rustium-pdf 0.1.3

Pure-Rust PDF page-primitive extraction and rendering: glyphs with geometry, vector paths, images.
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
//! Stream filters.
//!
//! Content and font streams decode fully here. Image-only filters (DCT, JPX, CCITT, JBIG2) are
//! passed through untouched with the filter name reported, because the pixel decoders live in
//! the renderer and callers extracting structure never need the pixels.

use crate::error::{Error, Result};
use crate::object::{Dict, Object};

/// The most a single stream may decompress to.
///
/// Flate and LZW both have compression ratios of a thousand to one and better, so a few kilobytes
/// of hostile input can otherwise ask for gigabytes of memory. Real streams are far smaller than
/// this; a stream that reaches the cap is truncated there, the same as a corrupt tail, rather than
/// failing the document.
const MAX_DECOMPRESSED: usize = 256 << 20;

/// The result of running a stream's filter chain.
pub struct Decoded {
    pub data: Vec<u8>,
    /// `Some(name)` when the chain ended at an image codec the byte-level filters do not
    /// decode; `data` is then that codec's compressed payload.
    pub image_filter: Option<String>,
}

/// Applies a stream's `/Filter` chain to `data`.
///
/// `resolve` maps indirect objects to direct ones, since `/Filter`, `/DecodeParms` and their
/// members may all be references.
pub fn decode(dict: &Dict, data: &[u8], resolve: &dyn Fn(&Object) -> Object) -> Result<Decoded> {
    let filters = filter_names(dict, resolve);
    let parms = decode_parms(dict, filters.len(), resolve);

    let mut data = data.to_vec();
    for (i, name) in filters.iter().enumerate() {
        let parm = parms.get(i).cloned().unwrap_or_default();
        match name.as_str() {
            "FlateDecode" | "Fl" => {
                data = apply_predictor(&parm, inflate(&data)?, resolve)?;
            }
            "LZWDecode" | "LZW" => {
                let early = parm
                    .get("EarlyChange")
                    .map(resolve)
                    .and_then(|o| o.as_int())
                    .unwrap_or(1)
                    != 0;
                data = apply_predictor(&parm, lzw_decode(&data, early), resolve)?;
            }
            "ASCIIHexDecode" | "AHx" => data = ascii_hex_decode(&data),
            "ASCII85Decode" | "A85" => data = ascii85_decode(&data),
            "RunLengthDecode" | "RL" => data = run_length_decode(&data),
            "Crypt" => {
                // Identity crypt filters appear on already-decrypted streams; anything else was
                // handled at the document layer before filters ran.
            }
            "DCTDecode" | "DCT" | "JPXDecode" | "CCITTFaxDecode" | "CCF" | "JBIG2Decode" => {
                return Ok(Decoded {
                    data,
                    image_filter: Some(name.clone()),
                });
            }
            other => {
                return Err(Error::Parse(format!("unknown filter {other}")));
            }
        }
    }
    Ok(Decoded {
        data,
        image_filter: None,
    })
}

fn filter_names(dict: &Dict, resolve: &dyn Fn(&Object) -> Object) -> Vec<String> {
    let entry = dict
        .get("Filter")
        .or_else(|| dict.get("F").filter(|o| !matches!(o, Object::String(_))));
    let Some(entry) = entry else {
        return Vec::new();
    };
    match resolve(entry) {
        Object::Name(n) => vec![n],
        Object::Array(items) => items
            .iter()
            .filter_map(|o| resolve(o).as_name().map(str::to_owned))
            .collect(),
        _ => Vec::new(),
    }
}

fn decode_parms(dict: &Dict, n: usize, resolve: &dyn Fn(&Object) -> Object) -> Vec<Dict> {
    let entry = dict.get("DecodeParms").or_else(|| dict.get("DP"));
    let Some(entry) = entry else {
        return vec![Dict::new(); n];
    };
    match resolve(entry) {
        Object::Dict(d) => {
            let mut v = vec![Dict::new(); n];
            if let Some(slot) = v.first_mut() {
                *slot = d;
            }
            v
        }
        Object::Array(items) => (0..n)
            .map(|i| {
                items
                    .get(i)
                    .map(resolve)
                    .and_then(|o| o.as_dict().cloned())
                    .unwrap_or_default()
            })
            .collect(),
        _ => vec![Dict::new(); n],
    }
}

/// Inflates zlib or raw deflate data, keeping whatever decoded cleanly when the tail is corrupt.
///
/// Truncated and trailing-garbage Flate streams are common enough in the wild that failing hard
/// would reject otherwise fine documents. Output stops at [`MAX_DECOMPRESSED`].
fn inflate(data: &[u8]) -> Result<Vec<u8>> {
    use flate2::{Decompress, FlushDecompress, Status};

