docxide-pdf 0.9.0

CLI and Library for converting DOCX files to PDF, matching Microsoft Word's output as closely as possible
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
use std::collections::HashMap;

use pdf_writer::{Content, Name, Rect, Str};

use crate::fonts::{FontEntry, encode_as_gids, font_key, font_key_buf, to_winansi_bytes};
use crate::model::{Alignment, Run, TabAlignment, TabStop, VertAlign};

/// Split text into (preceding_space_count, word) pairs.
/// Leading and inter-word spaces are counted. Trailing spaces (after the last word)
/// are handled separately by the caller.
fn split_preserving_spaces(text: &str) -> Vec<(usize, &str)> {
    let mut result = Vec::new();
    let mut chars = text.char_indices().peekable();
    let mut space_count: usize = 0;

    while let Some(&(i, c)) = chars.peek() {
        if c.is_whitespace() {
            space_count += 1;
            chars.next();
        } else {
            let start = i;
            while let Some(&(_, c)) = chars.peek() {
                if c.is_whitespace() {
                    break;
                }
                chars.next();
            }
            let end = chars.peek().map(|&(i, _)| i).unwrap_or(text.len());
            result.push((space_count, &text[start..end]));
            space_count = 0;
        }
    }

    result
}

pub(super) struct WordChunk {
    pub(super) pdf_font: String,
    pub(super) text: String,
    pub(super) font_size: f32,
    pub(super) color: Option<[u8; 3]>,
    pub(super) highlight: Option<[u8; 3]>,
    pub(super) x_offset: f32, // x relative to line start
    pub(super) width: f32,
    pub(super) underline: bool,
    pub(super) strikethrough: bool,
    pub(super) dstrike: bool,
    pub(super) char_spacing: f32,
    pub(super) text_scale: f32, // percentage, 100.0 = normal
    pub(super) y_offset: f32,   // vertical offset for superscript/subscript
    pub(super) hyperlink_url: Option<String>,
    pub(super) inline_image_name: Option<String>,
    pub(super) inline_image_height: f32,
}

pub(super) struct LinkAnnotation {
    pub(super) rect: Rect,
    pub(super) url: String,
}

pub(super) struct TextLine {
    pub(super) chunks: Vec<WordChunk>,
    pub(super) total_width: f32,
}

/// True when a paragraph has no visible text (may still have phantom font-info runs).
pub(super) fn is_text_empty(runs: &[Run]) -> bool {
    runs.iter()
        .all(|r| r.vanish || (r.text.is_empty() && !r.is_tab && r.inline_image.is_none()))
}

fn effective_font_size(run: &Run) -> f32 {
    let base = match run.vertical_align {
        VertAlign::Superscript | VertAlign::Subscript => run.font_size * 0.58,
        VertAlign::Baseline => run.font_size,
    };
    if run.small_caps {
        (base - 2.0).max(1.0)
    } else {
        base
    }
}

/// Returns the text to use for layout/rendering, applying caps/smallCaps transforms.
fn effective_text(run: &Run) -> std::borrow::Cow<'_, str> {
    if run.caps || run.small_caps {
        std::borrow::Cow::Owned(run.text.to_uppercase())
    } else {
        std::borrow::Cow::Borrowed(&run.text)
    }
}

fn vert_y_offset(run: &Run) -> f32 {
    match run.vertical_align {
        VertAlign::Superscript => run.font_size * 0.35,
        VertAlign::Subscript => -run.font_size * 0.14,
        VertAlign::Baseline => 0.0,
    }
}

const DEFAULT_TAB_INTERVAL: f32 = 36.0; // 0.5 inches

fn finish_line(chunks: &mut Vec<WordChunk>) -> TextLine {
    let total_width = chunks.last().map(|c| c.x_offset + c.width).unwrap_or(0.0);
    TextLine {
        chunks: std::mem::take(chunks),
        total_width,
    }
}

