Vakyartha documentation

1//! The paged projection the views draw (docs/design/72, P4) and the
2//! Structure listing (U4).
3
4#![allow(clippy::unwrap_used, clippy::expect_used)]
5
6use std::io::Cursor;
7
8use vak_ooxml::projection::{self, PAGE_BYTES};
9use vak_ooxml::{Limits, fixtures, read};
10
11fn read_bytes(bytes: &[u8]) -> read::Document {
12 read::read(Cursor::new(bytes.to_vec()), Limits::default()).unwrap()
13}
14
15#[test]
16fn pages_cover_every_unit_once_in_order() {
17 let document = read_bytes(&fixtures::docx());
18 let mut from = 0;
19 let mut anchors = Vec::new();
20 let mut pages = 0;
21 loop {
22 let page = projection::project(&document, from, 200);
23 assert_eq!(page.from, from);
24 assert!(!page.units.is_empty(), "a page always holds a unit");
25 anchors.extend(page.units.iter().map(|unit| unit.anchor.clone()));
26 pages += 1;
27 match page.next {
28 Some(next) => from = next,
29 None => break,
30 }
31 }
32 assert!(pages > 1, "a small budget pages the document");
33 let all: Vec<String> = document
34 .units
35 .iter()
36 .map(|unit| unit.anchor.clone())
37 .collect();
38 assert_eq!(anchors, all);
39
40 let whole = projection::project(&document, 0, PAGE_BYTES);
41 assert_eq!(whole.next, None);
42 assert_eq!(whole.total_units, document.units.len());
43 assert_eq!(whole.kind, "Word document");
44 assert_eq!(whole.extension, "docx");
45 for entry in &whole.outline {
46 let first = &document.units[entry.first_unit];
47 assert!(
48 entry.units == 0 || first.anchor.starts_with('p') || first.anchor == entry.anchor,
49 "{entry:?} points at {first:?}"
50 );
51 }
52
53 let labelled = projection::project(
54 &read_bytes(&fixtures::signed_labelled_docx()),
55 0,
56 PAGE_BYTES,
57 );
58 assert!(
59 labelled
60 .flags
61 .iter()
62 .any(|flag| flag.contains("digitally signed")),
63 "{:?}",
64 labelled.flags
65 );
66 assert_eq!(labelled.sensitivity_labels, ["Confidential"]);
67}
68
69#[test]
70fn a_sheet_page_carries_its_cells_and_a_slide_deck_its_outline() {
71 let workbook = projection::project(&read_bytes(&fixtures::xlsx()), 0, PAGE_BYTES);
72 assert_eq!(workbook.kind, "Excel workbook");
73 assert!(workbook.outline.iter().any(|entry| entry.title == "Budget"));
74 let row = workbook
75 .units
76 .iter()
77 .find(|unit| unit.cells.iter().any(|(address, _)| address == "B2"))
78 .expect("row 2");
79 assert!(row.anchor.starts_with("Budget!"));
80
81 let deck = projection::project(&read_bytes(&fixtures::pptx()), 0, PAGE_BYTES);
82 assert_eq!(deck.kind, "PowerPoint presentation");
83 assert!(
84 deck.outline
85 .iter()
86 .all(|entry| entry.anchor.starts_with("slide:"))
87 );
88 assert!(!deck.outline.is_empty());
89}
90
91#[test]
92fn a_unit_larger_than_the_page_is_cut_and_says_so() {
93 let long = "Long paragraph text. ".repeat(400);
94 let body = fixtures::MINIMAL_WORD_BODY.replace("Hello", &long);
95 let bytes = fixtures::word_with(fixtures::WORD_MAIN, &body, &[], &[], &[], &[]);
96 let page = projection::project(&read_bytes(&bytes), 0, 2048);
97 assert_eq!(page.units.len(), 1);
98 assert!(page.units[0].text.len() < long.len());
99 assert!(
100 page.units[0]
101 .labels
102 .iter()
103 .any(|label| label.starts_with("shown up to")),
104 "{:?}",
105 page.units[0].labels
106 );
107}
108
109#[test]
110fn structure_lists_parts_types_and_relationships() {
