xdoc-rs 0.1.1

//! Deterministic XML serialization.
//!
//! Empty elements are serialized as self-closing tags, for example `<Empty/>`.
//! Pretty serialization only inserts indentation for structural content. Elements
//! containing text, CDATA, or mixed content are serialized inline to avoid
//! changing semantically significant whitespace.

use crate::core::{Document, ElementData, ErrorKind, NodeKind, QName, XmlError, XmlResult};

const DEFAULT_ENCODING: &str = "UTF-8";

/// Controls XML serialization.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct WriterConfig {
    pretty: bool,
    include_declaration: bool,
    encoding: String,
    indent: String,
}

impl WriterConfig {
    pub fn compact() -> Self {
        Self {
            pretty: false,
            include_declaration: false,
            encoding: DEFAULT_ENCODING.to_owned(),
            indent: String::new(),
        }
    }

    pub fn pretty() -> Self {
        Self {
            pretty: true,
            include_declaration: false,
            encoding: DEFAULT_ENCODING.to_owned(),
            indent: "  ".to_owned(),
        }
    }

    pub fn with_xml_declaration(mut self, include_declaration: bool) -> Self {
        self.include_declaration = include_declaration;
        self
    }

    pub fn with_encoding(mut self, encoding: impl Into<String>) -> Self {
        self.encoding = encoding.into();
        self
    }

    pub fn with_indent(mut self, indent: impl Into<String>) -> Self {
        self.indent = indent.into();
        self
    }

    pub fn include_declaration(&self) -> bool {
        self.include_declaration
    }

    pub fn encoding(&self) -> &str {
        &self.encoding
    }

    pub fn indent(&self) -> &str {
        &self.indent
    }
}

impl Default for WriterConfig {
    fn default() -> Self {
        Self::compact()
    }
}

/// Serializes a document without extra whitespace or XML declaration.
pub fn to_string_compact(document: &Document) -> XmlResult<String> {
    to_string_with_config(document, &WriterConfig::compact())
}

/// Serializes a document using stable indentation.
pub fn to_string_pretty(document: &Document, config: WriterConfig) -> XmlResult<String> {
    to_string_with_config(
        document,
        &WriterConfig {
            pretty: true,
            ..config
        },
    )
}

/// Serializes a document with an explicit writer configuration.
pub fn to_string_with_config(document: &Document, config: &WriterConfig) -> XmlResult<String> {
    let root = document.root().ok_or_else(|| {
        XmlError::new(
            ErrorKind::InvalidOperation,
            "cannot serialize a document without a root element",
        )
    })?;

    let mut output = String::new();
    if config.include_declaration {
        output.push_str("<?xml version=\"1.0\" encoding=\"");
        output.push_str(&escape_attribute(&config.encoding));
        output.push_str("\"?>");
        if config.pretty {
            output.push('\n');
        }
    }

    write_node(document, root, config, 0, &mut output)?;
    Ok(output)
}

/// Escapes text node content.
pub fn escape_text(text: &str) -> String {
    let mut escaped = String::with_capacity(text.len());
    for ch in text.chars() {
        match ch {
            '&' => escaped.push_str("&amp;"),
            '<' => escaped.push_str("&lt;"),
            '>' => escaped.push_str("&gt;"),
            _ => escaped.push(ch),
        }
    }
    escaped
}

/// Escapes attribute values.
pub fn escape_attribute(value: &str) -> String {
    let mut escaped = String::with_capacity(value.len());
    for ch in value.chars() {
        match ch {
            '&' => escaped.push_str("&amp;"),
            '<' => escaped.push_str("&lt;"),
            '>' => escaped.push_str("&gt;"),
            '"' => escaped.push_str("&quot;"),
            _ => escaped.push(ch),
        }
    }
    escaped
}

fn write_node(
    document: &Document,
    node_id: crate::core::NodeId,
    config: &WriterConfig,
    depth: usize,
    output: &mut String,
) -> XmlResult<()> {
    match document.node(node_id)?.kind() {
        NodeKind::Element(element) => write_element(document, element, config, depth, output)?,
        NodeKind::Text(text) => output.push_str(&escape_text(text)),
        NodeKind::Comment(comment) => write_comment(comment, output)?,
        NodeKind::CData(cdata) => write_cdata(cdata, output)?,
        NodeKind::ProcessingInstruction { target, data } => {
            write_processing_instruction(target, data.as_deref(), output)?
        }
    }

    Ok(())
}

fn write_element(
    document: &Document,
    element: &ElementData,
    config: &WriterConfig,
    depth: usize,
    output: &mut String,
) -> XmlResult<()> {
    output.push('<');
    write_qname(element.name(), output);
    write_namespace_declarations(element, output);
    write_attributes(element, output);

    if element.children().is_empty() {
        output.push_str("/>");
        return Ok(());
    }

    output.push('>');
    if config.pretty {
        if has_textual_content(document, element)? {
            write_inline_children(document, element, config, depth, output)?;
        } else {
            write_pretty_children(document, element, config, depth, output)?;
        }
    } else {
        write_inline_children(document, element, config, depth, output)?;
    }

    output.push_str("</");
    write_qname(element.name(), output);
    output.push('>');
    Ok(())
}

fn has_textual_content(document: &Document, element: &ElementData) -> XmlResult<bool> {
    for child in element.children() {
        match document.node(*child)?.kind() {
            NodeKind::Text(_) | NodeKind::CData(_) => return Ok(true),
            NodeKind::Element(_)
            | NodeKind::Comment(_)
            | NodeKind::ProcessingInstruction { .. } => {}
        }
    }

