imprimatur 0.2.0 → 0.3.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (56) hide show
  1. checksums.yaml +4 -4
  2. data/CHANGELOG.md +22 -0
  3. data/Cargo.lock +123 -3
  4. data/Cargo.toml +1 -1
  5. data/README.md +98 -20
  6. data/ext/imprimatur_native/cli/Cargo.toml +1 -1
  7. data/ext/imprimatur_native/core/Cargo.toml +6 -1
  8. data/ext/imprimatur_native/core/README.md +12 -3
  9. data/ext/imprimatur_native/core/src/archive.rs +3 -16
  10. data/ext/imprimatur_native/core/src/doc/paragraph.rs +10 -3
  11. data/ext/imprimatur_native/core/src/doc/picture.rs +192 -0
  12. data/ext/imprimatur_native/core/src/doc/table.rs +21 -0
  13. data/ext/imprimatur_native/core/src/doc.rs +86 -2
  14. data/ext/imprimatur_native/core/src/docx/picture.rs +72 -0
  15. data/ext/imprimatur_native/core/src/docx.rs +91 -28
  16. data/ext/imprimatur_native/core/src/error.rs +12 -0
  17. data/ext/imprimatur_native/core/src/font.rs +229 -0
  18. data/ext/imprimatur_native/core/src/lib.rs +3 -0
  19. data/ext/imprimatur_native/core/src/pdf/layout.rs +38 -15
  20. data/ext/imprimatur_native/core/src/pdf/objects.rs +29 -5
  21. data/ext/imprimatur_native/core/src/pdf/picture.rs +197 -0
  22. data/ext/imprimatur_native/core/src/pdf.rs +31 -3
  23. data/ext/imprimatur_native/core/src/read/content.rs +75 -2
  24. data/ext/imprimatur_native/core/src/read/json.rs +94 -7
  25. data/ext/imprimatur_native/core/src/read/numbers.rs +65 -42
  26. data/ext/imprimatur_native/core/src/read/ods.rs +1 -2
  27. data/ext/imprimatur_native/core/src/read/pdf/layout/bidi.rs +212 -0
  28. data/ext/imprimatur_native/core/src/read/pdf/layout.rs +25 -119
  29. data/ext/imprimatur_native/core/src/read/xls.rs +1 -1
  30. data/ext/imprimatur_native/core/src/read/xlsx/formats.rs +155 -0
  31. data/ext/imprimatur_native/core/src/read/xlsx.rs +104 -19
  32. data/ext/imprimatur_native/core/src/read.rs +4 -1
  33. data/ext/imprimatur_native/core/src/script.rs +116 -0
  34. data/ext/imprimatur_native/core/src/style.rs +21 -0
  35. data/ext/imprimatur_native/core/src/test_support.rs +3 -2
  36. data/ext/imprimatur_native/core/src/xlsx/autofit.rs +60 -0
  37. data/ext/imprimatur_native/core/src/xlsx/date.rs +7 -1
  38. data/ext/imprimatur_native/core/src/xlsx/format.rs +22 -6
  39. data/ext/imprimatur_native/core/src/xlsx/package.rs +22 -8
  40. data/ext/imprimatur_native/core/src/xlsx/styles.rs +9 -8
  41. data/ext/imprimatur_native/core/src/xlsx/worksheet.rs +35 -7
  42. data/ext/imprimatur_native/core/src/xlsx.rs +40 -3
  43. data/ext/imprimatur_native/core/src/xml/reader.rs +10 -0
  44. data/ext/imprimatur_native/ffi/Cargo.toml +3 -2
  45. data/ext/imprimatur_native/ffi/src/archive.rs +1 -3
  46. data/ext/imprimatur_native/ffi/src/document.rs +48 -1
  47. data/ext/imprimatur_native/ffi/src/lib.rs +38 -1
  48. data/ext/imprimatur_native/ffi/src/read.rs +8 -11
  49. data/ext/imprimatur_native/ffi/src/sheets.rs +114 -0
  50. data/lib/imprimatur/document.rb +30 -6
  51. data/lib/imprimatur/reader.rb +2 -2
  52. data/lib/imprimatur/spreadsheet.rb +151 -0
  53. data/lib/imprimatur/version.rb +1 -1
  54. data/lib/imprimatur/xlsx.rb +25 -4
  55. data/lib/imprimatur.rb +15 -3
  56. metadata +11 -1
@@ -0,0 +1,192 @@
1
+ //! Pictures: PNG or JPEG files, read once for what PDF and DOCX need.
2
+
3
+ use std::sync::Arc;
4
+
5
+ use crate::style::Align;
6
+ use crate::{Error, Result};
7
+
8
+ /// Pixels per point at 96 pixels an inch: a picture's size when none is set.
9
+ const POINTS_PER_PIXEL: f64 = 0.75;
10
+ /// The largest side, in points, a picture may be set to: PDF's page limit.
11
+ const MAX_POINTS: f64 = 14_400.0;
12
+ /// The largest side, in pixels, a picture may have.
13
+ const MAX_PIXELS: u32 = 1 << 16;
14
+
15
+ /// A PNG or JPEG drawn in the document's flow.
16
+ ///
17
+ /// It is drawn at the width or height set (in points, the other following
18