111 let rels = r#"<?xml version="1.0"?><Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships"><Relationship Id="rId9" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/hyperlink" Target="https://example.com/" TargetMode="External"/></Relationships>"#;
112 let bytes = fixtures::with_parts(
113 &fixtures::docx(),
114 &[("word/_rels/document.xml.rels", rels.as_bytes())],
115 );
116 let structure = projection::structure(Cursor::new(bytes), Limits::default()).unwrap();
117 assert_eq!(structure.main_part, "word/document.xml");
118 let main = structure
119 .parts
120 .iter()
121 .find(|part| part.name == "word/document.xml")
122 .unwrap();
123 assert!(
124 main.content_type
125 .as_deref()
126 .is_some_and(|kind| kind.contains("wordprocessingml.document.main"))
127 );
128 assert!(main.size > 0);
129 assert!(
130 structure
131 .relationships
132 .iter()
133 .any(|relationship| relationship.source.is_empty()
134 && relationship.kind == "officeDocument"
135 && relationship.target == "word/document.xml")
136 );
137 assert!(
138 structure
139 .relationships
140 .iter()
141 .any(|relationship| relationship.external
142 && relationship.kind == "hyperlink"
143 && relationship.source == "word/document.xml"),
144 "{:?}",
145 structure.relationships
146 );
147 assert!(structure.untyped_parts.is_empty());
148}
149
150#[test]
151fn a_citation_finds_the_unit_it_names() {
152 let workbook = read_bytes(&fixtures::xlsx());
153 let rows: Vec<(usize, &read::Unit)> = workbook
154 .units
155 .iter()
156 .enumerate()
157 .filter(|(_, unit)| unit.kind == read::UnitKind::SheetRow)
158 .collect();
159 let (index, row) = rows[1];
160 let (cell, _) = &row.cells[1];
161 let sheet = row.anchor.split_once('!').unwrap().0;
162 assert_eq!(
163 projection::locate(&workbook, &row.anchor),
164 Some(index),
165 "the row by its own anchor"
166 );
167 assert_eq!(
168 projection::locate(&workbook, &format!("{sheet}!{cell}")),
169 Some(index),
170 "a cell finds its row"
171 );
172 assert_eq!(
173 projection::locate(&workbook, &format!("{sheet}!{}:Z99", cell.to_lowercase())),
174 Some(index),
175 "a range finds the row of its first cell"
176 );
177 assert_eq!(
178 projection::locate(&workbook, &format!("{sheet}!ZZ9999")),
179 None
180 );
181
182 let deck = read_bytes(&fixtures::pptx());
183 let slide = deck
184 .units
185 .iter()
186 .position(|unit| unit.kind == read::UnitKind::Shape)
187 .unwrap();
188 let slide_anchor = deck.units[slide].anchor.split('/').next().unwrap();
189 let located = projection::locate(&deck, slide_anchor).unwrap();
190 assert!(deck.units[located].anchor.starts_with(slide_anchor));
191 assert!(located <= slide);
192}
193
194#[test]
195fn a_cited_page_opens_at_the_start_of_its_section() {
196 let workbook = read_bytes(&fixtures::xlsx());
197 let rows: Vec<usize> = workbook
198 .units
199 .iter()
200 .enumerate()
201 .filter(|(_, unit)| unit.kind == read::UnitKind::SheetRow)
202 .map(|(index, _)| index)
203 .collect();
204 let cited = rows[1];
205 let sheet = workbook
206 .sections
207 .iter()
208 .find(|section| section.units.contains(&cited))
209 .unwrap();
210 assert_eq!(
211 projection::page_start(&workbook, cited, PAGE_BYTES),
212 sheet.units.start,
213 "a cited row opens with its sheet's first row, the header"
214 );
215 assert_eq!(
216 projection::page_start(&workbook, cited, 1),
217 cited,
218 "when the section does not fit, the page starts at the citation"
219 );
220}
221

1//! The paged projection the views draw (docs/design/72, P4) and the
2//! Structure listing (U4).
