Skip to main content

extend_docx/parse/
build.rs

1use crate::model::{DocModel, DocModelMetadata, DocumentNoteDefinition};
2use std::cell::RefCell;
3
4use crate::package::OoxmlPackage;
5
6use super::content_types::parse_content_types;
7use super::context::ParseContext;
8#[cfg(target_arch = "wasm32")]
9use super::document::parse_document_xml;
10use super::metadata::{
11    extract_document_open_tag, extract_section_properties_xml, parse_document_background_color,
12    parse_document_compatibility_settings, parse_document_default_tab_stop_twips,
13    parse_document_page_count_from_app_properties,
14};
15use super::numbering::parse_numbering_with_assets;
16use super::relationships::parse_part_relationships;
17use super::sections::{
18    parse_document_comments, parse_document_notes_from_part, parse_document_sections,
19    parse_footer_sections, parse_header_sections,
20};
21use super::styles::{acquire_document_style_sheet, release_document_style_metadata};
22
23pub fn build_doc_model(pkg: &OoxmlPackage) -> DocModel {
24    #[cfg(not(target_arch = "wasm32"))]
25    {
26        let bytes = pkg.parts.values().map(|part| part.content.len())
27            .chain(pkg.binary_assets.values().map(Vec::len))
28            .fold(0usize, usize::saturating_add);
29        if let Some(pool) = (bytes >= 64 * 1024)
30            .then(super::namespaces::worker_pool).flatten()
31        {
32            let mut identity = None;
33            let mut model = pool.in_place_scope(|scope| {
34                scope.spawn(|_| {
35                    identity = Some(crate::structured::package_identity(pkg));
36                });
37                build_doc_model_contents(pkg)
38            });
39            crate::structured::assign_source_ids(&mut model, &identity.expect("joined identity task"));
40            return model;
41        }
42    }
43    let identity = crate::structured::package_identity(pkg);
44    let mut model = build_doc_model_contents(pkg);
45    crate::structured::assign_source_ids(&mut model, &identity);
46    model
47}
48
49fn build_doc_model_contents(pkg: &OoxmlPackage) -> DocModel {
50    let _style_cache = super::style::run_style_cache_scope();
51    let binary_assets = &pkg.binary_assets;
52    let (normalized, indexes) = super::namespaces::normalized_package(pkg);
53    let pkg = &normalized;
54    let _relationship_cache = crate::package::relationship_cache_scope(pkg);
55    let _xml_index = crate::xml_index::scope(
56        pkg.parts
57            .values()
58            .map(|part| (part.name.as_str(), part.content.as_str())),
59        indexes,
60    );
61    let mut warnings = pkg.warnings.clone();
62    warnings.extend(
63        crate::source_graph::preserved_features(pkg)
64            .into_iter()
65            .map(|feature| {
66                format!(
67                    "{}: {} occurrence(s) of {} have support level {}",
68                    feature.part_name, feature.occurrences, feature.feature, feature.support
69                )
70            }),
71    );
72    let main_part = crate::package::main_document_part_name(pkg);
73    let source_part_names = ["footnotes", "endnotes", "comments"]
74        .into_iter()
75        .map(|kind| {
76            (
77                kind.to_string(),
78                crate::package::related_document_part_name(pkg, kind),
79            )
80        })
81        .chain(std::iter::once(("document".into(), main_part.clone())))
82        .collect();
83    let document_xml = pkg.parts.get(&main_part).map(|part| part.content.as_str());
84    if document_xml.is_none() {
85        warnings.push("Missing word/document.xml".to_string());
86    }
87
88    let resolved_document_xml = document_xml.unwrap_or("");
89    let document_open_tag = extract_document_open_tag(resolved_document_xml);
90    let document_background_color = parse_document_background_color(resolved_document_xml);
91    let document_page_count = parse_document_page_count_from_app_properties(pkg);
92    let compatibility = parse_document_compatibility_settings(pkg);
93    let default_tab_stop_twips = parse_document_default_tab_stop_twips(pkg);
94    let section_properties_xml = extract_section_properties_xml(resolved_document_xml);
95    let content_types = parse_content_types(pkg);
96    let style_handle = acquire_document_style_sheet(pkg);
97    let style_sheet: &super::context::ParsedStyleSheet = std::borrow::Borrow::borrow(&style_handle);
