Update go.yml

Download videos by hashtag; limit option; get just json data

.github/workflows/go.yml

@@ -1,5 +1,5 @@
 name: tiktok-dl_CI
-on: [push]
+on: [push, pull_request]
 jobs:
   build:
     strategy:

README.md

@@ -22,6 +22,8 @@ Clone this repository and run `go build` to build the executable.
 * `-batch-file` - File containing URLs/Usernames to download, one value per line. Lines starting with '#', are considered as comments and ignored.
 * `-deadline` - Sets the timout for scraper logic in seconds (used as a workaround for context deadline exceeded error) (default 1500)
 * `-quiet` - Supress output
+* `-json` - Returns whole data, that was scraped from TikTok, in json
+* `-limit` - Sets the max count of video that will be downloaded (default infinity)
 
 ## Acknowledgments
 This software uses the **chromedp** for web scraping, it can be found here: https://github.com/chromedp/chromedp \

client/executeClientAction.go

@@ -2,10 +2,12 @@ package client
 
 import (
 	"context"
+	"errors"
 	"github.com/chromedp/chromedp"
 	"io/ioutil"
 	"log"
 	"os"
+	"strings"
 	"time"
 
 	config "../models/config"
@@ -13,9 +15,11 @@ import (
 )
 
 // GetMusicUploads - Get all uploads by given music
-func executeClientAction(url string, jsAction string) string {
+func executeClientAction(url string, jsAction string) (string, error) {
 	dir, err := ioutil.TempDir("", "chromedp-example")
-	utils.CheckErr(err)
+	if err != nil {
+		return "", err
+	}
 	defer os.RemoveAll(dir)
 
 	opts := append(chromedp.DefaultExecAllocatorOptions[:],
@@ -36,15 +40,16 @@ func executeClientAction(url string, jsAction string) string {
 	ctx, cancel = context.WithTimeout(ctx, time.Duration(config.Config.Deadline)*time.Second)
 	defer cancel()
 
-	var jsOutput string
-	jsOutput = runScrapeWithInfo(ctx, jsAction, url)
-
-	return jsOutput
+	jsOutput, err := runScrapeWithInfo(ctx, jsAction, url)
+	if strings.HasPrefix(jsOutput, "\"ERR:") {
+		err = errors.New(jsOutput)
+	}
+	return jsOutput, err
 }
 
-func runScrapeQuiet(ctx context.Context, jsAction string, url string) string {
+func runScrapeQuiet(ctx context.Context, jsAction string, url string) (string, error) {
 	var jsOutput string
-	err := chromedp.Run(ctx,
+	if err := chromedp.Run(ctx,
 		// Navigate to user's page
 		chromedp.Navigate(url),
 		// Execute url grabber script
@@ -54,33 +59,40 @@ func runScrapeQuiet(ctx context.Context, jsAction string, url string) string {
 		chromedp.WaitVisible(`video_urls`),
 		// Grab url links from our element
 		chromedp.InnerHTML(`video_urls`, &jsOutput),
-	)
-	utils.CheckErr(err)
-	return jsOutput
+	); err != nil {
+		return "", err
+	}
+
+	return jsOutput, nil
 }
 
-func runScrapeWithInfo(ctx context.Context, jsAction string, url string) string {
+func runScrapeWithInfo(ctx context.Context, jsAction string, url string) (string, error) {
 	var jsOutput string
-	err := chromedp.Run(ctx,
+	if err := chromedp.Run(ctx,
 		// Navigate to user's page
 		chromedp.Navigate(url),
 		// Execute url grabber script
 		chromedp.EvaluateAsDevTools(utils.ReadFileAsString("scraper.js"), &jsOutput),
 		chromedp.EvaluateAsDevTools(jsAction, &jsOutput),
-	)
-	utils.CheckErr(err)
+	); err != nil {
+		return "", err
+	}
 
 	for {
-		err = chromedp.Run(ctx, chromedp.EvaluateAsDevTools("currentState.preloadCount.toString()", &jsOutput))
-		utils.CheckErr(err)
+		if err := chromedp.Run(ctx, chromedp.EvaluateAsDevTools("currentState.preloadCount.toString()", &jsOutput)); err != nil {
+			return "", err
+		}
+
 		if jsOutput != "0" {
-			utils.Logf("\rPreloading... Currently loaded %s items.", jsOutput)
+			utils.Logf("\rPreloading... %s items have been founded.", jsOutput)
 		} else {
 			utils.Logf("\rPreloading...")
 		}
 
