From e27fdabf78b8a2d03c51c76971fe3e6ce1937d7b Mon Sep 17 00:00:00 2001
From: derfenix <derfenix@gmail.com>
Date: Thu, 16 Nov 2023 22:22:48 +0300
Subject: [PATCH] Fix page meta retrieve

---
 adapters/processors/processors.go      | 23 ++++++++++++++++++++++-
 adapters/processors/processors_test.go | 26 ++++++++++++++++++++++++++
 2 files changed, 48 insertions(+), 1 deletion(-)
 create mode 100644 adapters/processors/processors_test.go

diff --git a/adapters/processors/processors.go b/adapters/processors/processors.go
index bab3c90..f17aff8 100644
--- a/adapters/processors/processors.go
+++ b/adapters/processors/processors.go
@@ -129,8 +129,29 @@ func (p *Processors) GetMeta(ctx context.Context, url string) (entity.Meta, erro
 		return entity.Meta{}, fmt.Errorf("parse response body: %w", err)
 	}
 
+	var fc *html.Node
+	for fc = htmlNode.FirstChild; fc != nil && fc.Data != "html"; fc = fc.NextSibling {
+	}
+
+	if fc == nil {
+		return entity.Meta{}, fmt.Errorf("failed to find html tag")
+	}
+
+	fc = fc.NextSibling
+	if fc == nil {
+		return entity.Meta{}, fmt.Errorf("failed to find html tag")
+	}
+
+	for fc = fc.FirstChild; fc != nil && fc.Data != "head"; fc = fc.NextSibling {
+		fmt.Println(fc.Data)
+	}
+
+	if fc == nil {
+		return entity.Meta{}, fmt.Errorf("failed to find html tag")
+	}
+
 	meta := entity.Meta{}
-	getMetaData(htmlNode, &meta)
+	getMetaData(fc, &meta)
 	meta.Encoding = encodingFromHeader(response.Header)
 
 	return meta, nil
diff --git a/adapters/processors/processors_test.go b/adapters/processors/processors_test.go
new file mode 100644
index 0000000..d226477
--- /dev/null
+++ b/adapters/processors/processors_test.go
@@ -0,0 +1,26 @@
+package processors
+
+import (
+	"context"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	"github.com/derfenix/webarchive/config"
+)
+
+func TestProcessors_GetMeta(t *testing.T) {
+	t.Parallel()
+
+	ctx := context.Background()
+	cfg, err := config.NewConfig(ctx)
+	require.NoError(t, err)
+
+	procs, err := NewProcessors(cfg)
+	require.NoError(t, err)
+
+	meta, err := procs.GetMeta(ctx, "https://habr.com/ru/companies/wirenboard/articles/722718/")
+	require.NoError(t, err)
+	assert.Equal(t, "Сколько стоит умный дом? Рассказываю, как строил свой и что получилось за 1000 руб./м² / Хабр", meta.Title)
+}