Made parsoid work

Cargo.toml

@@ -4,5 +4,5 @@ version = "0.1.0"
 edition = "2024"
 
 [dependencies]
-mwbot = "0.6.1"
+mwbot = { git = "https://gitlab.wikimedia.org/repos/mwbot-rs/mwbot.git", rev = "05cbb12188f18e2da710de158d89a9a4f1b42689" }
 tokio = { version = "1.46.1", features = ["rt", "rt-multi-thread", "macros"] }

src/main.rs

@@ -1,13 +1,16 @@
 use mwbot::{
     Bot,
-    generators::{CategoryMemberSort, Generator, SortDirection},
+    generators::{Generator, SortDirection, categories::CategoryMemberSort},
-    parsoid::{self, WikinodeIterator},
 };
 use std::{error::Error, path::Path};
 
+use crate::old_style::get_description;
+
+mod old_style;
+
 fn list_teacher_pages(bot: &Bot) -> tokio::sync::mpsc::Receiver<Result<mwbot::Page, mwbot::Error>> {
     let category_title = "Category:Teachers";
-    let pages = mwbot::generators::CategoryMembers::new(category_title)
+    let pages = mwbot::generators::categories::CategoryMembers::new(category_title)
         .dir(SortDirection::Descending)
         .sort(CategoryMemberSort::Timestamp);
 
@@ -17,18 +20,18 @@ fn list_teacher_pages(bot: &Bot) -> tokio::sync::mpsc::Receiver<Result<mwbot::Pa
 }
 
 async fn print_teachers(bot: &Bot) {
-    while let Some(page) = list_teacher_pages(bot).recv().await {
+    let mut v = list_teacher_pages(bot);
+    while let Some(page) = v.recv().await {
         let p = page.unwrap();
-        println!("- {}", p.as_title().dbkey());
+        println!(
-        print_wsdc_id(bot, &p).await;
+            "- {} [{}]",
-    }
+            p.as_title().dbkey(),
-}
+            old_style::get_wsdc_id(bot, &p)
-
+                .await
-async fn print_wsdc_id(bot: &Bot, page: &mwbot::Page) {
+                .map(|x| x.to_string())
-    println!("{}", page.wikitext().await.unwrap());
+                .unwrap_or("Unknown".to_string())
-    let x = page.html().await.unwrap().into_mutable();
+        );
-    for w in &x.filter_external_links() {
+        get_description(bot, &p).await;
-        println!("{}", w.text_contents());
     }
 }
 

src/old_style.rs

@@ -0,0 +1,47 @@
+use mwbot::{Bot, parsoid::WikinodeIterator as _};
+
+fn extract_number_from_url(url: &str) -> Option<u32> {
+    // Split the URL into parts using '/'
+    let parts: Vec<&str> = url.split('/').collect();
+
+    // Iterate over the parts to find the one that ends with ".html"
+    for part in parts {
+        if part.ends_with(".html") {
+            // Remove the ".html" suffix
+            let number_str = part.trim_end_matches(".html");
+            // Parse the remaining string into an integer
+            return number_str.parse().ok();
+        }
+    }
+
+    None
+}
+
+pub async fn get_wsdc_id(bot: &Bot, page: &mwbot::Page) -> Option<u32> {
+    let x = page.html().await.unwrap().into_mutable();
+    for w in &x.filter_external_links() {
+        if let Some(id) = extract_number_from_url(&w.target()) {
+            return Some(id);
+        }
+    }
+    None
+}
+
+pub async fn get_description(bot: &Bot, page: &mwbot::Page) -> Option<String> {
+    let x = page.html().await.unwrap().into_mutable();
+    for w in &x.iter_sections() {
+        dbg!(w);
+        let Some(h) = w.heading() else {
+            println!("No heading");
+            continue;
+        };
+        dbg!(&h);
+        let Some(t) = h.as_text() else {
+            println!("No text");
+            continue;
+        };
+
+        dbg!(t.borrow());
+    }
+    None
+}