mirror of
https://github.com/pikami/tiktok-dl.git
synced 2025-12-21 01:39:51 +00:00
Compare commits
11 Commits
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
feee0a9154 | ||
|
|
af7972685e | ||
|
|
f9d35e3bf2 | ||
|
|
9a65746fd4 | ||
|
|
70c605a696 | ||
|
|
208bffb846 | ||
|
|
7b9b7688a1 | ||
|
|
e77c904f89 | ||
|
|
68612282ee | ||
|
|
b6bb470064 | ||
|
|
f724f0f2a2 |
2
.github/workflows/go.yml
vendored
2
.github/workflows/go.yml
vendored
@@ -1,5 +1,5 @@
|
|||||||
name: tiktok-dl_CI
|
name: tiktok-dl_CI
|
||||||
on: [push]
|
on: [push, pull_request]
|
||||||
jobs:
|
jobs:
|
||||||
build:
|
build:
|
||||||
strategy:
|
strategy:
|
||||||
|
|||||||
@@ -16,11 +16,14 @@ You can download items listed in a text file by running `./tiktok-dl [OPTIONS] -
|
|||||||
Clone this repository and run `go build` to build the executable.
|
Clone this repository and run `go build` to build the executable.
|
||||||
|
|
||||||
## Available options
|
## Available options
|
||||||
* `-debug` - enables debug mode
|
* `-archive` - Download only videos not listed in the archive file. Record the IDs of all downloaded videos in it.
|
||||||
* `-output some_directory` - Output path (default "./downloads")
|
|
||||||
* `-metadata` - Write video metadata to a .json file
|
|
||||||
* `-batch-file` - File containing URLs/Usernames to download, one value per line. Lines starting with '#', are considered as comments and ignored.
|
* `-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)
|
* `-deadline` - Sets the timout for scraper logic in seconds (used as a workaround for context deadline exceeded error) (default 1500)
|
||||||
|
* `-debug` - enables debug mode
|
||||||
|
* `-json` - Returns whole data, that was scraped from TikTok, in json
|
||||||
|
* `-limit` - Sets the max count of video that will be downloaded (default infinity)
|
||||||
|
* `-metadata` - Write video metadata to a .json file
|
||||||
|
* `-output some_directory` - Output path (default "./downloads")
|
||||||
* `-quiet` - Supress output
|
* `-quiet` - Supress output
|
||||||
|
|
||||||
## Acknowledgments
|
## Acknowledgments
|
||||||
|
|||||||
@@ -3,15 +3,16 @@ package client
|
|||||||
import (
|
import (
|
||||||
"context"
|
"context"
|
||||||
"errors"
|
"errors"
|
||||||
"github.com/chromedp/chromedp"
|
|
||||||
"io/ioutil"
|
"io/ioutil"
|
||||||
"log"
|
|
||||||
"os"
|
"os"
|
||||||
"strings"
|
"strings"
|
||||||
"time"
|
"time"
|
||||||
|
|
||||||
|
"github.com/chromedp/chromedp"
|
||||||
|
|
||||||
config "../models/config"
|
config "../models/config"
|
||||||
utils "../utils"
|
utils "../utils"
|
||||||
|
log "../utils/log"
|
||||||
)
|
)
|
||||||
|
|
||||||
// GetMusicUploads - Get all uploads by given music
|
// GetMusicUploads - Get all uploads by given music
|
||||||
@@ -33,7 +34,7 @@ func executeClientAction(url string, jsAction string) (string, error) {
|
|||||||
|
|
||||||
ctx, cancel := chromedp.NewContext(
|
ctx, cancel := chromedp.NewContext(
|
||||||
allocCtx,
|
allocCtx,
|
||||||
chromedp.WithLogf(log.Printf),
|
chromedp.WithLogf(log.Logf),
|
||||||
)
|
)
|
||||||
defer cancel()
|
defer cancel()
|
||||||
|
|
||||||
@@ -84,9 +85,9 @@ func runScrapeWithInfo(ctx context.Context, jsAction string, url string) (string
|
|||||||
}
|
}
|
||||||
|
|
||||||
if jsOutput != "0" {
|
if jsOutput != "0" {
|
||||||
utils.Logf("\rPreloading... Currently loaded %s items.", jsOutput)
|
log.Logf("\rPreloading... %s items have been found.", jsOutput)
|
||||||
} else {
|
} else {
|
||||||
utils.Logf("\rPreloading...")
|
log.Logf("\rPreloading...")
|
||||||
}
|
}
|
||||||
|
|
||||||
if err := chromedp.Run(ctx, chromedp.EvaluateAsDevTools("currentState.finished.toString()", &jsOutput)); err != nil {
|
if err := chromedp.Run(ctx, chromedp.EvaluateAsDevTools("currentState.finished.toString()", &jsOutput)); err != nil {
|
||||||
@@ -100,7 +101,7 @@ func runScrapeWithInfo(ctx context.Context, jsAction string, url string) (string
|
|||||||
time.Sleep(50 * time.Millisecond)
|
time.Sleep(50 * time.Millisecond)
|
||||||
}
|
}
|
||||||
|
|
||||||
utils.Log("\nRetrieving items...")
|
log.Log("\nRetrieving items...")