    // Skip leading whitespace some producers leave before the zlib header.
    let start = data
        .iter()
        .position(|&b| !b.is_ascii_whitespace())
        .unwrap_or(0);
    let data = &data[start..];

    let zlib = data.first().is_some_and(|&b| b & 0x0f == 8);
    let mut inflater = Decompress::new(zlib);
    let mut out = Vec::with_capacity(data.len().saturating_mul(4).min(1 << 20));
    let mut buf = vec![0u8; 64 * 1024];
    loop {
        let before_in = inflater.total_in();
        let before_out = inflater.total_out();
        let status = inflater.decompress(
            &data[inflater.total_in() as usize..],
            &mut buf,
            FlushDecompress::None,
        );
        let produced = (inflater.total_out() - before_out) as usize;
        let room = MAX_DECOMPRESSED - out.len();
        out.extend_from_slice(&buf[..produced.min(room)]);
        if produced >= room {
            // A zip bomb, or a stream far larger than anything a page needs; keep the prefix.
            break;
        }
        match status {
            Ok(Status::StreamEnd) => break,
            Ok(_) => {
                let consumed = inflater.total_in() - before_in;
                if consumed == 0 && produced == 0 {
                    break; // No progress: truncated stream.
                }
            }
            Err(_) => break, // Corrupt tail: keep what we have.
        }
    }
    if out.is_empty() && !data.is_empty() {
        return Err(Error::Parse("flate stream decoded to nothing".into()));
    }
    Ok(out)
}

fn apply_predictor(
    parm: &Dict,
    data: Vec<u8>,
    resolve: &dyn Fn(&Object) -> Object,
) -> Result<Vec<u8>> {
    let get = |k: &str, default: i64| {
        parm.get(k)
            .map(resolve)
            .and_then(|o| o.as_int())
            .unwrap_or(default)
    };
    let predictor = get("Predictor", 1);
    if predictor <= 1 {
        return Ok(data);
    }
    let colors = get("Colors", 1).max(1) as usize;
    let bpc = get("BitsPerComponent", 8).max(1) as usize;
    let columns = get("Columns", 1).max(1) as usize;
    let bpp = (colors * bpc).div_ceil(8); // bytes per pixel, min 1
    let row_len = (colors * bpc * columns).div_ceil(8);

    if predictor == 2 {
        return Ok(tiff_predictor(data, colors, bpc, columns));
    }

    // PNG predictors: each row is prefixed with its filter type.
    let stride = row_len + 1;
    let rows = data.len() / stride;
    let mut out = Vec::with_capacity(rows * row_len);
    let mut prev = vec![0u8; row_len];
    for r in 0..rows {
        let row = &data[r * stride..(r + 1) * stride];
        let ft = row[0];
        let mut cur = row[1..].to_vec();
        match ft {
            0 => {}
            1 => {
                for i in bpp..row_len {
                    cur[i] = cur[i].wrapping_add(cur[i - bpp]);
                }
            }
            2 => {
                for i in 0..row_len {
                    cur[i] = cur[i].wrapping_add(prev[i]);
                }
            }
            3 => {
                for i in 0..row_len {
                    let left = if i >= bpp { cur[i - bpp] as u32 } else { 0 };
                    cur[i] = cur[i].wrapping_add(((left + prev[i] as u32) / 2) as u8);
                }
            }
            4 => {
                for i in 0..row_len {
                    let a = if i >= bpp { cur[i - bpp] as i32 } else { 0 };
                    let b = prev[i] as i32;
                    let c = if i >= bpp { prev[i - bpp] as i32 } else { 0 };
                    let p = a + b - c;
                    let (pa, pb, pc) = ((p - a).abs(), (p - b).abs(), (p - c).abs());
                    let pred = if pa <= pb && pa <= pc {
                        a
                    } else if pb <= pc {
                        b
                    } else {
                        c
                    };
                    cur[i] = cur[i].wrapping_add(pred as u8);
                }
            }
            _ => return Err(Error::Parse(format!("png predictor row type {ft}"))),
        }
        out.extend_from_slice(&cur);
        prev = cur;
    }
    Ok(out)
}

fn tiff_predictor(mut data: Vec<u8>, colors: usize, bpc: usize, columns: usize) -> Vec<u8> {
    if bpc != 8 {
        // Sub-byte TIFF prediction is vanishingly rare; return as-is rather than corrupt.
        return data;
    }
    let row_len = colors * columns;
    for row in data.chunks_mut(row_len) {
        for i in colors..row.len() {
            row[i] = row[i].wrapping_add(row[i - colors]);
        }
    }
    data
}

