Command line application and library for effortlessly uploading files to Arweave. Arweave enables you to store documents and applications forever.
Upload gigabytes of files with one command specifying a glob pattern to match files against. Files are read and posted to arweave.net asynchronously with computationally intensive bundle preparation performed in parallel across multiple threads.
Usage with SOL is currently essentially free. The cost per transaction is 10,000 lamports (~$0.002), including the Solana network fee of 5,000 lamports.
You can also install from crates.io once you have rust installed with the nightly toolchain.
rustup default nightly
cargo install arloader
Get an Arweave wallet json file here.
If you're going to use AR to pay for transactions, get AR tokens. I've been using gate.io despite the high withdrawal fees and KYC delays.
If you're going to use SOL, get a Solana wallet json file and transfer some SOL to it.
Put your assets and associated metadata file with .json extension in folder by themselves.
To take advantage of the discounted SOL transactions, run:
arloader upload-nfts <GLOB> --with-sol --ar-default-keypair
This will first upload your assets and log status to a newly created sub directory of the folder where the assets are located of the form arloader_<RANDOM_CHARS>.
A manifest file will be created from these logged statuses and uploaed to Arweave. A manifest is a special file that Arweave will use to access your files by their names relative to the id of the manifest transaction: https://arweave.net/<MANIFEST_ID>/<FILE_PATH>. You'll still be able to access your files at https://arweave.net/<BUNDLE_ITEM_ID>, but creating and uploading a manifest gives you the option of using either link.
Next your metadata files will be updated with the links to the uploaded assets. Arloader adds or replaces the image and files keys with the newly created links, defaulting to using the id link (https://arweave.net/<BUNDLE_ITEM_ID>) for the image key and updates the files key to include both links. If you prefer to use the file path based link for the image key, you can pass the --link-file flag to the upload-nfts command.
Then your metadata files will uploaded and a manifest created for them.
Once everything is done, you can use the manifest_<TXID>.json file in arloader_<RAND_CHAR>/metadata/ to get the links to the uploaded metadata to include in our tokens' on chain metadata.
json
{
"0.json": {
"id": "ScU9mEuKBbPX5o5nv8DZkDnZuJbzf84lyLk-uLVDqNk",
"files": [
{
"uri": "https://arweave.net/ScU9mEuKBbPX5o5nv8DZkDnZuJbzf84lyLk-uLVDqNk",
"type": "application/json"
},
{
"uri": "https://arweave.net/fo9P3OOq78REajk48vFWbKfIhw6mDzgjANQIh3L7Njs/0.json",
"type": "application/json"
}
]
},
"1.json": {
"id": "8APeQ5lW0-csTcBaGdPBDLAL2ci2AT9pTn2tppGPU_8",
"files": [
{
"uri": "https://arweave.net/8APeQ5lW0-csTcBaGdPBDLAL2ci2AT9pTn2tppGPU_8",
"type": "application/json"
},
{
"uri": "https://arweave.net/fo9P3OOq78REajk48vFWbKfIhw6mDzgjANQIh3L7Njs/1.json",
"type": "application/json"
}
]
},
If you happen to be creating your NFTs with the Metaplex Candy Machine, you can create a json file of links you can copy
and paste into your candy machine config by running the command below where <GLOB> is a pattern that will match your metadata files (something *.json).
arloader write-metaplex-items <GLOB> --manifest-path <MANIFEST_PATH> --log-dir <LOG_DIR>
This will write a file named metaplex_items_<MANIFIEST_ID>.json to <LOG_DIR> with the format below that you can copy into the items key in your candy machine config. Arloader defaults to using the id based link (https://arweave.net/<BUNDLE_ITEM_ID>), but
you can use the file based link (https://arweave.net/<MANIFEST_ID>/<FILE_PATH>), by passing the --link-file flag.
json
{
"0": {
"link": "uri link",
"name": "name",
"onChain": false
},
"1": {
"link": "uri link",
"name": "name",
"onChain": false
},
Before you creating your tokens, make sure that all of your transactions have been confirmed at least 25 times. Run the command below where <LOG_DIR> refers to the automatically created directory in your assets folder that begins with arloader_.
arloader update-nft-status <LOG_DIR>
``` Updating asset bundle statuses...
kmgLCgV-dB-DGML8cvFwuP3a-ZKedz7nyDuEsqYPTis 10 980 Confirmed 60 Updated 1 statuses.
Updating metadata bundle statuses...
kmgLCgV-dB-DGML8cvFwuP3a-ZKedz7nyDuEsqYPTis 10 980 Confirmed 60 Updated 1 statuses.
Updating asset manifest status...
URwQtoqrbYlc5183STNy3ZPwSCRY4o8goaF7MJay3xY Confirmed 60
Updating metadata manifest status...
fo9P3OOq78REajk48vFWbKfIhw6mDzgjANQIh3L7Njs Confirmed 57 ```
If you're uploading more than one file, you should pretty much always be using bundles. Bundles take multiple files and packages them together in a single transaction. This is better than uploading multiple individual files because you only have to wait for one transaction to be confirmed. Once the bundle transaction is confirmed, all of your files will be available. Larger transactions with larger rewards are more attractive to miners, which means a larger bundled transaction is more likely to get written quickly than a bunch of smaller individual ones.
