1..];
        let mut current = vec![0u8; row_bytes];

        for i in 0..row_bytes {
            let raw_byte = raw[i];
            let left = if i >= bytes_per_pixel {
                current[i - bytes_per_pixel]
            } else {
                0
            };
            let above = previous_row[i];
            let upper_left = if i >= bytes_per_pixel {
                previous_row[i - bytes_per_pixel]
            } else {
                0
            };

            current[i] = match filter {
                0 => raw_byte,
                1 => raw_byte.wrapping_add(left),
                2 => raw_byte.wrapping_add(above),
                3 => {
                    let average = (u16::from(left) + u16::from(above)) / 2;
                    raw_byte.wrapping_add(average as u8)
                }
                4 => {
                    let p = left as i16 + above as i16 - upper_left as i16;
                    let pa = (p - left as i16).abs();
                    let pb = (p - above as i16).abs();
                    let pc = (p - upper_left as i16).abs();

                    let predictor = if pa <= pb && pa <= pc {
                        left
                    } else if pb <= pc {
                        above
                    } else {
                        upper_left
                    };

                    raw_byte.wrapping_add(predictor)
                }
                _ => {
                    return Err(Error::Unsupported(
                        "unsupported PNG predictor filter".to_string(),
                    ))
                }
            };
        }

        decoded.extend_from_slice(&current);
        previous_row = current;
    }

    Ok(decoded)
}

/// Apply a single PNG filter to one byte and return the encoded (filtered) byte.
///
/// Shared between the optimum-cost loop and the per-row encoding loop so that
/// the filter logic exists in exactly one place.
#[inline]
fn png_filter_byte(filter: u8, raw: u8, left: u8, above: u8, upper_left: u8) -> u8 {
    match filter {
        0 => raw,
        1 => raw.wrapping_sub(left),
        2 => raw.wrapping_sub(above),
        3 => {
            let avg = ((u16::from(left) + u16::from(above)) / 2) as u8;
            raw.wrapping_sub(avg)
        }
        4 => {
            let p = i16::from(left) + i16::from(above) - i16::from(upper_left);
            let pa = (p - i16::from(left)).abs();
            let pb = (p - i16::from(above)).abs();
            let pc = (p - i16::from(upper_left)).abs();
            let predictor_val = if pa <= pb && pa <= pc {
                left
            } else if pb <= pc {
                above
            } else {
                upper_left
            };
            raw.wrapping_sub(predictor_val)
        }
        _ => unreachable!("png_filter_byte expects filter in 0..=4"),
    }
}

fn encode_png_predictor(
    bytes: &[u8],
    row_bytes: usize,
    bytes_per_pixel: usize,
    predictor: u8,
) -> Result<Vec<u8>> {
    if row_bytes == 0 || !bytes.len().is_multiple_of(row_bytes) {
        return Err(Error::Unsupported(
            "raw data not divisible by row_bytes".to_string(),
        ));
    }

    let row_count = bytes.len() / row_bytes;
    // Each encoded row = 1 filter byte + row_bytes of filtered data
    let mut encoded = Vec::with_capacity(row_count * (row_bytes + 1));
    let mut previous_row = vec![0u8; row_bytes];

    for raw_row in bytes.chunks_exact(row_bytes) {
        // Determine which filter byte to use for this row. For predictor 15
        // (Optimum) we accumulate all 5 filters' costs in a single pass over
        // the row to avoid iterating 5× and recomputing neighbor values for
        // each filter (libpng minimum-sum heuristic).
        let filter_byte = if predictor == 15 {
            let mut costs = [0u64; 5];
            for i in 0..row_bytes {
                let raw = raw_row[i];
                let left = if i >= bytes_per_pixel {
                    raw_row[i - bytes_per_pixel]
                } else {
                    0
                };
                let above = previous_row[i];
                let upper_left = if i >= bytes_per_pixel {
                    previous_row[i - bytes_per_pixel]
                } else {
                    0
                };
                for (f, cost) in costs.iter_mut().enumerate() {
                    let filtered = png_filter_byte(f as u8, raw, left, above, upper_left);
                    *cost += u64::from((filtered as i8).unsigned_abs());
                }
            }
            costs
                .iter()
                .enumerate()
                .min_by_key(|&(_, &c)| c)
                .map(|(i, _)| i as u8)
                .expect("costs has 5 entries")
        } else {
            // Fixed filter: Predictor 10→0, 11→1, 12→2, 13→3, 14→4
            predictor - 10
        };