|
||||||
if err := chromedp.Run(ctx,
|
if err := chromedp.Run(ctx,
|
||||||
// Wait until custom js finishes
|
// Wait until custom js finishes
|
||||||
chromedp.WaitVisible(`video_urls`),
|
chromedp.WaitVisible(`video_urls`),
|
||||||
|
|||||||
28
client/getHashtagUploads.go
Normal file
28
client/getHashtagUploads.go
Normal file
@@ -0,0 +1,28 @@
|
|||||||
|
package client
|
||||||
|
|
||||||
|
import (
|
||||||
|
"fmt"
|
||||||
|
|
||||||
|
models "../models"
|
||||||
|
config "../models/config"
|
||||||
|
)
|
||||||
|
|
||||||
|
// GetHashtagUploads - Get all uploads marked with given hashtag
|
||||||
|
func GetHashtagUploads(hashtagURL string) ([]models.Upload, error) {
|
||||||
|
actionOutput, err := GetHashtagUploadsJSON(hashtagURL)
|
||||||
|
if err != nil {
|
||||||
|
return nil, err
|
||||||
|
}
|
||||||
|
|
||||||
|
return models.ParseUploads(actionOutput), nil
|
||||||
|
}
|
||||||
|
|
||||||
|
// GetHashtagUploadsJSON - Get hashtag uploads scrape
|
||||||
|
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
|
||||||
|
}
|
||||||
@@ -1,14 +1,27 @@
|
|||||||
package client
|
package client
|
||||||
|
|
||||||
import (
|
import (
|
||||||
|
"fmt"
|
||||||
|
|
||||||
models "../models"
|
models "../models"
|
||||||
|
config "../models/config"
|
||||||
)
|
)
|
||||||
|
|
||||||
// GetMusicUploads - Get all uploads by given music
|
// GetMusicUploads - Get all uploads by given music
|
||||||
func GetMusicUploads(url string) ([]models.Upload, error) {
|
func GetMusicUploads(url string) ([]models.Upload, error) {
|
||||||
actionOutput, err := executeClientAction(url, "bootstrapIteratingVideos()")
|
actionOutput, err := GetMusicUploadsJSON(url)
|
||||||
if err != nil {
|
if err != nil {
|
||||||
return nil, err
|
return nil, err
|
||||||
}
|
}
|
||||||
return models.ParseUploads(actionOutput), nil
|
return models.ParseUploads(actionOutput), nil
|
||||||
}
|
}
|
||||||
|
|
||||||
|
// GetMusicUploadsJSON - Get music uploads scrape
|
||||||
|
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
|
||||||
|
}
|
||||||
|
|||||||
52
client/getRedirectUrl.go
Normal file
52
client/getRedirectUrl.go
Normal file
@@ -0,0 +1,52 @@
|
|||||||
|
package client
|
||||||
|
|
||||||
|
import (
|
||||||
|
"context"
|
||||||
|
"github.com/chromedp/chromedp"
|
||||||
|
"io/ioutil"
|
||||||
|
"os"
|
||||||
|
"time"
|
||||||
|
|
||||||
|
config "../models/config"
|
||||||
|
log "../utils/log"
|
||||||
|
)
|
||||||
|
|
||||||
|
func GetRedirectUrl(url string) (string, error) {
|
||||||
|
dir, err := ioutil.TempDir("", "chromedp-example")
|
||||||
|
if err != nil {
|
||||||
|
return "", err
|
||||||
|
}
|
||||||
|
defer os.RemoveAll(dir)
|
||||||
|
|
||||||
|
opts := append(chromedp.DefaultExecAllocatorOptions[:],
|
||||||
|
chromedp.DisableGPU,
|
||||||
|
chromedp.UserDataDir(dir),
|
||||||
|
chromedp.Flag("headless", !config.Config.Debug),
|
||||||
|
)
|
||||||
|
|
||||||
|
allocCtx, cancel := chromedp.NewExecAllocator(context.Background(), opts...)
|
||||||
|
defer cancel()
|
||||||
|
|
||||||
|
ctx, cancel := chromedp.NewContext(
|
||||||
|
allocCtx,
|
||||||
|
chromedp.WithLogf(log.Logf),
|
||||||
|
)
|
||||||
|
defer cancel()
|
||||||
|
|
||||||
|
ctx, cancel = context.WithTimeout(ctx, time.Duration(config.Config.Deadline)*time.Second)
|
||||||
|
defer cancel()
|
||||||
|
|
||||||
|
var jsOutput string
|
||||||
|
if err := chromedp.Run(ctx,
|
||||||
|
// Navigate to user's page
|
||||||
|
chromedp.Navigate(url),
|
||||||
|
// Wait until page loads
|
||||||
|
chromedp.WaitReady(`div`),
|
||||||
|
// Grab url links from our element
|
||||||
|
chromedp.EvaluateAsDevTools(`window.location.href`, &jsOutput),
|
||||||
|
); err != nil {
|
||||||
|
return "", err
|
||||||
|
}
|
||||||
|
|
||||||
|
return jsOutput, err
|
||||||
|
}
|
||||||
@@ -1,14 +1,27 @@
|
|||||||
package client
|
package client
|
||||||
|
|
||||||
import (
|
import (
|
||||||
|
"fmt"
|
||||||
|
|
||||||
models "../models"
|
models "../models"
|
||||||
|
config "../models/config"
|
||||||
)
|
)
|
||||||
|
|
||||||
// GetUserUploads - Get all uploads by user
|
// GetUserUploads - Get all uploads by user
|
||||||
func GetUserUploads(username string) ([]models.Upload, error) {
|
func GetUserUploads(username string) ([]models.Upload, error) {
|
||||||
actionOutput, err := executeClientAction(`https://www.tiktok.com/@`+username, "bootstrapIteratingVideos()")
|
actionOutput, err := GetUserUploadsJSON(username)
|
||||||
if err != nil {
|
if err != nil {
|
||||||
return nil, err
|
return nil, err
|
||||||
}
|
}
|
||||||
return models.ParseUploads(actionOutput), nil
|
return models.ParseUploads(actionOutput), nil
|
||||||
}
|
}
|
||||||
|
|
||||||
|
// GetUserUploadsJSON - Get user uploads scrape
|
||||||
|
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
|
||||||
|
}
|
||||||
|
|||||||
@@ -8,23 +8,29 @@ import (
|
|||||||
|
|
||||||
// Config - Runtime configuration
|
// Config - Runtime configuration
|
||||||
var Config struct {
|
var Config struct {
|
||||||
URL string
|
URL string
|
||||||
OutputPath string
|
OutputPath string
|
||||||
BatchFilePath string
|
BatchFilePath string
|
||||||
Debug bool
|
ArchiveFilePath string
|
||||||
MetaData bool
|
Debug bool
|
||||||
Quiet bool
|
MetaData bool
|
||||||
Deadline int
|
Quiet bool
|
||||||
|
JSONOnly bool
|
||||||
|
Deadline int
|
||||||
|
Limit int
|
||||||
}
|
}
|
||||||
|
|
||||||
// GetConfig - Returns Config object
|
// GetConfig - Returns Config object
|
||||||
func GetConfig() {
|
func GetConfig() {
|
||||||
outputPath := flag.String("output", "./downloads", "Output path")
|
outputPath := flag.String("output", "./downloads", "Output path")
|
||||||
batchFilePath := flag.String("batch-file", "", "File containing URLs/Usernames to download, one value per line. Lines starting with '#', are considered as comments and ignored.")
