Search refinements (#1337)

SilasMarvin · web-flow · commit 5905b570ac17 · 2024-03-01T11:13:17.000-08:00
diff --git a/pgml-dashboard/src/api/cms.rs b/pgml-dashboard/src/api/cms.rs
@@ -16,7 +16,7 @@ use crate::{
     guards::Cluster,
     responses::{Response, ResponseOk, Template},
     templates::docs::*,
-    utils::config,
+    utils::{config, markdown::SearchResult},
 };
 use serde::{Deserialize, Serialize};
 use std::fmt;
@@ -626,6 +626,40 @@ impl Collection {
 #[get("/search?<query>", rank = 20)]
 async fn search(query: &str, site_search: &State<crate::utils::markdown::SiteSearch>) -> ResponseOk {
     let results = site_search.search(query, None).await.expect("Error performing search");
+
+    let results: Vec<SearchResult> = results
+        .into_iter()
+        .map(|document| {
+            let snippet = if let Some(description) = document.description {
+                description
+            } else {
+                let author = document.author.unwrap_or_else(|| String::from("xzxzxz"));
+                // The heuristics used here are ok, not the best it will be better when we can just use the description field
+                document
+                    .contents
+                    .lines()
+                    .find(|l| !l.is_empty() && !l.contains(&document.title) && !l.contains(&author) && l.len() > 30)
+                    .unwrap_or("")
+                    .split(' ')
+                    .take(20)
+                    .collect::<Vec<&str>>()
+                    .join(" ")
+                    + "&nbsp;..."
+            };
+            let path = document
+                .path
+                .to_str()
+                .unwrap_or_default()
+                .replace(".md", "")
+                .replace(&config::static_dir().display().to_string(), "");
+            SearchResult {
+                title: document.title,
+                path,
+                snippet,
+            }
+        })
+        .collect();
+
     ResponseOk(
         Template(Search {
             query: query.to_string(),
diff --git a/pgml-dashboard/src/utils/markdown.rs b/pgml-dashboard/src/utils/markdown.rs
@@ -1,6 +1,7 @@
 use crate::api::cms::{DocType, Document};
 use crate::{templates::docs::TocLink, utils::config};
 use anyhow::Context;
+use comrak::{format_html_with_plugins, parse_document, ComrakPlugins};
 use std::cell::RefCell;
 use std::collections::HashMap;
 use std::path::PathBuf;
@@ -21,6 +22,9 @@ use std::fmt;
 use std::sync::Mutex;
 use url::Url;
 
+// Excluded paths in the pgml-cms directory
+const EXCLUDED_DOCUMENT_PATHS: [&str; 1] = ["blog/README.md"];
+
 pub struct MarkdownHeadings {
     header_map: Arc<Mutex<HashMap<String, usize>>>,
 }
@@ -1291,7 +1295,7 @@ impl SiteSearch {
             .collect()
     }
 
-    pub async fn search(&self, query: &str, doc_type: Option<DocType>) -> anyhow::Result<Vec<SearchResult>> {
+    pub async fn search(&self, query: &str, doc_type: Option<DocType>) -> anyhow::Result<Vec<Document>> {
         let mut search = serde_json::json!({
             "query": {
                 // "full_text_search": {
@@ -1323,29 +1327,19 @@ impl SiteSearch {
             "limit": 10
         });
         if let Some(doc_type) = doc_type {
-            search["query"]["filter"] = serde_json::json!({
-                "doc_type": {
-                    "$eq": doc_type
-                }
+            search["query"]["filter"]["doc_type"] = serde_json::json!({
+                "$eq": doc_type
             });
         }
         let results = self.collection.search_local(search.into(), &self.pipeline).await?;
 
         results["results"]
             .as_array()
             .context("Error getting results from search")?
-            .into_iter()
+            .iter()
             .map(|r| {
-                let SearchResultWithoutSnippet { title, contents, path } =
-                    serde_json::from_value(r["document"].clone())?;
-                let path = path
-                    .replace(".md", "")
-                    .replace(&config::static_dir().display().to_string(), "");
-                Ok(SearchResult {
-                    title,
-                    path,
-                    snippet: contents.split(' ').take(20).collect::<Vec<&str>>().join(" ") + "&nbsp;...",
-                })
+                let document: Document = serde_json::from_value(r["document"].clone())?;
+                Ok(document)
             })
             .collect()
     }
@@ -1358,6 +1352,24 @@ impl SiteSearch {
                 .map(|path| async move { Document::from_path(&path).await }),
         )
         .await?;
+        // Filter out documents who only have 1 line (this is usually just an empty document with the title as the first line)
+        // and documents that are in our excluded paths list
+        let documents: Vec<Document> = documents
+            .into_iter()
+            .filter(|f| {
+                !EXCLUDED_DOCUMENT_PATHS
+                    .iter()
+                    .any(|p| f.path == config::cms_dir().join(p))
+                    && !f
+                        .contents
+                        .lines()
+                        .skip(1)
+                        .collect::<Vec<&str>>()
+                        .join("")
+                        .trim()
+                        .is_empty()
+            })
+            .collect();
         let documents: Vec<pgml::types::Json> = documents
             .into_iter()
             .map(|d| {
diff --git a/pgml-dashboard/static/js/search.js b/pgml-dashboard/static/js/search.js
@@ -15,11 +15,16 @@ export default class extends Controller {
         this.target.addEventListener('shown.bs.modal', this.focusSearchInput)
         this.target.addEventListener('hidden.bs.modal', this.updateSearch)
         this.searchInput.addEventListener('input', (e) => this.search(e))
+
+        this.timer;
     }
 
     search(e) {
+        clearTimeout(this.timer);
         const query = e.currentTarget.value
-        this.searchFrame.src = `/search?query=${query}`
+        this.timer = setTimeout(() => {
+            this.searchFrame.src = `/search?query=${query}`
+        }, 250);
     }
 
     focusSearchInput = (e) => {
