pdf_oxide 0.3.59

The fastest Rust PDF library with text extraction: 0.8ms mean, 100% pass rate on 3,830 PDFs. 5× faster than pdf_extract, 17× faster than oxidize_pdf. Extract, create, and edit 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
//! Split a PDF into multiple PDFs at outline (bookmark) boundaries
//! (issue #482).
//!
//! This module is pure orchestration: it turns a parsed outline +
//! page count into a deterministic set of half-open page-range
//! segments with collision-free, filesystem-safe file stems. It does
//! **no** low-level PDF writing — byte production delegates to the
//! existing editor page-range extraction (wired in a later increment).
//!
//! The three pure stages — `flatten_outline`, `collect_split_points`,
//! `build_segments` — are independently unit-tested without any PDF
//! fixture (SRP / testability).

use std::collections::HashMap;

use crate::error::{Error, Result};
use crate::filename::{slugify_title_with, DEFAULT_MAX_SLUG_BYTES};
use crate::outline::{Destination, OutlineItem};

/// Which outline depth(s) become split points.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
pub enum BookmarkLevel {
    /// Only top-level (depth 1) items. Default (qpdf "level-0").
    #[default]
    TopLevel,
    /// Items at depth `<= n` (1-based; `UpTo(1)` == `TopLevel`).
    UpTo(u32),
    /// Every item at any depth.
    All,
}

impl BookmarkLevel {
    /// Map the CLI/binding `u32` (`0` = all, `1` = top-level, `n` =
    /// up-to-depth-n) to a [`BookmarkLevel`].
    pub fn from_u32(n: u32) -> Self {
        match n {
            0 => BookmarkLevel::All,
            1 => BookmarkLevel::TopLevel,
            n => BookmarkLevel::UpTo(n),
        }
    }

    /// Whether a (1-based) outline `depth` is selected by this level.
    fn includes(self, depth: u32) -> bool {
        match self {
            BookmarkLevel::TopLevel => depth == 1,
            BookmarkLevel::UpTo(n) => depth <= n.max(1),
            BookmarkLevel::All => true,
        }
    }
}

/// Options controlling a bookmark split. Construct with [`Default`]
/// (front-matter on, 80-byte slug budget).
#[derive(Debug, Clone)]
pub struct SplitByBookmarksOptions {
    /// Only split at bookmarks whose trimmed title starts with this
    /// prefix. `None` => every selected-level bookmark.
    pub title_prefix: Option<String>,
    /// Case-insensitive prefix match.
    pub ignore_case: bool,
    /// Outline depth selection.
    pub level: BookmarkLevel,
    /// Emit the pages before the first split point as a leading
    /// `front-matter` segment (only when non-empty).
    pub include_front_matter: bool,
    /// Use page labels (§12.4.2) for filenames instead of ordinals.
    /// (Label population is a later increment; the flag is plumbed.)
    pub use_page_labels: bool,
    /// Max slug byte length.
    pub max_slug_bytes: usize,
}

impl Default for SplitByBookmarksOptions {
    fn default() -> Self {
        Self {
            title_prefix: None,
            ignore_case: false,
            level: BookmarkLevel::TopLevel,
            include_front_matter: true,
            use_page_labels: false,
            max_slug_bytes: DEFAULT_MAX_SLUG_BYTES,
        }
    }
}

/// One planned output segment (also a `--dry-run` row).
///
/// `Serialize`/`Deserialize` so every binding (WASM, C-ABI, Go, C#)
/// marshals it through the one JSON shape and the parity goldens
/// compare a single source of truth (DRY — foundation §5.3).
#[derive(Debug, Clone, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
pub struct BookmarkSegment {
    /// 1-based segment ordinal across the whole document.
    pub index: usize,
    /// Inclusive start page (0-based).
    pub start_page: usize,
    /// Exclusive end page (0-based); range is `start_page..end_page`.
    pub end_page: usize,
    /// Source bookmark title (`None` for the front-matter segment).
    pub title: Option<String>,
    /// Final de-duplicated, sanitised file stem (no extension).
    pub file_stem: String,
    /// Page label of `start_page`, if available and requested.
    pub page_label: Option<String>,
}