        encoded.push(filter_byte);

        for i in 0..row_bytes {
            let raw = raw_row[i];
            let left = if i >= bytes_per_pixel {
                raw_row[i - bytes_per_pixel]
            } else {
                0
            };
            let above = previous_row[i];
            let upper_left = if i >= bytes_per_pixel {
                previous_row[i - bytes_per_pixel]
            } else {
                0
            };
            encoded.push(png_filter_byte(filter_byte, raw, left, above, upper_left));
        }

        // Reuse previous_row's buffer instead of allocating a fresh Vec each
        // row. previous_row was initialised with `row_bytes` zeros and
        // raw_row is row_bytes long (guaranteed by `chunks_exact(row_bytes)`),
        // so the lengths always match.
        previous_row.copy_from_slice(raw_row);
    }

    Ok(encoded)
}

fn apply_single_filter_decode(
    filter_name: &[u8],
    stream_data: &[u8],
    decode_params: Option<&Object>,
) -> std::result::Result<Vec<u8>, String> {
    if filter_name == b"FlateDecode" {
        let mut decoded = Vec::new();
        let mut decoder = ZlibDecoder::new(stream_data);
        decoder
            .read_to_end(&mut decoded)
            .map_err(|error| error.to_string())?;
        return Ok(decoded);
    }

    if filter_name == b"LZWDecode" {
        // EarlyChange (default 1 per PDF spec): when 1, the code width increases
        // one symbol *before* the table fills; when 0, it increases *after*.
        let early_change = match decode_params {
            Some(Object::Dictionary(params)) => match params.get("EarlyChange") {
                Some(Object::Integer(v)) => *v != 0,
                _ => true, // default EarlyChange = 1
            },
            _ => true, // no DecodeParms → default EarlyChange = 1
        };
        return lzw_decode(stream_data, early_change);
    }

    if filter_name == b"ASCII85Decode" {
        return ascii85::decode(stream_data);
    }

    if filter_name == b"ASCIIHexDecode" {
        return ascii_hex::decode(stream_data);
    }

    if filter_name == b"RunLengthDecode" {
        return run_length::decode(stream_data);
    }

    Err(format!(
        "unsupported stream filter: {}",
        std::str::from_utf8(filter_name).unwrap_or("<binary>")
    ))
}

/// Decode an LZW-compressed byte stream as defined by PDF §7.4.4 (LZWDecode).
///
/// PDF's LZW variant:
/// - Starts at 9-bit codes; maximum code width is 12 bits.
/// - Code 256 = ClearTable (resets the table to the initial state and next code
///   width to 9 bits).
/// - Code 257 = EOD (end of data; any remaining bits in the current byte are
///   discarded).
/// - `early_change`: when `true` (PDF default / EarlyChange=1), the code width
///   increments one code *before* the table is full (i.e. when the next code
///   to be added would exceed the current code width capacity).  When `false`
///   (EarlyChange=0), the width increments *after* the table fills.
fn lzw_decode(data: &[u8], early_change: bool) -> std::result::Result<Vec<u8>, String> {
    const CLEAR_CODE: u16 = 256;
    const EOD_CODE: u16 = 257;
    const FIRST_CODE: u16 = 258;
    const MAX_BITS: u32 = 12;
    const MAX_TABLE_SIZE: usize = 1 << MAX_BITS; // 4096

    // The string table: each entry is the byte sequence it represents.
    // Entries 0–255 are the literal single-byte strings; 256 and 257 are
    // control codes (not stored in the table as strings).
    let mut table: Vec<Vec<u8>> = (0u8..=255).map(|b| vec![b]).collect();
    // Pad to index 258 so FIRST_CODE aligns with the next push.
    table.push(vec![]); // slot 256 (ClearCode sentinel — never looked up)
    table.push(vec![]); // slot 257 (EOD sentinel — never looked up)

    let mut output = Vec::new();
    let mut bit_buf: u64 = 0;
    let mut bits_in_buf: u32 = 0;
    let mut code_bits: u32 = 9;