+ /// its proportions), or else at 96 pixels an inch, never larger than the
19
+ /// page leaves room for.
20
+ ///
21
+ /// ```no_run
22
+ /// use imprimatur::doc::{Document, Picture};
23
+ /// use imprimatur::style::Align;
24
+ ///
25
+ /// let mut doc = Document::new();
26
+ /// doc.picture(Picture::new(std::fs::read("logo.png")?)?.width(144.0)?.align(Align::Center));
27
+ /// # Ok::<(), imprimatur::Error>(())
28
+ /// ```
29
+ #[derive(Debug, Clone, PartialEq)]
30
+ pub struct Picture {
31
+ pub(crate) data: Arc<[u8]>,
32
+ pub(crate) kind: Kind,
33
+ pub(crate) pixels: (u32, u32),
34
+ width: Option<f64>,
35
+ height: Option<f64>,
36
+ pub(crate) align: Option<Align>,
37
+ }
38
+
39
+ /// The format, and what writing it takes.
40
+ #[derive(Debug, Clone, PartialEq, Eq)]
41
+ pub enum Kind {
42
+ Png(Png),
43
+ /// Gray (1), RGB (3) or CMYK (4) components; Adobe's CMYK is inverted.
44
+ Jpeg {
45
+ components: u8,
46
+ inverted: bool,
47
+ },
48
+ }
49
+
50
+ /// A PNG's pixel layout and its compressed pixels (its `IDAT` chunks).
51
+ #[derive(Debug, Clone, PartialEq, Eq)]
52
+ pub struct Png {
53
+ pub depth: u8,
54
+ /// 0 gray, 2 RGB, 3 palette, 4 gray and alpha, 6 RGB and alpha.
55
+ pub color: u8,
56
+ pub palette: Vec<u8>,
57
+ /// The `tRNS` chunk: a palette's alphas, or the transparent color.
58
+ pub transparency: Vec<u8>,
59
+ pub compressed: Vec<u8>,
60
+ }
61
+
62
+ impl Picture {
63
+ /// # Errors
64
+ /// [`Error::InvalidImage`] for data that is no PNG or JPEG, a damaged
65
+ /// one, an interlaced PNG, a JPEG other than 8 bits a sample, or one
66
+ /// larger than 65 536 pixels a side.
67
+ pub fn new(data: impl Into<Vec<u8>>) -> Result<Self> {
68
+ let data: Arc<[u8]> = data.into().into();
69
+ let (kind, pixels) = if data.starts_with(b"\x89PNG\r\n\x1a\n") {
70
+ png(&data)?
71
+ } else if data.starts_with(&[0xFF, 0xD8, 0xFF]) {
72
+ jpeg(&data)?
73
+ } else {
74
+ return Err(Error::InvalidImage("not a PNG or JPEG image"));
75
+ };
76
+ let valid = (1..=MAX_PIXELS).contains(&pixels.0) && (1..=MAX_PIXELS).contains(&pixels.1);
77
+ valid.ok_or(Error::InvalidImage("a picture of 1 to 65 536 pixels a side"))?;
78
+ Ok(Self { data, kind, pixels, width: None, height: None, align: None })
79
+ }
80
+
81
+ /// Its width in points; the height follows unless set too.
82
+ ///
83
+ /// # Errors
84
+ /// [`Error::InvalidImage`] outside 1 to 14 400 points.
85
+ pub fn width(mut self, points: f64) -> Result<Self> {
86
+ self.width = Some(points_in_range(points)?);
87
+ Ok(self)
88
+ }
89
+
90
+ /// Its height in points; the width follows unless set too.
91
+ ///
92
+ /// # Errors
93
+ /// [`Error::InvalidImage`] outside 1 to 14 400 points.
94
+ pub fn height(mut self, points: f64) -> Result<Self> {
95
+ self.height = Some(points_in_range(points)?);
96
+ Ok(self)
97
+ }
98
+
99
+ #[must_use]
100
+ pub const fn align(mut self, align: Align) -> Self {
101
+ self.align = Some(align);
102
+ self
103
+ }
104
+
105
+ /// The size it is drawn at, in points, fitted into `room` (the width
106
+ /// and height between the margins).
107
+ pub(crate) fn size(&self, room: (f64, f64)) -> (f64, f64) {
108
+ let (pixels_wide, pixels_high) = (f64::from(self.pixels.0), f64::from(self.pixels.1));
109
+ let (width, height) = match (self.width, self.height) {
110
+ (Some(width), Some(height)) => (width, height),
111
+ (Some(width), None) => (width, width * pixels_high / pixels_wide),
112
+ (None, Some(height)) => (height * pixels_wide / pixels_high, height),
113
+ (None, None) => (pixels_wide * POINTS_PER_PIXEL, pixels_high * POINTS_PER_PIXEL),
114
+ };
115
+ let scale = (room.0 / width).min(room.1 / height).min(1.0);
116
+ (width * scale, height * scale)
117
+ }
118
+ }
119
+
120
+ fn points_in_range(points: f64) -> Result<f64> {
121
+ (1.0..=MAX_POINTS)
122
+ .contains(&points)
123
+ .then_some(points)
124
+ .ok_or(Error::InvalidImage("a picture of 1 to 14 400 points a side"))
