前言
下半年出于工作变动，Endless的更新频次犹如人的心情般漂浮不定。原本计划分享近两年来在光学卫星辐射校正方向的一些研究心得，实际就是，忙忙忙....... 现在做了个产品经理的角色，也算是新体验，每天与开发打架，你质疑我，我质疑你。
需求场景
言归正传，在某个系统开发中，需要获取每个Sentinel-2数据成像时其太阳及卫星角度参数。在标准的.SAFE格式下，其卫星成像时的角度参数存放在GRANULE/L1C_XXXXXX_XXXXXXXXXXXXXX下的MTD_TL.xml文件中。那么如何在无须下载整个产品.SAFE格式数据下批量获取呢？
算法实现逻辑
涉及这种产品级的访问，当然优先考虑OData ，但距离接触OData已是两年前，因此我选择去看OData API官方文档寻找我想要的答案，能不能直接通过API访问至产品内部的各级文件？有Listing product nodes章节这就好办了。
OData API官方文档：https://documentation.dataspace.copernicus.eu/APIs/OData.html#query-by-geographic-criteria
问题：为什么不通过GEE这类云计算平台获取？
回答：该MTD_TL.xml文件中的角度参数只有官方.SAFE标准格式中有，里面存放着每个波段、每个传感器焦平面成像的角度参数。
通过输入ROI坐标、成像时间、卫星名称、云量等参数访问至产品级；
问题：为什么访问查询至产品时，只保留第一个产品？
回答：因为对于我们查询的ROI，该地点可能会位于两景影像重叠处，这就意味着有两个返回，为了方便处理，直接取列表的第一个产品。
对于一个产品，其API响应的JSON数据体所包含16个属性：
其各属性解释如下：
|
|
|
|
|---|---|---|
@odata.mediaContentType |
application/octet-stream |
.SAFE 文件）
|
Id |
184e8dbe-cd72-477f-a555-ae55ad2ab95d |
|
Name |
S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE |
|
ContentType |
application/octet-stream |
|
ContentLength |
603,498,308
|
|
OriginDate |
2025-05-03T12:26:34Z |
|
PublicationDate |
2025-05-04T10:32:37.407Z |
|
ModificationDate |
2025-05-04T10:32:37.407Z |
|
Online |
True |
False 则需“归档恢复”）
|
EvictionDate |
9999-12-31T23:59:59.999Z |
|
S3Path |
/eodata/Sentinel-2/MSI/L2A/2025/05/03/... |
|
Checksum |
|
|
ContentDate |
2025-05-03T07:56:31.025Z
|
|
Footprint |
|
|
在上一步API响应的JSON数据体中获取到了产品ID,即产品唯一标识符，我们用该产品ID与odata固定API拼接，然后进行requests请求；
产品ID：
184e8dbe-cd72-477f-a555-ae55ad2ab95d固定请求API：url = f"https://download.dataspace.copernicus.eu/odata/v1/Products({product_uuid})/Nodes"
'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes'
.SAFE 产品的目录结构元数据响应
结果返回的是一个 Sentinel-2 .SAFE 产品的目录结构元数据响应，表示该产品是一个“容器”（文件夹），包含多个子节点（即内部的文件和子文件夹）；
{'result': [{'ChildrenNumber': 8, 'ContentLength': 0, 'Id': 'S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE', 'Name': 'S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE', 'Nodes': {'uri': 'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes'}}]}
result |
|
|---|---|
ChildrenNumber
|
.SAFE 文件夹内有 8 个直接子项（可能是文件夹或文件）
|
ContentLength
|
|
Id
Name
|
.SAFE 包名
|
Nodes.uri |
|
具体的举例，即访问至产品的目录层级，对应的是S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE这个目录，现在我们需要利用在响应头中Nodes.uri获取下一级，即文件夹下八个文件的url;
图5 产品目录层级
.SAFE 产品的目录内结构元数据响应
'Nodes': {'uri': 'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes'}
one_namedir= data['result'][0]['Nodes']['uri']
resp1 = requests.get(one_namedir, headers=headers)
data1 = resp1.json()
响应头返回如下内容：
{'result': [{'ChildrenNumber': 6, 'ContentLength': 0, 'Id': 'HTML', 'Name': 'HTML', 'Nodes': {'uri': 'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes(HTML)/Nodes'}}, {'ChildrenNumber': 0, 'ContentLength': 18903, 'Id': 'INSPIRE.xml', 'Name': 'INSPIRE.xml', 'Nodes': {'uri': 'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes(INSPIRE.xml)/Nodes'}}, {'ChildrenNumber': 0, 'ContentLength': 58791, 'Id': 'MTD_MSIL2A.xml', 'Name': 'MTD_MSIL2A.xml', 