    // The "next table index" at which the width bumps.  With early_change=true
    // the bump triggers when next_entry == (1 << code_bits) - 1 (one slot before
    // the table fills); with early_change=false it triggers when
    // next_entry == (1 << code_bits) (table exactly full).
    let early_offset: usize = if early_change { 1 } else { 0 };

    let mut prev_entry: Option<Vec<u8>> = None;

    let mut byte_pos = 0usize;

    loop {
        // Fill the bit buffer until it has at least `code_bits` bits.
        while bits_in_buf < code_bits {
            if byte_pos >= data.len() {
                // Ran out of input before EOD code — treat as implicit EOD.
                return Ok(output);
            }
            bit_buf = (bit_buf << 8) | u64::from(data[byte_pos]);
            bits_in_buf += 8;
            byte_pos += 1;
        }

        // Extract the next `code_bits`-wide code from the MSB side.
        let shift = bits_in_buf - code_bits;
        let code = ((bit_buf >> shift) & ((1u64 << code_bits) - 1)) as u16;
        bits_in_buf -= code_bits;
        bit_buf &= (1u64 << bits_in_buf) - 1;

        if code == EOD_CODE {
            break;
        }

        if code == CLEAR_CODE {
            // Reset: truncate the table back to the 256 literals + 2 sentinels.
            table.truncate(FIRST_CODE as usize);
            code_bits = 9;
            prev_entry = None;
            continue;
        }

        // Resolve the code to its string.
        let entry: Vec<u8> = if (code as usize) < table.len() {
            table[code as usize].clone()
        } else if code as usize == table.len() {
            // The classic "KwKwK" case: the code being added is the one we're
            // currently processing.  Its string is prev + first_byte(prev).
            match &prev_entry {
                Some(prev) => {
                    let mut s = prev.clone();
                    s.push(prev[0]);
                    s
                }
                None => {
                    return Err(format!(
                        "LZWDecode: code {code} is one past table end but no previous entry"
                    ))
                }
            }
        } else {
            return Err(format!(
                "LZWDecode: code {code} out of range (table size {})",
                table.len()
            ));
        };

        output.extend_from_slice(&entry);

        // Add a new entry = prev_string + first_byte(current_string).
        if let Some(ref prev) = prev_entry {
            if table.len() < MAX_TABLE_SIZE {
                let mut new_entry = prev.clone();
                new_entry.push(entry[0]);
                table.push(new_entry);

                // Bump code width when the table reaches the trigger threshold.
                if code_bits < MAX_BITS && table.len() == (1usize << code_bits) - early_offset {
                    code_bits += 1;
                }
            }
        }

        prev_entry = Some(entry);
    }

    Ok(output)
}

fn apply_single_filter_encode(
    filter_name: &[u8],
    stream_data: &[u8],
) -> std::result::Result<Vec<u8>, String> {
    if filter_name == b"FlateDecode" {
        let mut encoder = ZlibEncoder::new(Vec::new(), Compression::default());
        encoder
            .write_all(stream_data)
            .map_err(|error| error.to_string())?;
        let encoded = encoder.finish().map_err(|error| error.to_string())?;
        return Ok(encoded);
    }

    if filter_name == b"ASCII85Decode" {
        return Ok(ascii85::encode(stream_data));
    }

    if filter_name == b"ASCIIHexDecode" {
        return Ok(ascii_hex::encode(stream_data));
    }

    if filter_name == b"RunLengthDecode" {
        return Ok(run_length::encode(stream_data));
    }