125
+ }
126
+
127
+ /// A PNG's chunks; interlaced ones are refused.
128
+ fn png(data: &[u8]) -> Result<(Kind, (u32, u32))> {
129
+ let damaged = || Error::InvalidImage("a damaged PNG");
130
+ let mut png = Png {
131
+ depth: 0,
132
+ color: 0,
133
+ palette: Vec::new(),
134
+ transparency: Vec::new(),
135
+ compressed: Vec::new(),
136
+ };
137
+ let (mut pixels, mut at) = (None, 8);
138
+ while let Some(head) = data.get(at..at + 8) {
139
+ let length = u32::from_be_bytes([head[0], head[1], head[2], head[3]]) as usize;
140
+ let body = data.get(at + 8..at + 8 + length).ok_or_else(damaged)?;
141
+ match &head[4..8] {
142
+ b"IHDR" if body.len() == 13 => {
143
+ let side = |at: usize| {
144
+ u32::from_be_bytes([body[at], body[at + 1], body[at + 2], body[at + 3]])
145
+ };
146
+ pixels = Some((side(0), side(4)));
147
+ (png.depth, png.color) = (body[8], body[9]);
148
+ (body[12] == 0).ok_or(Error::InvalidImage("an interlaced PNG"))?;
149
+ }
150
+ b"PLTE" => png.palette = body.to_vec(),
151
+ b"tRNS" => png.transparency = body.to_vec(),
152
+ b"IDAT" => png.compressed.extend_from_slice(body),
153
+ b"IEND" => break,
154
+ _ => {}
155
+ }
156
+ at += 12 + length;
157
+ }
158
+ let layout = matches!(
159
+ (png.color, png.depth),
160
+ (0, 1 | 2 | 4 | 8 | 16) | (3, 1 | 2 | 4 | 8) | (2 | 4 | 6, 8 | 16)
161
+ );
162
+ let palette = png.color != 3 || !png.palette.is_empty();
163
+ let pixels = pixels.filter(|_| layout && palette && !png.compressed.is_empty());
164
+ Ok((Kind::Png(png), pixels.ok_or_else(damaged)?))
165
+ }
166
+
167
+ /// A JPEG's frame header (`SOF`), and Adobe's mark, which inverts CMYK.
168
+ fn jpeg(data: &[u8]) -> Result<(Kind, (u32, u32))> {
169
+ let damaged = || Error::InvalidImage("a damaged JPEG");
170
+ let (mut at, mut inverted) = (2, false);
171
+ loop {
172
+ let &[0xFF, marker, high, low] = data.get(at..at + 4).ok_or_else(damaged)? else {
173
+ return Err(damaged());
174
+ };
175
+ let length = usize::from(u16::from_be_bytes([high, low]));
176
+ if marker == 0xEE && data.get(at + 4..at + 9) == Some(b"Adobe") {
177
+ inverted = true;
178
+ }
179
+ // SOF0 to SOF15, but DHT, JPG and DAC share the range.
180
+ if (0xC0..=0xCF).contains(&marker) && !matches!(marker, 0xC4 | 0xC8 | 0xCC) {
181
+ let frame = data.get(at + 4..at + 10).ok_or_else(damaged)?;
182
+ (frame[0] == 8).ok_or(Error::InvalidImage("a JPEG of 8 bits a sample"))?;
183
+ let height = u32::from(u16::from_be_bytes([frame[1], frame[2]]));
184
+ let width = u32::from(u16::from_be_bytes([frame[3], frame[4]]));
185
+ let components = frame[5];
186
+ matches!(components, 1 | 3 | 4).ok_or_else(damaged)?;
187
+ let kind = Kind::Jpeg { components, inverted: inverted && components == 4 };
188
+ return Ok((kind, (width, height)));
189
+ }
190
+ at += 2 + length;
191
+ }
192
+ }
@@ -1,5 +1,12 @@
1
+ use crate::script::Counts;
2
+ use crate::style::TextDirection;
3
+
1
4
  /// A grid of text cells; columns share the page width equally. Short rows
2
5
  /// are padded with empty cells.
6
+ ///
7
+ /// Each cell reads in the direction of its first letter, a cell of figures
8
+ /// in the table's; a table mostly right to left has its first column on the
9
+ /// right.
3
10
  #[derive(Debug, Clone, PartialEq, Eq)]
4
11
  pub struct Table {
5
12
  pub(crate) rows: Vec<Vec<String>>,
@@ -38,4 +45,18 @@ impl Table {
38
45
  pub(crate) const fn has_header(&self) -> bool {
39
46
  self.header
40
47
  }
48
+
49
+ /// The direction most of the table's letters read in; `None` without a
50
+ /// letter.
51
+ pub(crate) fn direction(&self) -> Option<TextDirection> {
52
+ let mut counts = Counts::default();
53
+ self.count(&mut counts);
54
+ counts.direction()
55
+ }
56
+
57
+ pub(crate) fn count(&self, counts: &mut Counts) {
58
+ for cell in self.rows.iter().flatten() {
59
+ counts.add(cell);
60
+ }
61
+ }
41
62
  }
@@ -22,6 +22,7 @@
22
22
  //! ```
23
23
 
