73 lines
1.6 KiB
Go
Raw Normal View History

package scraper
import (
2022-03-13 17:09:43 -07:00
"fmt"
"path"
"sort"
)
type VideoID int64
// TODO video-source-user: extract source user information (e.g., someone shares a video
// from someone else).
2021-08-01 15:52:04 -07:00
type Video struct {
2022-03-13 17:09:43 -07:00
ID VideoID
TweetID TweetID
Width int
Height int
RemoteURL string
LocalFilename string
2021-12-23 15:12:01 -05:00
2022-03-13 17:09:43 -07:00
ThumbnailRemoteUrl string
ThumbnailLocalPath string `db:"thumbnail_local_filename"`
Duration int // milliseconds
ViewCount int
2021-12-23 15:12:01 -05:00
2022-03-13 17:09:43 -07:00
IsDownloaded bool
IsGif bool
}
func ParseAPIVideo(apiVideo APIExtendedMedia, tweet_id TweetID) Video {
2022-03-13 17:09:43 -07:00
variants := apiVideo.VideoInfo.Variants
sort.Sort(variants)
2022-03-13 17:09:43 -07:00
var view_count int
2021-12-24 16:26:34 -05:00
2022-03-13 17:09:43 -07:00
r := apiVideo.Ext.MediaStats.R
2021-12-24 16:26:34 -05:00
2022-03-13 17:09:43 -07:00
switch r.(type) {
case string:
view_count = 0
case map[string]interface{}:
OK_entry, ok := r.(map[string]interface{})["ok"]
if !ok {
panic("No 'ok' value found in the R!")
}
view_count_str, ok := OK_entry.(map[string]interface{})["viewCount"]
view_count = int_or_panic(view_count_str.(string))
if !ok {
panic("No 'viewCount' value found in the OK!")
}
}
2021-12-24 16:26:34 -05:00
2022-03-13 17:09:43 -07:00
local_filename := fmt.Sprintf("%d.mp4", tweet_id)
2022-03-13 17:09:43 -07:00
return Video{
ID: VideoID(apiVideo.ID),
TweetID: tweet_id,
Width: apiVideo.OriginalInfo.Width,
Height: apiVideo.OriginalInfo.Height,
RemoteURL: variants[0].URL,
LocalFilename: local_filename,
2021-12-23 15:12:01 -05:00
2022-03-13 17:09:43 -07:00
ThumbnailRemoteUrl: apiVideo.MediaURLHttps,
ThumbnailLocalPath: path.Base(apiVideo.MediaURLHttps),
Duration: apiVideo.VideoInfo.Duration,
ViewCount: view_count,
2021-12-23 15:12:01 -05:00
2022-03-13 17:09:43 -07:00
IsDownloaded: false,
IsGif: apiVideo.Type == "animated_gif",
}
}