Arloader accepts file glob patterns and defaults to creating a bundle for your files.
Arloader will create as many bundles as necessary to upload all of your files. Your files are read asynchronously, bundled in parallel across multiple threads and then posted to arweave.net. Arloader supports bundle sizes up to 200 MB, but the default bundle size is 10 MB, which makes it possible to post full bundle size payloads to the /tx endpoint instead of in 256 KB chunks to the /chunk endpoint. This should work fine for individual files up to 10 MB. If your files sizes are bigger than 10 MB (but smaller than 200 MB), you can specify a larger bundle size with the --bundles-size argument - --bundle-size 100000000 to specify a size of 100 MB, for example.
arloader estimate "<GLOB>"
Make sure to include quotes around your glob patterns, otherwise your shell will expand them into a list of files. Arloader expects a glob pattern, not a list of files.
arloader upload "<GLOB>" --log-dir "<LOG_DIR>"
This kicks off the process of uploading a stream of bundles created from your files. The default bundle size is 10 MB. The example output below had a bundle size of 5000 bytes.
```
QGPFcZq91lQgmmz2l7rQHkSQpgfJi-Vhv47oTqIYLm4 2 3 Submitted 0 -bhdsi4irDEWz8R9wXT-1c06WVQVSMAmQxVF9OkW94 2 3 Submitted 0 -OAWdFiGS4NKOZXVJG3yZ0yN4xydGOhfQGX2FCdlG88 2 3 Submitted 0 UBWGFKyTrUVaCa7wi181FjAd545vdoHmBQEdlaVdA4 2 3 Submitted 0 qzQlASZrQXNF9HYIOTPjEZL9uy1U9Ou086kCkQWqld0 2 3 Submitted 0 ```
A json status object gets written to LOG_DIR for each uploaded bundle with a file name of <TXID>.json. It has the transaction id, reward, creation time and ids and paths of the files included in the bundle.
json
{
"id": "_-bhdsi4irDEWz8R9wXT-1c06WVQVSMAmQxVF9OkW94",
"status": "Submitted",
"file_paths": {
"tests/fixtures/8.png": {
"id": "0jd-NTQUZhmnKRY-kMt2vEcmSqgzKOLX_P3QYw6CaNE"
},
"tests/fixtures/9.png": {
"id": "1XdiLkoZ5POHsNx7eLyRgisjnxTLzW8SxGsRcb22j84"
}
},
"number_of_files": 2,
"data_size": 3546,
"created_at": "2021-11-23T05:47:41.948103600Z",
"last_modified": "2021-11-23T05:47:41.948107100Z",
"reward": 50947968
}
arloader update-status --log-dir "<LOG_DIR>"
This will read the files in <LOG_DIR>, looking for a valid transaction id as a file stem, and then go out to the network to update the status of each. The example below contained two sets of bundles, one still pending and one with 45 confirmations.
```
-OAWdFiGS4NKOZXVJG3yZ0yN4xydGOhfQGX2FCdlG88 2 3 Pending 0 -bhdsi4irDEWz8R9wXT-1c06WVQVSMAmQxVF9OkW94 2 3 Pending 0 qzQlASZrQXNF9HYIOTPjEZL9uy1U9Ou086kCkQWqld0 2 3 Pending 0 QGPFcZq91lQgmmz2l7rQHkSQpgfJi-Vhv47oTqIYLm4 2 3 Pending 0 UBWGFKyTrUVaCa7wi181FjAd545vdoHmBQEdlaVdA4 2 3 Pending 0 KuuEZpbfCbw6izMeN3knWlpzmaFhnrDL9dUKCW2LQHw 2 3 Confirmed 45 IRToYYvsftCiR71sfW5qt8XCzBFotwoDFBoEMEtrMrU 2 3 Confirmed 45 M2QZYxUqw3ZJ2KXzU4pfw9fFIkVOSrJbSpE7NAvHLvo 2 3 Confirmed 45 qvci4i6Mfr-5NHI1bL-Omv16QEUw3iiirzv4fXefnM 2 3 Confirmed 45 NAP2vTKQdMGeKyKBYz3876T4yBFl4oYFYqwwwnHbFA 2 3 Confirmed 45 ```
arloader upload-manifest --log-dir "<LOG_DIR>"
where <LOG_DIR> is the directory containing your bundle status json files. This will go through and consolidate the paths from each of the bundles, create a consolidated manifest, upload it to Arweave and then write a file named manifest_<TXID>.jsonto <LOG_DIR>. Once the transaction uploading the manifest has been confirmed, you will be able to access your files and both https://arweave.net/<BUNDLE_ITEM_ID> and https://arweave.net/<MANIFEST_ID>/<FILE_PATH> where MANIFEST_ID is the id of the manifest transaction and FILE_PATH is the relative path of the file included in the GLOB pattern you specified with the upload command.