24
24
  mod paragraph;
25
+ pub(crate) mod picture;
25
26
  mod table;
26
27
  #[cfg(test)]
27
28
  mod tests;
@@ -31,24 +32,49 @@ use std::io::{BufWriter, Write};
31
32
  use std::path::Path;
32
33
 
33
34
  pub use paragraph::{Paragraph, Run};
35
+ pub use picture::Picture;
34
36
  pub use table::Table;
35
37
 
38
+ use crate::script::Counts;
36
39
  use crate::{Error, Result};
37
40
 
38
- /// Paper size; pages are portrait.
41
+ /// Paper size; A4 and Letter pages are portrait.
39
42
  #[derive(Debug, Clone, Copy, Default, PartialEq, Eq)]
40
43
  pub enum PageSize {
41
44
  #[default]
42
45
  A4,
43
46
  Letter,
47
+ /// Width and height in hundredths of a point; see [`PageSize::custom`].
48
+ Custom(u32, u32),
44
49
  }
45
50
 
46
51
  impl PageSize {
52
+ /// A page of this width and height, in points (1/72 inch): 36 to 1584
53
+ /// (half an inch to 22 inches), as both PDF and Word take them.
54
+ ///
55
+ /// # Errors
56
+ /// [`Error::InvalidPageSize`] outside that range.
57
+ pub fn custom(width: f64, height: f64) -> Result<Self> {
58
+ let side = |points: f64| {
59
+ #[expect(
60
+ clippy::cast_possible_truncation,
61
+ clippy::cast_sign_loss,
62
+ reason = "36 to 1584"
63
+ )]
64
+ (36.0..=1584.0).contains(&points).then(|| (points * 100.0).round() as u32)
65
+ };
66
+ side(width)
67
+ .zip(side(height))
68
+ .map(|(width, height)| Self::Custom(width, height))
69
+ .ok_or(Error::InvalidPageSize(width, height))
70
+ }
71
+
47
72
  /// Width and height in PDF points (1/72 inch).
48
- pub(crate) const fn points(self) -> (f64, f64) {
73
+ pub(crate) fn points(self) -> (f64, f64) {
49
74
  match self {
50
75
  Self::A4 => (595.276, 841.89),
51
76
  Self::Letter => (612.0, 792.0),
77
+ Self::Custom(width, height) => (f64::from(width) / 100.0, f64::from(height) / 100.0),
52
78
  }
53
79
  }
54
80
 
@@ -57,6 +83,7 @@ impl PageSize {
57
83
  match self {
58
84
  Self::A4 => (11_906, 16_838),
59
85
  Self::Letter => (12_240, 15_840),
86
+ Self::Custom(width, height) => (width / 5, height / 5),
60
87
  }
61
88
  }
62
89
  }
@@ -68,6 +95,7 @@ pub enum Block {
68
95
  Heading { level: u8, text: String },
69
96
  Paragraph(Paragraph),
70
97
  Table(Table),
98
+ Picture(Picture),
71
99
  PageBreak,
72
100
  }
73
101
 
@@ -78,6 +106,8 @@ pub struct Document {
78
106
  blocks: Vec<Block>,
79
107
  page_size: PageSize,
80
108
  page_numbers: bool,
109
+ /// Hundredths of a point; one inch when unset.
110
+ margin: Option<u32>,
81
111
  }
82
112
 
83
113
  impl Document {
@@ -91,6 +121,43 @@ impl Document {
91
121
  self
92
122
  }
93
123
 