'Nodes': {'uri': 'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes(MTD_MSIL2A.xml)/Nodes'}}, {'ChildrenNumber': 0, 'ContentLength': 3461, 'Id': 'S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713-ql.jpg', 'Name': 'S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713-ql.jpg', 'Nodes': {'uri': 'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713-ql.jpg)/Nodes'}}, {'ChildrenNumber': 0, 'ContentLength': 69144, 'Id': 'manifest.safe', 'Name': 'manifest.safe', 'Nodes': {'uri': 'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes(manifest.safe)/Nodes'}}, {'ChildrenNumber': 3, 'ContentLength': 0, 'Id': 'rep_info', 'Name': 'rep_info', 'Nodes': {'uri': 'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes(rep_info)/Nodes'}}, {'ChildrenNumber': 1, 'ContentLength': 0, 'Id': 'DATASTRIP', 'Name': 'DATASTRIP', 'Nodes': {'uri': 'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes(DATASTRIP)/Nodes'}}, {'ChildrenNumber': 1, 'ContentLength': 0, 'Id': 'GRANULE', 'Name': 'GRANULE', 'Nodes': {'uri': 'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes(GRANULE)/Nodes'}}]}
上述一大串东西的就是目录下八个文件/文件夹层级的url,即我们进入产品的第二层，产品内部的八个文件/文件夹的url;
因为本次算法的是获取MTD_TL.xml文件，因此我们需要继续访问GRANULE文件夹，即使用GRANULE的url；
{'ChildrenNumber': 1, 'ContentLength': 0, 'Id': 'GRANULE', 'Name': 'GRANULE', 'Nodes': {'uri': 'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes(GRANULE)/Nodes'}}
two_dir = next((item['Nodes']['uri'] for item in data1['result'] if item['Id'] == 'GRANULE'), None)
resp2 = requests.get(two_dir, headers=headers)
data2 = resp2.json()
GRANULE文件夹结构元数据响应
响应头返回如下内容：
{'result': [{'ChildrenNumber': 4, 'ContentLength': 0, 'Id': 'L2A_T37QBG_A003437_20250503T081037', 'Name': 'L2A_T37QBG_A003437_20250503T081037', 'Nodes': {'uri': 'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes(GRANULE)/Nodes(L2A_T37QBG_A003437_20250503T081037)/Nodes'}}]}
返回的是GRANULE文件夹下一个存放影像数据文件夹，即L2A_T37QBG_A003437_20250503T081037目录；
图8 示例存放影像数据文件夹
访问至GRANULE 目录下的一个Granule子目录的元数据信息。这是 .SAFE 包中实际影像数据的存放位置；
其标准格式为：
L2A_T37QBG_A003437_20250503T081037/
├── IMG_DATA/ # 实际影像波段（.jp2 格式）
├── AUX_DATA/ # 辅助数据（如大气模型）
├── QI_DATA/ # 质量信息（如云掩膜、雪掩膜）
├── MTD_TL.xml # 像元级元数据文件
pr_dir = data2['result'][0]['Nodes']['uri']
resp3 = requests.get(pr_dir, headers=headers)
data3 = resp3.json()
L2A_xxxxxx_xxxxxxx_xxxxxxxxxxxxxxxxx文件夹内结构元数据响应
响应头返回如下内容：
{'result': [{'ChildrenNumber': 2, 'ContentLength': 0, 'Id': 'AUX_DATA', 'Name': 'AUX_DATA', 'Nodes': {'uri': 'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes(GRANULE)/Nodes(L2A_T37QBG_A003437_20250503T081037)/Nodes(AUX_DATA)/Nodes'}}, {'ChildrenNumber': 3, 'ContentLength': 0, 'Id': 'IMG_DATA', 'Name': 'IMG_DATA', 'Nodes': {'uri': 'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes(GRANULE)/Nodes(L2A_T37QBG_A003437_20250503T081037)/Nodes(IMG_DATA)/Nodes'}}, {'ChildrenNumber': 0, 'ContentLength': 368363, 'Id': 'MTD_TL.xml', 'Name': 'MTD_TL.xml', 'Nodes': {'uri': 'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes(GRANULE)/Nodes(L2A_T37QBG_A003437_20250503T081037)/Nodes(MTD_TL.xml)/Nodes'}}, {'ChildrenNumber': 37, 'ContentLength': 0, 'Id': 'QI_DATA', 'Name': 'QI_DATA', 'Nodes': {'uri': 'https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes(GRANULE)/Nodes(L2A_T37QBG_A003437_20250503T081037)/Nodes(QI_DATA)/Nodes'}}]}