json
{
"tests/fixtures/0.png": {
"files": [
{
"type": "image/png",
"uri": "https://arweave.net/BSvIAiwthQu_xwQBHn9FcgACaZ8ko4py5mqMNP4r-jM/tests/fixtures/0.png"
},
{
"type": "image/png",
"uri": "https://arweave.net/JQbz5py065lqaS_8R7NCtLcK2b-pSkkG6Je0OT8379c"
}
],
"id": "JQbz5py065lqaS_8R7NCtLcK2b-pSkkG6Je0OT8379c"
},
"tests/fixtures/1.png": {
"files": [
{
"type": "image/png",
"uri": "https://arweave.net/BSvIAiwthQu_xwQBHn9FcgACaZ8ko4py5mqMNP4r-jM/tests/fixtures/1.png"
},
{
"type": "image/png",
"uri": "https://arweave.net/Os-tEyRqdjwwyNo1mpLaPGu8_r3KbV-iNRH-aPtJFOw"
}
],
"id": "Os-tEyRqdjwwyNo1mpLaPGu8_r3KbV-iNRH-aPtJFOw"
},
You can run the following command to get an update on the status of your manifest transaction.
arloader get-status `<MANIFEST_ID>`
You can use SOL to pay for your transactions without going through the hassle of procuring AR tokens.
Arloader usage is pretty much exactly the same as above, with the addition of the --with-sol flag.
arloader estimate "<GLOB>" --with-sol
arloader upload "<GLOB>" --log-dir "<LOG_DIR>" --with sol --ar-default-keypair
This will create the same stream of bundles that gets created without using SOL and then goes out to an api to get your transactions signed. Once the SOL payment transaction has gone through, the signature comes back from the api and gets added to your bundle transaction. Then the transaction gets uploaded directly to the arweave.net gateway from your computer.
Arweave is limited to approximately 1,000 transactions every two minutes so if you happen to submit your transaction at a time when there are a lot of pending transactions, it may take longer to get written, or if there are enough more attractive transaction, i.e, with higher rewards, it may not get written at all. To check the current number of pending transactions, run
arloader pending
and that will print the number of pending transactions every second for one minute.
```
118 | ▥▥▥ 123 | ▥▥▥ 124 | ▥▥▥ 224 | ▥▥▥▥▥ 125 | ▥▥▥ 326 | ▥▥▥▥▥▥▥ 128 | ▥▥▥ ```
Given that Arloader bundles by default, your transaction is hopefully relatively attractive and you don't need to increase the reward to get it written in a timely fashion. However, if you see that there are a lot of transactions pending and you want to be sure your transaction goes through quickly, you can adjust the reward with --reward-multiplier followed by something tha can be parsed as a float between 0.0 and 10.0. The reward included in your transaction will then be multiplied by this factor when it gets submitted. Similar to the --with-sol flag, you can add --reward-multiplier to both estimate and upload commands.
You can add the --no-bundle flag if for some reason you want to create individual transactions. This works with both estimate and upload commands. In that case individual status objects are written to LOG_DIR and you can run update-status to update them from the network and status-report for a count of transactions by status.
The table below shows the average duration required to create transactions across a range of file sizes and numbers of files. Detailed statistical analyses and charts can be found here (numbers may vary slightly from those below).
For an NFT project with 10,000 tokens it would take 20 seconds to process the images if they were 256 KB. If the images were 4 MB, it would take approximately two minutes.
| File Size | Num Files | Total Size | Data Item | Data Items | Bundle | Transaction | Total | Per 1,000 | | --------: | --------: | ---------: | --------: | ---------: | -----: | ----------: | ----: | --------: | | 32 KB | 500 | 16 | 4 | 430 | 30 | 40 | 0.5 | 1.0 | | 256 KB | 500 | 128 | 4 | 493 | 179 | 326 | 1.0 | 2.0 | | 1 MB | 500 | 512 | 5 | 616 | 903 | 1033 | 2.6 | 4.1 | | 4 MB | 150 | 614 | 11 | 360 | 1050 | 1554 | 3.0 | 10.4 | | 16 MB | 50 | 819 | 35 | 393 | 1403 | 2058 | 3.9 | 77.1 |
Benchmarks include only processing activity and exclude reading files from disk and uploading them to the network. Benchmarks were performed on an Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz processor with 6 cores.
| Column | Description | | --- | --- | | Total Size | File size x number of files in megabytes.| | Data Item | Time in milliseconds required to create a single data item of the file size. The entails creating a merkle tree data root, generating an id from the deep hash algorithm and signing it.| | Data Items | Time in milliseconds required to create data items for the number of files. Data items are processed in parallel using all available cores.| | Bundle | Time in milliseconds required to create a single bundle from the data items. This entails serializing each of the data items and packing them together.| | Transaction |Time in milliseconds required to create a transaction from the bundle. This entails creating a merkle tree data root, generating an id from the deep hash algorithm and signing it.| | Total | Sum of the time required to create data items, bundle and transaction in seconds.| | Per 1,000 | Extrapolation of total to 1,000 files.|