98    super::style::bind_style_cache(style_sheet);
99    let numbering_definitions = parse_numbering_with_assets(pkg, &content_types, binary_assets);
100    let document_relationships = parse_part_relationships(pkg, &main_part);
101
102    let mut part_cache =
103        super::sections::PartNodeCache::new(pkg, &content_types, style_sheet, binary_assets);
104    let header_sections = parse_header_sections(
105        pkg,
106        resolved_document_xml,
107        &document_relationships,
108        &mut warnings,
109        &mut part_cache,
110    );
111    let footer_sections = parse_footer_sections(
112        pkg,
113        resolved_document_xml,
114        &document_relationships,
115        &mut warnings,
116        &mut part_cache,
117    );
118    let sections = parse_document_sections(
119        resolved_document_xml,
120        &document_relationships,
121        &mut warnings,
122        &mut part_cache,
123    );
124    drop(part_cache);
125    let footnotes = parse_document_notes_from_part(
126        pkg,
127        &crate::package::related_document_part_name(pkg, "footnotes"),
128        "w:footnote",
129        &content_types,
130        style_sheet,
131        &mut warnings,
132        binary_assets,
133    );
134    let endnotes = parse_document_notes_from_part(
135        pkg,
136        &crate::package::related_document_part_name(pkg, "endnotes"),
137        "w:endnote",
138        &content_types,
139        style_sheet,
140        &mut warnings,
141        binary_assets,
142    );
143    let comments = parse_document_comments(pkg, &content_types, style_sheet, binary_assets);
144
145    let context = ParseContext {
146        relationships: document_relationships.clone(),
147        content_types: content_types.clone(),
148        parts: &pkg.parts,
149        binary_assets,
150        style_sheet: style_sheet,
151        warnings: RefCell::new(Vec::new()),
152    };
153
154    #[cfg(not(target_arch = "wasm32"))]
155    let nodes =
156        super::document::parse_document_xml_parallel(resolved_document_xml, &context, &_xml_index);
157    #[cfg(target_arch = "wasm32")]
158    let nodes = parse_document_xml(resolved_document_xml, &context);
159    warnings.extend(context.warnings.borrow().iter().cloned());
160
161    let (paragraph_styles, default_paragraph_style_id) =
162        release_document_style_metadata(style_handle);
163    let model = DocModel {
164        nodes,
165        metadata: DocModelMetadata {
166            document_id: None,
167            source_part_names: Some(source_part_names),
168            source_parts: pkg.parts.len() as i64,
169            warnings,
170            document_page_count,
171            document_open_tag,
172            document_background_color,
173            section_properties_xml,
174            sections: Some(sections),
175            header_sections,
176            footer_sections,
177            paragraph_styles,
178            default_paragraph_style_id,
179            default_tab_stop_twips,
180            document_settings_imported: crate::package::related_document_part(pkg, "settings")
181                .is_some()
182                .then_some(true),
183            line_spacing_compatibility_imported: crate::package::related_document_part(
184                pkg, "settings",
185            )
186            .is_some()
187            .then_some(true),
188            numbering_definitions,
189            compatibility,
190            footnotes: clone_notes_if_non_empty(footnotes),
191            endnotes: clone_notes_if_non_empty(endnotes),
192            comments: if comments.is_empty() {
193                None
194            } else {
195                Some(comments)
196            },
197        },
198    };
199    model
200}
201
202fn clone_notes_if_non_empty(
203    notes: Vec<DocumentNoteDefinition>,
204) -> Option<Vec<DocumentNoteDefinition>> {
205    if notes.is_empty() {
206        return None;
207    }
208
209    Some(notes)
210}
211
212#[cfg(test)]
213mod tests {
214    use std::collections::HashMap;
215
216    use super::*;
217    use crate::package::OoxmlPart;
218
219    #[test]
220    fn build_doc_model_warns_when_document_missing() {
221        let pkg = OoxmlPackage {
222            parts: HashMap::new(),
223            binary_assets: HashMap::new(),
224            warnings: Vec::new(),
225        };
