From a5a1a974fc2d0cbb0601d627b9c5cd65021bc0a6 Mon Sep 17 00:00:00 2001
From: Elias Kohout <elias@kohout.de>
Date: Wed, 22 Jan 2025 09:02:32 +0100
Subject: [PATCH] add summarization of article, when collected

---
 src/cmd/frontend/main.go                      | 47 ++++++++++++-------
 src/internal/app/article.go                   |  2 +-
 .../model/database/articlerepository.go       | 15 +++---
 .../model/database/documentrepository.go      | 23 ++++++++-
 4 files changed, 61 insertions(+), 26 deletions(-)

diff --git a/src/cmd/frontend/main.go b/src/cmd/frontend/main.go
index 8fc7f25..65d91a2 100644
--- a/src/cmd/frontend/main.go
+++ b/src/cmd/frontend/main.go
@@ -19,31 +19,46 @@ func main() {
 		log.Fatal("failed to connect to database due to", err.Error())
 	}
 
-    // summarize documents
-    documents := &database.DocumentRepository{DB: db}
+	// summarize documents
+	documents := &database.DocumentRepository{DB: db}
 
-    go documents.Map(func(doc *model.Document) *model.Document {
-        if doc.Summary == "" {
-            summaryText, err := util.Summarize(doc.Content)
-            if err == nil {
-                doc.Summary = summaryText
-                return doc
-            }
-            log.Println(err.Error())
-        }
-        return doc
-    })
+	sumDoc := func(doc *model.Document) *model.Document {
+		if doc.Summary == "" {
+			summaryText, err := util.Summarize(doc.Content)
+			if err == nil {
+				doc.Summary = summaryText
+				return doc
+			}
+			log.Println(err.Error())
+		}
+		return doc
+	}
+	go documents.Map(sumDoc)
 
 	// run web crawlers
-    articles := &database.ArticleRepository{DB: db}
+	articles := &database.ArticleRepository{DB: db}
 	crawler := crawler.CrawlerFacade{}
 	crawler.Init()
 
 	crawler.SubscribeToSpiegelFeed(func(a *model.Article) {
-        articles.Insert(a)
+		id, err := articles.Insert(a)
+		if err == nil {
+			doc, err := documents.ByArticleId(id)
+			if err == nil {
+				doc = sumDoc(doc)
+				documents.Update(doc)
+			}
+		}
 	})
 	crawler.SubscribeToZeitFeed(func(a *model.Article) {
-        articles.Insert(a)
+		id, err := articles.Insert(a)
+		if err == nil {
+			doc, err := documents.ByArticleId(id)
+			if err == nil {
+				doc = sumDoc(doc)
+				documents.Update(doc)
+			}
+		}
 	})
 
 	// define app
diff --git a/src/internal/app/article.go b/src/internal/app/article.go
index 028c666..81f56e2 100644
--- a/src/internal/app/article.go
+++ b/src/internal/app/article.go
@@ -17,7 +17,7 @@ func (app *App) Article(w http.ResponseWriter, req *http.Request) {
 	}
 
 	// get articles
-	article, err := app.articles.ById(int(id))
+	article, err := app.articles.ById(int64(id))
 	if err != nil {
 		// treat as no result
 		http.Error(w, err.Error(), http.StatusInternalServerError)
diff --git a/src/internal/model/database/articlerepository.go b/src/internal/model/database/articlerepository.go
index 467a670..8c6c0a2 100644
--- a/src/internal/model/database/articlerepository.go
+++ b/src/internal/model/database/articlerepository.go
@@ -94,7 +94,7 @@ func (m *ArticleRepository) Search(query string) ([]*model.Article, error) {
 
 // Will return an article given an id. This may fail if the connection to the
 // database fails or there is no aritcle with the given id.
-func (m *ArticleRepository) ById(id int) (*model.Article, error) {
+func (m *ArticleRepository) ById(id int64) (*model.Article, error) {
 	stmt := `
 		SELECT a.id, a.title, a.sourceurl, a.content, a.publishdate, a.fetchDate
 		FROM articles a
@@ -112,15 +112,16 @@ func (m *ArticleRepository) ById(id int) (*model.Article, error) {
 }
 
 // Inserts a new article into the database. The id attribute of the given
-// article will be ignored. May throw an error if the execution of the database
-// query fails.
-func (m *ArticleRepository) Insert(a *model.Article) error {
-	// insert article
+// article will be ignored. Returns the id of the last inserted element. May
+// throw an error if the execution of the database query fails.
+func (m *ArticleRepository) Insert(a *model.Article) (int64, error) {
 	stmt := `INSERT INTO articles (title, sourceUrl, content, publishDate, fetchDate)
              VALUES ($1, $2, $3, $4, $5)
+			 RETURNING id
     `
-	_, err := m.DB.Exec(stmt, a.Title, a.SourceUrl, a.Content, a.PublishDate, a.FetchDate)
-	return err
+	var lastInsertID int64
+	err := m.DB.QueryRow(stmt, a.Title, a.SourceUrl, a.Content, a.PublishDate, a.FetchDate).Scan(&lastInsertID)
+	return lastInsertID, err
 }
 
 // Update an article in the database. Will use the id that is set in the article
diff --git a/src/internal/model/database/documentrepository.go b/src/internal/model/database/documentrepository.go
index 0bb6adc..a9e0618 100644
--- a/src/internal/model/database/documentrepository.go
+++ b/src/internal/model/database/documentrepository.go
@@ -40,12 +40,31 @@ func (d *DocumentRepository) All(limit int, offset int) ([]*model.Document, erro
 	return docs, nil
 }
 
-// Will return an article given an id. This may fail if the connection to the
+// Will return an document given an id. This may fail if the connection to the
 // database fails or there is no aritcle with the given id.
-func (m *DocumentRepository) ById(id int) (*model.Document, error) {
+func (m *DocumentRepository) ById(id int64) (*model.Document, error) {
 	stmt := `
 		SELECT id, content, summary
         FROM documents
+		WHERE id = $1
+	`
+
+	rows := m.DB.QueryRow(stmt, id)
+
+	d := &model.Document{}
+	if err := rows.Scan(&d.Id, &d.Content, &d.Summary); err != nil {
+		return nil, err
+	}
+
+	return d, nil
+}
+
+// Will return an document given an id of an article. This may fail if the
+// connection to the database fails or there is no aritcle with the given id.
+func (m *DocumentRepository) ByArticleId(id int64) (*model.Document, error) {
+	stmt := `
+		SELECT d.id, d.content, d.summary
+        FROM documents d JOIN articles a ON d.id = a.document_id
 		WHERE a.id = $1
 	`