|
batchFilePath := flag.String("batch-file", "", "File containing URLs/Usernames to download, one value per line. Lines starting with '#', are considered as comments and ignored.")
|
||||||
|
archive := flag.String("archive", "", "Download only videos not listed in the archive file. Record the IDs of all downloaded videos in it.")
|
||||||
debug := flag.Bool("debug", false, "Enables debug mode")
|
debug := flag.Bool("debug", false, "Enables debug mode")
|
||||||
metadata := flag.Bool("metadata", false, "Write video metadata to a .json file")
|
metadata := flag.Bool("metadata", false, "Write video metadata to a .json file")
|
||||||
quiet := flag.Bool("quiet", false, "Supress output")
|
quiet := flag.Bool("quiet", false, "Supress output")
|
||||||
|
jsonOnly := flag.Bool("json", false, "Just get JSON data from scraper (without video downloading)")
|
||||||
deadline := flag.Int("deadline", 1500, "Sets the timout for scraper logic in seconds (used as a workaround for 'context deadline exceeded' error)")
|
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)")
|
||||||
flag.Parse()
|
flag.Parse()
|
||||||
|
|
||||||
args := flag.Args()
|
args := flag.Args()
|
||||||
@@ -41,8 +47,14 @@ func GetConfig() {
|
|||||||
}
|
}
|
||||||
Config.OutputPath = *outputPath
|
Config.OutputPath = *outputPath
|
||||||
Config.BatchFilePath = *batchFilePath
|
Config.BatchFilePath = *batchFilePath
|
||||||
|
Config.ArchiveFilePath = *archive
|
||||||
Config.Debug = *debug
|
Config.Debug = *debug
|
||||||
Config.MetaData = *metadata
|
Config.MetaData = *metadata
|
||||||
Config.Quiet = *quiet
|
Config.Quiet = *quiet
|
||||||
|
if *jsonOnly {
|
||||||
|
Config.Quiet = true
|
||||||
|
}
|
||||||
|
Config.JSONOnly = *jsonOnly
|
||||||
Config.Deadline = *deadline
|
Config.Deadline = *deadline
|
||||||
|
Config.Limit = *limit
|
||||||
}
|
}
|
||||||
|
|||||||
@@ -1,11 +1,13 @@
|
|||||||
package models
|
package models
|
||||||
|
|
||||||
import (
|
import (
|
||||||
res "../resources"
|
|
||||||
utils "../utils"
|
|
||||||
"encoding/json"
|
"encoding/json"
|
||||||
"os"
|
"os"
|
||||||
"strings"
|
"strings"
|
||||||
|
|
||||||
|
res "../resources"
|
||||||
|
checkErr "../utils/checkErr"
|
||||||
|
log "../utils/log"
|
||||||
)
|
)
|
||||||
|
|
||||||
// Upload - Upload object
|
// Upload - Upload object
|
||||||
@@ -47,16 +49,16 @@ func (u Upload) GetUploadID() string {
|
|||||||
func (u Upload) WriteToFile(outputPath string) {
|
func (u Upload) WriteToFile(outputPath string) {
|
||||||
bytes, err := json.Marshal(u)
|
bytes, err := json.Marshal(u)
|
||||||
if err != nil {
|
if err != nil {
|
||||||
utils.Logf(res.ErrorCouldNotSerializeJSON, u.GetUploadID())
|
log.Logf(res.ErrorCouldNotSerializeJSON, u.GetUploadID())
|
||||||
panic(err)
|
panic(err)
|
||||||
}
|
}
|
||||||
|
|
||||||
// Create the file
|
// Create the file
|
||||||
out, err := os.Create(outputPath)
|
out, err := os.Create(outputPath)
|
||||||
utils.CheckErr(err)
|
checkErr.CheckErr(err)
|
||||||
defer out.Close()
|
defer out.Close()
|
||||||
|
|
||||||
// Write to file
|
// Write to file
|
||||||
_, err = out.Write(bytes)
|
_, err = out.Write(bytes)
|
||||||
utils.CheckErr(err)
|
checkErr.CheckErr(err)
|
||||||
}
|
}
|
||||||
|
|||||||
@@ -1,10 +1,11 @@
|
|||||||
package models
|
package models
|
||||||
|
|
||||||
import (
|
import (
|
||||||
testUtil "../unitTestUtil"
|
|
||||||
utils "../utils"
|
|
||||||
"os"
|
"os"
|
||||||
"testing"
|
"testing"
|
||||||
|
|
||||||
|
testUtil "../unitTestUtil"
|
||||||
|
fileio "../utils/fileio"
|
||||||
)
|
)
|
||||||
|
|
||||||
func TestParseUploads(t *testing.T) {
|
func TestParseUploads(t *testing.T) {
|
||||||
@@ -62,7 +63,7 @@ func TestWriteToFile(t *testing.T) {
|
|||||||
|
|
||||||
upload.WriteToFile(filePath)
|
upload.WriteToFile(filePath)
|
||||||
|
|
||||||
actual := utils.ReadFileToString(filePath)
|
