Skip to main content

document_svg/document/
xpdl.rs

1//! Bounded XPDL workflow-definition previews.
2//!
3//! XPDL exchanges workflow models between tools. The preview reports process
4//! structure only; expressions, scripts, participant values and external
5//! applications are never executed or resolved.
6
7use 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_XPDL_BYTES: u64 = 128 * 1024 * 1024;
16const MAX_XPDL_EVENTS: usize = 1_000_000;
17const MAX_XPDL_NODES: usize = 500_000;
18const MAX_XPDL_DEPTH: usize = 128;
19const MAX_XPDL_TEXT_BYTES: usize = 32 * 1024 * 1024;
20const MAX_XPDL_ROWS: usize = 200_000;
21const MAX_XPDL_DISPLAY_BYTES: usize = 512;
22
23#[derive(Default)]
24struct Summary {
25    version: String,
26    packages: usize,
27    processes: usize,
28    activities: usize,
29    transitions: usize,
30    applications: usize,
31    participants: usize,
32    data_fields: usize,
33    pools: usize,
34    lanes: usize,
35    connectors: usize,
36    external_refs: usize,
37    rows: Vec<Vec<String>>,
38}
39
40struct XpdlPageSink<'a> {
41    inner: &'a mut dyn PageConsumer,
42    warnings: &'a [String],
43}
44
45impl PageConsumer for XpdlPageSink<'_> {
46    fn consume(&mut self, mut page: crate::ir::Page) -> Result<()> {
47        page.source_format = "xpdl".into();
48        if page.title.is_empty() {
49            page.title = "XPDL workflow definition".into();
50        }
51        page.description =
52            "XPDL workflow structure is rendered as bounded inert metadata; expressions, scripts and external applications are not executed".into();
53        for warning in self.warnings {
54            page.warn(warning.clone());
55        }
56        self.inner.consume(page)
57    }
58}
59
60pub(crate) fn looks_like_prefix(prefix: &[u8]) -> bool {
61    crate::geospatial::xml_tree::looks_like_root(prefix, b"Package", None)
62        && String::from_utf8_lossy(prefix)
63            .to_ascii_lowercase()
64            .contains("wfmc.org")
65}
66
67pub(crate) fn convert(
68    path: &Path,
69    options: &ConvertOptions,
70    sink: &mut dyn PageConsumer,
71) -> Result<Vec<String>> {
72    let bytes = read_limited_file(
73        path,
74        options.max_input_bytes.min(MAX_XPDL_BYTES),
75        "XPDL input",
76    )?;
77    let root = parse_xml_tree(
78        &bytes,
79        &XmlLimits {
80            max_events: options.max_xml_events.min(MAX_XPDL_EVENTS),
81            max_nodes: MAX_XPDL_NODES,
82            max_depth: MAX_XPDL_DEPTH,
83            max_text_bytes: MAX_XPDL_TEXT_BYTES,
84        },
85        "XPDL",
86    )?;
87    if !root.name.eq_ignore_ascii_case("Package") {
88        return Err(Error::InvalidInput("XPDL root must be <Package>".into()));
89    }
90    if root
91        .namespace
92        .as_deref()
93        .is_none_or(|namespace| !namespace.to_ascii_lowercase().contains("wfmc.org"))
94    {
95        return Err(Error::InvalidInput(
96            "XPDL root uses an unsupported namespace".into(),
97        ));
98    }
99    let mut summary = Summary {
100        version: attr_local(&root, "xpdlVersion")
101            .or_else(|| attr_local(&root, "version"))
102            .map(truncate)
103            .unwrap_or_default(),
104        packages: 1,
105        processes: count_named(&root, "WorkflowProcess"),
106        activities: count_named(&root, "Activity"),
107        transitions: count_named(&root, "Transition"),
108        applications: count_named(&root, "Application"),
109        participants: count_named(&root, "Participant"),
110        data_fields: count_named(&root, "DataField") + count_named(&root, "FormalParameter"),
111        pools: count_named(&root, "Pool"),
112        lanes: count_named(&root, "Lane"),
