update

Cargo.toml

@@ -12,5 +12,6 @@ ego-tree = "0.10.0"
 env_logger = "0.11.8"
 log = "0.4.27"
 merlin_env_helper = { version = "0.2.0", registry = "merlin" }
+regex = "1.11.1"
 reqwest = {version="0.12.15", features=["blocking"]}
 scraper = "0.23.1"

src/main.rs

@@ -1,10 +1,11 @@
 mod types;
 
-use std::{fs::File, io::Read, net::Incoming, ops::Deref};
+use std::{fs::File, io::Read, net::Incoming, ops::Deref, time::Duration};
 
 use ego_tree::NodeRef;
+use regex::Regex;
 use scraper::{Element, ElementRef, Node};
-use types::Icon;
+use types::{Icon, ParseType};
 
 fn main() -> Result<(), Box<dyn std::error::Error>> {
     env_logger::init();
@@ -98,7 +99,39 @@ fn parse(html: &str) -> Result<bool, Box<dyn std::error::Error>> {
     }
 
     let elt_verwendung = elt_verwendung.unwrap();
+    let mut elt_verwendung = elt_verwendung.parent().unwrap();
 
+    let mut elt_ul = None;
+
+    while elt_verwendung.next_sibling().is_some() {
+        elt_verwendung = elt_verwendung.next_sibling().unwrap();
+
+        if !elt_verwendung.value().is_element() {
+            continue;
+        }
+
+        let elt = elt_verwendung.value().as_element().unwrap();
+        if elt.name() == "ul" {
+            elt_ul = Some(elt_verwendung);
+            break;
+        }
+
+        if elt.name() == "h2" {
+            eprintln!("Found 'h2' element, stopping search for 'ul'");
+            break;
+        }
+    }
+
+    if elt_ul.is_none() {
+        eprintln!("No second 'ul' element found after '#Verwendung'");
+        return Ok(false);
+    }
+    let elt_ul = ElementRef::wrap(elt_ul.unwrap()).unwrap();
+    let li = elt_ul.select(&selector_li);
+
+    for item in li {
+        parse_li_to_resource(item);
+    }
     Ok(true)
 }
 
@@ -108,21 +141,137 @@ fn parse_li_to_resource(item: ElementRef<'_>) {
         return;
     }
 
-    let mut resource_items: Vec<NodeRef<'_, Node>> = Vec::new();
+    let mut resource_items: Vec<ParseType> = Vec::new();
+
     let selector = scraper::Selector::parse("span,a,img,small").unwrap();
     let iter = item.select(&selector);
 