actual := fileio.ReadFileToString(filePath)
|
||||||
tu.AssertString(actual, expected, "File content")
|
tu.AssertString(actual, expected, "File content")
|
||||||
|
|
||||||
os.Remove(filePath)
|
os.Remove(filePath)
|
||||||
|
|||||||
22
scraper.js
22
scraper.js
@@ -1,7 +1,7 @@
|
|||||||
optStrings = {
|
optStrings = {
|
||||||
selectors: {
|
selectors: {
|
||||||
feedLoading: 'div.tiktok-loading.feed-loading',
|
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',
|
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',
|
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',
|
modalShareInput: '.copy-link-container > input',
|
||||||
@@ -36,6 +36,7 @@ optStrings = {
|
|||||||
currentState = {
|
currentState = {
|
||||||
preloadCount: 0,
|
preloadCount: 0,
|
||||||
finished: false,
|
finished: false,
|
||||||
|
limit: 0
|
||||||
};
|
};
|
||||||
|
|
||||||
checkForErrors = function() {
|
checkForErrors = function() {
|
||||||
@@ -65,8 +66,14 @@ buldVidUrlArray = function(finishCallback) {
|
|||||||
var videoArray = [];
|
var videoArray = [];
|
||||||
var intervalID = window.setInterval(x => {
|
var intervalID = window.setInterval(x => {
|
||||||
videoArray.push(getCurrentModalVideo());
|
videoArray.push(getCurrentModalVideo());
|
||||||
|
if(currentState.limit > 0) {
|
||||||
var arrowRight = document.querySelectorAll(optStrings.selectors.modalArrowLeft)[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)) {
|
if (arrowRight.classList.contains(optStrings.classes.modalCloseDisabled)) {
|
||||||
window.clearInterval(intervalID);
|
window.clearInterval(intervalID);
|
||||||
document.querySelector(optStrings.selectors.modalClose).click();
|
document.querySelector(optStrings.selectors.modalClose).click();
|
||||||
@@ -127,6 +134,12 @@ scrollWhileNew = function(finishCallback) {
|
|||||||
var intervalID = window.setInterval(x => {
|
var intervalID = window.setInterval(x => {
|
||||||
var oldCount = state.count;
|
var oldCount = state.count;
|
||||||
state.count = document.getElementsByClassName(optStrings.classes.feedVideoItem).length;
|
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()) {
|
if(checkForErrors()) {
|
||||||
window.clearInterval(intervalID);
|
window.clearInterval(intervalID);
|
||||||
return;
|
return;
|
||||||
@@ -145,7 +158,8 @@ scrollWhileNew = function(finishCallback) {
|
|||||||
}, 1000);
|
}, 1000);
|
||||||
};
|
};
|
||||||
|
|
||||||
bootstrapIteratingVideos = function() {
|
bootstrapIteratingVideos = function(limit) {
|
||||||
|
currentState.limit = limit;
|
||||||
scrollWhileNew(buldVidUrlArray);
|
scrollWhileNew(buldVidUrlArray);
|
||||||
return 'bootstrapIteratingVideos';
|
return 'bootstrapIteratingVideos';
|
||||||
};
|
};
|
||||||
|
|||||||
54
utils/archive.go
Normal file
54
utils/archive.go
Normal file
@@ -0,0 +1,54 @@
|
|||||||
|
package utils
|
||||||
|
|
||||||
|
import (
|
||||||
|
models "../models"
|
||||||
|
config "../models/config"
|
||||||
|
fileio "./fileio"
|
||||||
|
log "./log"
|
||||||
|
)
|
||||||
|
|
||||||
|
// IsItemInArchive - Checks if the item is already archived
|
||||||
|
func IsItemInArchive(upload models.Upload) bool {
|
||||||
|
if len(RemoveArchivedItems([]models.Upload{upload})) == 0 {
|
||||||
|
return true
|
||||||
|
}
|
||||||
|
return false
|
||||||
|
}
|
||||||
|
|
||||||
|
// RemoveArchivedItems - Returns items slice without archived items
|
||||||
|
func RemoveArchivedItems(uploads []models.Upload) []models.Upload {
|
||||||
|
archiveFilePath := config.Config.ArchiveFilePath
|
||||||
|
|
||||||
|
if archiveFilePath == "" || !fileio.CheckIfExists(archiveFilePath) {
|
||||||
|
return uploads
|
||||||
|
}
|
||||||
|
|
||||||
|
removeArchivedItemsDelegate := func(archivedItem string) {
|
||||||
|
for i, upload := range uploads {
|
||||||
|
if upload.GetUploadID() == archivedItem {
|
||||||
|
uploads = append(uploads[:i], uploads[i+1:]...)