/// Layout runs into wrapped lines.
/// Handles cross-run contiguous text correctly: no space is inserted between
/// runs unless the preceding text ended with whitespace or the new run starts
/// with whitespace (e.g., "bold" + ", " → "bold," not "bold ,").
pub(super) fn build_paragraph_lines(
    runs: &[Run],
    seen_fonts: &HashMap<String, FontEntry>,
    max_width: f32,
    first_line_hanging: f32,
    inline_image_names: &HashMap<usize, String>,
) -> Vec<TextLine> {
    let mut lines: Vec<TextLine> = Vec::new();
    let mut current_chunks: Vec<WordChunk> = Vec::new();
    let mut current_x: f32 = 0.0;
    let mut pending_space_w: f32 = 0.0;
    let mut key_buf = String::new();

    for (run_idx, run) in runs.iter().enumerate() {
        if run.vanish || run.is_tab {
            continue; // vanished runs hidden; tabs handled in build_tabbed_line
        }

        // Handle inline images as single block elements in the line
        if let Some(img) = &run.inline_image {
            if let Some(pdf_name) = inline_image_names.get(&run_idx) {
                let img_w = img.display_width;
                let need_space = !current_chunks.is_empty() && pending_space_w > 0.0;
                let proposed_x = if need_space {
                    current_x + pending_space_w
                } else {
                    current_x
                };

                let line_max = if lines.is_empty() {
                    max_width + first_line_hanging
                } else {
                    max_width
                };
                if !current_chunks.is_empty() && proposed_x + img_w > line_max {
                    lines.push(finish_line(&mut current_chunks));
                    current_x = 0.0;
                } else {
                    current_x = proposed_x;
                }
                pending_space_w = 0.0;

                current_chunks.push(WordChunk {
                    pdf_font: String::new(),
                    text: String::new(),
                    font_size: run.font_size,
                    color: None,
                    highlight: None,
                    x_offset: current_x,
                    width: img_w,
                    underline: false,
                    strikethrough: false,
                    dstrike: false,
                    char_spacing: 0.0,
                    text_scale: 100.0,
                    y_offset: 0.0,
                    hyperlink_url: None,
                    inline_image_name: Some(pdf_name.clone()),
                    inline_image_height: img.display_height,
                });
                current_x += img_w;
            }
            continue;
        }

        let key = font_key_buf(run, &mut key_buf);
        let entry = seen_fonts.get(key).expect("font registered");
        let eff_fs = effective_font_size(run);
        let space_w = entry.space_width(eff_fs);
        let text = effective_text(run);
        let y_off = vert_y_offset(run);

        let cs = run.char_spacing;
        let ts = run.text_scale / 100.0;
        let space_w_cs = space_w * ts + cs;

        for (space_count, word) in split_preserving_spaces(&text) {
            pending_space_w += space_count as f32 * space_w_cs;

            let char_count = word.chars().count();
            let kern = run.kern_threshold.is_some_and(|t| eff_fs >= t);
            let ww = entry.word_width(word, eff_fs, kern) * ts + cs * char_count as f32;

            let need_space = !current_chunks.is_empty() && pending_space_w > 0.0;

            let proposed_x = if need_space {
                current_x + pending_space_w
            } else {
                current_x
            };

            let line_max = if lines.is_empty() {
                max_width + first_line_hanging
            } else {
                max_width
            };
            if !current_chunks.is_empty() && proposed_x + ww > line_max {
                lines.push(finish_line(&mut current_chunks));
                current_x = 0.0;
            } else {
                current_x = proposed_x;
            }
            pending_space_w = 0.0;

            current_chunks.push(WordChunk {
                pdf_font: entry.pdf_name.clone(),
                text: word.to_string(),
                font_size: eff_fs,
                color: run.color,
                highlight: run.highlight,
                x_offset: current_x,
                width: ww,
                underline: run.underline,
                strikethrough: run.strikethrough,
                dstrike: run.dstrike,
                char_spacing: cs,
                text_scale: run.text_scale,
                y_offset: y_off,
                hyperlink_url: run.hyperlink_url.clone(),
                inline_image_name: None,
                inline_image_height: 0.0,
            });
            current_x += ww;
        }