    Err(format!(
        "unsupported stream filter: {}",
        std::str::from_utf8(filter_name).unwrap_or("<binary>"),
    ))
}

fn object_debug_repr(object: &Object) -> &'static str {
    match object {
        Object::Name(name) if name == b"FlateDecode" => "FlateDecode",
        Object::Name(name) if name == b"ASCII85Decode" => "ASCII85Decode",
        Object::Name(name) if name == b"ASCIIHexDecode" => "ASCIIHexDecode",
        Object::Name(name) if name == b"LZWDecode" => "LZWDecode",
        Object::Name(name) if name == b"RunLengthDecode" => "RunLengthDecode",
        _ => "unsupported",
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::security::primitives::rc4;
    use crate::security::standard::StringCipher;
    use aes::{Aes128, Aes256};
    use cbc::cipher::{block_padding::Pkcs7, BlockEncryptMut, KeyIvInit};
    use cbc::Encryptor;

    fn flate_dict() -> Dictionary {
        let mut dict = Dictionary::new();
        dict.insert("Filter", Object::Name(b"FlateDecode".to_vec()));
        dict
    }

    fn aes128_stream(key: &[u8; 16], iv: &[u8; 16], plaintext: &[u8]) -> Vec<u8> {
        let enc = <Encryptor<Aes128> as KeyIvInit>::new(key.into(), iv.into());
        let mut buf = plaintext.to_vec();
        let msg_len = buf.len();
        buf.resize(msg_len + 16, 0);
        let ciphertext = enc.encrypt_padded_mut::<Pkcs7>(&mut buf, msg_len).unwrap();
        let mut out = iv.to_vec();
        out.extend_from_slice(ciphertext);
        out
    }

    fn aes256_stream(key: &[u8; 32], iv: &[u8; 16], plaintext: &[u8]) -> Vec<u8> {
        let enc = <Encryptor<Aes256> as KeyIvInit>::new(key.into(), iv.into());
        let mut buf = plaintext.to_vec();
        let msg_len = buf.len();
        buf.resize(msg_len + 16, 0);
        let ciphertext = enc.encrypt_padded_mut::<Pkcs7>(&mut buf, msg_len).unwrap();
        let mut out = iv.to_vec();
        out.extend_from_slice(ciphertext);
        out
    }

    #[test]
    fn decode_stream_data_with_decryption_decrypts_before_flate() {
        let dict = flate_dict();
        let plaintext = b"stream plaintext after flate";
        let mut encrypted = encode_stream_data(&dict, plaintext).unwrap();
        rc4(b"Key", &mut encrypted).unwrap();

        let decoded = decode_stream_data_with_decryption(
            &dict,
            &encrypted,
            StringCipher::Rc4 { key: b"Key" },
        )
        .unwrap();

        assert_eq!(decoded, plaintext);
    }

    #[test]
    fn decode_stream_data_with_decryption_handles_no_filter_rc4_streams() {
        let dict = Dictionary::new();
        let mut encrypted = b"raw encrypted stream".to_vec();
        rc4(b"Key", &mut encrypted).unwrap();

        let decoded = decode_stream_data_with_decryption(
            &dict,
            &encrypted,
            StringCipher::Rc4 { key: b"Key" },
        )
        .unwrap();

        assert_eq!(decoded, b"raw encrypted stream");
    }

    #[test]
    fn decode_stream_data_with_decryption_handles_aes_streams() {
        let dict = Dictionary::new();
        let aes128_key = [0x11; 16];
        let aes128_iv = [0x22; 16];
        let aes128 = aes128_stream(&aes128_key, &aes128_iv, b"AES-128 stream");
        let decoded = decode_stream_data_with_decryption(
            &dict,
            &aes128,
            StringCipher::Aes128 { key: &aes128_key },
        )
        .unwrap();
        assert_eq!(decoded, b"AES-128 stream");

        let aes256_key = [0x33; 32];
        let aes256_iv = [0x44; 16];
        let aes256 = aes256_stream(&aes256_key, &aes256_iv, b"AES-256 stream");
        let decoded = decode_stream_data_with_decryption(
            &dict,
            &aes256,
            StringCipher::Aes256 { key: &aes256_key },
        )
        .unwrap();
        assert_eq!(decoded, b"AES-256 stream");
    }