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
lenBeforeRemoval := len(uploads)
|
||||||
|
fileio.ReadFileLineByLine(archiveFilePath, removeArchivedItemsDelegate)
|
||||||
|
|
||||||
|
removedCount := lenBeforeRemoval - len(uploads)
|
||||||
|
if removedCount > 0 {
|
||||||
|
log.Logf("%d items, found in archive. Skipping...\n", removedCount)
|
||||||
|
}
|
||||||
|
|
||||||
|
return uploads
|
||||||
|
}
|
||||||
|
|
||||||
|
// AddItemToArchive - Adds item to archived list
|
||||||
|
func AddItemToArchive(uploadID string) {
|
||||||
|
archiveFilePath := config.Config.ArchiveFilePath
|
||||||
|
|
||||||
|
if archiveFilePath == "" {
|
||||||
|
return
|
||||||
|
}
|
||||||
|
|
||||||
|
fileio.AppendToFile(uploadID, archiveFilePath)
|
||||||
|
}
|
||||||
@@ -4,21 +4,23 @@ import (
|
|||||||
"io"
|
"io"
|
||||||
"net/http"
|
"net/http"
|
||||||
"os"
|
"os"
|
||||||
|
|
||||||
|
checkErr "./checkErr"
|
||||||
)
|
)
|
||||||
|
|
||||||
// DownloadFile - Downloads content from `url` and stores it in `outputPath`
|
// DownloadFile - Downloads content from `url` and stores it in `outputPath`
|
||||||
func DownloadFile(outputPath string, url string) {
|
func DownloadFile(outputPath string, url string) {
|
||||||
// Get the data
|
// Get the data
|
||||||
resp, err := http.Get(url)
|
resp, err := http.Get(url)
|
||||||
CheckErr(err)
|
checkErr.CheckErr(err)
|
||||||
defer resp.Body.Close()
|
defer resp.Body.Close()
|
||||||
|
|
||||||
// Create the file
|
// Create the file
|
||||||
out, err := os.Create(outputPath)
|
out, err := os.Create(outputPath)
|
||||||
CheckErr(err)
|
checkErr.CheckErr(err)
|
||||||
defer out.Close()
|
defer out.Close()
|
||||||
|
|
||||||
// Write the body to file
|
// Write the body to file
|
||||||
_, err = io.Copy(out, resp.Body)
|
_, err = io.Copy(out, resp.Body)
|
||||||
CheckErr(err)
|
checkErr.CheckErr(err)
|
||||||
}
|
}
|
||||||
|
|||||||
@@ -4,6 +4,8 @@ import (
|
|||||||
"bufio"
|
"bufio"
|
||||||
"io/ioutil"
|
"io/ioutil"
|
||||||
"os"
|
"os"
|
||||||
|
|
||||||
|
checkErr "../checkErr"
|
||||||
)
|
)
|
||||||
|
|
||||||
type delegateString func(string)
|
type delegateString func(string)
|
||||||
@@ -37,7 +39,7 @@ func ReadFileToString(path string) string {
|
|||||||
// ReadFileLineByLine - Reads file line by line and calls delegate
|
// ReadFileLineByLine - Reads file line by line and calls delegate
|
||||||
func ReadFileLineByLine(path string, delegate delegateString) {
|
func ReadFileLineByLine(path string, delegate delegateString) {
|
||||||
file, err := os.Open(path)
|
file, err := os.Open(path)
|
||||||
CheckErr(err)
|
checkErr.CheckErr(err)
|
||||||
defer file.Close()
|
defer file.Close()
|
||||||
|
|
||||||
scanner := bufio.NewScanner(file)
|
scanner := bufio.NewScanner(file)
|
||||||
@@ -49,3 +51,14 @@ func ReadFileLineByLine(path string, delegate delegateString) {
|
|||||||
panic(err)
|
panic(err)
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
// AppendToFile - Appends line to file
|
||||||
|
func AppendToFile(str string, filePath string) {
|
||||||
|
f, err := os.OpenFile(filePath, os.O_APPEND|os.O_CREATE|os.O_WRONLY, 0644)
|
||||||
|
checkErr.CheckErr(err)
|
||||||
|
|
||||||
|
defer f.Close()
|
||||||
|
if _, err := f.WriteString(str + "\n"); err != nil {
|
||||||
|
checkErr.CheckErr(err)
|
||||||
|
}
|
||||||
|
}
|
||||||
16
utils/getHashtag.go
Normal file
16
utils/getHashtag.go
Normal file
@@ -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))
|
||||||
|
}
|
||||||
@@ -1,9 +1,10 @@
|
|||||||
package utils
|
package utils
|
||||||
|
|
||||||
import (
|
import (
|
||||||
config "../models/config"
|
|
||||||
"fmt"
|
"fmt"
|
||||||
"os"
|
"os"
|
||||||
|
|
||||||
|
config "../../models/config"
|
||||||
)
|
)
|
||||||
|
|
||||||
// Log - Write to std out
|
// Log - Write to std out
|
||||||
@@ -2,11 +2,13 @@ package utils
|
|||||||
|
|
||||||
import (
|
import (
|
||||||
"io/ioutil"
|
"io/ioutil"
|
||||||
|
|
||||||
|
checkErr "./checkErr"
|
||||||
)
|
)
|
||||||
|
|
||||||
// ReadFileAsString - Returns contents of given file
|
// ReadFileAsString - Returns contents of given file
|
||||||
func ReadFileAsString(fileName string) string {
|
func ReadFileAsString(fileName string) string {
|
||||||
content, err := ioutil.ReadFile(fileName)
|
content, err := ioutil.ReadFile(fileName)
|
||||||
CheckErr(err)
|