3
4#![allow(clippy::unwrap_used, clippy::expect_used)]
5
6use std::io::Cursor;
7
8use vak_ooxml::projection::{self, PAGE_BYTES};
9use vak_ooxml::{Limits, fixtures, read};
10
11fn read_bytes(bytes: &[u8]) -> read::Document {
12 read::read(Cursor::new(bytes.to_vec()), Limits::default()).unwrap()
13}
14
15#[test]
16fn pages_cover_every_unit_once_in_order() {
17 let document = read_bytes(&fixtures::docx());
18 let mut from = 0;
19 let mut anchors = Vec::new();
20 let mut pages = 0;
21 loop {
22 let page = projection::project(&document, from, 200);
23 assert_eq!(page.from, from);
24 assert!(!page.units.is_empty(), "a page always holds a unit");
25 anchors.extend(page.units.iter().map(|unit| unit.anchor.clone()));
26 pages += 1;
27 match page.next {
28 Some(next) => from = next,
29 None => break,
30 }
31 }
32 assert!(pages > 1, "a small budget pages the document");
33 let all: Vec<String> = document
34 .units
35 .iter()
36 .map(|unit| unit.anchor.clone())
37 .collect();
38 assert_eq!(anchors, all);
39
40 let whole = projection::project(&document, 0, PAGE_BYTES);
41 assert_eq!(whole.next, None);
42 assert_eq!(whole.total_units, document.units.len());
43 assert_eq!(whole.kind, "Word document");
44 assert_eq!(whole.extension, "docx");
45 for entry in &whole.outline {
46 let first = &document.units[entry.first_unit];
47 assert!(
48 entry.units == 0 || first.anchor.starts_with('p') || first.anchor == entry.anchor,
49 "{entry:?} points at {first:?}"
50 );
51 }
52
53 let labelled = projection::project(
54 &read_bytes(&fixtures::signed_labelled_docx()),
55 0,
56 PAGE_BYTES,
57 );
58 assert!(
59 labelled
60 .flags
61 .iter()
62 .any(|flag| flag.contains("digitally signed")),
63 "{:?}",
64 labelled.flags
65 );
66 assert_eq!(labelled.sensitivity_labels, ["Confidential"]);
67}
68
69#[test]
70fn a_sheet_page_carries_its_cells_and_a_slide_deck_its_outline() {
71 let workbook = projection::project(&read_bytes(&fixtures::xlsx()), 0, PAGE_BYTES);
72 assert_eq!(workbook.kind, "Excel workbook");
73 assert!(workbook.outline.iter().any(|entry| entry.title == "Budget"));
74 let row = workbook
75 .units
76 .iter()
77 .find(|unit| unit.cells.iter().any(|(address, _)| address == "B2"))
78 .expect("row 2");
79 assert!(row.anchor.starts_with("Budget!"));
80
81 let deck = projection::project(&read_bytes(&fixtures::pptx()), 0, PAGE_BYTES);
82 assert_eq!(deck.kind, "PowerPoint presentation");
83 assert!(
84 deck.outline
85 .iter()
86 .all(|entry| entry.anchor.starts_with("slide:"))
87 );
88 assert!(!deck.outline.is_empty());
89}
90
91#[test]
92fn a_unit_larger_than_the_page_is_cut_and_says_so() {
93 let long = "Long paragraph text. ".repeat(400);
94 let body = fixtures::MINIMAL_WORD_BODY.replace("Hello", &long);
95 let bytes = fixtures::word_with(fixtures::WORD_MAIN, &body, &[], &[], &[], &[]);
96 let page = projection::project(&read_bytes(&bytes), 0, 2048);
97 assert_eq!(page.units.len(), 1);
98 assert!(page.units[0].text.len() < long.len());
99 assert!(
100 page.units[0]
101 .labels
102 .iter()
103 .any(|label| label.starts_with("shown up to")),
104 "{:?}",
105 page.units[0].labels
106 );
107}
108
109#[test]
110fn structure_lists_parts_types_and_relationships() {
111 let rels = r#"<?xml version="1.0"?><Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships"><Relationship Id="rId9" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/hyperlink" Target="https://example.com/" TargetMode="External"/></Relationships>"#;