226        let model = build_doc_model(&pkg);
227        assert_eq!(model.nodes.len(), 1);
228        assert!(model
229            .metadata
230            .warnings
231            .iter()
232            .any(|warning| warning.contains("Missing word/document.xml")));
233    }
234
235    #[test]
236    fn build_doc_model_parses_empty_document() {
237        let mut parts = HashMap::new();
238        parts.insert(
239            "word/document.xml".to_string(),
240            OoxmlPart {
241                name: "word/document.xml".to_string(),
242                content: "<w:document><w:body/></w:document>".to_string(),
243            },
244        );
245        let pkg = OoxmlPackage {
246            parts,
247            binary_assets: HashMap::new(),
248            warnings: Vec::new(),
249        };
250        let model = build_doc_model(&pkg);
251        assert_eq!(model.nodes.len(), 1);
252        assert!(model.metadata.sections.is_some());
253    }
254
255    #[test]
256    fn build_doc_model_preserves_adapter_warnings() {
257        let pkg = OoxmlPackage {
258            parts: HashMap::new(),
259            binary_assets: HashMap::new(),
260            warnings: vec!["Legacy DOC adapter warning".to_string()],
261        };
262        let model = build_doc_model(&pkg);
263        assert!(model
264            .metadata
265            .warnings
266            .iter()
267            .any(|warning| warning == "Legacy DOC adapter warning"));
268    }
269}
270
271#[cfg(test)]
272mod package_preparation_tests {
273    #[test]
274    fn parallel_build_preserves_identity_and_normalized_source_count() {
275        let mut parts = std::collections::HashMap::new();
276        for index in 0..6 {
277            let name = format!("part-{index}.xml");
278            parts.insert(name.clone(), crate::package::OoxmlPart { name, content: format!("<w:root xmlns:w=\"http://schemas.openxmlformats.org/wordprocessingml/2006/main\"><w:t>{}</w:t></w:root>", "text &amp; detail ".repeat(1000)) });
279        }
280        let package = crate::package::OoxmlPackage {
281            parts,
282            binary_assets: std::collections::HashMap::from([("asset".into(), vec![5; 32 * 1024])]),
283            warnings: vec!["prior warning".into()],
284        };
285        let expected_identity = crate::structured::package_identity(&package);
286        let (expected, _) = crate::parse::namespaces::normalized_package(&package);
287        let mut serial = super::build_doc_model_contents(&package);
288        crate::structured::assign_source_ids(&mut serial, &expected_identity);
289        let actual = super::build_doc_model(&package);
290        assert_eq!(actual.metadata.document_id.as_deref(), Some(expected_identity.as_str()));
291        assert_eq!(actual.metadata.source_parts, expected.parts.len() as i64);
292        assert_eq!(serde_json::to_string(&actual).unwrap(), serde_json::to_string(&serial).unwrap());
293    }
294}
295
296#[cfg(all(test, not(target_arch = "wasm32")))]
297mod identity_overlap_tests {
298    use super::*;
299    use std::collections::HashMap;
300    use crate::package::OoxmlPart;
301
302    #[test]
303    fn overlapped_identity_matches_serial_construction() {
304        for asset_size in [0, 65536, 262145] {
305            let document = r#"<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main"><w:body><w:p><w:r><w:t>content</w:t></w:r></w:p></w:body></w:document>"#;
306            let part = OoxmlPart { name: "word/document.xml".into(), content: document.into() };
307            let mut package = OoxmlPackage { parts: HashMap::from([(part.name.clone(), part)]), binary_assets: HashMap::new(), warnings: Vec::new() };
308            package.binary_assets.insert("word/media/payload.bin".into(), (0..asset_size).map(|n| n as u8).collect());
309            let mut serial = build_doc_model_contents(&package);
310            crate::structured::assign_source_ids(&mut serial, &crate::structured::package_identity(&package));
311            let parallel = build_doc_model(&package);
312            assert_eq!(serde_json::to_string(&serial).unwrap(), serde_json::to_string(&parallel).unwrap());
313            package.binary_assets.get_mut("word/media/payload.bin").unwrap().push(7);
314            let changed = build_doc_model(&package);
315            assert_ne!(parallel.metadata.document_id, changed.metadata.document_id);
316        }
317    }
318}