/// Decodes LZW, stopping at [`MAX_DECOMPRESSED`] output bytes.
fn lzw_decode(data: &[u8], early_change: bool) -> Vec<u8> {
    const CLEAR: u16 = 256;
    const EOD: u16 = 257;

    let mut out = Vec::with_capacity(data.len().saturating_mul(3).min(1 << 20));
    let mut table: Vec<Vec<u8>> = Vec::new();
    let reset = |table: &mut Vec<Vec<u8>>| {
        table.clear();
        for b in 0u16..256 {
            table.push(vec![b as u8]);
        }
        table.push(Vec::new()); // 256: clear
        table.push(Vec::new()); // 257: eod
    };
    reset(&mut table);

    let mut bits = 9usize;
    let mut acc = 0u32;
    let mut nbits = 0usize;
    let mut prev: Option<u16> = None;

    for &byte in data {
        acc = (acc << 8) | byte as u32;
        nbits += 8;
        while nbits >= bits {
            let code = ((acc >> (nbits - bits)) & ((1 << bits) - 1)) as u16;
            nbits -= bits;

            if code == CLEAR {
                reset(&mut table);
                bits = 9;
                prev = None;
                continue;
            }
            if code == EOD {
                return out;
            }
            let entry: Vec<u8> = if (code as usize) < table.len() {
                table[code as usize].clone()
            } else if let Some(p) = prev {
                // KwKwK case.
                let mut e = table[p as usize].clone();
                if let Some(&f) = table[p as usize].first() {
                    e.push(f);
                }
                e
            } else {
                return out; // Corrupt.
            };
            let room = MAX_DECOMPRESSED - out.len();
            if entry.len() >= room {
                // A bomb, or a stream far larger than anything a page needs; keep the prefix.
                out.extend_from_slice(&entry[..room]);
                return out;
            }
            out.extend_from_slice(&entry);
            if let Some(p) = prev {
                let mut ne = table[p as usize].clone();
                if let Some(&f) = entry.first() {
                    ne.push(f);
                }
                table.push(ne);
            }
            prev = Some(code);

            let limit = if early_change { 1 } else { 0 };
            if table.len() + limit >= (1 << bits) && bits < 12 {
                bits += 1;
            }
        }
    }
    out
}

fn ascii_hex_decode(data: &[u8]) -> Vec<u8> {
    let mut out = Vec::with_capacity(data.len() / 2);
    let mut hi: Option<u8> = None;
    for &b in data {
        if b == b'>' {
            break;
        }
        let Some(v) = crate::lexer::hex_val(b) else {
            continue;
        };
        match hi.take() {
            Some(h) => out.push(h * 16 + v),
            None => hi = Some(v),
        }
    }
    if let Some(h) = hi {
        out.push(h * 16);
    }
    out
}

fn ascii85_decode(data: &[u8]) -> Vec<u8> {
    let mut out = Vec::with_capacity(data.len() * 4 / 5);
    let mut group = [0u8; 5];
    let mut n = 0usize;
    let mut i = 0usize;
    // Optional <~ prefix.
    if data.len() >= 2 && &data[0..2] == b"<~" {
        i = 2;
    }
    while i < data.len() {
        let b = data[i];
        i += 1;
        match b {
            b'~' => break,
            b'z' if n == 0 => out.extend_from_slice(&[0, 0, 0, 0]),
            b'!'..=b'u' => {
                group[n] = b - b'!';
                n += 1;
                if n == 5 {
                    let mut v = 0u32;
                    for &g in &group {
                        v = v.wrapping_mul(85).wrapping_add(g as u32);
                    }
                    out.extend_from_slice(&v.to_be_bytes());
                    n = 0;
                }
            }
            _ => {} // Whitespace and junk.
        }
    }
    if n > 0 {
        // Pad with 'u' (84) and keep n-1 bytes.
        for slot in group.iter_mut().skip(n) {
            *slot = 84;
        }
        let mut v = 0u32;
        for &g in &group {
            v = v.wrapping_mul(85).wrapping_add(g as u32);
        }
        out.extend_from_slice(&v.to_be_bytes()[..n - 1]);
    }
    out
}

fn run_length_decode(data: &[u8]) -> Vec<u8> {
    let mut out = Vec::with_capacity(data.len() * 2);
    let mut i = 0usize;
    while i < data.len() {
        let len = data[i];
        i += 1;
        match len {
            0..=127 => {
                let take = (len as usize + 1).min(data.len() - i);
                out.extend_from_slice(&data[i..i + take]);
                i += take;
            }
            128 => break,
            _ => {
                if i < data.len() {
                    out.extend(std::iter::repeat_n(data[i], 257 - len as usize));
                    i += 1;
                }
            }
        }
    }
    out
}

