1use std::path::Path;
8
9use crate::convert::{ConvertOptions, PageConsumer, read_limited_file};
10use crate::document::html::{HtmlBlock, render_blocks_to_pages};
11use crate::error::{Error, Result};
12use crate::geospatial::xml_tree::{XmlElement, XmlLimits, parse_xml_tree};
13use crate::table::{TableAlign, TableData};
14
15const MAX_STEPXML_BYTES: u64 = 128 * 1024 * 1024;
16const MAX_STEPXML_EVENTS: usize = 1_000_000;
17const MAX_STEPXML_NODES: usize = 500_000;
18const MAX_STEPXML_DEPTH: usize = 128;
19const MAX_STEPXML_TEXT_BYTES: usize = 32 * 1024 * 1024;
20const MAX_STEPXML_ROWS: usize = 200_000;
21const MAX_STEPXML_DISPLAY_BYTES: usize = 512;
22
23#[derive(Default)]
24struct Summary {
25 version: String,
26 products: usize,
27 product_defs: usize,
28 parts: usize,
29 assemblies: usize,
30 occurrences: usize,
31 representations: usize,
32 geometry: usize,
33 points: usize,
34 directions: usize,
35 properties: usize,
36 external_refs: usize,
37 rows: Vec<Vec<String>>,
38}
39
40struct StepXmlPageSink<'a> {
41 inner: &'a mut dyn PageConsumer,
42 warnings: &'a [String],
43}
44
45impl PageConsumer for StepXmlPageSink<'_> {
46 fn consume(&mut self, mut page: crate::ir::Page) -> Result<()> {
47 page.source_format = "stepxml".into();
48 if page.title.is_empty() {
49 page.title = "STEP-XML product data".into();
50 }
51 page.description = "STEP-XML entity metadata is rendered as bounded inert rows; EXPRESS rules, external schemas and referenced geometry are not evaluated".into();
52 for warning in self.warnings {
53 page.warn(warning.clone());
54 }
55 self.inner.consume(page)
56 }
57}
58
59pub(crate) fn looks_like_prefix(prefix: &[u8]) -> bool {
60 let text = String::from_utf8_lossy(prefix).to_ascii_lowercase();
61 text.contains("iso_10303_28")
62 && text.contains("xmlschema")
63 && (text.contains("product_definition") || text.contains("cartesian_point"))
64}
65
66pub(crate) fn convert(
67 path: &Path,
68 options: &ConvertOptions,
69 sink: &mut dyn PageConsumer,
70) -> Result<Vec<String>> {
71 let bytes = read_limited_file(
72 path,
73 options.max_input_bytes.min(MAX_STEPXML_BYTES),
74 "STEP-XML input",
75 )?;
76 let root = parse_xml_tree(
77 &bytes,
78 &XmlLimits {
79 max_events: options.max_xml_events.min(MAX_STEPXML_EVENTS),
80 max_nodes: MAX_STEPXML_NODES,
81 max_depth: MAX_STEPXML_DEPTH,
82 max_text_bytes: MAX_STEPXML_TEXT_BYTES,
83 },
84 "STEP-XML",
85 )?;
86 if !root.name.to_ascii_lowercase().contains("iso_10303_28") {
87 return Err(Error::InvalidInput(
88 "STEP-XML document must have an iso_10303_28 root".into(),
89 ));
90 }
91 if !root.namespace.as_deref().is_some_and(|namespace| {
92 let namespace = namespace.to_ascii_lowercase();
93 namespace.contains("10303") && namespace.contains("xmlschema")
94 }) {
95 return Err(Error::InvalidInput(
96 "STEP-XML root uses an unsupported namespace".into(),
97 ));
98 }
99 let mut summary = Summary {
100 version: attr_local(&root, "version").unwrap_or_default().to_owned(),
101 ..Summary::default()
102 };
103 summary.products = count_named(&root, "product");
104 summary.product_defs = count_named(&root, "product_definition");
105 summary.parts = count_named(&root, "product_definition_formation");
106 summary.assemblies = count_named(&root, "next_assembly_usage_occurrence")
107 .saturating_add(count_named(&root, "assembly_component"));
108 summary.occurrences = count_named(&root, "product_definition_relationship")
109 .saturating_add(count_named(&root, "component_occurrence"));
110 summary.representations = count_named(&root, "shape_representation")
111 .saturating_add(count_named(&root, "representation"));
112 summary.geometry = count_named(&root, "geometric_representation_item")
113 .saturating_add(count_named(&root, "advanced_brep_shape_representation"))