    Ok(false)
}

fn write_inline_children(
    document: &Document,
    element: &ElementData,
    config: &WriterConfig,
    depth: usize,
    output: &mut String,
) -> XmlResult<()> {
    for child in element.children() {
        write_node(document, *child, config, depth + 1, output)?;
    }

    Ok(())
}

fn write_pretty_children(
    document: &Document,
    element: &ElementData,
    config: &WriterConfig,
    depth: usize,
    output: &mut String,
) -> XmlResult<()> {
    for child in element.children() {
        output.push('\n');
        write_indent(config, depth + 1, output);
        write_node(document, *child, config, depth + 1, output)?;
    }
    output.push('\n');
    write_indent(config, depth, output);
    Ok(())
}

fn write_namespace_declarations(element: &ElementData, output: &mut String) {
    for declaration in element.namespace_declarations() {
        output.push(' ');
        match declaration.prefix() {
            Some(prefix) => {
                output.push_str("xmlns:");
                output.push_str(prefix.as_str());
            }
            None => output.push_str("xmlns"),
        }
        output.push_str("=\"");
        output.push_str(&escape_attribute(declaration.uri().as_str()));
        output.push('"');
    }
}

fn write_attributes(element: &ElementData, output: &mut String) {
    for attribute in element.attributes() {
        output.push(' ');
        write_qname(attribute.name(), output);
        output.push_str("=\"");
        output.push_str(&escape_attribute(attribute.value()));
        output.push('"');
    }
}

fn write_qname(name: &QName, output: &mut String) {
    output.push_str(&name.lexical_name());
}

fn write_comment(comment: &str, output: &mut String) -> XmlResult<()> {
    if comment.contains("--") {
        return Err(XmlError::new(
            ErrorKind::InvalidOperation,
            "XML comments cannot contain `--`",
        ));
    }

    output.push_str("<!--");
    output.push_str(comment);
    output.push_str("-->");
    Ok(())
}

fn write_cdata(cdata: &str, output: &mut String) -> XmlResult<()> {
    if cdata.contains("]]>") {
        return Err(XmlError::new(
            ErrorKind::InvalidOperation,
            "CDATA sections cannot contain `]]>`",
        ));
    }

    output.push_str("<![CDATA[");
    output.push_str(cdata);
    output.push_str("]]>");
    Ok(())
}

fn write_processing_instruction(
    target: &str,
    data: Option<&str>,
    output: &mut String,
) -> XmlResult<()> {
    if data.is_some_and(|data| data.contains("?>")) {
        return Err(XmlError::new(
            ErrorKind::InvalidOperation,
            "processing instruction data cannot contain `?>`",
        ));
    }

    output.push_str("<?");
    output.push_str(target);
    if let Some(data) = data {
        output.push(' ');
        output.push_str(data);
    }
    output.push_str("?>");
    Ok(())
}

fn write_indent(config: &WriterConfig, depth: usize, output: &mut String) {
    for _ in 0..depth {
        output.push_str(config.indent());
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::core::{Attribute, NamespaceDeclaration, QName};
    use crate::testing::assert_xml_eq;

    const SIMPLE_GOLDEN: &str = include_str!("../../tests/golden/writer_simple.xml");
    const NAMESPACES_GOLDEN: &str = include_str!("../../tests/golden/writer_namespaces.xml");
    const PRETTY_GOLDEN: &str = include_str!("../../tests/golden/writer_pretty.xml");
    const PRETTY_MIXED_GOLDEN: &str = include_str!("../../tests/golden/writer_pretty_mixed.xml");
    const PRETTY_CDATA_GOLDEN: &str = include_str!("../../tests/golden/writer_pretty_cdata.xml");
    const PRETTY_STRUCTURAL_MISC_GOLDEN: &str =
        include_str!("../../tests/golden/writer_pretty_structural_misc.xml");

    fn qname(local: &str) -> QName {
        QName::new(local).expect("valid qname")
    }

    fn simple_document() -> Document {
        let mut document = Document::new();
        let root = document.add_root_element(qname("Root")).expect("root");
        let child = document.add_element(root, qname("Child")).expect("child");
        document.add_text(child, "value").expect("text");
        document
    }