/// DFS pre-order flatten of the outline, applying the depth filter.
/// Returns `(title, Some(page_index))` for resolved destinations and
/// `(title, None)` for unresolvable ones (the caller drops those).
/// Children are always traversed; only items at a selected depth are
/// emitted.
pub fn flatten_outline(
    items: &[OutlineItem],
    level: BookmarkLevel,
) -> Vec<(String, Option<usize>)> {
    fn walk(
        items: &[OutlineItem],
        depth: u32,
        level: BookmarkLevel,
        out: &mut Vec<(String, Option<usize>)>,
    ) {
        for it in items {
            if level.includes(depth) {
                let page = match &it.dest {
                    Some(Destination::PageIndex(i)) => Some(*i),
                    _ => None,
                };
                out.push((it.title.clone(), page));
            }
            if !it.children.is_empty() {
                walk(&it.children, depth + 1, level, out);
            }
        }
    }
    let mut out = Vec::new();
    walk(items, 1, level, &mut out);
    out
}

/// Apply the prefix filter, drop unresolvable destinations, sort by
/// page ascending (stable — document order breaks ties), and dedupe
/// by page (first title in document order wins).
pub fn collect_split_points(
    flat: Vec<(String, Option<usize>)>,
    opts: &SplitByBookmarksOptions,
) -> Vec<(usize, String)> {
    let matches_prefix = |title: &str| -> bool {
        match &opts.title_prefix {
            None => true,
            Some(p) => {
                let t = title.trim();
                if opts.ignore_case {
                    t.to_lowercase().starts_with(&p.to_lowercase())
                } else {
                    t.starts_with(p.as_str())
                }
            },
        }
    };

    let mut points: Vec<(usize, String)> = flat
        .into_iter()
        .filter(|(t, _)| matches_prefix(t))
        .filter_map(|(t, p)| p.map(|page| (page, t)))
        .collect();

    // Stable sort by page keeps document order for equal pages, so
    // the first-seen title wins the dedupe below.
    points.sort_by_key(|(page, _)| *page);
    let mut seen = std::collections::HashSet::new();
    points.retain(|(page, _)| seen.insert(*page));
    points
}

/// Form half-open page-range segments from sorted-unique split points
/// over a `page_count`-page document, adding an optional leading
/// front-matter segment and collision-free sanitised file stems.
///
/// # Errors
/// [`Error::InvalidOperation`] if no usable split point remains (all
/// filtered out, unresolvable, or past EOF) — the caller turns this
/// into the user-facing "matched nothing" / "no outline" message.
pub fn build_segments(
    points: &[(usize, String)],
    page_count: usize,
    opts: &SplitByBookmarksOptions,
) -> Result<Vec<BookmarkSegment>> {
    // Drop points at/after EOF (invalid/dangling destinations).
    let pts: Vec<&(usize, String)> = points.iter().filter(|(p, _)| *p < page_count).collect();
    if pts.is_empty() {
        return Err(Error::InvalidOperation(
            "no resolvable bookmark split points (empty outline, prefix matched \
             nothing, or all destinations unresolvable/past EOF)"
                .to_string(),
        ));
    }

    let mut segs: Vec<BookmarkSegment> = Vec::new();

    // Optional leading front-matter [0, p0).
    if opts.include_front_matter && pts[0].0 > 0 {
        segs.push(BookmarkSegment {
            index: 0, // re-numbered below
            start_page: 0,
            end_page: pts[0].0,
            title: None,
            file_stem: "front-matter".to_string(),
            page_label: None,
        });
    }

    for (i, (start, title)) in pts.iter().map(|p| (p.0, &p.1)).enumerate() {
        let end = pts.get(i + 1).map(|p| p.0).unwrap_or(page_count);
        segs.push(BookmarkSegment {
            index: 0,
            start_page: start,
            end_page: end,
            title: Some(title.clone()),
            file_stem: slugify_title_with(title, opts.max_slug_bytes),
            page_label: None,
        });
    }