-		err = chromedp.Run(ctx, chromedp.EvaluateAsDevTools("currentState.finished.toString()", &jsOutput))
-		utils.CheckErr(err)
+		if err := chromedp.Run(ctx, chromedp.EvaluateAsDevTools("currentState.finished.toString()", &jsOutput)); err != nil {
+			return "", err
+		}
+
 		if jsOutput == "true" {
 			break
 		}
@@ -89,13 +101,14 @@ func runScrapeWithInfo(ctx context.Context, jsAction string, url string) string
 	}
 
 	utils.Log("\nRetrieving items...")
-	err = chromedp.Run(ctx,
+	if err := chromedp.Run(ctx,
 		// Wait until custom js finishes
 		chromedp.WaitVisible(`video_urls`),
 		// Grab url links from our element
 		chromedp.InnerHTML(`video_urls`, &jsOutput),
-	)
-	utils.CheckErr(err)
+	); err != nil {
+		return "", err
+	}
 
-	return jsOutput
+	return jsOutput, nil
 }

client/getHashtagUploads.go

@@ -0,0 +1,26 @@
+package client
+
+import (
+	models "../models"
+	config "../models/config"
+	"fmt"
+)
+
+// GetUserUploads - Get all uploads marked with given hashtag
+func GetHashtagUploads(hashtagURL string) ([]models.Upload, error) {
+	jsMethod := fmt.Sprintf("bootstrapIteratingVideos(%d)", config.Config.Limit)
+	actionOutput, err := executeClientAction(hashtagURL, jsMethod)
+	if err != nil {
+		return nil, err
+	}
+	return models.ParseUploads(actionOutput), nil
+}
+
+func GetHashtagUploadsJson(hashtagURL string) (string, error) {
+	jsMethod := fmt.Sprintf("bootstrapIteratingVideos(%d)", config.Config.Limit)
+	actionOutput, err := executeClientAction(hashtagURL, jsMethod)
+	if err != nil {
+		return "", err
+	}
+	return actionOutput, nil
+}

client/getMusicUploads.go

@@ -2,10 +2,25 @@ package client
 
 import (
 	models "../models"
+	config "../models/config"
+	"fmt"
 )
 
 // GetMusicUploads - Get all uploads by given music
-func GetMusicUploads(url string) []models.Upload {
-	actionOutput := executeClientAction(url, "bootstrapIteratingVideos()")
-	return models.ParseUploads(actionOutput)
+func GetMusicUploads(url string) ([]models.Upload, error) {
+	jsMethod := fmt.Sprintf("bootstrapIteratingVideos(%d)", config.Config.Limit)
+	actionOutput, err := executeClientAction(url, jsMethod)
+	if err != nil {
+		return nil, err
+	}
+	return models.ParseUploads(actionOutput), nil
+}
+
+func GetMusicUploadsJson(url string) (string, error) {
+	jsMethod := fmt.Sprintf("bootstrapIteratingVideos(%d)", config.Config.Limit)
+	actionOutput, err := executeClientAction(url, jsMethod)
+	if err != nil {
+		return "", err
+	}
+	return actionOutput, nil
 }

client/getUserUploads.go

@@ -2,10 +2,25 @@ package client
 
 import (
 	models "../models"
+	config "../models/config"
+	"fmt"
 )
 
 // GetUserUploads - Get all uploads by user
-func GetUserUploads(username string) []models.Upload {
-	actionOutput := executeClientAction(`https://www.tiktok.com/@`+username, "bootstrapIteratingVideos()")
-	return models.ParseUploads(actionOutput)
+func GetUserUploads(username string) ([]models.Upload, error) {
+	jsMethod := fmt.Sprintf("bootstrapIteratingVideos(%d)", config.Config.Limit)
+	actionOutput, err := executeClientAction(`https://www.tiktok.com/@`+username, jsMethod)
+	if err != nil {
+		return nil, err
+	}
+	return models.ParseUploads(actionOutput), nil
+}
+
+func GetUserUploadsJson(username string) (string, error) {
+	jsMethod := fmt.Sprintf("bootstrapIteratingVideos(%d)", config.Config.Limit)
+	actionOutput, err := executeClientAction(`https://www.tiktok.com/@`+username, jsMethod)
+	if err != nil {
+		return "", err
+	}
+	return actionOutput, nil
 }

client/getVideoDetails.go

@@ -5,7 +5,10 @@ import (
 )
 