    #[test]
    fn decode_stream_data_without_decryption_keeps_plaintext_behavior() {
        let dict = flate_dict();
        let plaintext = b"legacy plaintext flate";
        let encoded = encode_stream_data(&dict, plaintext).unwrap();

        let decoded = decode_stream_data(&dict, &encoded).unwrap();

        assert_eq!(decoded, plaintext);
    }

    // ----- ASCIIHexDecode filter integration tests -----

    fn ascii_hex_dict() -> Dictionary {
        let mut dict = Dictionary::new();
        dict.insert("Filter", Object::Name(b"ASCIIHexDecode".to_vec()));
        dict
    }

    #[test]
    fn decode_stream_data_ascii_hex_round_trip() {
        let dict = ascii_hex_dict();
        let plaintext = b"Hello from ASCIIHexDecode filter!";

        let encoded = encode_stream_data(&dict, plaintext).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();

        assert_eq!(decoded, plaintext.as_slice());
    }

    #[test]
    fn decode_stream_data_ascii_hex_empty() {
        let dict = ascii_hex_dict();
        let plaintext = b"";

        let encoded = encode_stream_data(&dict, plaintext).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();

        assert_eq!(decoded, plaintext.as_slice());
    }

    #[test]
    fn decode_stream_data_ascii_hex_odd_length_data() {
        let dict = ascii_hex_dict();
        // 3 bytes → odd nibble count in inner encoding only if we provide raw odd data;
        // encode always emits two hex chars per byte so no padding needed on decode
        let plaintext = b"ABC";

        let encoded = encode_stream_data(&dict, plaintext).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();

        assert_eq!(decoded, plaintext.as_slice());
    }

    // ----- ASCII85Decode filter integration tests -----

    fn ascii85_dict() -> Dictionary {
        let mut dict = Dictionary::new();
        dict.insert("Filter", Object::Name(b"ASCII85Decode".to_vec()));
        dict
    }

    #[test]
    fn decode_stream_data_ascii85_round_trip() {
        let dict = ascii85_dict();
        let plaintext = b"Hello from ASCII85Decode filter!";

        let encoded = encode_stream_data(&dict, plaintext).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();

        assert_eq!(decoded, plaintext.as_slice());
    }

    #[test]
    fn decode_stream_data_ascii85_empty() {
        let dict = ascii85_dict();
        let plaintext = b"";

        let encoded = encode_stream_data(&dict, plaintext).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();

        assert_eq!(decoded, plaintext.as_slice());
    }

    #[test]
    fn decode_stream_data_ascii85_zero_block() {
        let dict = ascii85_dict();
        // A 4-byte all-zero block triggers the 'z' shorthand in the encoder
        let plaintext = [0u8; 8]; // two complete zero blocks → encoder emits "zz~>"

        let encoded = encode_stream_data(&dict, &plaintext).unwrap();
        // Verify the encoder actually used the 'z' shorthand
        assert!(
            encoded.contains(&b'z'),
            "encoder should emit 'z' for 4-byte zero block"
        );
        let decoded = decode_stream_data(&dict, &encoded).unwrap();

        assert_eq!(decoded, plaintext.as_slice());
    }

    #[test]
    fn decode_stream_data_ascii85_short_final_group() {
        let dict = ascii85_dict();
        // Test all three short-final-group lengths: 1, 2, 3 bytes remainder
        for plaintext in [b"M".as_slice(), b"Ma", b"Man"] {
            let encoded = encode_stream_data(&dict, plaintext).unwrap();
            let decoded = decode_stream_data(&dict, &encoded).unwrap();
            assert_eq!(
                decoded,
                plaintext,
                "short final group round-trip failed for {} bytes",
                plaintext.len()
            );
        }
    }

    #[test]
    fn decode_stream_data_ascii85_rejects_invalid_byte() {
        let dict = ascii85_dict();
        // 'v' (0x76) is above the valid range '!'..'u' (0x21..=0x75)
        // Feed a hand-crafted stream: "9jqov~>" where 'v' is out-of-range
        let invalid_stream = b"9jqov~>";

        let result = decode_stream_data(&dict, invalid_stream);

        assert!(result.is_err(), "expected error for out-of-range byte");
        let msg = result.unwrap_err().to_string();
        assert!(
            msg.contains("ASCII85Decode"),
            "error message should contain 'ASCII85Decode', got: {msg}"
        );
    }