    // 1-based ordinal + collision-free stems (` (2)`, ` (3)` …, the
    // CoolUtils-style human-readable form; suffix applied *after*
    // slugify so the base is already filesystem-safe).
    let mut counts: HashMap<String, usize> = HashMap::new();
    for (idx, seg) in segs.iter_mut().enumerate() {
        seg.index = idx + 1;
        let n = counts.entry(seg.file_stem.clone()).or_insert(0);
        *n += 1;
        if *n > 1 {
            seg.file_stem = format!("{} ({})", seg.file_stem, *n);
        }
    }

    Ok(segs)
}

/// Plan the split for `doc` without producing any bytes — cheap;
/// ideal for `--dry-run` and for binding callers that drive their own
/// writes.
///
/// # Errors
/// [`Error::InvalidOperation`] if the document has no outline, or no
/// selected bookmark resolves to a usable split point.
pub fn plan_split_by_bookmarks(
    doc: &crate::document::PdfDocument,
    opts: &SplitByBookmarksOptions,
) -> Result<Vec<BookmarkSegment>> {
    let outline = doc.get_outline()?.ok_or_else(|| {
        Error::InvalidOperation(
            "document has no bookmarks/outline; use a plain per-page split instead".to_string(),
        )
    })?;
    let flat = flatten_outline(&outline, opts.level);
    let points = collect_split_points(flat, opts);
    let page_count = doc.page_count()?;
    build_segments(&points, page_count, opts)
}

/// Execute the split: returns each segment paired with its PDF bytes
/// (parallel to [`plan_split_by_bookmarks`]). The source is not
/// modified. Segment ranges are `[start_page, end_page)` and are fed
/// directly to the editor's half-open page-range extraction.
///
/// # Errors
/// Propagates planning errors and any extraction failure.
pub fn split_by_bookmarks_to_bytes(
    src_bytes: &[u8],
    opts: &SplitByBookmarksOptions,
) -> Result<Vec<(BookmarkSegment, Vec<u8>)>> {
    let doc = crate::document::PdfDocument::from_bytes(src_bytes.to_vec())?;
    let segments = plan_split_by_bookmarks(&doc, opts)?;

    let mut editor = crate::editor::DocumentEditor::from_bytes(src_bytes.to_vec())?;
    let ranges: Vec<(usize, usize)> = segments
        .iter()
        .map(|s| (s.start_page, s.end_page))
        .collect();
    let blobs = editor.extract_page_ranges_to_bytes(&ranges)?;

    Ok(segments.into_iter().zip(blobs).collect())
}

/// Split `src_path` and write each segment to `dir` as `{file_stem}.pdf`,
/// returning the written paths in document order. Creates `dir` if
/// missing. The CLI uses this.
///
/// NOTE (deviation from #482 plan §4): the planned `password` parameter
/// is omitted in v0.3.50. This helper `std::fs::read`s `src_path` and
/// operates on **raw bytes** — it does NOT decrypt, so encrypted inputs
/// are unsupported here. The CLI does not silently ignore a password:
/// `pdf-oxide split --by-bookmarks` rejects `--password` fail-closed
/// with an actionable message. A caller needing to split an encrypted
/// document must decrypt it first and pass the decrypted bytes/file.
/// Keeping this a pure, testable byte→files step is intentional;
/// encrypted-output handling is out of scope per plan §9.
///
/// # Errors
/// Propagates planning errors and any filesystem error
/// ([`Error::Io`]).
pub fn split_by_bookmarks_to_dir(
    src_path: &std::path::Path,
    dir: &std::path::Path,
    opts: &SplitByBookmarksOptions,
) -> Result<Vec<std::path::PathBuf>> {
    let bytes = std::fs::read(src_path)?;
    let parts = split_by_bookmarks_to_bytes(&bytes, opts)?;
    std::fs::create_dir_all(dir)?;
    let mut written = Vec::with_capacity(parts.len());
    for (seg, blob) in parts {
        let path = dir.join(format!("{}.pdf", seg.file_stem));
        std::fs::write(&path, blob)?;
        written.push(path);
    }
    Ok(written)
}