分别对应四个文件/文件夹的URL,对于我们需要的MTD_TLxml,已经访问至文件本身，即可通过其提供的URL进行下载；
MTD_TL.xml完整URL
MTD_TL = next((item['Nodes']['uri'] for item in data3['result'] if item['Id'] == 'MTD_TL.xml'), None)
MTD_TL ='https://download.dataspace.copernicus.eu/odata/v1/Products(184e8dbe-cd72-477f-a555-ae55ad2ab95d)/Nodes(S2C_MSIL2A_20250503T075631_N0511_R035_T37QBG_20250503T113713.SAFE)/Nodes(GRANULE)/Nodes(L2A_T37QBG_A003437_20250503T081037)/Nodes(MTD_TL.xml)/Nodes'
已经获取到文件本身URL，直接通过API拼接即可完成下载。
图11 实际运行效果
完整代码
"""
该脚本为批量输入ROI的.geojson文件，下载Sentinel-2产品多层级下的单个文件
OData API官方文档
https://documentation.dataspace.copernicus.eu/APIs/OData.html#query-by-geographic-criteria
Access token官方文档
https://documentation.dataspace.copernicus.eu/APIs/Token.html
# 响应头所包含的属性
"@odata.context": "$metadata#Products",
"value": [
{
"@odata.mediaContentType": "application/octet-stream",
"Id": "b30f2bbf-7f02-494a-b3f4-3ea11cd3d14d",
"Name": "S3A_REP1_AUX_POEORB_POD__20161216T083833_V20160226T215943_20160227T235943_DGNS.EOF",
"ContentType": "application/octet-stream",
"ContentLength": 4410152,
"OriginDate": "2021-04-17T09:08:14.628Z",
"PublicationDate": "2023-10-25T16:30:52.016Z",
"ModificationDate": "2023-10-25T18:03:45.644Z",
"Online": true,
"EvictionDate": "",
"S3Path": "/eodata/Sentinel-3/AUX/AUX_POEORB/2016/02/26/S3A_REP1_AUX_POEORB_POD__20161216T083833_V20160226T215943_20160227T235943_DGNS.EOF",
"Checksum": [
{
"Value": "32d1b4ee5cf9255ac0536d658255c1fe",
"Algorithm": "MD5",
"ChecksumDate": "2023-10-25T18:03:44.760291Z"
},
{
"Value": "56a78e26853a6b69021cc17c7cf495c4473ecb415e857a3bf62aa55beb60d5cd",
"Algorithm": "BLAKE3",
"ChecksumDate": "2023-10-25T18:03:44.779812Z"
}
],
"ContentDate": {
"Start": "2016-02-26T21:59:43.000Z",
"End": "2016-02-27T23:59:43.000Z"
},
"Footprint": null,
"GeoFootprint": null
},
"""
import os
import sys
import json
import hashlib
import requests
import tqdm
from osgeo import ogr, gdal
import pandas as pd
import re
from datetime import datetime, timedelta
# -------------------- 配置参数 --------------------
EMAIL ="longchｘｘuo@163.com"
PASSWORD = "ｘｘｘｘｘｘｘｘｘｘ"
OUTPUT_DIR = r"C:\Users\ｘｘｘｘ\Desktop\下载Sentinel-2单个系数\xml"
INPUT_DIR = r"C:\Users\ｘｘｘｘｘ\Desktop\下载Sentinel-2单个系数\roi" # 存放geojson文件的文件夹
SATELLITE_NAME = "SENTINEL-2"
PRODUCT_TYPE = "MSIL2A"
CLOUD_COVER = 100
# -------------------- 工具函数 --------------------
def GetFileMd5(filename):
ifnot os.path.isfile(filename):
returnNone
myHash = hashlib.md5()
with open(filename, 'rb') as f:
whileTrue:
b = f.read(8096)
ifnot b:
break
myHash.update(b)
return myHash.hexdigest()
def bbox(xmin, xmax, ymin, ymax):
returnf"{xmin} {ymin}, {xmax} {ymin}, {xmax} {ymax}, {xmin} {ymax}, {xmin} {ymin}"