 // GetVideoDetails - returns details of video
-func GetVideoDetails(videoURL string) models.Upload {
-	actionOutput := executeClientAction(videoURL, "bootstrapGetCurrentVideo()")
-	return models.ParseUpload(actionOutput)
+func GetVideoDetails(videoURL string) (models.Upload, error) {
+	actionOutput, err := executeClientAction(videoURL, "bootstrapGetCurrentVideo()")
+	if err != nil {
+		return models.Upload{}, err
+	}
+	return models.ParseUpload(actionOutput), nil
 }

models/config/config.go

@@ -15,6 +15,8 @@ var Config struct {
 	MetaData      bool
 	Quiet         bool
 	Deadline      int
+	Limit         int
+	JSONOnly      bool
 }
 
 // GetConfig - Returns Config object
@@ -25,6 +27,8 @@ func GetConfig() {
 	metadata := flag.Bool("metadata", false, "Write video metadata to a .json file")
 	quiet := flag.Bool("quiet", false, "Supress output")
 	deadline := flag.Int("deadline", 1500, "Sets the timout for scraper logic in seconds (used as a workaround for 'context deadline exceeded' error)")
+	limit := flag.Int("limit", 0, "Sets the videos count limit (useful when there too many videos from the user or by hashtag)")
+	jsonOnly := flag.Bool("json", false, "Just get JSON data from scraper (without video downloading)")
 	flag.Parse()
 
 	args := flag.Args()
@@ -44,5 +48,10 @@ func GetConfig() {
 	Config.Debug = *debug
 	Config.MetaData = *metadata
 	Config.Quiet = *quiet
+	if *jsonOnly {
+		Config.Quiet = true
+	}
 	Config.Deadline = *deadline
+	Config.Limit = *limit
+	Config.JSONOnly = *jsonOnly
 }

resources/strings.go

@@ -6,5 +6,8 @@ var ErrorCouldNotSerializeJSON = "Could not serialize json for video: %s\n"
 // ErrorCouldNotRecogniseURL -
 var ErrorCouldNotRecogniseURL = "Could not recognise URL format of string %s"
 
+// ErrorCouldNotGetUserUploads -
+var ErrorCouldNotGetUserUploads = "Failed to get user uploads: %s\n"
+
 // ErrorPathNotFound -
 var ErrorPathNotFound = "File path %s not found."

scraper.js

@@ -1,7 +1,7 @@
 optStrings = {
     selectors: {
         feedLoading: 'div.tiktok-loading.feed-loading',
-        modalArrowLeft: 'div.video-card-modal > div > img.arrow-right',
+        modalArrowRight: 'div.video-card-modal > div > img.arrow-right',
         modalClose: '.video-card-modal > div > div.close',
         modalPlayer: 'div > div > main > div.video-card-modal > div > div.video-card-big > div.video-card-container > div > div > video',
         modalShareInput: '.copy-link-container > input',
@@ -17,6 +17,7 @@ optStrings = {
     classes: {
         feedVideoItem: 'video-feed-item-wrapper',
         modalCloseDisabled: 'disabled',
+        titleMessage: 'title',
     },
     tags: {
         resultTag: 'video_urls',
@@ -25,11 +26,30 @@ optStrings = {
     attributes: {
         src: "src",
     },
+    tiktokMessages: [
+        "Couldn't find this account",
+        "No videos yet",
+        "Video currently unavailable",
+    ],
 };
 
 currentState = {
     preloadCount: 0,
     finished: false,
+    limit: 0
+};
+
+checkForErrors = function() {
+    var titles = document.getElementsByClassName(optStrings.classes.titleMessage);
+    debugger;
+    if (titles && titles.length) {
+        var error = Array.from(titles).find(x => optStrings.tiktokMessages.includes(x.textContent)).textContent;
+        if (error) {
+            createVidUrlElement("ERR: " + error);
+            return true;
+        }
+    }
+    return false;
 };
 
 createVidUrlElement = function(outputObj) {
@@ -37,7 +57,7 @@ createVidUrlElement = function(outputObj) {
     urlSetElement.innerText = JSON.stringify(outputObj);
     document.getElementsByTagName(optStrings.tags.resultParentTag)[0].appendChild(urlSetElement);
     currentState.finished = true;
-}
+};
 