checkErr.CheckErr(err)
|
||||||
return string(content)
|
return string(content)
|
||||||
}
|
}
|
||||||
|
|||||||
@@ -2,7 +2,8 @@ package workflows
|
|||||||
|
|
||||||
import (
|
import (
|
||||||
res "../resources"
|
res "../resources"
|
||||||
utils "../utils"
|
fileio "../utils/fileio"
|
||||||
|
log "../utils/log"
|
||||||
)
|
)
|
||||||
|
|
||||||
// CanUseDownloadBatchFile - Check's if DownloadBatchFile can be used
|
// CanUseDownloadBatchFile - Check's if DownloadBatchFile can be used
|
||||||
@@ -12,11 +13,11 @@ func CanUseDownloadBatchFile(batchFilePath string) bool {
|
|||||||
|
|
||||||
// DownloadBatchFile - Download items from batch file
|
// DownloadBatchFile - Download items from batch file
|
||||||
func DownloadBatchFile(batchFilePath string) {
|
func DownloadBatchFile(batchFilePath string) {
|
||||||
if !utils.CheckIfExists(batchFilePath) {
|
if !fileio.CheckIfExists(batchFilePath) {
|
||||||
utils.LogFatal(res.ErrorPathNotFound, batchFilePath)
|
log.LogFatal(res.ErrorPathNotFound, batchFilePath)
|
||||||
}
|
}
|
||||||
|
|
||||||
utils.ReadFileLineByLine(batchFilePath, downloadItem)
|
fileio.ReadFileLineByLine(batchFilePath, downloadItem)
|
||||||
}
|
}
|
||||||
|
|
||||||
func downloadItem(batchItem string) {
|
func downloadItem(batchItem string) {
|
||||||
|
|||||||
52
workflows/downloadHashtag.go
Normal file
52
workflows/downloadHashtag.go
Normal file
@@ -0,0 +1,52 @@
|
|||||||
|
package workflows
|
||||||
|
|
||||||
|
import (
|
||||||
|
"fmt"
|
||||||
|
"strings"
|
||||||
|
|
||||||
|
client "../client"
|
||||||
|
config "../models/config"
|
||||||
|
res "../resources"
|
||||||
|
utils "../utils"
|
||||||
|
fileio "../utils/fileio"
|
||||||
|
log "../utils/log"
|
||||||
|
)
|
||||||
|
|
||||||
|
// 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 {
|
||||||
|
log.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
|
||||||
|
return
|
||||||
|
}
|
||||||
|
|
||||||
|
uploads = utils.RemoveArchivedItems(uploads)
|
||||||
|
uploadCount := len(uploads)
|
||||||
|
|
||||||
|
hashtag := utils.GetHashtagFromURL(url)
|
||||||
|
downloadDir := fmt.Sprintf("%s/%s", config.Config.OutputPath, hashtag)
|
||||||
|
|
||||||
|
fileio.InitOutputDirectory(downloadDir)
|
||||||
|
|
||||||
|
for index, upload := range uploads {
|
||||||
|
downloadVideo(upload, downloadDir)
|
||||||
|
log.Logf("\r[%d/%d] Downloaded", index+1, uploadCount)
|
||||||
|
}
|
||||||
|
log.Log()
|
||||||
|
}
|
||||||
|
|
||||||
|
// GetHashtagJSON - Prints scraped info from hashtag
|
||||||
|
func GetHashtagJSON(url string) {
|
||||||
|
uploads, err := client.GetHashtagUploadsJSON(url)
|
||||||
|
if err != nil {
|
||||||
|
log.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
|
||||||
|
return
|
||||||
|
}
|
||||||
|
fmt.Printf("%s", uploads)
|
||||||
|
}
|
||||||
@@ -1,12 +1,15 @@
|
|||||||
package workflows
|
package workflows
|
||||||
|
|
||||||
import (
|
import (
|
||||||
|
"fmt"
|
||||||
|
"regexp"
|
||||||
|
|
||||||
client "../client"
|
client "../client"
|
||||||
config "../models/config"
|
config "../models/config"
|
||||||
res "../resources"
|
res "../resources"
|
||||||
utils "../utils"
|
utils "../utils"
|
||||||
"fmt"
|
fileio "../utils/fileio"
|
||||||
"regexp"
|
log "../utils/log"
|
||||||
)
|
)
|
||||||
|
|
||||||
// CanUseDownloadMusic - Check's if DownloadMusic can be used for parameter
|
// CanUseDownloadMusic - Check's if DownloadMusic can be used for parameter
|
||||||
@@ -19,18 +22,30 @@ func CanUseDownloadMusic(url string) bool {
|
|||||||
func DownloadMusic(url string) {
|
func DownloadMusic(url string) {
|
||||||
uploads, err := client.GetMusicUploads(url)
|
uploads, err := client.GetMusicUploads(url)
|
||||||
if err != nil {
|
if err != nil {
|
||||||
utils.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
|
log.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
|
||||||
return
|
return
|
||||||
}
|
}
|
||||||
|
|
||||||
|
uploads = utils.RemoveArchivedItems(uploads)
|
||||||
uploadCount := len(uploads)
|
uploadCount := len(uploads)
|
||||||
|
|
||||||
for index, upload := range uploads {
|
for index, upload := range uploads {
|
||||||
username := utils.GetUsernameFromString(upload.Uploader)
|
username := utils.GetUsernameFromString(upload.Uploader)
|
||||||
downloadDir := fmt.Sprintf("%s/%s", config.Config.OutputPath, username)
|