        // Accumulate trailing whitespace for the next run
        let trailing_spaces = text.chars().rev().take_while(|c| c.is_whitespace()).count();
        pending_space_w += trailing_spaces as f32 * space_w_cs;
    }

    if !current_chunks.is_empty() {
        lines.push(finish_line(&mut current_chunks));
    }

    if lines.is_empty() {
        lines.push(TextLine {
            chunks: vec![],
            total_width: 0.0,
        });
    }
    lines
}

fn find_next_tab_stop(current_x: f32, tab_stops: &[TabStop], indent_left: f32) -> TabStop {
    let abs_x = current_x + indent_left;
    for stop in tab_stops {
        if stop.position > abs_x + 0.5 {
            return stop.clone();
        }
    }
    let next_default = ((abs_x / DEFAULT_TAB_INTERVAL).floor() + 1.0) * DEFAULT_TAB_INTERVAL;
    TabStop {
        position: next_default,
        alignment: TabAlignment::Left,
        leader: None,
    }
}

fn segment_width(runs: &[&Run], seen_fonts: &HashMap<String, FontEntry>) -> f32 {
    let mut w: f32 = 0.0;
    let mut first = true;
    let mut key_buf = String::new();
    for run in runs {
        let key = font_key_buf(run, &mut key_buf);
        let entry = seen_fonts.get(key).expect("font registered");
        let eff_fs = effective_font_size(run);
        let ts = run.text_scale / 100.0;
        let space_w = entry.space_width(eff_fs) * ts;
        let text = effective_text(run);
        for (i, word) in text.split_whitespace().enumerate() {
            if !first || i > 0 {
                w += space_w;
            }
            let kern = run.kern_threshold.is_some_and(|t| eff_fs >= t);
            w += entry.word_width(word, eff_fs, kern) * ts;
            first = false;
        }
    }
    w
}

fn decimal_before_width(runs: &[&Run], seen_fonts: &HashMap<String, FontEntry>) -> f32 {
    let texts: Vec<std::borrow::Cow<'_, str>> = runs.iter().map(|r| effective_text(r)).collect();
    let full_text: String = texts.iter().map(|t| t.as_ref()).collect();
    let before = if let Some(dot_pos) = full_text.find('.') {
        &full_text[..dot_pos]
    } else {
        &full_text
    };
    let mut w: f32 = 0.0;
    let mut chars_remaining = before.len();
    let mut key_buf = String::new();
    for (run, text) in runs.iter().zip(texts.iter()) {
        let key = font_key_buf(run, &mut key_buf);
        let entry = seen_fonts.get(key).expect("font registered");
        let eff_fs = effective_font_size(run);
        let text_to_measure = if text.len() <= chars_remaining {
            chars_remaining -= text.len();
            text.as_ref()
        } else {
            let s = &text[..chars_remaining];
            chars_remaining = 0;
            s
        };
        let kern = run.kern_threshold.is_some_and(|t| eff_fs >= t);
        w += entry.word_width(text_to_measure, eff_fs, kern);
        if chars_remaining == 0 {
            break;
        }
    }
    w
}

/// Build TextLines for a paragraph that contains tab characters.
/// Wraps to new lines when content exceeds `max_width`.
pub(super) fn build_tabbed_line(
    runs: &[Run],
    seen_fonts: &HashMap<String, FontEntry>,
    tab_stops: &[TabStop],
    indent_left: f32,
    max_width: f32,
    first_line_hanging: f32,
    inline_image_names: &HashMap<usize, String>,
) -> Vec<TextLine> {
    // Split runs into segments at tab markers, tracking original run indices
    let mut segments: Vec<(Vec<&Run>, Vec<usize>, Option<TabStop>)> = Vec::new();
    let mut current_seg: Vec<&Run> = Vec::new();
    let mut current_indices: Vec<usize> = Vec::new();
    let mut pending_tab: Option<TabStop> = None;

