add scraping

Cargo.toml

@@ -7,9 +7,11 @@ edition = "2021"
 default = []
 
 inventory = ["dep:inventory"]
+scraping = ["dep:scraper"]
 
 [dependencies]
 inventory = { version = "0.3", optional = true }
+scraper = { version = "0.13.0", optional = true }
 tracing = "0.1.35"
 
 [[example]]

src/lib.rs

@@ -16,6 +16,9 @@ mod tests;
 #[macro_use]
 pub mod inventory;
 
+#[cfg(feature = "scraping")]
+pub mod scraping;
+
 /// used to generate css out of classes
 ///
 /// contains shorthands and replacements that can be modified

src/scraping.rs

@@ -0,0 +1,47 @@
+use scraper::{ElementRef, Html};
+
+/// Gets all classes from an html
+///
+/// ```
+/// # use zephyr::{*, scraping::*};
+/// # fn main() {
+/// let c = get_classes("<h1 class=\"m[1rem]\">Hello world!</h1>");
+/// let z = Zephyr::new();
+/// let css = z.generate_classes(c.iter().map(String::as_str));
+/// # }
+/// ```
+pub fn get_classes(html: &str) -> Vec<String> {
+    let document = Html::parse_document(html);
+
+    let mut classes = vec![];
+    let mut queue: Vec<ElementRef> = vec![document.root_element()];
+
+    while let Some(handle) = queue.pop() {
+        let el = handle.value();
+        if let Some(c) = el.attr("class") {
+            classes.push(c.to_string());
+        }
+
+        for child in handle.children() {
+            if let Some(child) = ElementRef::wrap(child) {
+                queue.push(child);
+            }
+        }
+    }
+    classes
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_parse() {
+        let c = get_classes(
+            "<h1 class=\"hey hello\">Hello, <i class=\"hiii\">world!</i></h1>
+<h1 class=\"hey hello\">Hello, <i class=\"hiii\">world!</i></h1>",
+        );
+
+        assert_eq!(c, vec!["hey hello", "hiii", "hey hello", "hiii",]);
+    }
+}