124
+ /// The margin on every side, in points: one inch (72) unless set.
125
+ ///
126
+ /// # Errors
127
+ /// [`Error::InvalidMargin`] outside 0 to 144 points.
128
+ pub fn set_margin(&mut self, points: f64) -> Result<&mut Self> {
129
+ #[expect(clippy::cast_possible_truncation, clippy::cast_sign_loss, reason = "0 to 144")]
130
+ let margin = (0.0..=144.0)
131
+ .contains(&points)
132
+ .then(|| (points * 100.0).round() as u32)
133
+ .ok_or(Error::InvalidMargin(points))?;
134
+ self.margin = Some(margin);
135
+ Ok(self)
136
+ }
137
+
138
+ /// The margin on every side, in points.
139
+ #[must_use]
140
+ pub fn margin(&self) -> f64 {
141
+ self.margin.map_or(72.0, |margin| f64::from(margin) / 100.0)
142
+ }
143
+
144
+ /// The width and height between the margins, in points.
145
+ ///
146
+ /// # Errors
147
+ /// [`Error::InvalidMargin`] when the margins leave no room.
148
+ pub(crate) fn room(&self) -> Result<(f64, f64)> {
149
+ let (width, height) = self.page_size.points();
150
+ let margin = self.margin();
151
+ let room = (2.0f64.mul_add(-margin, width), 2.0f64.mul_add(-margin, height));
152
+ (room.0 >= 1.0 && room.1 >= 1.0).then_some(room).ok_or(Error::InvalidMargin(margin))
153
+ }
154
+
155
+ /// A picture, in the flow of the text.
156
+ pub fn picture(&mut self, picture: Picture) -> &mut Self {
157
+ self.blocks.push(Block::Picture(picture));
158
+ self
159
+ }
160
+
94
161
  /// Numbers pages `1 / 3` at the bottom of each one.
95
162
  pub const fn set_page_numbers(&mut self, enabled: bool) -> &mut Self {
96
163
  self.page_numbers = enabled;
@@ -144,6 +211,23 @@ impl Document {
144
211
  self
145
212
  }
146
213
 
214
+ /// Whether most of the document's letters are right to left: its pages
215
+ /// are then laid out right to left.
216
+ pub(crate) fn reads_right_to_left(&self) -> bool {
217
+ let mut counts = Counts::default();
218
+ for block in &self.blocks {
219
+ match block {
220
+ Block::Heading { text, .. } => counts.add(text),
221
+ Block::Paragraph(paragraph) => {
222
+ paragraph.runs.iter().for_each(|run| counts.add(&run.text));
223
+ }
224
+ Block::Table(table) => table.count(&mut counts),
225
+ Block::Picture(_) | Block::PageBreak => {}
226
+ }
227
+ }
228
+ counts.right_to_left()
229
+ }
230
+
147
231
  #[must_use]
148
232
  pub fn blocks(&self) -> &[Block] {
149
233
  &self.blocks
@@ -0,0 +1,72 @@
1
+ //! Pictures as inline drawings, their files in `word/media`.
2
+
3
+ use crate::doc::Picture;
4
+ use crate::doc::picture::Kind;
5
+ use crate::xml::XmlWriter;
6
+
7
+ pub const WP_NS: &str = "http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing";
8
+ pub const A_NS: &str = "http://schemas.openxmlformats.org/drawingml/2006/main";
9
+ pub const PIC_NS: &str = "http://schemas.openxmlformats.org/drawingml/2006/picture";
10
+ /// English Metric Units a point, as drawings measure.
11
+ const EMU_PER_POINT: f64 = 12_700.0;
12
+
13
+ /// Its file's extension.
14
+ pub const fn extension(picture: &Picture) -> &'static str {
15
+ match picture.kind {
16
+ Kind::Png(_) => "png",
17
+ Kind::Jpeg { .. } => "jpeg",
18
+ }
19
+ }
20
+
21
+ /// Its part in `word/media`, by its number from 1.
22
+ pub fn file_name(picture: &Picture, number: usize) -> String {
23
+ format!("media/image{number}.{}", extension(picture))
24
+ }
25
+
26
+ /// A paragraph holding picture `number`, its file `relationship`, at its
27
+ /// size within `room`.
28
+ pub fn paragraph(
29
+ xml: &mut XmlWriter,
30
+ picture: &Picture,
31
+ room: (f64, f64),
32
+ number: usize,
33
+ relationship: &str,
34
+ ) {
35
+ let (width, height) = picture.size(room);
36
+ #[expect(clippy::cast_possible_truncation, clippy::cast_sign_loss, reason = "under 2**31")]
37
+ let emu = |points: f64| ((points * EMU_PER_POINT).round() as u64).to_string();
38
+ let (cx, cy, id, name) =
39
+ (emu(width), emu(height), number.to_string(), format!("Picture {number}"));
40
+ let extent = [("cx", cx.as_str()), ("cy", cy.as_str())];
41
+ super::open_paragraph(xml, None, false, picture.align);
42
+ xml.open("w:r", &[]);
43
+ xml.open("w:drawing", &[]);
44
+ xml.open("wp:inline", &[("distT", "0"), ("distB", "0"), ("distL", "0"), ("distR", "0")]);
45
+ xml.empty("wp:extent", &extent);
46
+ xml.empty("wp:docPr", &[("id", &id), ("name", &name)]);
47
+ xml.open("a:graphic", &[]);
48
+ xml.open("a:graphicData", &[("uri", PIC_NS)]);
49
+ xml.open("pic:pic", &[]);
50
+ xml.open("pic:nvPicPr", &[]);
51
+ xml.empty("pic:cNvPr", &[("id", &id), ("name", &name)]);
52
+ xml.empty("pic:cNvPicPr", &[]);
53
+ xml.close("pic:nvPicPr");
54
+ xml.open("pic:blipFill", &[]);
55
+ xml.empty("a:blip", &[("r:embed", relationship)]);
56
+ xml.open("a:stretch", &[]);
57
+ xml.empty("a:fillRect", &[]);
58
+ xml.close("a:stretch");
59
+ xml.close("pic:blipFill");
60
+ xml.open("pic:spPr", &[]);
61
+ xml.open("a:xfrm", &[]);
62
+ xml.empty("a:off", &[("x", "0"), ("y", "0")]);
63
+ xml.empty("a:ext", &extent);
64
+ xml.close("a:xfrm");
65
+ xml.open("a:prstGeom", &[("prst", "rect")]);
66
+ xml.empty("a:avLst", &[]);
67
+ xml.close("a:prstGeom");
68
+ xml.close("pic:spPr");
69
+ for tag in ["pic:pic", "a:graphicData", "a:graphic", "wp:inline", "w:drawing", "w:r", "w:p"] {
70
+ xml.close(tag);
71
+ }
72
+ }
@@ -2,13 +2,14 @@
2
2
  //! Word paragraphs, runs and tables, with real Heading 1-3 styles so Word's
3
3
  //! navigation pane and table of contents see them.
4
4
 