    for (global_idx, run) in runs.iter().enumerate() {
        if run.vanish {
            continue;
        }
        if run.is_tab {
            segments.push((
                std::mem::take(&mut current_seg),
                std::mem::take(&mut current_indices),
                pending_tab.take(),
            ));
            pending_tab = Some(TabStop {
                position: 0.0, // placeholder, resolved below
                alignment: TabAlignment::Left,
                leader: None,
            });
        } else {
            current_seg.push(run);
            current_indices.push(global_idx);
        }
    }
    segments.push((
        std::mem::take(&mut current_seg),
        std::mem::take(&mut current_indices),
        pending_tab.take(),
    ));

    let mut result_lines: Vec<TextLine> = Vec::new();
    let mut all_chunks: Vec<WordChunk> = Vec::new();
    let mut current_x: f32 = 0.0;
    let mut key_buf = String::new();
    let mut is_first_line = true;

    for (seg_idx, (seg_runs, seg_indices, tab_before)) in segments.iter().enumerate() {
        let line_max = if is_first_line {
            max_width + first_line_hanging
        } else {
            max_width
        };
        let line_indent = if is_first_line {
            indent_left - first_line_hanging
        } else {
            indent_left
        };

        if seg_idx > 0 {
            let stop = find_next_tab_stop(current_x, tab_stops, line_indent);
            let tab_target = stop.position - line_indent;

            // Calculate where segment text will start based on alignment
            let mut seg_start = match stop.alignment {
                TabAlignment::Left => tab_target.max(current_x),
                TabAlignment::Center => {
                    let sw = segment_width(seg_runs, seen_fonts);
                    (tab_target - sw / 2.0).max(current_x)
                }
                TabAlignment::Right => {
                    let sw = segment_width(seg_runs, seen_fonts);
                    (tab_target - sw).max(current_x)
                }
                TabAlignment::Decimal => {
                    let bw = decimal_before_width(seg_runs, seen_fonts);
                    (tab_target - bw).max(current_x)
                }
            };

            // Wrap to new line if tab pushes past line_max
            if seg_start > line_max && !all_chunks.is_empty() {
                let total_width = all_chunks
                    .last()
                    .map(|c| c.x_offset + c.width)
                    .unwrap_or(0.0);
                result_lines.push(TextLine {
                    chunks: std::mem::take(&mut all_chunks),
                    total_width,
                });
                current_x = 0.0;
                is_first_line = false;
                // Re-resolve the tab stop from position 0 on the new line
                let new_stop = find_next_tab_stop(0.0, tab_stops, indent_left);
                let new_target = new_stop.position - indent_left;
                seg_start = match new_stop.alignment {
                    TabAlignment::Left => new_target.max(0.0),
                    TabAlignment::Center => {
                        let sw = segment_width(seg_runs, seen_fonts);
                        (new_target - sw / 2.0).max(0.0)
                    }
                    TabAlignment::Right => {
                        let sw = segment_width(seg_runs, seen_fonts);
                        (new_target - sw).max(0.0)
                    }
                    TabAlignment::Decimal => {
                        let bw = decimal_before_width(seg_runs, seen_fonts);
                        (new_target - bw).max(0.0)
                    }
                };
            }

            // Draw leader fill between end of previous text and start of aligned text
            if tab_before.is_some() {
                let abs_x = current_x + line_indent;
                let leader = tab_stops
                    .iter()
                    .find(|s| s.position > abs_x + 0.5)
                    .and_then(|s| s.leader);