    #[test]
    fn writer_serializes_compact_xml() {
        let document = simple_document();

        let xml = to_string_compact(&document).expect("serialized XML");

        assert_eq!(xml, "<Root><Child>value</Child></Root>");
    }

    #[test]
    fn golden_simple_xml_matches_expected_file() {
        let document = simple_document();

        let xml = to_string_compact(&document).expect("serialized XML");

        assert_xml_eq(SIMPLE_GOLDEN, &xml);
    }

    #[test]
    fn golden_namespaces_xml_matches_expected_file() {
        let mut document = Document::new();
        let root = document
            .add_root_element(QName::qualified("doc", "Root", "urn:doc").expect("qname"))
            .expect("root");
        document
            .add_namespace_declaration(
                root,
                NamespaceDeclaration::default("urn:default").expect("namespace"),
            )
            .expect("namespace declaration");
        document
            .add_namespace_declaration(
                root,
                NamespaceDeclaration::prefixed("doc", "urn:doc").expect("namespace"),
            )
            .expect("namespace declaration");
        document
            .add_attribute(
                root,
                Attribute::new(
                    QName::qualified("doc", "id", "urn:doc").expect("qname"),
                    "123",
                ),
            )
            .expect("attribute");

        let xml = to_string_compact(&document).expect("serialized XML");

        assert_xml_eq(NAMESPACES_GOLDEN, &xml);
    }

    #[test]
    fn golden_pretty_xml_matches_expected_file() {
        let document = simple_document();

        let xml = to_string_pretty(&document, WriterConfig::pretty()).expect("serialized XML");

        assert_xml_eq(PRETTY_GOLDEN, &xml);
    }

    #[test]
    fn golden_pretty_preserves_mixed_content() {
        let mut document = Document::new();
        let root = document.add_root_element(qname("Paragraph")).expect("root");
        document.add_text(root, "Hello ").expect("text");
        let bold = document.add_element(root, qname("Bold")).expect("bold");
        document.add_text(bold, "world").expect("bold text");
        document.add_text(root, "!").expect("tail text");

        let xml = to_string_pretty(&document, WriterConfig::pretty()).expect("serialized XML");

        assert_xml_eq(PRETTY_MIXED_GOLDEN, &xml);
    }

    #[test]
    fn golden_pretty_preserves_cdata_content() {
        let mut document = Document::new();
        let root = document.add_root_element(qname("Script")).expect("root");
        document
            .add_cdata(root, "if (a < b) { keep(); }")
            .expect("cdata");

        let xml = to_string_pretty(&document, WriterConfig::pretty()).expect("serialized XML");

        assert_xml_eq(PRETTY_CDATA_GOLDEN, &xml);
    }

    #[test]
    fn golden_pretty_indents_structural_comments_and_processing_instructions() {
        let mut document = Document::new();
        let root = document.add_root_element(qname("Root")).expect("root");
        document.add_comment(root, "generated").expect("comment");
        document
            .add_processing_instruction(root, "xml-stylesheet", Some("href=\"style.xsl\""))
            .expect("processing instruction");
        let child = document.add_element(root, qname("Child")).expect("child");
        document.add_text(child, "value").expect("text");

        let xml = to_string_pretty(&document, WriterConfig::pretty()).expect("serialized XML");

        assert_xml_eq(PRETTY_STRUCTURAL_MISC_GOLDEN, &xml);
    }

    #[test]
    fn escaping_text_escapes_xml_special_characters() {
        let mut document = Document::new();
        let root = document.add_root_element(qname("Root")).expect("root");
        document.add_text(root, "a & b < c > d").expect("text");

        let xml = to_string_compact(&document).expect("serialized XML");

        assert_eq!(xml, "<Root>a &amp; b &lt; c &gt; d</Root>");
    }

    #[test]
    fn escaping_attributes_escapes_xml_special_characters() {
        let mut document = Document::new();
        let root = document.add_root_element(qname("Root")).expect("root");
        document
            .add_attribute(root, Attribute::new(qname("value"), "a & b < c > \"d\""))
            .expect("attribute");

        let xml = to_string_compact(&document).expect("serialized XML");

        assert_eq!(xml, "<Root value=\"a &amp; b &lt; c &gt; &quot;d&quot;\"/>");
    }

    #[test]
    fn writer_config_controls_xml_declaration_and_default_encoding() {
        let document = simple_document();
        let config = WriterConfig::compact().with_xml_declaration(true);

        let xml = to_string_with_config(&document, &config).expect("serialized XML");

        assert!(xml.starts_with("<?xml version=\"1.0\" encoding=\"UTF-8\"?>"));
        assert_eq!(config.encoding(), "UTF-8");
    }

    #[test]
    fn writer_serializes_empty_elements_as_self_closing_tags() {
        let mut document = Document::new();
        document.add_root_element(qname("Empty")).expect("root");

        let xml = to_string_compact(&document).expect("serialized XML");

        assert_eq!(xml, "<Empty/>");
    }

    #[test]
    fn writer_keeps_output_deterministic_and_does_not_modify_document() {
        let document = simple_document();

        let first = to_string_compact(&document).expect("serialized XML");
        let second = to_string_compact(&document).expect("serialized XML");

        assert_eq!(first, second);
        assert_eq!(
            document
                .path(document.root().expect("root"))
                .unwrap()
                .to_string(),
            "/Root"
        );
    }
}