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

    fn item(title: &str, page: Option<usize>, children: Vec<OutlineItem>) -> OutlineItem {
        OutlineItem {
            title: title.to_string(),
            dest: page.map(Destination::PageIndex),
            children,
        }
    }

    #[test]
    fn flatten_depth_filter() {
        let tree = vec![
            item("Ch1", Some(0), vec![item("S1.1", Some(1), vec![])]),
            item(
                "Ch2",
                Some(3),
                vec![item("S2.1", Some(4), vec![item("S2.1.1", Some(5), vec![])])],
            ),
        ];
        let top: Vec<_> = flatten_outline(&tree, BookmarkLevel::TopLevel)
            .into_iter()
            .map(|(t, _)| t)
            .collect();
        assert_eq!(top, vec!["Ch1", "Ch2"]);

        let upto2: Vec<_> = flatten_outline(&tree, BookmarkLevel::UpTo(2))
            .into_iter()
            .map(|(t, _)| t)
            .collect();
        assert_eq!(upto2, vec!["Ch1", "S1.1", "Ch2", "S2.1"]);

        let all: Vec<_> = flatten_outline(&tree, BookmarkLevel::All)
            .into_iter()
            .map(|(t, _)| t)
            .collect();
        assert_eq!(all, vec!["Ch1", "S1.1", "Ch2", "S2.1", "S2.1.1"]);
    }

    #[test]
    fn from_u32_mapping() {
        assert_eq!(BookmarkLevel::from_u32(0), BookmarkLevel::All);
        assert_eq!(BookmarkLevel::from_u32(1), BookmarkLevel::TopLevel);
        assert_eq!(BookmarkLevel::from_u32(3), BookmarkLevel::UpTo(3));
    }

    #[test]
    fn collect_prefix_filter_case_and_sort_dedupe() {
        let flat = vec![
            ("Chapter 2".to_string(), Some(5)),
            ("chapter 1".to_string(), Some(2)),
            ("Appendix".to_string(), Some(8)),
            ("Chapter 1 (dup page)".to_string(), Some(2)), // same page as ch1
            ("Bad".to_string(), None),                     // unresolvable -> dropped
        ];
        let opts = SplitByBookmarksOptions {
            title_prefix: Some("Chapter".to_string()),
            ..Default::default()
        };
        // Case-sensitive: lowercase "chapter 1" excluded; both
        // "Chapter …" entries kept (incl. the page-2 dup), sorted by
        // page ascending.
        let cs = collect_split_points(flat.clone(), &opts);
        assert_eq!(
            cs,
            vec![
                (2, "Chapter 1 (dup page)".to_string()),
                (5, "Chapter 2".to_string())
            ]
        );

        // Case-insensitive: both chapters; sorted by page; page 2
        // de-duped to the first document-order title.
        let ci = collect_split_points(
            flat,
            &SplitByBookmarksOptions {
                ignore_case: true,
                ..opts.clone()
            },
        );
        assert_eq!(ci, vec![(2, "chapter 1".to_string()), (5, "Chapter 2".to_string())]);
    }