5
+ mod picture;
5
6
  #[cfg(test)]
6
7
  mod tests;
7
8
 
8
9
  use std::io::Write;
9
10
 
10
11
  use crate::Result;
11
- use crate::doc::{Block, Document, Paragraph, Run, Table};
12
+ use crate::doc::{Block, Document, Paragraph, Picture, Run, Table};
12
13
  use crate::opc;
13
14
  use crate::style::{Align, Color, TextDirection};
14
15
  use crate::xml::XmlWriter;
@@ -16,39 +17,58 @@ use crate::zip::ZipWriter;
16
17
 
17
18
  const W_NS: &str = "http://schemas.openxmlformats.org/wordprocessingml/2006/main";
18
19
  const WORD_TYPE: &str = "application/vnd.openxmlformats-officedocument.wordprocessingml";
19
- /// One inch, Word's default margin, in twips.
20
- const MARGIN: u32 = 1_440;
21
20
  /// A header row's shading.
22
21
  const HEADER_FILL: &str = "D9E2F3";
23
22
 
24
23
  pub fn write<W: Write>(document: &Document, out: W) -> Result<W> {
24
+ let room = document.room()?;
25
25
  let footer = document.has_page_numbers();
26
- let mut parts = vec![("styles", "styles.xml")];
26
+ let mut parts = vec![("styles", "styles.xml".to_owned())];
27
27
  if footer {
28
- parts.push(("footer", "footer1.xml"));
28
+ parts.push(("footer", "footer1.xml".to_owned()));
29
+ }
30
+ // Pictures' files follow, their relationships numbered from here.
31
+ let first_media = parts.len() + 1;
32
+ let pictures: Vec<&Picture> = document
33
+ .blocks()
34
+ .iter()
35
+ .filter_map(|block| if let Block::Picture(picture) = block { Some(picture) } else { None })
36
+ .collect();
37
+ for (number, picture) in (1..).zip(&pictures) {
38
+ parts.push(("image", picture::file_name(picture, number)));
29
39
  }
30
40
  let mut zip = ZipWriter::new(out);
31
- zip.add("[Content_Types].xml", content_types(footer).as_bytes())?;
41
+ zip.add("[Content_Types].xml", content_types(footer, &pictures).as_bytes())?;
32
42
  let package = opc::relationships(&[("officeDocument", "word/document.xml")]);
33
43
  zip.add("_rels/.rels", package.as_bytes())?;
34
- zip.add("word/document.xml", body(document).as_bytes())?;
44
+ zip.add("word/document.xml", body(document, room, first_media).as_bytes())?;
35
45
  zip.add("word/_rels/document.xml.rels", opc::relationships(&parts).as_bytes())?;
36
46
  zip.add("word/styles.xml", styles().as_bytes())?;
37
47
  if footer {
38
48
  zip.add("word/footer1.xml", page_number_footer().as_bytes())?;
39
49
  }
50
+ for (number, picture) in (1..).zip(&pictures) {
51
+ zip.add(&format!("word/{}", picture::file_name(picture, number)), &picture.data)?;
52
+ }
40
53
  zip.finish()
41
54
  }
42
55
 
