130 lines
4.4 KiB
Rust
130 lines
4.4 KiB
Rust
use regex::Regex;
|
|
|
|
/// Convert the Xpath selectors to CSS selector
|
|
///
|
|
/// ## Things it supports
|
|
/// Any tag (including *), contains(), text(), indexed, any attribute (@id, @class, @anything).
|
|
///
|
|
/// ## Things it does not support
|
|
/// Move up to a parent tag (//a/../p) and maybe something else I'm not aware of...
|
|
///
|
|
/// ## Examples
|
|
///
|
|
/// Basic usage:
|
|
/// ```
|
|
/// let result = cssifier::cssifier("//a/p");
|
|
/// assert_eq!(result, Some("a p".to_string()));
|
|
///
|
|
/// let result = cssifier::cssifier("//a/p[@id='hello']");
|
|
/// assert_eq!(result, Some("a p#hello".to_string()));
|
|
///
|
|
/// let result = cssifier::cssifier("//a/p[contains(text(), 'hello')]");
|
|
/// assert_eq!(result, Some("a p:contains(hello)".to_string()));
|
|
///
|
|
/// let result = cssifier::cssifier("*random selector//*"); // Invalid selectors throw a empty string (WIP)
|
|
/// assert_eq!(result, Some("".to_string()));
|
|
/// ```
|
|
pub fn cssifier<S: AsRef<str>>(xpath: S) -> Option<String> {
|
|
// Trait to &str
|
|
let xpath = xpath.as_ref();
|
|
|
|
// Ultra magic regex to parse XPath selectors
|
|
let reg = Regex::new(r#"(?P<node>(^id\(["']?(?P<idvalue>\s*[\w/:][-/\w\s,:;.]*)["']?\)|(?P<nav>//?)(?P<tag>([a-zA-Z][a-zA-Z0-9]{0,10}|\*))(\[((?P<matched>(?P<mattr>@?[.a-zA-Z_:][-\w:.]*(\(\))?)=["'](?P<mvalue>\s*[\w/:][-/\w\s,:;.]*))["']|(?P<contained>contains\((?P<cattr>@?[.a-zA-Z_:][-\w:.]*(\(\))?),\s*["'](?P<cvalue>\s*[\w/:][-/\w\s,:;.]*)["']\)))\])?(\[(?P<nth>\d)\])?))"#).unwrap();
|
|
let mut css = String::new();
|
|
let mut position = 0;
|
|
|
|
while position < xpath.len() {
|
|
let node = reg.captures(&xpath[position..])?;
|
|
let find = reg.find(&xpath[position..])?;
|
|
|
|
// See the nav identifier
|
|
let nav = match position {
|
|
0 => "",
|
|
_ => {
|
|
if node.name("nav")?.as_str() != "//" {
|
|
" "
|
|
} else {
|
|
" > "
|
|
}
|
|
}
|
|
};
|
|
|
|
// See the tag name
|
|
let tag = if node.name("tag")?.as_str() == "*" {
|
|
""
|
|
} else {
|
|
match node.name("tag") {
|
|
Some(tag) => tag.as_str(),
|
|
_ => "",
|
|
}
|
|
};
|
|
|
|
// See the idenfitier attribute of the tag
|
|
let attr = if node.name("idvalue").is_some() {
|
|
format!("#{}", node.name("idvalue")?.as_str().replace(" ", "#"))
|
|
} else if node.name("matched").is_some() {
|
|
let mattr = node.name("mattr")?.as_str();
|
|
let mvalue = node.name("mvalue")?.as_str();
|
|
|
|
if mattr == "@id" {
|
|
format!("#{}", mvalue.replace(" ", "#"))
|
|
} else if mattr == "@class" {
|
|
format!(".{}", mvalue.replace(" ", "."))
|
|
} else if mattr == "text()" || mattr == "." {
|
|
format!(":contains(^{}$)", mvalue)
|
|
} else if !mattr.is_empty() {
|
|
let new_mvalue = if mvalue.contains(' ') {
|
|
format!("\"{}\"", mvalue)
|
|
} else {
|
|
mvalue.to_string()
|
|
};
|
|
format!("[{}={}]", mattr.replace("@", ""), new_mvalue)
|
|
} else {
|
|
String::from("")
|
|
}
|
|
} else if node.name("contained").is_some() {
|
|
let cattr = node.name("cattr")?.as_str();
|
|
let cvalue = node.name("cvalue")?.as_str();
|
|
|
|
if cattr.starts_with('@') {
|
|
format!("[{}*={}]", cattr.replace("@", ""), cvalue)
|
|
} else if cattr == "text()" {
|
|
format!(":contains({})", cvalue)
|
|
} else {
|
|
String::from("")
|
|
}
|
|
} else {
|
|
String::from("")
|
|
};
|
|
|
|
// See the child type
|
|
let nth = if node.name("nth").is_some() {
|
|
format!(":nth-of-type({})", node.name("nth")?.as_str())
|
|
} else {
|
|
String::from("")
|
|
};
|
|
|
|
// Paste all the magic :sparkles:
|
|
css = format!("{}{}{}{}{}", css, nav, tag, attr, nth);
|
|
position += find.end();
|
|
}
|
|
|
|
Some(css)
|
|
}
|
|
|
|
#[cfg(test)]
|
|
mod tests {
|
|
use super::cssifier;
|
|
|
|
#[test]
|
|
fn it_works() {
|
|
assert_eq!(cssifier("//a/b").unwrap(), "a b");
|
|
assert_eq!(cssifier("//a/b[@id='hello']").unwrap(), "a b#hello");
|
|
assert_eq!(
|
|
cssifier("//a/b[contains(text(), 'hello')]").unwrap(),
|
|
"a b:contains(hello)"
|
|
);
|
|
assert_eq!(cssifier("*random shit//*").unwrap(), "");
|
|
}
|
|
}
|