first interface for generated examples

author: Alex Auvolat <alex@adnab.me> 2024-03-11 00:37:59 +0100
committer: Alex Auvolat <alex@adnab.me> 2024-03-11 00:37:59 +0100
commit: ec546004d3b1619c12574f5c0aa0fc4272b7201f (patch)
tree: dbc99900e483fb565a8c06c94eed793334d944c3 /src
parent: d80afc4b77a2d34272e29914280859c836d9efd0 (diff)
download: datagengo-ec546004d3b1619c12574f5c0aa0fc4272b7201f.tar.gz
datagengo-ec546004d3b1619c12574f5c0aa0fc4272b7201f.zip
1 files changed, 102 insertions, 21 deletions
diff --git a/src/server.rs b/src/server.rs
index 58c4dee..c90ee40 100644
--- a/src/server.rs
+++ b/src/server.rs
@@ -1,8 +1,8 @@
 use std::fs;
 use std::io::Write;
-use std::sync::Arc;
 
-use anyhow::anyhow;
+use futures::stream::TryStreamExt;
+use anyhow::{anyhow, Result};
 use rand::prelude::*;
 
 use http_types::mime;
@@ -10,6 +10,7 @@ use tide::Request;
 
 use crate::datafiles::*;
 use crate::format::*;
+use crate::example::*;
 use crate::*;
 
 pub async fn server_main() -> tide::Result<()> {
@@ -64,7 +65,7 @@ pub async fn server_main() -> tide::Result<()> {
 
     // ---- setup http server ----
 
-    let state = Arc::new(StateStruct {
+    let state: State = Box::leak(Box::new(StateStruct {
         jmdict_raw,
         jmdict_xml,
         jmdict_idx,
@@ -73,7 +74,7 @@ pub async fn server_main() -> tide::Result<()> {
         examples,
         example_freq,
         furigana_overrides,
-    });
+    }));
 
     let mut app = tide::with_state(state);
     app.with(tide::log::LogMiddleware::new());
@@ -81,6 +82,8 @@ pub async fn server_main() -> tide::Result<()> {
     app.at("/").get(home_page);
     app.at("/index.html").get(home_page);
     app.at("/style.css").serve_file("static/style.css")?;
+    app.at("/script.js").serve_file("static/script.js")?;
+    app.at("/jquery.js").serve_file("static/jquery.js")?;
     app.at("/about.html").get(about_page);
     app.at("/ex/:start/:end").get(gen_examples_page);
     app.at("/:batch").get(batch_page);
@@ -93,7 +96,7 @@ pub async fn server_main() -> tide::Result<()> {
     Ok(())
 }
 
-type State = Arc<StateStruct>;
+type State = &'static StateStruct;
 #[allow(dead_code)]
 struct StateStruct {
     jmdict_raw: &'static str,
@@ -172,18 +175,92 @@ async fn gen_examples_page(req: Request<State>) -> tide::Result {
             .flatten(),
     );
 
-    let mut examples = gen_examples(&req.state(), &allowed_chars, &needed_chars, 50);
-    for ex in examples.iter_mut() {
-        ex.gen_furigana(&req.state().jmdict_idx, &req.state().furigana_overrides);
-    }
+    let (tx, rx) = async_channel::unbounded();
+
+    let state: State = req.state();
+    std::thread::spawn(move || {
+        tx.send_blocking(Ok(format!(r#"
+        <!DOCTYPE html>
+        <html>
+            <head>
+                <meta charset=\"UTF-8\" />
+                <title>{:03} - {:03} practice</title>
+                <link rel="stylesheet" type="text/css" href="/style.css" />
+                <script src="/jquery.js"></script>
+                <script src="/script.js"></script>
+            </head>
+            <body>
+              <div class="batch_page">
+                <p><a href="index.html">index</a></p>
+                <p>Practice for {:03} - {:03}</p>
+                <hr />
+                <div id="gen_section">
+                  <div id="gen_ex_cnt">
+                  </div>
+                  <div id="gen_ex_display">
+                  </div>
+                  <div id="gen_ex_en">
+                  </div>
+                  <div id="gen_ex_words" class="vocabtable">
+                  </div>
+                </div>
+              </div>
+            </body>
+        "#, first_level, last_level, first_level, last_level).into_bytes()))?;
+
+        gen_examples(state, &allowed_chars, &needed_chars, 50, |mut ex| {
+            ex.gen_furigana(&req.state().jmdict_idx, &req.state().furigana_overrides);
+
+            let mut expl = "<table>".to_string();
+            for word in ex.expl.split(|c| c == ' ' || c == '~') {
+                let (keb, reb) = expl_clean_word(word);
+                let wchars = Charset::new(keb);
+                if !wchars.intersects(&allowed_chars) {
+                    continue;
+                }
+                if let Some(ents) = state.jmdict_idx.get(keb) {
+                    for ent in ents.iter() {
+                        let ent_r_ele = ent.children().find(|x| x.has_tag_name("r_ele")).unwrap();
+                        let ent_reb = ent_r_ele.children().find(|x| x.has_tag_name("reb")).unwrap();
+                        let ent_reb = ent_reb.text().unwrap().trim();
+                        if reb.map(|x| x != ent_reb).unwrap_or(false) {
+                            continue;
+                        }
+                        expl += &format!(r#"<tr><td style="word-break: keep-all">&nbsp;&nbsp;<span class="tab_large font_ja">{}</span>&nbsp;&nbsp;</td><td width="50%">"#, keb);
+
+                        for sense in ent.children().filter(|x| x.has_tag_name("sense")) {
+                            if let Some(s) = sense.children().find(|x| x.has_tag_name("gloss")) {
+                                if !expl.ends_with('>') {
+                                    expl += "; ";
+                                }
+                                expl += s.text().unwrap().trim();
+                            }
+                        }
+                        expl += &format!(r#"</td><td style="word-break: keep-all" class="tab_large font_ja">{}</td></tr>"#, ent_reb);
+                    }
+                }
+            }
 