43
- /// One override per part: the document, its styles and the footer.
44
- fn content_types(footer: bool) -> String {
56
+ /// One override per part: the document, its styles and the footer; the
57
+ /// pictures' types by extension.
58
+ fn content_types(footer: bool, pictures: &[&Picture]) -> String {
45
59
  let mut parts = vec![("/word/document.xml", "document.main"), ("/word/styles.xml", "styles")];
46
60
  if footer {
47
61
  parts.push(("/word/footer1.xml", "footer"));
48
62
  }
49
63
  let overrides: Vec<(&str, String)> =
50
64
  parts.into_iter().map(|(part, kind)| (part, format!("{WORD_TYPE}.{kind}+xml"))).collect();
51
- opc::content_types(&[], &overrides)
65
+ let mut defaults = Vec::new();
66
+ for (extension, content_type) in [("png", "image/png"), ("jpeg", "image/jpeg")] {
67
+ if pictures.iter().any(|picture| picture::extension(picture) == extension) {
68
+ defaults.push((extension, content_type));
69
+ }
70
+ }
71
+ opc::content_types(&defaults, &overrides)
52
72
  }
53
73
 
54
74
  /// `word/footer1.xml`: a centered `PAGE / NUMPAGES`, which Word computes.
@@ -82,18 +102,34 @@ fn page_number_footer() -> String {
82
102
  xml.finish()
83
103
  }
84
104
 
85
- fn body(document: &Document) -> String {
105
+ fn body(document: &Document, room: (f64, f64), first_media: usize) -> String {
106
+ // What reads in no direction of its own (figures, punctuation) takes the
107
+ // document's.
108
+ let rtl = document.reads_right_to_left();
86
109
  let mut xml = XmlWriter::new();
87
- xml.open("w:document", &[("xmlns:w", W_NS), ("xmlns:r", opc::RELATIONSHIPS)]);
110
+ let namespaces = [
111
+ ("xmlns:w", W_NS),
112
+ ("xmlns:r", opc::RELATIONSHIPS),
113
+ ("xmlns:wp", picture::WP_NS),
114
+ ("xmlns:a", picture::A_NS),
115
+ ("xmlns:pic", picture::PIC_NS),
116
+ ];
117
+ xml.open("w:document", &namespaces);
88
118
  xml.open("w:body", &[]);
119
+ let mut pictures = 0;
89
120
  for block in document.blocks() {
90
121
  match block {
91
122
  Block::Heading { level, text } => {
92
123
  let style = format!("Heading{level}");
93
- simple_paragraph(&mut xml, Some(&style), text, Look::default());
124
+ simple_paragraph(&mut xml, Some(&style), text, Look::default(), rtl);
125
+ }
126
+ Block::Paragraph(content) => paragraph(&mut xml, content, rtl),
127
+ Block::Table(content) => table(&mut xml, content, document, rtl),
128
+ Block::Picture(content) => {
129
+ pictures += 1;
130
+ let relationship = format!("rId{}", first_media + pictures - 1);
131
+ picture::paragraph(&mut xml, content, room, pictures, &relationship);
94
132
  }
95
- Block::Paragraph(content) => paragraph(&mut xml, content),
96
- Block::Table(content) => table(&mut xml, content, document),
97
133
  Block::PageBreak => {
98
134
  xml.open("w:p", &[]);
99
135
  xml.open("w:r", &[]);
@@ -107,25 +143,37 @@ fn body(document: &Document) -> String {
107
143
  if matches!(document.blocks().last(), Some(Block::Table(_))) {
108
144
  xml.empty("w:p", &[]);
109
145
  }
110
- section(&mut xml, document);
146
+ section(&mut xml, document, rtl);
111
147
  xml.close("w:body");
112
148
  xml.close("w:document");
113
149
  xml.finish()
114
150
  }
115
151
 
116
- fn paragraph(xml: &mut XmlWriter, content: &Paragraph) {
117
- let rtl = content.direction == Some(TextDirection::RightToLeft);
152
+ /// Each run reads in the direction of its own first letter, as Word sets
153
+ /// them: the Latin runs of a Hebrew paragraph left to right, the Arabic
154
+ /// ones of an English paragraph right to left.
155
+ fn paragraph(xml: &mut XmlWriter, content: &Paragraph, document_rtl: bool) {
156
+ let rtl = right_to_left(content.effective_direction(), document_rtl);
118
157
  open_paragraph(xml, None, rtl, content.align);
119
158
  for item in &content.runs {
120
- run(xml, &item.text, Look::of(item), rtl);
159
+ run(xml, &item.text, Look::of(item), right_to_left(TextDirection::of(&item.text), rtl));
121
160
  }
122
161
  xml.close("w:p");
123
162
  }
124
163
 