def read_area_interest(insterest_path):
"""
读取兴趣区域文件（geojson / shp），输出 bbox 坐标字符串
"""
if insterest_path.endswith(".geojson"):
with open(insterest_path, "r", encoding="utf-8") as f:
data = json.load(f)
geom_type = data["features"][0]["geometry"]["type"]
coords = []
if geom_type == "Polygon":
coords = data["features"][0]["geometry"]["coordinates"][0]
elif geom_type == "MultiPolygon":
for poly in data["features"][0]["geometry"]["coordinates"]:
coords.extend(poly[0])
else:
raise ValueError(f"不支持的几何类型: {geom_type}")
lons = [c[0] for c in coords]
lats = [c[1] for c in coords]
return bbox(str(min(lons)), str(max(lons)), str(min(lats)), str(max(lats)))
elif insterest_path.endswith(".shp"):
gdal.SetConfigOption("GDAL_FILENAME_IS_UTF8", "NO")
ogr.RegisterAll()
dataset = ogr.Open(insterest_path)
if dataset isNone:
raise IOError(f"无法打开 {insterest_path}")
layer = dataset.GetLayerByIndex(0)
if layer isNone:
raise IOError(f"无法读取 {insterest_path} 图层")
xmin, xmax, ymin, ymax = layer.GetExtent()
return bbox(str(xmin), str(xmax), str(ymin), str(ymax))
else:
raise ValueError(f"无法读取该文件 {insterest_path}，仅支持 .geojson / .shp")
# -------------------- Copernicus Access Token --------------------
def get_access_token(username: str, password: str) -> str:
data = {
"client_id": "cdse-public",
"username": username,
"password": password,
"grant_type": "password",
}
r = requests.post(
"https://identity.dataspace.copernicus.eu/auth/realms/CDSE/protocol/openid-connect/token",
data=data
)
r.raise_for_status()
return r.json()["access_token"]
# -------------------- Sentinel 产品搜索 --------------------
def Sentinel_api(satellite, area_interest, startDate, endDate, cloudCover=None, ProductType=None):
base_url = "https://catalogue.dataspace.copernicus.eu/odata/v1/Products?$filter="
satellite_collection = f"Collection/Name eq '{satellite}'"
roi = f"OData.CSC.Intersects(area=geography'SRID=4326;POLYGON(({area_interest}))')"
time_range = f"ContentDate/Start gt {startDate}T00:00:00.000Z and ContentDate/Start lt {endDate}T00:00:00.000Z"
search_lim = "&$top=1000"
expand_assets = "&$expand=Assets"
if ProductType:
attr_product = f"contains(Name,'{ProductType}')"
else:
attr_product = "true"
if cloudCover:
cloud_cover_filter = f"Attributes/OData.CSC.DoubleAttribute/any(att:att/Name eq 'cloudCover' and att/OData.CSC.DoubleAttribute/Value le {cloudCover})"
request_url = f"{base_url}{attr_product} and {cloud_cover_filter} and {roi} and {satellite_collection} and {time_range}{search_lim}{expand_assets}"
else:
request_url = f"{base_url}{attr_product} and {roi} and {satellite_collection} and {time_range}{search_lim}{expand_assets}"
resp = requests.get(request_url)
resp.raise_for_status()
data = resp.json()
df = pd.DataFrame.from_dict(data['value'])
return df
# -------------------- 递归列出节点 --------------------
def list_nodes(product_uuid, access_token):
headers = {"Authorization": f"Bearer {access_token}"}
