feat: implement task queue statistics and add UI templates for job management and visualization.

pkg/crawler/google/google.go

@@ -13,6 +13,7 @@ import (
 	"strings"
 	"time"
 
+	md "github.com/JohannesKaufmann/html-to-markdown/v2"
 	"github.com/hireme/hireme/internal/crawler"
 	"golang.org/x/net/html"
 )
@@ -130,17 +131,27 @@ func (g *GoogleCrawler) FetchJobDetail(ctx context.Context, applyURL string) (*c
 		location = "See Job Posting"
 	}
 
-	// Raw content: render the <main> container HTML
-	var buf bytes.Buffer
+	// Raw content: extract <main> HTML, convert to Markdown for LLM efficiency
+	var rawMarkdown string
 	doc, parseErr := html.Parse(strings.NewReader(htmlContent))
 	if parseErr == nil {
-		if main := findElement(doc, "main"); main != nil {
-			_ = html.Render(&buf, main)
+		if mainEl := findElement(doc, "main"); mainEl != nil {
+			var buf bytes.Buffer
+			_ = html.Render(&buf, mainEl)
+			converted, convErr := md.ConvertString(buf.String())
+			if convErr == nil && len(converted) > 50 {
+				rawMarkdown = converted
+			}
 		}
 	}
-	if buf.Len() < 100 {
-		buf.Reset()
-		buf.WriteString(htmlContent)
+	if rawMarkdown == "" {
+		// Fallback: convert the whole page
+		converted, convErr := md.ConvertString(htmlContent)
+		if convErr == nil {
+			rawMarkdown = converted
+		} else {
+			rawMarkdown = htmlContent
+		}
 	}
 
 	return &crawler.ExtractedJob{
@@ -149,7 +160,7 @@ func (g *GoogleCrawler) FetchJobDetail(ctx context.Context, applyURL string) (*c
 		Company:     company,
 		Location:    location,
 		ApplyURL:    applyURL,
-		RawContent:  buf.String(),
+		RawContent:  rawMarkdown,
 	}, nil
 }