                if let Some(leader_char) = leader {
                    let font_run = seg_runs.first().or_else(|| {
                        segments[..seg_idx]
                            .iter()
                            .rev()
                            .flat_map(|(r, _, _)| r.last())
                            .next()
                    });
                    if let Some(run) = font_run {
                        let key = font_key_buf(run, &mut key_buf);
                        let entry = seen_fonts.get(key).expect("font registered");
                        let eff_fs = effective_font_size(run);
                        {
                            let char_w = entry.char_width_1000(leader_char) * eff_fs / 1000.0;
                            let leader_gap = seg_start - current_x;
                            if char_w > 0.0 && leader_gap > char_w * 2.0 {
                                let count = ((leader_gap - char_w) / char_w).floor() as usize;
                                if count > 0 {
                                    let leader_text: String =
                                        std::iter::repeat_n(leader_char, count).collect();
                                    let leader_w = count as f32 * char_w;
                                    let leader_start = seg_start - leader_w;
                                    all_chunks.push(WordChunk {
                                        pdf_font: entry.pdf_name.clone(),
                                        text: leader_text,
                                        font_size: eff_fs,
                                        color: run.color,
                                        highlight: None,
                                        x_offset: leader_start,
                                        width: leader_w,
                                        underline: false,
                                        strikethrough: false,
                                        dstrike: false,
                                        char_spacing: 0.0,
                                        text_scale: 100.0,
                                        y_offset: 0.0,
                                        hyperlink_url: None,
                                        inline_image_name: None,
                                        inline_image_height: 0.0,
                                    });
                                }
                            }
                        }
                    }
                }
            }

            current_x = seg_start;
        }

        // Layout text in this segment from current_x
        let mut prev_ws = false;
        for (local_idx, run) in seg_runs.iter().enumerate() {
            // Handle inline images (same pattern as build_paragraph_lines)
            if let Some(img) = &run.inline_image {
                if let Some(pdf_name) = inline_image_names.get(&seg_indices[local_idx]) {
                    all_chunks.push(WordChunk {
                        pdf_font: String::new(),
                        text: String::new(),
                        font_size: run.font_size,
                        color: None,
                        highlight: None,
                        x_offset: current_x,
                        width: img.display_width,
                        underline: false,
                        strikethrough: false,
                        dstrike: false,
                        char_spacing: 0.0,
                        text_scale: 100.0,
                        y_offset: 0.0,
                        hyperlink_url: None,
                        inline_image_name: Some(pdf_name.clone()),
                        inline_image_height: img.display_height,
                    });
                    current_x += img.display_width;
                }
                continue;
            }

            let key = font_key_buf(run, &mut key_buf);
            let entry = seen_fonts.get(key).expect("font registered");
            let eff_fs = effective_font_size(run);
            let space_w = entry.space_width(eff_fs);
            let y_off = vert_y_offset(run);
            let text = effective_text(run);

            let cs = run.char_spacing;
            let ts = run.text_scale / 100.0;
            for (i, word) in text.split_whitespace().enumerate() {
                let char_count = word.chars().count();
                let kern = run.kern_threshold.is_some_and(|t| eff_fs >= t);
                let ww = entry.word_width(word, eff_fs, kern) * ts + cs * char_count as f32;
                if !all_chunks.is_empty()
                    && (i > 0 || prev_ws || text.starts_with(char::is_whitespace))
                {
                    current_x += space_w * ts + cs;
                }
                let cur_line_max = if is_first_line {
                    max_width + first_line_hanging
                } else {
                    max_width
                };
                // Wrap word to new line if it exceeds max_width
                if current_x + ww > cur_line_max && !all_chunks.is_empty() {
                    let total_width = all_chunks
                        .last()
                        .map(|c| c.x_offset + c.width)
                        .unwrap_or(0.0);
                    result_lines.push(TextLine {
                        chunks: std::mem::take(&mut all_chunks),
                        total_width,
                    });
                    current_x = 0.0;
                    is_first_line = false;
                }
                all_chunks.push(WordChunk {
                    pdf_font: entry.pdf_name.clone(),
                    text: word.to_string(),
                    font_size: eff_fs,
                    color: run.color,
                    highlight: run.highlight,
                    x_offset: current_x,
                    width: ww,
                    underline: run.underline,
                    strikethrough: run.strikethrough,
                    dstrike: run.dstrike,
                    char_spacing: cs,
                    text_scale: run.text_scale,
                    y_offset: y_off,
                    hyperlink_url: run.hyperlink_url.clone(),
                    inline_image_name: None,
                    inline_image_height: 0.0,
                });
                current_x += ww;
            }
            prev_ws = text.ends_with(char::is_whitespace);
        }
    }