url = f"https://download.dataspace.copernicus.eu/odata/v1/Products({product_uuid})/Nodes"
resp = requests.get(url, headers=headers)
resp.raise_for_status()
data = resp.json()
one_namedir= data['result'][0]['Nodes']['uri']
resp1 = requests.get(one_namedir, headers=headers)
data1 = resp1.json()
two_dir = next((item['Nodes']['uri'] for item in data1['result'] if item['Id'] == 'GRANULE'), None)
resp2 = requests.get(two_dir, headers=headers)
data2 = resp2.json()
pr_dir = data2['result'][0]['Nodes']['uri']
resp3 = requests.get(pr_dir, headers=headers)
data3 = resp3.json()
MTD_TL = next((item['Nodes']['uri'] for item in data3['result'] if item['Id'] == 'MTD_TL.xml'), None)
return MTD_TL
# -------------------- 下载文件 --------------------
def download_file(file_path, url, access_token, overwrite=False):
os.makedirs(os.path.dirname(file_path), exist_ok=True)
headers = {"Authorization": f"Bearer {access_token}"}
mode = "wb"
downloaded_bytes = 0
if os.path.exists(file_path) andnot overwrite:
print(f"文件已存在，跳过: {file_path}")
return
with requests.get(url, headers=headers, stream=True) as r:
total_size = int(r.headers.get("Content-Length", 0)) + downloaded_bytes
with tqdm.tqdm(total=total_size, unit='B', unit_scale=True, unit_divisor=1024,
desc=os.path.basename(file_path), initial=downloaded_bytes) as pbar:
with open(file_path, mode) as f:
for chunk in r.iter_content(chunk_size=1024):
if chunk:
f.write(chunk)
pbar.update(len(chunk))
# -------------------- 主程序 --------------------
if __name__ == "__main__":
geojson_files = [f for f in os.listdir(INPUT_DIR) if f.endswith(".geojson")]
# 获取 xml 下载链接
access_token = get_access_token(EMAIL, PASSWORD)
for geojson in geojson_files:
geojson_path = os.path.join(INPUT_DIR, geojson)
print(f"\n>>> 处理文件: {geojson_path}")
# 从文件名提取日期
date_str = geojson.split("_")[0] # 例如 "20250130"
date_obj = datetime.strptime(date_str, "%Y%m%d")
START_DATE = date_obj.strftime("%Y-%m-%d")
END_DATE = (date_obj + timedelta(days=1)).strftime("%Y-%m-%d")
# 读取ROI
roi_bbox = read_area_interest(geojson_path)
# 查询产品
df_products = Sentinel_api(SATELLITE_NAME, roi_bbox, START_DATE, END_DATE,
cloudCover=CLOUD_COVER, ProductType=PRODUCT_TYPE)
if df_products.empty:
print("未查询到数据，跳过。")
continue
# 只保留第一个产品
row = df_products.iloc[0]
product_uuid = row['Id']
product_name = row['Name']
print(f"选定产品: {product_name} | UUID: {product_uuid}")
node_url = list_nodes(product_uuid, access_token)
matches = re.findall(r'\((.*?)\)', node_url)
if len(matches) < 5:
print("未能解析节点URL，跳过。")
continue
product_uuid, safe_name, granule, granule_id, file_name = matches
xml_name = product_name[:-5] + '.xml'
local_path = os.path.join(OUTPUT_DIR, xml_name)
url_part1 = "https://zipper.dataspace.copernicus.eu/odata/v1/Products%28"
url_part2 = "/$value"
data_url = url_part1 + product_uuid + '%29/Nodes%28' + safe_name + '%29/Nodes%28GRANULE%29/Nodes%28' + granule_id + '%29/Nodes%28MTD_TL.xml%29' + url_part2
print(f"下载URL: {data_url}")
download_file(local_path, data_url, access_token)