    #[test]
    fn build_segments_ranges_frontmatter_and_eof() {
        let points = vec![(2, "Ch1".to_string()), (5, "Ch2".to_string())];
        let segs = build_segments(&points, 9, &SplitByBookmarksOptions::default()).unwrap();
        // front-matter [0,2), Ch1 [2,5), Ch2 [5,9)
        assert_eq!(segs.len(), 3);
        assert_eq!((segs[0].start_page, segs[0].end_page), (0, 2));
        assert_eq!(segs[0].file_stem, "front-matter");
        assert_eq!(segs[0].title, None);
        assert_eq!((segs[1].start_page, segs[1].end_page), (2, 5));
        assert_eq!(segs[1].file_stem, "Ch1");
        assert_eq!((segs[2].start_page, segs[2].end_page), (5, 9));
        assert_eq!((segs[0].index, segs[1].index, segs[2].index), (1, 2, 3));

        // No front-matter when first point is page 0.
        let s2 = build_segments(
            &[(0, "A".to_string()), (3, "B".to_string())],
            6,
            &SplitByBookmarksOptions::default(),
        )
        .unwrap();
        assert_eq!(s2.len(), 2);
        assert_eq!((s2[0].start_page, s2[0].end_page), (0, 3));

        // include_front_matter=false drops the cover.
        let s3 = build_segments(
            &points,
            9,
            &SplitByBookmarksOptions {
                include_front_matter: false,
                ..Default::default()
            },
        )
        .unwrap();
        assert_eq!(s3.len(), 2);
        assert_eq!(s3[0].file_stem, "Ch1");
    }

    #[test]
    fn build_segments_dedupes_stems_and_drops_past_eof() {
        let points = vec![
            (0, "Record".to_string()),
            (2, "Record".to_string()),
            (4, "Record".to_string()),
            (99, "Past EOF".to_string()), // page >= count -> dropped
        ];
        let segs = build_segments(&points, 6, &SplitByBookmarksOptions::default()).unwrap();
        assert_eq!(segs.len(), 3);
        assert_eq!(segs[0].file_stem, "Record");
        assert_eq!(segs[1].file_stem, "Record (2)");
        assert_eq!(segs[2].file_stem, "Record (3)");
        assert_eq!((segs[2].start_page, segs[2].end_page), (4, 6));
    }

    #[test]
    fn build_segments_empty_is_invalid_operation() {
        let err = build_segments(&[], 10, &SplitByBookmarksOptions::default()).unwrap_err();
        assert!(matches!(err, Error::InvalidOperation(_)));
        // All points past EOF -> same error.
        let err2 = build_segments(&[(20, "X".to_string())], 5, &SplitByBookmarksOptions::default())
            .unwrap_err();
        assert!(matches!(err2, Error::InvalidOperation(_)));
    }

    /// Cross-binding wire-shape golden. Every #482 surface (C-ABI →
    /// Go/C#/Node, WASM, Python) marshals `BookmarkSegment` through
    /// this exact JSON shape; a field rename/reorder would silently
    /// break all of them at once. This is the single source of truth
    /// the binding parity tests compare against (foundation §5.3).
    #[test]
    fn bookmark_segment_json_wire_shape_is_frozen() {
        let titled = BookmarkSegment {
            index: 2,
            start_page: 3,
            end_page: 7,
            title: Some("Chapter 1".to_string()),
            file_stem: "Chapter-1".to_string(),
            page_label: None,
        };
        assert_eq!(
            serde_json::to_string(&titled).unwrap(),
            r#"{"index":2,"start_page":3,"end_page":7,"title":"Chapter 1","file_stem":"Chapter-1","page_label":null}"#
        );
        // Front-matter segment: null title.
        let fm = BookmarkSegment {
            index: 1,
            start_page: 0,
            end_page: 3,
            title: None,
            file_stem: "front-matter".to_string(),
            page_label: Some("iii".to_string()),
        };
        assert_eq!(
            serde_json::to_string(&fm).unwrap(),
            r#"{"index":1,"start_page":0,"end_page":3,"title":null,"file_stem":"front-matter","page_label":"iii"}"#
        );
        // Round-trips (Deserialize parity for any binding that reads back).
        let back: BookmarkSegment =
            serde_json::from_str(&serde_json::to_string(&titled).unwrap()).unwrap();
        assert_eq!(back, titled);
    }
}