 buldVidUrlArray = function(finishCallback) {
     var feedItem = document.getElementsByClassName(optStrings.classes.feedVideoItem)[0];
@@ -46,8 +66,14 @@ buldVidUrlArray = function(finishCallback) {
     var videoArray = [];
     var intervalID = window.setInterval(x => {
         videoArray.push(getCurrentModalVideo());
-
-        var arrowRight = document.querySelectorAll(optStrings.selectors.modalArrowLeft)[0];
+        if(currentState.limit > 0) {
+            if (videoArray.length >= currentState.limit) {
+                window.clearInterval(intervalID);
+                document.querySelector(optStrings.selectors.modalClose).click();
+                finishCallback(videoArray);
+            }
+        }
+        var arrowRight = document.querySelectorAll(optStrings.selectors.modalArrowRight)[0];
         if (arrowRight.classList.contains(optStrings.classes.modalCloseDisabled)) {
             window.clearInterval(intervalID);
             document.querySelector(optStrings.selectors.modalClose).click();
@@ -78,9 +104,10 @@ getCurrentModalVideo = function() {
             link: soundHref,
         },
     };
-}
+};
 
 getCurrentVideo = function() {
+    if(checkForErrors()) return;
     var player = document.querySelector(optStrings.selectors.videoPlayer);
     var vidUrl = player.getAttribute(optStrings.attributes.src);
     var shareLink = document.querySelector(optStrings.selectors.videoShareInput).value;
@@ -100,13 +127,23 @@ getCurrentVideo = function() {
             link: soundHref,
         },
     };
-}
+};
 
 scrollWhileNew = function(finishCallback) {
     var state = { count: 0 };
     var intervalID = window.setInterval(x => {
         var oldCount = state.count;
         state.count = document.getElementsByClassName(optStrings.classes.feedVideoItem).length;
+        if(currentState.limit > 0) {
+            if (currentState.preloadCount >= currentState.limit || state.count >= currentState.limit) {
+                finishCallback(createVidUrlElement);
+                window.clearInterval(intervalID);
+            }
+        }
+        if(checkForErrors()) {
+            window.clearInterval(intervalID);
+            return;
+        }
         if (oldCount !== state.count) {
             currentState.preloadCount = state.count;
             window.scrollTo(0, document.body.scrollHeight);
@@ -121,7 +158,8 @@ scrollWhileNew = function(finishCallback) {
     }, 1000);
 };
 
-bootstrapIteratingVideos = function() {
+bootstrapIteratingVideos = function(limit) {
+    currentState.limit = limit;
     scrollWhileNew(buldVidUrlArray);
     return 'bootstrapIteratingVideos';
 };
@@ -130,7 +168,7 @@ bootstrapGetCurrentVideo = function() {
     var video = getCurrentVideo();
     createVidUrlElement(video);
     return 'bootstrapGetCurrentVideo';
-}
+};
 
 init = () => {
     const newProto = navigator.__proto__;

utils/getHashtag.go

@@ -0,0 +1,16 @@
+package utils
+
+import (
+	res "../resources"
+	"fmt"
+	"strings"
+)
+
+// GetHashtagFromURL - Get's tag name from passed url
+func GetHashtagFromURL(str string) string {
+	if match := strings.Contains(str, "/tag/"); match {
+		return strings.Split(str, "/tag/")[1]
+	}
+
+	panic(fmt.Sprintf(res.ErrorCouldNotRecogniseURL, str))
+}

utils/log.go

@@ -3,6 +3,7 @@ package utils
 import (
 	config "../models/config"
 	"fmt"
+	"os"
 )
 
 // Log - Write to std out
@@ -23,3 +24,8 @@ func Logf(format string, a ...interface{}) {
 func LogFatal(format string, a ...interface{}) {
 	panic(fmt.Sprintf(format, a...))
 }
+
+// LogErr - Write error
+func LogErr(format string, a ...interface{}) {
+	fmt.Fprintf(os.Stderr, format, a...)
+}

workflows/downloadHashtag.go

@@ -0,0 +1,45 @@
+package workflows
+
+import (
+	client "../client"
+	config "../models/config"
+	res "../resources"
+	utils "../utils"
+	"fmt"
+	"strings"
+)
+
+// CanUseDownloadHashtag - Test's if this workflow can be used for parameter
+func CanUseDownloadHashtag(url string) bool {
+	match := strings.Contains(url, "/tag/")
+	return match
+}
+
+// DownloadHashtag - Download videos marked with given hashtag
+func DownloadHashtag(url string) {
+	uploads, err := client.GetHashtagUploads(url)
+	if err != nil {
+		utils.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
+		return
+	}
+	uploadCount := len(uploads)
+	hashtag := utils.GetHashtagFromURL(url)
+	downloadDir := fmt.Sprintf("%s/%s", config.Config.OutputPath, hashtag)
+
+	utils.InitOutputDirectory(downloadDir)
+
+	for index, upload := range uploads {
+		downloadVideo(upload, downloadDir)
+		utils.Logf("\r[%d/%d] Downloaded", index+1, uploadCount)
+	}
+	utils.Log()
+}
+
+func GetHashtagJson(url string) {
+	uploads, err := client.GetHashtagUploads(url)
+	if err != nil {
+		utils.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
+		return
+	}
+	fmt.Printf("%s", uploads)
+}

workflows/downloadMusic.go

@@ -3,6 +3,7 @@ package workflows
 import (
 	client "../client"
 	config "../models/config"
+	res "../resources"
 	utils "../utils"
 	"fmt"
 	"regexp"
@@ -16,7 +17,11 @@ func CanUseDownloadMusic(url string) bool {
 