downloadDir := fmt.Sprintf("%s/%s", config.Config.OutputPath, username)
|
||||||
|
|
||||||
utils.InitOutputDirectory(downloadDir)
|
fileio.InitOutputDirectory(downloadDir)
|
||||||
downloadVideo(upload, downloadDir)
|
downloadVideo(upload, downloadDir)
|
||||||
utils.Logf("\r[%d/%d] Downloaded", index+1, uploadCount)
|
log.Logf("\r[%d/%d] Downloaded", index+1, uploadCount)
|
||||||
}
|
}
|
||||||
utils.Log()
|
log.Log()
|
||||||
|
}
|
||||||
|
|
||||||
|
// GetMusicJSON - Prints scraped info from music
|
||||||
|
func GetMusicJSON(url string) {
|
||||||
|
uploads, err := client.GetMusicUploadsJSON(url)
|
||||||
|
if err != nil {
|
||||||
|
log.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
|
||||||
|
return
|
||||||
|
}
|
||||||
|
fmt.Printf("%s", uploads)
|
||||||
}
|
}
|
||||||
|
|||||||
27
workflows/downloadShareLink.go
Normal file
27
workflows/downloadShareLink.go
Normal file
@@ -0,0 +1,27 @@
|
|||||||
|
package workflows
|
||||||
|
|
||||||
|
import (
|
||||||
|
client "../client"
|
||||||
|
res "../resources"
|
||||||
|
log "../utils/log"
|
||||||
|
"regexp"
|
||||||
|
)
|
||||||
|
|
||||||
|
// CanUseDownloadShareLink - Check's if DownloadShareLink can be used
|
||||||
|
func CanUseDownloadShareLink(url string) bool {
|
||||||
|
match, _ := regexp.MatchString("vm.tiktok.com\\/.+", url)
|
||||||
|
return match
|
||||||
|
}
|
||||||
|
|
||||||
|
// DownloadShareLink - Download item by share link
|
||||||
|
func DownloadShareLink(url string) {
|
||||||
|
log.Logf("Resolving share link: %s\n", url)
|
||||||
|
|
||||||
|
finalURL, err := client.GetRedirectUrl(url)
|
||||||
|
if err != nil {
|
||||||
|
log.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
|
||||||
|
return
|
||||||
|
}
|
||||||
|
|
||||||
|
StartWorkflowByParameter(finalURL)
|
||||||
|
}
|
||||||
@@ -1,13 +1,16 @@
|
|||||||
package workflows
|
package workflows
|
||||||
|
|
||||||
import (
|
import (
|
||||||
|
"fmt"
|
||||||
|
"regexp"
|
||||||
|
"strings"
|
||||||
|
|
||||||
client "../client"
|
client "../client"
|
||||||
config "../models/config"
|
config "../models/config"
|
||||||
res "../resources"
|
res "../resources"
|
||||||
utils "../utils"
|
utils "../utils"
|
||||||
"fmt"
|
fileio "../utils/fileio"
|
||||||
"regexp"
|
log "../utils/log"
|
||||||
"strings"
|
|
||||||
)
|
)
|
||||||
|
|
||||||
// CanUseDownloadUser - Test's if this workflow can be used for parameter
|
// CanUseDownloadUser - Test's if this workflow can be used for parameter
|
||||||
@@ -21,17 +24,30 @@ func CanUseDownloadUser(url string) bool {
|
|||||||
func DownloadUser(username string) {
|
func DownloadUser(username string) {
|
||||||
uploads, err := client.GetUserUploads(username)
|
uploads, err := client.GetUserUploads(username)
|
||||||
if err != nil {
|
if err != nil {
|
||||||
utils.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
|
log.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
|
||||||
return
|
return
|
||||||
}
|
}
|
||||||
|
|
||||||
|
uploads = utils.RemoveArchivedItems(uploads)
|
||||||
uploadCount := len(uploads)
|
uploadCount := len(uploads)
|
||||||
|
|
||||||
downloadDir := fmt.Sprintf("%s/%s", config.Config.OutputPath, username)
|
downloadDir := fmt.Sprintf("%s/%s", config.Config.OutputPath, username)
|
||||||
|
|
||||||
utils.InitOutputDirectory(downloadDir)
|
fileio.InitOutputDirectory(downloadDir)
|
||||||
|
|
||||||
for index, upload := range uploads {
|
for index, upload := range uploads {
|
||||||
downloadVideo(upload, downloadDir)
|
downloadVideo(upload, downloadDir)
|
||||||
utils.Logf("\r[%d/%d] Downloaded", index+1, uploadCount)
|
log.Logf("\r[%d/%d] Downloaded", index+1, uploadCount)
|
||||||
}
|
}
|
||||||
utils.Log()
|
log.Log()
|
||||||
|
}
|
||||||
|
|
||||||
|
// GetUserVideosJSON - Prints scraped info from user
|
||||||
|
func GetUserVideosJSON(username string) {
|
||||||
|
uploads, err := client.GetUserUploadsJSON(username)
|
||||||
|
if err != nil {
|
||||||
|
log.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
|
||||||
|
return
|
||||||
|
}
|
||||||
|
fmt.Printf("%s", uploads)
|
||||||
}
|
}
|
||||||
|
|||||||
@@ -1,13 +1,16 @@
|
|||||||
package workflows
|
package workflows
|
||||||
|
|
||||||
import (
|
import (
|
||||||
|
"fmt"
|
||||||
|
"regexp"
|
||||||
|
|
||||||
client "../client"
|
client "../client"
|
||||||
models "../models"
|