     for child in iter {
-        println!(
+        let elem = child.value();
-            "name: {}, text: {}, text1: {}",
+        let name = elem.name();
-            child.value().name(),
+
-            get_text(child.deref()),
+        if name == "a"
-            get_text1(child.deref())
+            && elem.attrs().find(|attr| attr.0 == "href").is_some()
-        );
+            && elem.attrs().find(|attr| attr.0 == "title").is_some()
+        {
+            resource_items.push(ParseType::Link {
+                url: elem
+                    .attrs()
+                    .find(|attr| attr.0 == "href")
+                    .unwrap()
+                    .1
+                    .to_string(),
+                title: elem
+                    .attrs()
+                    .find(|attr| attr.0 == "title")
+                    .unwrap()
+                    .1
+                    .to_string(),
+            });
+
+            let txt = get_text_next(&child);
+
+            if !txt.is_empty() {
+                parse_text(&txt, &mut resource_items);
+            }
+        } else if name == "img"
+            && elem.attrs().find(|attr| attr.0 == "data-src").is_some()
+            && elem.attrs().find(|attr| attr.0 == "width").is_some()
+            && elem.attrs().find(|attr| attr.0 == "height").is_some()
+            && elem
+                .attrs()
+                .find(|attr| attr.0 == "data-image-name")
+                .is_some()
+        {
+            let url = elem.attrs().find(|attr| attr.0 == "data-src").unwrap().1;
+            let name = elem
+                .attrs()
+                .find(|attr| attr.0 == "data-image-name")
+                .unwrap()
+                .1;
+            let width = elem
+                .attrs()
+                .find(|attr| attr.0 == "width")
+                .unwrap()
+                .1
+                .parse()
+                .unwrap_or(0);
+            let height = elem
+                .attrs()
+                .find(|attr| attr.0 == "height")
+                .unwrap()
+                .1
+                .parse()
+                .unwrap_or(0);
+
+            resource_items.push(ParseType::Img(Icon {
+                name: name.to_string(),
+                url: url.to_string(),
+                width,
+                height,
+                content_type: "image/png".to_string(), // Assuming PNG, adjust as needed
+            }));
+        } else if name == "small" {
+            let txt = get_text(&child);
+            parse_text(&txt, &mut resource_items);
+        }
+    }
+    for item in resource_items.iter() {
+        match item {
+            ParseType::Link { url, title } => {
+                println!("Link: {} - {}", url, title);
+            }
+            ParseType::Img(icon) => {
+                println!(
+                    "Image: {} ({}x{}) - {}",
+                    icon.name, icon.width, icon.height, icon.url
+                );
+            }
+            ParseType::Count(count) => {
+                println!("Count: {}", count);
+            }
+            ParseType::ResourceAdd => {
+                println!("ResourceAdd");
+            }
+            ParseType::ResourceLast => {
+                println!("ResourceLast");
+            }
+            ParseType::Duration { duration, unit } => {
+                println!("Duration: {} {}", duration, unit);
+            }
+        }
     }
     println!("======================");
 }
 
+fn parse_text(text: &str, resource_items: &mut Vec<ParseType>) {
+    let reg1 = Regex::new(r"^\s*x(?<count>\d+)\s+(?<end>[→+])\s*$").unwrap();
+    let reg2 = Regex::new(r"^\s*\(.*(?<duration>\d+,\d+)\ssek\./(?<unit>\w+)\s*\)$").unwrap();
+
+    if let Some(res) = reg1.captures(text) {
+        let count = res.name("count").unwrap().as_str().parse().unwrap_or(0);
+
+        resource_items.push(ParseType::Count(count));
+
+        let end = res.name("end").unwrap().as_str().to_string();
+
+        if end == "+" {
+            resource_items.push(ParseType::ResourceAdd);
+        } else {
+            resource_items.push(ParseType::ResourceLast);
+        }
+    }
+
+    if let Some(res) = reg2.captures(text) {
+        let duration_str = res.name("duration").unwrap().as_str();
+        let duration: f64 = duration_str.replace(',', ".").parse().unwrap_or(0.0);
+        let unit = res.name("unit").unwrap().as_str().to_string();
+        let duration: u64 = (duration * 1000.0) as u64; // Convert to milliseconds
+
+        resource_items.push(ParseType::Duration { duration, unit });
+    }
+}
+
 fn get_text(node: &NodeRef<'_, Node>) -> String {
     if node.value().is_text() {
         return node.value().as_text().unwrap().text.trim().to_string();
@@ -154,7 +303,7 @@ fn get_text(node: &NodeRef<'_, Node>) -> String {
     String::new()
 }
 
-fn get_text1(node: &NodeRef<'_, Node>) -> String {
+fn get_text_next(node: &NodeRef<'_, Node>) -> String {
     if node.value().is_text() {
         return node.value().as_text().unwrap().text.trim().to_string();
     }

src/types/types.rs

@@ -18,3 +18,12 @@ impl Clone for Icon {
         }
     }
 }
+#[derive(Debug, PartialEq, Eq)]
+pub enum ParseType {
+    Link { url: String, title: String },
+    Img(Icon),
+    Count(u32),
+    ResourceAdd,
+    ResourceLast,
+    Duration { duration: u64, unit: String },
+}