#[cfg(test)]
mod tests {
    use super::*;

    fn no_resolve(o: &Object) -> Object {
        o.clone()
    }

    #[test]
    fn flate_round_trip() {
        use flate2::write::ZlibEncoder;
        use flate2::Compression;
        use std::io::Write;
        let payload = b"BT /F1 12 Tf (Hello) Tj ET".repeat(100);
        let mut enc = ZlibEncoder::new(Vec::new(), Compression::default());
        enc.write_all(&payload).unwrap();
        let compressed = enc.finish().unwrap();

        let mut dict = Dict::new();
        dict.insert("Filter", Object::Name("FlateDecode".into()));
        let d = decode(&dict, &compressed, &no_resolve).unwrap();
        assert_eq!(d.data, payload);
        assert!(d.image_filter.is_none());
    }

    #[test]
    fn truncated_flate_keeps_prefix() {
        use flate2::write::ZlibEncoder;
        use flate2::Compression;
        use std::io::Write;
        let payload = vec![7u8; 100_000];
        let mut enc = ZlibEncoder::new(Vec::new(), Compression::default());
        enc.write_all(&payload).unwrap();
        let mut compressed = enc.finish().unwrap();
        compressed.truncate(compressed.len() / 2);

        let mut dict = Dict::new();
        dict.insert("Filter", Object::Name("FlateDecode".into()));
        let d = decode(&dict, &compressed, &no_resolve).unwrap();
        assert!(!d.data.is_empty());
        assert!(d.data.iter().all(|&b| b == 7));
    }

    #[test]
    fn flate_bomb_is_capped() {
        use flate2::write::ZlibEncoder;
        use flate2::Compression;
        use std::io::Write;

        // Half a gigabyte of zeros in about a megabyte of stream: the shape of a zip bomb.
        let chunk = vec![0u8; 1 << 20];
        let mut enc = ZlibEncoder::new(Vec::new(), Compression::best());
        for _ in 0..512 {
            enc.write_all(&chunk).unwrap();
        }
        let compressed = enc.finish().unwrap();
        assert!(compressed.len() < 1 << 21, "test bomb should be small");

        let mut dict = Dict::new();
        dict.insert("Filter", Object::Name("FlateDecode".into()));
        let d = decode(&dict, &compressed, &no_resolve).unwrap();
        assert_eq!(d.data.len(), MAX_DECOMPRESSED);
    }

    #[test]
    fn ascii_filters() {
        assert_eq!(ascii_hex_decode(b"48656C6C6F>"), b"Hello");
        assert_eq!(ascii85_decode(b"87cURD_*#T~>"), b"Hello, w");
        assert_eq!(ascii85_decode(b"z~>"), [0, 0, 0, 0]);
    }

    #[test]
    fn run_length() {
        // 2 literal bytes, then 0xFF repeated 4 times (257-253), then EOD.
        assert_eq!(
            run_length_decode(&[1, b'a', b'b', 253, 0xFF, 128]),
            b"ab\xff\xff\xff\xff"
        );
    }

    #[test]
    fn lzw_known_vector() {
        // The example from the PDF spec, 7.4.4.2: decodes to 45 45 45 45 45 65 45 45 45 66.
        let encoded = [0x80, 0x0B, 0x60, 0x50, 0x22, 0x0C, 0x0C, 0x85, 0x01];
        assert_eq!(
            lzw_decode(&encoded, true),
            [45, 45, 45, 45, 45, 65, 45, 45, 45, 66]
        );
    }

    #[test]
    fn png_up_predictor() {
        // Two rows of 3 bytes, filter type 2 (Up).
        let mut dict = Dict::new();
        dict.insert("Filter", Object::Name("FlateDecode".into()));
        let mut parms = Dict::new();
        parms.insert("Predictor", Object::Int(12));
        parms.insert("Columns", Object::Int(3));
        dict.insert("DecodeParms", Object::Dict(parms));

        use flate2::write::ZlibEncoder;
        use flate2::Compression;
        use std::io::Write;
        let raw = [2u8, 1, 2, 3, 2, 1, 1, 1];
        let mut enc = ZlibEncoder::new(Vec::new(), Compression::default());
        enc.write_all(&raw).unwrap();
        let compressed = enc.finish().unwrap();

        let d = decode(&dict, &compressed, &no_resolve).unwrap();
        assert_eq!(d.data, [1, 2, 3, 2, 3, 4]);
    }

    #[test]
    fn dct_passes_through() {
        let mut dict = Dict::new();
        dict.insert("Filter", Object::Name("DCTDecode".into()));
        let d = decode(&dict, b"\xff\xd8jpegdata", &no_resolve).unwrap();
        assert_eq!(d.image_filter.as_deref(), Some("DCTDecode"));
        assert_eq!(d.data, b"\xff\xd8jpegdata");
    }
}