125
- /// A one-run paragraph straight from borrowed text: headings and table cells.
126
- fn simple_paragraph(xml: &mut XmlWriter, style: Option<&str>, text: &str, look: Look) {
127
- open_paragraph(xml, style, false, None);
128
- run(xml, text, look, false);
164
+ /// Whether `direction` is right to left; `otherwise` when there is none:
165
+ /// text with no letter (figures, punctuation) takes the direction of what
166
+ /// holds it.
167
+ fn right_to_left(direction: Option<TextDirection>, otherwise: bool) -> bool {
168
+ direction.map_or(otherwise, |direction| direction == TextDirection::RightToLeft)
169
+ }
170
+
171
+ /// A one-run paragraph straight from borrowed text: headings and table
172
+ /// cells, in the direction of their first letter, or of what holds them.
173
+ fn simple_paragraph(xml: &mut XmlWriter, style: Option<&str>, text: &str, look: Look, rtl: bool) {
174
+ let rtl = right_to_left(TextDirection::of(text), rtl);
175
+ open_paragraph(xml, style, rtl, None);
176
+ run(xml, text, look, rtl);
129
177
  xml.close("w:p");
130
178
  }
131
179
 
@@ -207,16 +255,20 @@ fn run(xml: &mut XmlWriter, text: &str, look: Look, rtl: bool) {
207
255
  xml.close("w:r");
208
256
  }
209
257
 
210
- fn table(xml: &mut XmlWriter, content: &Table, document: &Document) {
258
+ fn table(xml: &mut XmlWriter, content: &Table, document: &Document, document_rtl: bool) {
259
+ let rtl = right_to_left(content.direction(), document_rtl);
211
260
  let columns = content.columns();
212
261
  let (page_width, _) = document.page_size().twips();
213
- let usable = page_width - 2 * MARGIN;
262
+ let usable = page_width - 2 * margin(document);
214
263
  // Never 0: a table has at least one column.
215
264
  let column_width = (usable / u32::try_from(columns).unwrap_or(1).max(1)).to_string();
216
265
 
217
266
  xml.open("w:tbl", &[]);
218
267
  xml.open("w:tblPr", &[]);
219
268
  xml.empty("w:tblStyle", &[("w:val", "TableGrid")]);
269
+ if rtl {
270
+ xml.empty("w:bidiVisual", &[]);
271
+ }
220
272
  xml.empty("w:tblW", &[("w:w", "5000"), ("w:type", "pct")]);
221
273
  xml.close("w:tblPr");
222
274
  xml.open("w:tblGrid", &[]);
@@ -244,7 +296,7 @@ fn table(xml: &mut XmlWriter, content: &Table, document: &Document) {
244
296
  }
245
297
  xml.close("w:tcPr");
246
298
  let text = row.get(column).map_or("", String::as_str);
247
- simple_paragraph(xml, None, text, Look { bold: header, ..Look::default() });
299
+ simple_paragraph(xml, None, text, Look { bold: header, ..Look::default() }, rtl);
248
300
  xml.close("w:tc");
249
301
  }
250
302
  xml.close("w:tr");
@@ -252,9 +304,17 @@ fn table(xml: &mut XmlWriter, content: &Table, document: &Document) {
252
304
  xml.close("w:tbl");
253
305
  }
254
306
 
255
- fn section(xml: &mut XmlWriter, document: &Document) {
307
+ /// The document's margin in twips.
308
+ fn margin(document: &Document) -> u32 {
309
+ #[expect(clippy::cast_possible_truncation, clippy::cast_sign_loss, reason = "0 to 2880")]
310
+ let twips = (document.margin() * 20.0).round() as u32;
311
+ twips
312
+ }
313
+
314
+ fn section(xml: &mut XmlWriter, document: &Document, rtl: bool) {
256
315
  let (width, height) = document.page_size().twips();
257
- let (width, height, margin) = (width.to_string(), height.to_string(), MARGIN.to_string());
316
+ let (width, height, margin) =
317
+ (width.to_string(), height.to_string(), margin(document).to_string());
258
318
  xml.open("w:sectPr", &[]);
259
319
  if document.has_page_numbers() {
260
320
  // rId2: the footer, after the styles (see `write`).
@@ -264,6 +324,9 @@ fn section(xml: &mut XmlWriter, document: &Document) {
264
324
  let margins =
265
325
  [("w:top", &margin), ("w:right", &margin), ("w:bottom", &margin), ("w:left", &margin)];
266
326
  xml.empty("w:pgMar", &margins.map(|(key, value)| (key, value.as_str())));
327
+ if rtl {
328
+ xml.empty("w:bidi", &[]);
329
+ }
267
330
  xml.close("w:sectPr");
268
331
  }
269
332