models "../models"
|
||||||
config "../models/config"
|
config "../models/config"
|
||||||
res "../resources"
|
res "../resources"
|
||||||
utils "../utils"
|
utils "../utils"
|
||||||
"fmt"
|
fileio "../utils/fileio"
|
||||||
"regexp"
|
log "../utils/log"
|
||||||
)
|
)
|
||||||
|
|
||||||
// CanUseDownloadSingleVideo - Check's if DownloadSingleVideo can be used for parameter
|
// CanUseDownloadSingleVideo - Check's if DownloadSingleVideo can be used for parameter
|
||||||
@@ -21,14 +24,18 @@ func DownloadSingleVideo(url string) {
|
|||||||
username := utils.GetUsernameFromString(url)
|
username := utils.GetUsernameFromString(url)
|
||||||
upload, err := client.GetVideoDetails(url)
|
upload, err := client.GetVideoDetails(url)
|
||||||
if err != nil {
|
if err != nil {
|
||||||
utils.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
|
log.LogErr(res.ErrorCouldNotGetUserUploads, err.Error())
|
||||||
|
return
|
||||||
|
}
|
||||||
|
|
||||||
|
if utils.IsItemInArchive(upload) {
|
||||||
return
|
return
|
||||||
}
|
}
|
||||||
downloadDir := fmt.Sprintf("%s/%s", config.Config.OutputPath, username)
|
downloadDir := fmt.Sprintf("%s/%s", config.Config.OutputPath, username)
|
||||||
|
|
||||||
utils.InitOutputDirectory(downloadDir)
|
fileio.InitOutputDirectory(downloadDir)
|
||||||
downloadVideo(upload, downloadDir)
|
downloadVideo(upload, downloadDir)
|
||||||
utils.Log("[1/1] Downloaded\n")
|
log.Log("[1/1] Downloaded\n")
|
||||||
}
|
}
|
||||||
|
|
||||||
// DownloadVideo - Downloads one video
|
// DownloadVideo - Downloads one video
|
||||||
@@ -36,7 +43,7 @@ func downloadVideo(upload models.Upload, downloadDir string) {
|
|||||||
uploadID := upload.GetUploadID()
|
uploadID := upload.GetUploadID()
|
||||||
downloadPath := fmt.Sprintf("%s/%s.mp4", downloadDir, uploadID)
|
downloadPath := fmt.Sprintf("%s/%s.mp4", downloadDir, uploadID)
|
||||||
|
|
||||||
if utils.CheckIfExists(downloadPath) {
|
if fileio.CheckIfExists(downloadPath) {
|
||||||
return
|
return
|
||||||
}
|
}
|
||||||
|
|
||||||
@@ -46,4 +53,6 @@ func downloadVideo(upload models.Upload, downloadDir string) {
|
|||||||
metadataPath := fmt.Sprintf("%s/%s.json", downloadDir, uploadID)
|
metadataPath := fmt.Sprintf("%s/%s.json", downloadDir, uploadID)
|
||||||
upload.WriteToFile(metadataPath)
|
upload.WriteToFile(metadataPath)
|
||||||
}
|
}
|
||||||
|
|
||||||
|
utils.AddItemToArchive(upload.GetUploadID())
|
||||||
}
|
}
|
||||||
|
|||||||
@@ -1,8 +1,10 @@
|
|||||||
package workflows
|
package workflows
|
||||||
|
|
||||||
import (
|
import (
|
||||||
|
config "../models/config"
|
||||||
res "../resources"
|
res "../resources"
|
||||||
utils "../utils"
|
utils "../utils"
|
||||||
|
log "../utils/log"
|
||||||
)
|
)
|
||||||
|
|
||||||
// StartWorkflowByParameter - Start needed workflow by given parameter
|
// StartWorkflowByParameter - Start needed workflow by given parameter
|
||||||
@@ -10,7 +12,11 @@ func StartWorkflowByParameter(url string) {
|
|||||||
|
|
||||||
// Music
|
// Music
|
||||||
if CanUseDownloadMusic(url) {
|
if CanUseDownloadMusic(url) {
|
||||||
DownloadMusic(url)
|
if config.Config.JSONOnly {
|
||||||
|
GetMusicJSON(url)
|
||||||
|
} else {
|
||||||
|
DownloadMusic(url)
|
||||||
|
}
|
||||||
return
|
return
|
||||||
}
|
}
|
||||||
|
|
||||||
@@ -22,9 +28,30 @@ func StartWorkflowByParameter(url string) {
|
|||||||
|
|
||||||
// Tiktok user
|
// Tiktok user
|
||||||
if CanUseDownloadUser(url) {
|
if CanUseDownloadUser(url) {
|
||||||
DownloadUser(utils.GetUsernameFromString(url))
|
if config.Config.JSONOnly {
|
||||||
|
GetUserVideosJSON(utils.GetUsernameFromString(url))
|
||||||
|
} else {
|
||||||
|
DownloadUser(utils.GetUsernameFromString(url))
|
||||||
|
}
|
||||||
|
|
||||||
return
|
return
|
||||||
}
|
}
|
||||||
|
|
||||||
utils.LogFatal(res.ErrorCouldNotRecogniseURL, url)
|
// Tiktok hashtag
|
||||||
|
if CanUseDownloadHashtag(url) {
|
||||||
|
if config.Config.JSONOnly {
|
||||||
|
GetHashtagJSON(url)
|
||||||
|
} else {
|
||||||
|
DownloadHashtag(url)
|
||||||
|
}
|
||||||
|
return
|
||||||
|
}
|
||||||
|
|
||||||
|
// Share URL
|
||||||
|
if CanUseDownloadShareLink(url) {
|
||||||
|
DownloadShareLink(url)
|
||||||
|
return
|
||||||
|
}
|
||||||
|
|
||||||
|
log.LogFatal(res.ErrorCouldNotRecogniseURL, url)
|
||||||
}
|
}
|
||||||
|
|||||||
Reference in New Issue
Block a user