113        connectors: count_named(&root, "Connector") + count_named(&root, "ExtendedAttribute"),
114        external_refs: count_named(&root, "ExternalReference")
115            + count_named(&root, "ExternalPackage"),
116        ..Summary::default()
117    };
118    push_row(
119        &mut summary.rows,
120        "Package",
121        &summary.packages.to_string(),
122        &format!("version={}", display_or_dash(&summary.version)),
123    )?;
124    push_row(
125        &mut summary.rows,
126        "Workflow",
127        &summary.processes.to_string(),
128        &format!(
129            "activities={} transitions={}",
130            summary.activities, summary.transitions
131        ),
132    )?;
133    push_row(
134        &mut summary.rows,
135        "Roles",
136        &summary.participants.to_string(),
137        &format!("pools={} lanes={}", summary.pools, summary.lanes),
138    )?;
139    push_row(
140        &mut summary.rows,
141        "Integration",
142        &summary.applications.to_string(),
143        &format!(
144            "dataFields={} connectors={} externalRefs={}",
145            summary.data_fields, summary.connectors, summary.external_refs
146        ),
147    )?;
148    let blocks = vec![
149        HtmlBlock::Heading {
150            level: 1,
151            text: "XPDL workflow definition".into(),
152        },
153        HtmlBlock::Paragraph {
154            text: "Workflow Process Definition Language structure is summarized for model review; expressions, scripts, applications and process execution remain inert.".into(),
155        },
156        HtmlBlock::Table(TableData {
157            headers: vec!["Kind".into(), "Value".into(), "Detail".into()],
158            rows: summary.rows,
159            alignments: vec![TableAlign::Left; 3],
160            raw_source: String::new(),
161        }),
162    ];
163    let warnings = vec![
164        "XPDL process IDs, labels, expressions, scripts, URLs, application parameters and private values are omitted or redacted; only bounded structural counts are shown".into(),
165        "XPDL external packages, application services, scripts, expressions, schemas and workflow engines are never opened or executed".into(),
166    ];
167    let mut page_sink = XpdlPageSink {
168        inner: sink,
169        warnings: &warnings,
170    };
171    render_blocks_to_pages(&blocks, &mut page_sink, options)?;
172    Ok(warnings)
173}
174
175fn count_named(element: &XmlElement, name: &str) -> usize {
176    element
177        .children
178        .iter()
179        .map(|child| usize::from(child.name.eq_ignore_ascii_case(name)) + count_named(child, name))
180        .sum()
181}
182
183fn attr_local<'a>(element: &'a XmlElement, name: &str) -> Option<&'a str> {
184    element.attributes.iter().find_map(|(key, value)| {
185        key.rsplit(':')
186            .next()
187            .filter(|local| local.eq_ignore_ascii_case(name))
188            .map(|_| value.as_str())
189    })
190}
191
192fn display_or_dash(value: &str) -> &str {
193    if value.is_empty() { "—" } else { value }
194}
195
196fn push_row(rows: &mut Vec<Vec<String>>, kind: &str, value: &str, detail: &str) -> Result<()> {
197    if rows.len() >= MAX_XPDL_ROWS {
198        return Err(Error::LimitExceeded(format!(
199            "XPDL rows exceed {MAX_XPDL_ROWS}"
200        )));
201    }
202    rows.push(vec![truncate(kind), truncate(value), truncate(detail)]);
203    Ok(())
204}
205
206fn truncate(value: &str) -> String {
207    if value.len() <= MAX_XPDL_DISPLAY_BYTES {
208        value.to_owned()
209    } else {
210        let mut end = MAX_XPDL_DISPLAY_BYTES;
211        while end > 0 && !value.is_char_boundary(end) {
212            end -= 1;
213        }
214        format!("{}…", &value[..end])
215    }
216}