112 let bytes = fixtures::with_parts(
113 &fixtures::docx(),
114 &[("word/_rels/document.xml.rels", rels.as_bytes())],
115 );
116 let structure = projection::structure(Cursor::new(bytes), Limits::default()).unwrap();
117 assert_eq!(structure.main_part, "word/document.xml");
118 let main = structure
119 .parts
120 .iter()
121 .find(|part| part.name == "word/document.xml")
122 .unwrap();
123 assert!(
124 main.content_type
125 .as_deref()
126 .is_some_and(|kind| kind.contains("wordprocessingml.document.main"))
127 );
128 assert!(main.size > 0);
129 assert!(
130 structure
131 .relationships
132 .iter()
133 .any(|relationship| relationship.source.is_empty()
134 && relationship.kind == "officeDocument"
135 && relationship.target == "word/document.xml")
136 );
137 assert!(
138 structure
139 .relationships
140 .iter()
141 .any(|relationship| relationship.external
142 && relationship.kind == "hyperlink"
143 && relationship.source == "word/document.xml"),
144 "{:?}",
145 structure.relationships
146 );
147 assert!(structure.untyped_parts.is_empty());
148}
149
150#[test]
151fn a_citation_finds_the_unit_it_names() {
152 let workbook = read_bytes(&fixtures::xlsx());
153 let rows: Vec<(usize, &read::Unit)> = workbook
154 .units
155 .iter()
156 .enumerate()
157 .filter(|(_, unit)| unit.kind == read::UnitKind::SheetRow)
158 .collect();
159 let (index, row) = rows[1];
160 let (cell, _) = &row.cells[1];
161 let sheet = row.anchor.split_once('!').unwrap().0;
162 assert_eq!(
163 projection::locate(&workbook, &row.anchor),
164 Some(index),
165 "the row by its own anchor"
166 );
167 assert_eq!(
168 projection::locate(&workbook, &format!("{sheet}!{cell}")),
169 Some(index),
170 "a cell finds its row"
171 );
172 assert_eq!(
173 projection::locate(&workbook, &format!("{sheet}!{}:Z99", cell.to_lowercase())),
174 Some(index),
175 "a range finds the row of its first cell"
176 );
177 assert_eq!(
178 projection::locate(&workbook, &format!("{sheet}!ZZ9999")),
179 None
180 );
181
182 let deck = read_bytes(&fixtures::pptx());
183 let slide = deck
184 .units
185 .iter()
186 .position(|unit| unit.kind == read::UnitKind::Shape)
187 .unwrap();
188 let slide_anchor = deck.units[slide].anchor.split('/').next().unwrap();
189 let located = projection::locate(&deck, slide_anchor).unwrap();
190 assert!(deck.units[located].anchor.starts_with(slide_anchor));
191 assert!(located <= slide);
192}
193
194#[test]
195fn a_cited_page_opens_at_the_start_of_its_section() {
196 let workbook = read_bytes(&fixtures::xlsx());
197 let rows: Vec<usize> = workbook
198 .units
199 .iter()
200 .enumerate()
201 .filter(|(_, unit)| unit.kind == read::UnitKind::SheetRow)
202 .map(|(index, _)| index)
203 .collect();
204 let cited = rows[1];
205 let sheet = workbook
206 .sections
207 .iter()
208 .find(|section| section.units.contains(&cited))
209 .unwrap();
210 assert_eq!(
211 projection::page_start(&workbook, cited, PAGE_BYTES),
212 sheet.units.start,
213 "a cited row opens with its sheet's first row, the header"
214 );
215 assert_eq!(
216 projection::page_start(&workbook, cited, 1),
217 cited,
218 "when the section does not fit, the page starts at the citation"
219 );
220}
221

Projection

Indexing the workspace…

Projection