114 .saturating_add(count_named(&root, "manifold_surface_shape_representation"));
115 summary.points = count_named(&root, "cartesian_point");
116 summary.directions = count_named(&root, "direction");
117 summary.properties = count_named(&root, "property_definition")
118 .saturating_add(count_named(&root, "property_definition_representation"));
119 summary.external_refs = count_named(&root, "external_reference")
120 .saturating_add(count_named(&root, "document_file"));
121 push_row(
122 &mut summary.rows,
123 "Document",
124 "STEP-XML",
125 &format!(
126 "version={} units/context=not evaluated",
127 display_or_dash(&summary.version)
128 ),
129 )?;
130 push_row(
131 &mut summary.rows,
132 "Products",
133 &summary.products.to_string(),
134 &format!(
135 "product_definitions={} formations={}",
136 summary.product_defs, summary.parts
137 ),
138 )?;
139 push_row(
140 &mut summary.rows,
141 "Assembly",
142 &summary.assemblies.to_string(),
143 &format!(
144 "occurrences={} relationships={}",
145 summary.occurrences, summary.occurrences
146 ),
147 )?;
148 push_row(
149 &mut summary.rows,
150 "Geometry",
151 &summary.representations.to_string(),
152 &format!(
153 "items={} points={} directions={}",
154 summary.geometry, summary.points, summary.directions
155 ),
156 )?;
157 push_row(
158 &mut summary.rows,
159 "Properties",
160 &summary.properties.to_string(),
161 &format!("externalReferences={}", summary.external_refs),
162 )?;
163 let blocks = vec![
164 HtmlBlock::Heading {
165 level: 1,
166 text: "STEP-XML product data".into(),
167 },
168 HtmlBlock::Paragraph {
169 text: "ISO 10303-28 XML represents EXPRESS product data. This preview reports bounded structure and never evaluates schemas or referenced CAD geometry.".into(),
170 },
171 HtmlBlock::Table(TableData {
172 headers: vec!["Kind".into(), "Value".into(), "Detail".into()],
173 rows: summary.rows,
174 alignments: vec![TableAlign::Left; 3],
175 raw_source: String::new(),
176 }),
177 ];
178 let warnings = vec![
179 "STEP-XML product, assembly, representation and geometry counts are shown; entity IDs, attribute values, URLs and external references are omitted or redacted".into(),
180 "EXPRESS/schema validation, external schema fetches, CAD geometry resolution, formulas and network operations never run".into(),
181 ];
182 let mut page_sink = StepXmlPageSink {
183 inner: sink,
184 warnings: &warnings,
185 };
186 render_blocks_to_pages(&blocks, &mut page_sink, options)?;
187 Ok(warnings)
188}
189
190fn attr_local<'a>(element: &'a XmlElement, name: &str) -> Option<&'a str> {
191 element.attributes.iter().find_map(|(key, value)| {
192 key.rsplit(':')
193 .next()
194 .filter(|local| local.eq_ignore_ascii_case(name))
195 .map(|_| value.as_str())
196 })
197}
198
199fn count_named(element: &XmlElement, name: &str) -> usize {
200 element
201 .children
202 .iter()
203 .map(|child| usize::from(child.name.eq_ignore_ascii_case(name)) + count_named(child, name))
204 .sum()
205}
206
207fn display_or_dash(value: &str) -> String {
208 if value.is_empty() {
209 "-".into()
210 } else {
211 truncate(value)
212 }
213}
214
215fn truncate(value: &str) -> String {
216 if value.len() <= MAX_STEPXML_DISPLAY_BYTES {
217 value.to_owned()
218 } else {
219 let mut end = MAX_STEPXML_DISPLAY_BYTES;
220 while !value.is_char_boundary(end) {
221 end -= 1;
222 }
223 format!("{}…", &value[..end])
224 }
225}
226
227fn push_row(rows: &mut Vec<Vec<String>>, kind: &str, value: &str, detail: &str) -> Result<()> {
228 if rows.len() >= MAX_STEPXML_ROWS {
229 return Err(Error::LimitExceeded(format!(
230 "STEP-XML rows exceed {MAX_STEPXML_ROWS}"
231 )));
232 }
233 rows.push(vec![truncate(kind), truncate(value), truncate(detail)]);
234 Ok(())
235}