-    let mut buf: Vec<u8> = vec![];
-    for ex in examples.iter() {
-        write!(&mut buf, "<p>{}</p>", ex.furigana_markup())?;
-    }
+            let item = serde_json::json!({
+                "ja": ex.ja,
+                "en": ex.en,
+                "furi": ex.furigana_markup(),
+                "vocab": expl + "</table>",
+            });
+            tx.send_blocking(Ok(format!("<script> add_example({}); </script>\n", serde_json::to_string(&item)?).into_bytes()))?;
+            Ok(())
+        })?;
+
+        tx.send_blocking(Ok(br#"
+            </body>
+            </html>
+        "#.to_vec()))?;
+
+        Ok::<_, anyhow::Error>(())
+    });
 
     Ok(tide::Response::builder(200)
-        .body(buf)
+        .body(tide::Body::from_reader(Box::pin(rx).into_async_read(), None))
         .content_type(mime::HTML)
         .build())
 }
@@ -200,14 +277,16 @@ fn calc_example_freq(examples: &[Example]) -> HashMap<char, usize> {
     ret
 }
 
-fn gen_examples(
+fn gen_examples<F>(
     data: &StateStruct,
     allowed_chars: &Charset,
     needed_chars: &Charset,
     count: usize,
-) -> Vec<Example> {
+    mut callback: F,
+) -> Result<()>
+where F: FnMut(Example) -> Result<()> {
     let mut rng = thread_rng();
-    let mut ret = vec![];
+    let mut generated = 0;
 
     let mut candidates = data
         .examples
@@ -219,7 +298,7 @@ fn gen_examples(
 
     let mut have_chars = Charset::new("");
     println!("Ex\tMinCnt\tChars\tNeeded\tAllowed\tCandidates\tChars");
-    while ret.len() < count {
+    while generated < count {
         let mut selection = None;
         let mut total_weight = 0f64;
 
@@ -245,11 +324,11 @@ fn gen_examples(
             let (ex, _) = candidates.remove(i);
             remaining_needed = remaining_needed.diff(&ex.chars);
             have_chars = have_chars.union(&ex.chars);
-            ret.push(ex.clone());
 
+            generated += 1;
             println!(
                 "{}\t{}\t{}\t{}\t{}\t{}\t{}",
-                ret.len(),
+                generated,
                 f,
                 have_chars.len(),
                 remaining_needed.len(),
@@ -257,10 +336,12 @@ fn gen_examples(
                 counted,
                 ex.chars.to_string()
             );
+
+            callback(ex.clone())?;
         } else {
             break;
         }
     }
 
-    ret
+    Ok(())
 }
author	Alex Auvolat <alex@adnab.me>	2024-03-11 00:37:59 +0100
committer	Alex Auvolat <alex@adnab.me>	2024-03-11 00:37:59 +0100
commit	ec546004d3b1619c12574f5c0aa0fc4272b7201f (patch)
tree	dbc99900e483fb565a8c06c94eed793334d944c3 /src
parent	d80afc4b77a2d34272e29914280859c836d9efd0 (diff)
download	datagengo-ec546004d3b1619c12574f5c0aa0fc4272b7201f.tar.gz datagengo-ec546004d3b1619c12574f5c0aa0fc4272b7201f.zip