    // Finalize remaining chunks into the last line
    if !all_chunks.is_empty() {
        let total_width = all_chunks
            .last()
            .map(|c| c.x_offset + c.width)
            .unwrap_or(0.0);
        result_lines.push(TextLine {
            chunks: all_chunks,
            total_width,
        });
    } else if result_lines.is_empty() {
        result_lines.push(TextLine {
            chunks: vec![],
            total_width: 0.0,
        });
    }

    result_lines
}

fn encode_text_for_pdf(
    text: &str,
    pdf_font: &str,
    pdf_name_to_entry: &HashMap<&str, &FontEntry>,
) -> Vec<u8> {
    match pdf_name_to_entry
        .get(pdf_font)
        .and_then(|e| e.char_to_gid.as_ref())
    {
        Some(map) => encode_as_gids(text, map),
        None => to_winansi_bytes(text),
    }
}

/// Render pre-built lines applying the paragraph alignment.
/// `total_line_count` is the full paragraph line count (for justify: last line stays left-aligned).
pub(super) fn render_paragraph_lines(
    content: &mut Content,
    lines: &[TextLine],
    alignment: &Alignment,
    margin_left: f32,
    text_width: f32,
    first_baseline_y: f32,
    line_pitch: f32,
    total_line_count: usize,
    first_line_index: usize,
    links: &mut Vec<LinkAnnotation>,
    first_line_hanging: f32,
    seen_fonts: &HashMap<String, FontEntry>,
) {
    let mut current_color: Option<[u8; 3]> = None;
    let mut cur_font_name = String::new();
    let mut cur_font_size: f32 = -1.0;
    let mut cur_char_spacing: f32 = 0.0;
    let mut cur_text_scale: f32 = 100.0;

    let pdf_name_to_entry: HashMap<&str, &FontEntry> = seen_fonts
        .values()
        .map(|e| (e.pdf_name.as_str(), e))
        .collect();

    // Pre-compute per-line y offsets accounting for inline images making lines taller
    let mut line_y_offsets: Vec<f32> = Vec::with_capacity(lines.len());
    let mut cumulative_y = 0.0f32;
    for (i, line) in lines.iter().enumerate() {
        line_y_offsets.push(cumulative_y);
        let img_h = line
            .chunks
            .iter()
            .map(|c| c.inline_image_height)
            .fold(0.0f32, f32::max);
        cumulative_y += if img_h > line_pitch {
            img_h
        } else {
            line_pitch
        };
        // First line offset is always 0
        if i == 0 {
            cumulative_y = line_pitch.max(img_h);
            line_y_offsets[0] = 0.0;
        }
    }

    let last_line_idx = total_line_count.saturating_sub(1);
    for (line_num, line) in lines.iter().enumerate() {
        let y = first_baseline_y - line_y_offsets[line_num];
        let global_line_idx = first_line_index + line_num;

        let is_justified = *alignment == Alignment::Justify
            && global_line_idx != last_line_idx
            && line.chunks.len() > 1;

        let (eff_margin, eff_width) = if global_line_idx == 0 && first_line_hanging.abs() > 0.001 {
            (
                margin_left - first_line_hanging,
                text_width + first_line_hanging,
            )
        } else {
            (margin_left, text_width)
        };

        let line_start_x = match alignment {
            Alignment::Center => eff_margin + (eff_width - line.total_width) / 2.0,
            Alignment::Right => eff_margin + eff_width - line.total_width,
            Alignment::Left | Alignment::Justify => eff_margin,
        };

        let extra_per_gap = if is_justified {
            (eff_width - line.total_width) / (line.chunks.len() - 1) as f32
        } else {
            0.0
        };

        let mut decorations: Vec<(f32, f32, f32, f32, Option<[u8; 3]>)> = Vec::new();
        let mut image_draws: Vec<(f32, f32, f32, f32, &str)> = Vec::new();