    // ----- RunLengthDecode filter integration tests -----

    fn run_length_dict() -> Dictionary {
        let mut dict = Dictionary::new();
        dict.insert("Filter", Object::Name(b"RunLengthDecode".to_vec()));
        dict
    }

    #[test]
    fn decode_stream_data_run_length_round_trip() {
        let dict = run_length_dict();
        let plaintext = b"Hello from RunLengthDecode filter!";

        let encoded = encode_stream_data(&dict, plaintext).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();

        assert_eq!(decoded, plaintext.as_slice());
    }

    #[test]
    fn decode_stream_data_run_length_empty() {
        let dict = run_length_dict();
        let plaintext = b"";

        let encoded = encode_stream_data(&dict, plaintext).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();

        assert_eq!(decoded, plaintext.as_slice());
    }

    #[test]
    fn decode_stream_data_run_length_with_repeats() {
        let dict = run_length_dict();
        // Data with prominent repeat runs (triggers repeat-run encoding).
        let mut plaintext = vec![0x42u8; 100]; // 100 'B' bytes
        plaintext.extend(b"literal");
        plaintext.extend(vec![0xCCu8; 50]); // 50 0xCC bytes

        let encoded = encode_stream_data(&dict, &plaintext).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();

        assert_eq!(decoded, plaintext.as_slice());
    }

    #[test]
    fn decode_stream_data_run_length_rejects_truncated_literal() {
        let dict = run_length_dict();
        // Hand-crafted truncated stream: header says 6 literals (l=5) but only 3 follow.
        let truncated_stream = vec![0x05u8, b'A', b'B', b'C']; // 3 bytes instead of 6

        let result = decode_stream_data(&dict, &truncated_stream);

        assert!(
            result.is_err(),
            "expected error for truncated literal stream"
        );
        let msg = result.unwrap_err().to_string();
        assert!(
            msg.contains("RunLengthDecode"),
            "error message should contain 'RunLengthDecode', got: {msg}"
        );
    }

    // ----- Array filter chain round-trip tests (regression for flpdf-fh8) -----
    //
    // Per ISO 32000-1 §7.4.2, the /Filter array names the filters in the order
    // they must be applied to *decode* the stream. The encoder therefore has
    // to apply them in reverse so that `decode(encode(x))` round-trips for any
    // multi-element filter chain.

    fn array_filter_dict(filters: &[&[u8]]) -> Dictionary {
        let mut dict = Dictionary::new();
        let names: Vec<Object> = filters.iter().map(|f| Object::Name(f.to_vec())).collect();
        dict.insert("Filter", Object::Array(names));
        dict
    }

    #[test]
    fn encode_stream_data_array_chain_round_trips_ascii85_then_flate() {
        // Decoder order: ASCII85Decode, then FlateDecode.
        // Encoder must therefore apply FlateDecode first, then ASCII85Decode.
        let dict = array_filter_dict(&[b"ASCII85Decode", b"FlateDecode"]);
        let plaintext = b"Round-trip me through ASCII85 over Flate, please!";

        let encoded = encode_stream_data(&dict, plaintext).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();

        assert_eq!(decoded, plaintext.as_slice());
    }

    #[test]
    fn encode_stream_data_array_chain_round_trips_ascii_hex_then_flate() {
        let dict = array_filter_dict(&[b"ASCIIHexDecode", b"FlateDecode"]);
        let plaintext: Vec<u8> = (0u8..=200u8).collect();

        let encoded = encode_stream_data(&dict, &plaintext).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();

        assert_eq!(decoded, plaintext);
    }

    #[test]
    fn encode_stream_data_array_chain_single_filter_matches_name_form() {
        // /Filter [/FlateDecode] should behave identically to /Filter /FlateDecode.
        let array_dict = array_filter_dict(&[b"FlateDecode"]);
        let name_dict = flate_dict();
        let plaintext = b"single-filter array form";

        let encoded_array = encode_stream_data(&array_dict, plaintext).unwrap();
        let encoded_name = encode_stream_data(&name_dict, plaintext).unwrap();

        assert_eq!(
            encoded_array, encoded_name,
            "Array form with one filter should produce the same bytes as the Name form"
        );
    }

