2019-04-17 20:26:44 +01:00
|
|
|
package main
|
|
|
|
|
|
|
|
import (
|
|
|
|
"encoding/json"
|
|
|
|
"fmt"
|
|
|
|
"path/filepath"
|
2019-05-05 17:51:51 +01:00
|
|
|
"strings"
|
2019-04-17 20:26:44 +01:00
|
|
|
|
|
|
|
. "./fileio"
|
|
|
|
. "./helpers"
|
|
|
|
. "./structs"
|
|
|
|
"github.com/mmcdole/gofeed"
|
|
|
|
)
|
|
|
|
|
|
|
|
func main() {
|
2019-04-20 20:30:42 +01:00
|
|
|
GetConfig()
|
2019-04-17 20:26:44 +01:00
|
|
|
|
|
|
|
fp := gofeed.NewParser()
|
2019-04-20 20:30:42 +01:00
|
|
|
LogInfo("Downloading " + Config.FeedURL)
|
|
|
|
feed, _ := fp.ParseURL(Config.FeedURL)
|
2019-04-17 20:26:44 +01:00
|
|
|
|
2019-04-20 20:30:42 +01:00
|
|
|
outputDir := Config.OutputPath + "/" + ToCleanString(feed.Title)
|
2019-04-17 20:26:44 +01:00
|
|
|
InitOutputDirectory(outputDir)
|
|
|
|
|
2019-04-20 19:48:28 +01:00
|
|
|
feedInfoPath := outputDir + "/feed_details.json"
|
|
|
|
LogInfo("Writing feed details as JSON to " + feedInfoPath)
|
|
|
|
WriteToFile(feedInfoPath, GrabFeedDetailsJSON(feed))
|
|
|
|
|
2019-05-05 18:06:09 +01:00
|
|
|
feedImagePath := outputDir + "/image" + filepath.Ext(feed.Image.URL)
|
|
|
|
DownloadFile(feedImagePath, feed.Image.URL)
|
|
|
|
|
2019-04-17 20:26:44 +01:00
|
|
|
for _, item := range feed.Items {
|
|
|
|
itemOutputFilename := ToCleanString(
|
|
|
|
item.PublishedParsed.Format("20060102") + "_" + item.Title)
|
|
|
|
itemOutputDir := outputDir + "/" + itemOutputFilename
|
|
|
|
|
|
|
|
if CheckIfExists(itemOutputDir) {
|
2019-04-20 19:48:28 +01:00
|
|
|
fmt.Println("Item '" + item.Title + "' already downloaded, skipping")
|
2019-04-17 20:26:44 +01:00
|
|
|
continue
|
|
|
|
}
|
|
|
|
|
2019-04-20 19:48:28 +01:00
|
|
|
LogInfo("Downloading feed item '" + item.Title + "' to " + itemOutputDir)
|
2019-04-17 20:26:44 +01:00
|
|
|
InitOutputDirectory(itemOutputDir)
|
2019-04-20 19:48:28 +01:00
|
|
|
|
|
|
|
itemDetailsPath := itemOutputDir + "/details.json"
|
|
|
|
LogInfo("Writing details to " + itemDetailsPath)
|
2019-04-17 20:26:44 +01:00
|
|
|
WriteToFile(
|
2019-04-20 19:48:28 +01:00
|
|
|
itemDetailsPath,
|
2019-04-17 20:26:44 +01:00
|
|
|
GrabFeedItemJSON(item))
|
2019-04-20 19:48:28 +01:00
|
|
|
|
|
|
|
itemImagePath := itemOutputDir + "/image" + filepath.Ext(item.Image.URL)
|
|
|
|
LogInfo("Downloading image to " + itemImagePath)
|
2019-04-17 20:26:44 +01:00
|
|
|
DownloadFile(
|
2019-04-20 19:48:28 +01:00
|
|
|
itemImagePath,
|
2019-04-17 20:26:44 +01:00
|
|
|
item.Image.URL)
|
2019-04-20 19:48:28 +01:00
|
|
|
|
2019-04-17 20:26:44 +01:00
|
|
|
for _, enclosure := range item.Enclosures {
|
2019-05-05 17:51:51 +01:00
|
|
|
filename := strings.Split(filepath.Base(enclosure.URL), "?")[0]
|
|
|
|
LogInfo("Downloading attachment '" + filename + "'")
|
2019-04-17 20:26:44 +01:00
|
|
|
DownloadFile(
|
2019-05-05 17:51:51 +01:00
|
|
|
itemOutputDir+"/"+filename,
|
2019-04-17 20:26:44 +01:00
|
|
|
enclosure.URL)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// GrabFeedDetailsJSON - Returns a feed summary json sring
|
|
|
|
func GrabFeedDetailsJSON(feed *gofeed.Feed) string {
|
|
|
|
feedParsed := &FeedDetails{
|
|
|
|
Title: feed.Title,
|
|
|
|
Description: feed.Description,
|
|
|
|
Categories: feed.Categories,
|
|
|
|
Language: feed.Language,
|
|
|
|
Link: feed.Link,
|
2019-05-05 18:06:09 +01:00
|
|
|
FeedLink: feed.FeedLink,
|
2019-04-17 20:26:44 +01:00
|
|
|
Updated: feed.Updated,
|
|
|
|
}
|
|
|
|
|
|
|
|
feedDetails, _ := json.Marshal(feedParsed)
|
|
|
|
return string(feedDetails)
|
|
|
|
}
|
|
|
|
|
|
|
|
// GrabFeedItemJSON - Returns a feed summary json sring
|
|
|
|
func GrabFeedItemJSON(item *gofeed.Item) string {
|
|
|
|
itemParsed := &ItemDetails{
|
|
|
|
Title: item.Title,
|
|
|
|
Description: item.Description,
|
|
|
|
Content: item.Content,
|
|
|
|
Link: item.Link,
|
|
|
|
Updated: item.Updated,
|
|
|
|
Published: item.Published,
|
|
|
|
GUID: item.GUID,
|
|
|
|
Categories: item.Categories,
|
|
|
|
}
|
|
|
|
|
|
|
|
itemDetails, _ := json.Marshal(itemParsed)
|
|
|
|
return string(itemDetails)
|
|
|
|
}
|