 // DownloadMusic - Download all videos by given music
 func DownloadMusic(url string) {
-	uploads := client.GetMusicUploads(url)
+	uploads, err := client.GetMusicUploads(url)
+	if err != nil {
+		utils.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
+		return
+	}
 	uploadCount := len(uploads)
 
 	for index, upload := range uploads {
@@ -29,3 +34,12 @@ func DownloadMusic(url string) {
 	}
 	utils.Log()
 }
+
+func GetMusicJson(url string) {
+	uploads, err := client.GetMusicUploadsJson(url)
+	if err != nil {
+		utils.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
+		return
+	}
+	fmt.Printf("%s", uploads)
+}

workflows/downloadUser.go

@@ -3,20 +3,27 @@ package workflows
 import (
 	client "../client"
 	config "../models/config"
+	res "../resources"
 	utils "../utils"
 	"fmt"
+	"regexp"
 	"strings"
 )
 
 // CanUseDownloadUser - Test's if this workflow can be used for parameter
 func CanUseDownloadUser(url string) bool {
-	match := strings.Contains(url, "/")
-	return !match
+	isURL := strings.Contains(url, "/")
+	match, _ := regexp.MatchString(".+com\\/@[^\\/]+", url)
+	return !isURL || match
 }
 
 // DownloadUser - Download all user's videos
 func DownloadUser(username string) {
-	uploads := client.GetUserUploads(username)
+	uploads, err := client.GetUserUploads(username)
+	if err != nil {
+		utils.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
+		return
+	}
 	uploadCount := len(uploads)
 	downloadDir := fmt.Sprintf("%s/%s", config.Config.OutputPath, username)
 
@@ -28,3 +35,12 @@ func DownloadUser(username string) {
 	}
 	utils.Log()
 }
+
+func GetUserVideosJson(username string) {
+	uploads, err := client.GetUserUploadsJson(username)
+	if err != nil {
+		utils.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
+		return
+	}
+	fmt.Printf("%s", uploads)
+}

workflows/downloadVideo.go

@@ -4,6 +4,7 @@ import (
 	client "../client"
 	models "../models"
 	config "../models/config"
+	res "../resources"
 	utils "../utils"
 	"fmt"
 	"regexp"
@@ -18,7 +19,11 @@ func CanUseDownloadSingleVideo(url string) bool {
 // DownloadSingleVideo - Downloads single video
 func DownloadSingleVideo(url string) {
 	username := utils.GetUsernameFromString(url)
-	upload := client.GetVideoDetails(url)
+	upload, err := client.GetVideoDetails(url)
+	if err != nil {
+		utils.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
+		return
+	}
 	downloadDir := fmt.Sprintf("%s/%s", config.Config.OutputPath, username)
 
 	utils.InitOutputDirectory(downloadDir)

workflows/startWorkflowByParameter.go

@@ -1,6 +1,7 @@
 package workflows
 
 import (
+	config "../models/config"
 	res "../resources"
 	utils "../utils"
 )
@@ -10,7 +11,11 @@ func StartWorkflowByParameter(url string) {
 
 	// Music
 	if CanUseDownloadMusic(url) {
-		DownloadMusic(url)
+		if config.Config.JSONOnly {
+			GetMusicJson(url)
+		} else {
+			DownloadMusic(url)
+		}
 		return
 	}
 
@@ -22,7 +27,22 @@ func StartWorkflowByParameter(url string) {
 
 	// Tiktok user
 	if CanUseDownloadUser(url) {
-		DownloadUser(utils.GetUsernameFromString(url))
+		if config.Config.JSONOnly {
+			GetUserVideosJson(utils.GetUsernameFromString(url))
+		} else {
+			DownloadUser(utils.GetUsernameFromString(url))
+		}
+
+		return
+	}
+
+	// Tiktok hashtag
+	if CanUseDownloadHashtag(url) {
+		if config.Config.JSONOnly {
+			GetHashtagJson(url)
+		} else {
+			DownloadHashtag(url)
+		}
 		return
 	}