From 93d6d36eb697cd9452eb4aab446151a1a33ed245 Mon Sep 17 00:00:00 2001
From: Adam Mathes <adam@trenchant.org>
Date: Mon, 23 Jan 2017 20:04:03 -0800
Subject: neko v2 initial commit

---
 crawler/crawler.go | 63 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 63 insertions(+)
 create mode 100644 crawler/crawler.go

(limited to 'crawler')

diff --git a/crawler/crawler.go b/crawler/crawler.go
new file mode 100644
index 0000000..e3e4aeb
--- /dev/null
+++ b/crawler/crawler.go
@@ -0,0 +1,63 @@
+package crawler
+
+import (
+	"log"
+	"neko/models/feed"
+	"neko/models/item"
+	"net/http"
+	"time"
+	"github.com/SlyMarbo/rss"
+)
+
+
+func Crawl() {
+
+	ch := make(chan string)
+
+	feeds,err := feed.All()
+	if err != nil {
+		log.Fatal(err)
+	}
+	for _, f := range feeds {
+		log.Printf("crawling %s", f.Url)
+		go CrawlFeed(f, ch)
+	}
+
+	for i := 0; i < len(feeds); i++ {
+		log.Println(<-ch)
+	}
+}
+
+/*
+ TODO: sanitize input on crawl
+*/
+func CrawlFeed(f *feed.Feed, ch chan<- string) {
+	c := &http.Client{
+		// give up after 5 seconds
+		Timeout: 5 * time.Second,
+	}
+
+	feed, err := rss.FetchByClient(f.Url, c)
+	if err != nil {
+		log.Print(err)
+		ch <- "failed to fetch and parse for " + f.Url
+		return
+	}
+
+	f.Title = feed.Title
+	f.Update()
+
+	for _, i := range feed.Items {
+		log.Printf("storing item: %s", i.Title)
+		var item item.Item
+		item.Title = i.Title
+		item.Url = i.Link
+		item.Description = i.Content
+		if item.Description == "" {
+			item.Description = i.Summary
+		}
+		item.FeedId = f.Id
+		item.Create()
+	}
+	ch <- "successfully crawled " + f.Url
+}
-- 
cgit v1.2.3