        // Draw run highlights as merged spans (contiguous same-color highlights)
        {
            let mut hl_start_x = 0.0f32;
            let mut hl_color: Option<[u8; 3]> = None;
            let mut hl_end_x = 0.0f32;
            let mut hl_fs = 0.0f32;

            let flush_hl =
                |content: &mut Content, color: [u8; 3], sx: f32, ex: f32, fs: f32, y: f32| {
                    let hl_bottom = y - fs * 0.2;
                    let hl_height = fs * 1.15;
                    content.save_state();
                    content.set_fill_rgb(
                        color[0] as f32 / 255.0,
                        color[1] as f32 / 255.0,
                        color[2] as f32 / 255.0,
                    );
                    content.rect(sx, hl_bottom, ex - sx, hl_height);
                    content.fill_nonzero();
                    content.restore_state();
                };

            for (chunk_idx, chunk) in line.chunks.iter().enumerate() {
                let x = line_start_x + chunk.x_offset + chunk_idx as f32 * extra_per_gap;
                if chunk.highlight == hl_color && hl_color.is_some() {
                    hl_end_x = x + chunk.width;
                    hl_fs = hl_fs.max(chunk.font_size);
                } else {
                    if let Some(c) = hl_color {
                        flush_hl(content, c, hl_start_x, hl_end_x, hl_fs, y);
                    }
                    if let Some(c) = chunk.highlight {
                        hl_start_x = x;
                        hl_end_x = x + chunk.width;
                        hl_fs = chunk.font_size;
                        hl_color = Some(c);
                    } else {
                        hl_color = None;
                    }
                }
            }
            if let Some(c) = hl_color {
                flush_hl(content, c, hl_start_x, hl_end_x, hl_fs, y);
            }
        }

        let has_text_chunks = line
            .chunks
            .iter()
            .any(|c| c.inline_image_name.is_none() && !c.text.is_empty());

        if has_text_chunks {
            content.begin_text();
            let mut td_x = 0.0_f32;
            let mut td_y = 0.0_f32;

            for (chunk_idx, chunk) in line.chunks.iter().enumerate() {
                if chunk.inline_image_name.is_some() {
                    continue;
                }

                let x = line_start_x + chunk.x_offset + chunk_idx as f32 * extra_per_gap;
                let cy = y + chunk.y_offset;

                if chunk.color != current_color {
                    if let Some([r, g, b]) = chunk.color {
                        content.set_fill_rgb(r as f32 / 255.0, g as f32 / 255.0, b as f32 / 255.0);
                    } else {
                        content.set_fill_gray(0.0);
                    }
                    current_color = chunk.color;
                }

                if chunk.char_spacing != cur_char_spacing {
                    content.set_char_spacing(chunk.char_spacing);
                    cur_char_spacing = chunk.char_spacing;
                }
                if chunk.text_scale != cur_text_scale {
                    content.set_horizontal_scaling(chunk.text_scale);
                    cur_text_scale = chunk.text_scale;
                }

                if cur_font_name != chunk.pdf_font || cur_font_size != chunk.font_size {
                    content.set_font(Name(chunk.pdf_font.as_bytes()), chunk.font_size);
                    cur_font_name.clear();
                    cur_font_name.push_str(&chunk.pdf_font);
                    cur_font_size = chunk.font_size;
                }

                content.next_line(x - td_x, cy - td_y);
                td_x = x;
                td_y = cy;

                let text_bytes =
                    encode_text_for_pdf(&chunk.text, &chunk.pdf_font, &pdf_name_to_entry);
                content.show(Str(&text_bytes));

                if chunk.underline {
                    let thick = (chunk.font_size * 0.05).max(0.5);
                    let ul_y = if chunk.hyperlink_url.is_some() {
                        y - chunk.font_size * 0.08
                    } else {
                        y - chunk.font_size * 0.12
                    };
                    decorations.push((x, ul_y - thick, chunk.width, thick, chunk.color));
                }
                if chunk.strikethrough {
                    let thick = (chunk.font_size * 0.05).max(0.5);
                    let st_y = y + chunk.font_size * 0.3;
                    decorations.push((x, st_y, chunk.width, thick, chunk.color));
                }
                if chunk.dstrike {
                    let thick = (chunk.font_size * 0.05).max(0.5);
                    let gap = thick * 1.5;
                    let mid_y = y + chunk.font_size * 0.3;
                    decorations.push((x, mid_y - gap / 2.0, chunk.width, thick, chunk.color));
                    decorations.push((x, mid_y + gap / 2.0, chunk.width, thick, chunk.color));
                }

