R1tschY / serde_xml_adapt

serde-explicit-xml

More explicit XML Serialization and Deserialization

Credits

This crate uses quick-xml as XML parser and writer.

Most code of this crate is based on quick-xml and serde-xml-rs. The main difference is that attributes are explicitly annotated (name starts with @):

#[derive(Serialize, Deserialize)]
struct Struct {
    #[serde(rename = "@attribute")]
    attribute: String,
    element: String,
}

<root attribute="attribute content">
    <element>element content</element>
</root>

It is therefore not a drop-in replacement for quick-xml or serde-xml-rs.

Other differences:

• $value works for serialisation
• Only true, false, 1, 0 accepted for boolean
• Offset information for deserialization errors

Examples

Also look at the serde examples for inspiration: https://serde.rs/examples.html

Attribute

#[derive(Serialize, Deserialize)]
struct Struct {
    #[serde(rename = "@string")]
    string: String,
    #[serde(rename = "@maybe_string")]
    maybe_string: Option<String>, // will be omitted when None
}

<root string="attribute content"/>

Sequence

#[derive(Serialize, Deserialize)]
struct Struct {
    #[serde(rename = "string")]
    strings: Vec<String>,
}

<root>
    <string>one</string>
    <string>two</string>
</root>

Inner value

Use the special $value as field name the model the inner value of an element:

#[derive(Serialize, Deserialize)]
struct Struct {
    #[serde(rename = "@attr")]
    attr: String,
    #[serde(rename = "$value")]
    value: String,
}

<root attr="">inner value</root>

String enumeration

To model a string with only a strict set of values use an enum with the inner value ($value) as tag. For the example for the values one, two, three:

#[derive(Serialize, Deserialize)]
#[serde(rename_all = "lowercase", tag = "$value")]
enum Enum {
    One,
    Two,
    Three,
}

<root>one</root>

Roadmap

• Panic free