    // ----- PNG predictor encode round-trip tests -----

    fn png_predictor_dict(predictor: i64, columns: i64) -> Dictionary {
        let mut dict = Dictionary::new();
        dict.insert("Filter", Object::Name(b"FlateDecode".to_vec()));
        let mut parms = Dictionary::new();
        parms.insert("Predictor", Object::Integer(predictor));
        parms.insert("Columns", Object::Integer(columns));
        dict.insert("DecodeParms", Object::Dictionary(parms));
        dict
    }

    fn png_predictor_dict_rgb(predictor: i64, columns: i64) -> Dictionary {
        let mut dict = Dictionary::new();
        dict.insert("Filter", Object::Name(b"FlateDecode".to_vec()));
        let mut parms = Dictionary::new();
        parms.insert("Predictor", Object::Integer(predictor));
        parms.insert("Columns", Object::Integer(columns));
        parms.insert("Colors", Object::Integer(3));
        parms.insert("BitsPerComponent", Object::Integer(8));
        dict.insert("DecodeParms", Object::Dictionary(parms));
        dict
    }

    /// Simple 2-row, 4-column grayscale raw data for predictor round-trip tests.
    fn sample_raw_4x2() -> Vec<u8> {
        vec![
            10, 20, 30, 40, // row 0
            50, 60, 70, 80, // row 1
        ]
    }

    #[test]
    fn encode_stream_data_png_predictor_10_round_trip() {
        let dict = png_predictor_dict(10, 4);
        let raw = sample_raw_4x2();
        let encoded = encode_stream_data(&dict, &raw).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();
        assert_eq!(decoded, raw);
    }

    #[test]
    fn encode_stream_data_png_predictor_11_round_trip() {
        let dict = png_predictor_dict(11, 4);
        let raw = sample_raw_4x2();
        let encoded = encode_stream_data(&dict, &raw).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();
        assert_eq!(decoded, raw);
    }

    #[test]
    fn encode_stream_data_png_predictor_12_round_trip() {
        let dict = png_predictor_dict(12, 4);
        let raw = sample_raw_4x2();
        let encoded = encode_stream_data(&dict, &raw).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();
        assert_eq!(decoded, raw);
    }

    #[test]
    fn encode_stream_data_png_predictor_13_round_trip() {
        let dict = png_predictor_dict(13, 4);
        let raw = sample_raw_4x2();
        let encoded = encode_stream_data(&dict, &raw).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();
        assert_eq!(decoded, raw);
    }

    #[test]
    fn encode_stream_data_png_predictor_14_round_trip() {
        let dict = png_predictor_dict(14, 4);
        let raw = sample_raw_4x2();
        let encoded = encode_stream_data(&dict, &raw).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();
        assert_eq!(decoded, raw);
    }

    #[test]
    fn encode_stream_data_png_predictor_15_round_trip() {
        let dict = png_predictor_dict(15, 4);
        let raw = sample_raw_4x2();
        let encoded = encode_stream_data(&dict, &raw).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();
        assert_eq!(decoded, raw);
    }

    #[test]
    fn encode_stream_data_png_predictor_handles_multi_row() {
        // row_bytes=8, rows=4 → 32 bytes total
        let dict = png_predictor_dict(12, 8);
        let raw: Vec<u8> = (0u8..32).collect();
        let encoded = encode_stream_data(&dict, &raw).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();
        assert_eq!(decoded, raw);
    }

    #[test]
    fn encode_stream_data_png_predictor_rgb_fixture_round_trip() {
        // Colors=3, BitsPerComponent=8, Columns=4 → row_bytes=12, rows=4 → 48 bytes
        let dict = png_predictor_dict_rgb(15, 4);
        let raw: Vec<u8> = (0u8..48).collect();
        let encoded = encode_stream_data(&dict, &raw).unwrap();
        let decoded = decode_stream_data(&dict, &encoded).unwrap();
        assert_eq!(decoded, raw);
    }
}