                if let Some(ref url) = chunk.hyperlink_url {
                    let bottom = y - chunk.font_size * 0.2;
                    let top = y + chunk.font_size * 0.8;
                    let merged = links
                        .last_mut()
                        .filter(|prev| prev.url == *url && (prev.rect.y1 - bottom).abs() < 1.0);
                    if let Some(prev) = merged {
                        prev.rect.x2 = x + chunk.width;
                    } else {
                        links.push(LinkAnnotation {
                            rect: Rect::new(x, bottom, x + chunk.width, top),
                            url: url.clone(),
                        });
                    }
                }
            }
            if cur_char_spacing != 0.0 {
                content.set_char_spacing(0.0);
                cur_char_spacing = 0.0;
            }
            if cur_text_scale != 100.0 {
                content.set_horizontal_scaling(100.0);
                cur_text_scale = 100.0;
            }
            content.end_text();
        }

        // Collect inline image draws
        for (chunk_idx, chunk) in line.chunks.iter().enumerate() {
            if let Some(ref img_name) = chunk.inline_image_name {
                let x = line_start_x + chunk.x_offset + chunk_idx as f32 * extra_per_gap;
                let img_bottom = y - (chunk.inline_image_height - chunk.font_size);
                image_draws.push((
                    x,
                    img_bottom,
                    chunk.width,
                    chunk.inline_image_height,
                    img_name,
                ));
            }
        }

        // Draw inline images outside text block
        for &(ix, iy, iw, ih, ref img_name) in &image_draws {
            content.save_state();
            content.transform([iw, 0.0, 0.0, ih, ix, iy]);
            content.x_object(Name(img_name.as_bytes()));
            content.restore_state();
        }

        for &(dx, dy, dw, dh, dcolor) in &decorations {
            if dcolor != current_color {
                if let Some([r, g, b]) = dcolor {
                    content.set_fill_rgb(r as f32 / 255.0, g as f32 / 255.0, b as f32 / 255.0);
                } else {
                    content.set_fill_gray(0.0);
                }
                current_color = dcolor;
            }
            content.rect(dx, dy, dw, dh).fill_nonzero();
        }
    }
    if current_color.is_some() {
        content.set_fill_gray(0.0);
    }
}

pub(super) fn font_metric(
    runs: &[Run],
    seen_fonts: &HashMap<String, FontEntry>,
    get: impl Fn(&FontEntry) -> Option<f32>,
) -> Option<f32> {
    runs.first()
        .map(font_key)
        .and_then(|k| seen_fonts.get(&k))
        .and_then(get)
}

/// Compute the effective font_size, line_h_ratio, and ascender_ratio for a set of runs
/// by picking the run that produces the tallest visual ascent (font_size * ascender_ratio).
pub(super) fn tallest_run_metrics(
    runs: &[Run],
    seen_fonts: &HashMap<String, FontEntry>,
) -> (f32, Option<f32>, Option<f32>) {
    let mut best_font_size = runs.first().map_or(12.0, |r| r.font_size);
    let mut best_ascent = 0.0f32;
    let mut best_line_h_ratio: Option<f32> = None;
    let mut best_ascender_ratio: Option<f32> = None;
    let mut key_buf = String::new();

    for run in runs {
        let key = font_key_buf(run, &mut key_buf);
        let entry = seen_fonts.get(key);
        let ar = entry.and_then(|e| e.ascender_ratio).unwrap_or(0.75);
        let ascent = run.font_size * ar;
        if ascent > best_ascent {
            best_ascent = ascent;
            best_font_size = run.font_size;
            best_ascender_ratio = entry.and_then(|e| e.ascender_ratio);
            best_line_h_ratio = entry.and_then(|e| e.line_h_ratio);
        }
    }
    (best_font_size, best_line_h_ratio, best_ascender_ratio)
}