API
通过 anyparse-cli api 启动服务器：请查看 CLI 工具
OpenAI 代理
- 基础 URL:
http://your_ip:your_port/anyparse/openai/v1 - API Key:
your_custom_secret_key
模型列表和文件 MIME 类型映射
from openai import OpenAI
client = OpenAI(
base_url="http://localhost:18007/anyparse/openai/v1",
api_key="your_custom_secret_key",
)
print(client.models.list())
非流式解析文件
import base64
with open("1.pdf", "r", encoding="utf-8") as f:
text_content = f.read()
encoded_bytes = base64.b64encode(text_content.encode('utf-8'))
base64_str = encoded_bytes.decode('utf-8')
response = client.chat.completions.create(
model="anyparse",
messages=[
{
"role": "user",
"content": [
{
"type": "file",
"file": {
"file_data": f"data:application/pdf;base64,{base64_str}"
}
}
]
}
], # data:application/pdf;base64 前缀遵循: client.models.list().data[0].allow_mimetypes
# extra_body={
# "runtimes_args": {
# "use_doc_layout": True
# }
# }
)
print(response.choices[0].message.content)
extra_body 参数
通过 runtime_args 传递上述调用参数。
输出格式
{
'code': 2000,
'data': {
'metadata': {
'file_idx': 'task-bb581aed6e4a469694613c454bce72e9',
'file_md5': '459b4c627b72e50a8e649c6329f1f1a8',
'file_type': 'pdf',
'file_name': '11.pdf',
'file_size': '3.99/KB'},
'pages': [ # 页面列表
{
'id': 1, # 页面 ID
'content': 'dsdddddddd', # 页面内容
'layout': [
{
'order_id': 0, # 阅读顺序
'label': 'text', # 布局类型
'box': [578, 1019, 972, 1095], # 布局框
'content': 'dsdddddddd' # 布局内容
}
],
'elapse_times': 2.4340291023254395 # 页面耗时
},
...
],
'content': 'dsdddddddd\n我也', # 文件内容
'elapse_times': 3.8336679935455322 # 文件耗时
},
'msg': 'success'
}
流式解析文件
import base64
with open("1.pdf", "r", encoding="utf-8") as f:
text_content = f.read()
encoded_bytes = base64.b64encode(text_content.encode('utf-8'))
base64_str = encoded_bytes.decode('utf-8')
response = client.chat.completions.create(
model="anyparse",
stream=True,
messages=[
{
"role": "user",
"content": [
{
"type": "file",
"file": {
"file_data": f"data:application/pdf;base64,{base64_str}"
}
}
]
}
], # data:application/pdf;base64 前缀遵循: client.models.list().data[0].allow_mimetypes
# extra_body={
# "runtimes_args": {
# "use_doc_layout": True
# }
# }
)
for x in response:
print(x)
创建上传文件
response = client.files.create(
file=open("11.pdf", "rb"),
purpose="assistants",
# extra_body = {
# "maxsize": 100.0,
# "file_read_size": 1024
# }
)
获取文件信息
client.files.retrieve(
file_id="your file id",
)
获取文件内容
response = client.files.content(
file_id="your file id"
)
with open("your file", "wb") as f:
for chunk in response.iter_bytes(chunk_size=1024*1024): # 每次读 1MB
f.write(chunk)
# 详情请查看 headers
print(response.response.headers)
删除文件
client.files.delete(
file_id="your file id",
)
通过文件 ID 解析文件
response = client.chat.completions.create(
model="anyparse",
messages=[
{
"role": "user",
"content": [
{
"type": "file_id",
"file": {
"file_data": "your file id"
}
}
]
}
],
stream=False,
# extra_body = {
# "runtimes_args": {
# "maxsize": 50
# }
# }
)
通过 OpenAI responses 接口解析文件
response = client.responses.create(
model="anyparse",
input=[
{
"role": "user",
"content": [
{
"type": "file_id",
"file": {
"file_data": "your file id"
}
}
]
}
],
stream=False,
# extra_body = {}
)
print(response.output_text)
RESTful API
- 基础 URL:
http://your_ip:your_port/anyparse - API Key:
your_custom_secret_key
文件类型
url = 'http://your_ip:your_port/anyparse/filetypes/v1'
headers = {
"Authorization": f"Bearer your_custom_secret_key",
}
res = httpx.get(url, headers=headers)
print(res.json())
解析文件
import httpx
args = {
"use_doc_cls": False,
"use_doc_rectifier": False,
"use_doc_layout": True,
"stream": False
}
url = f'http://your_ip:your_port/anyparse/invoke/v1'
headers = {
"Authorization": f"Bearer your_custom_secret_key",
}
files = {
'file': open('11.pdf','rb')
}
res = httpx.post(url, data=args, headers=headers, files = files)
print(res.json())
API 参数
通过 args 传递上述调用参数。
输出格式
{
'code': 2000,
'data': {
'metadata': {
'file_idx': 'task-bb581aed6e4a469694613c454bce72e9',
'file_md5': '459b4c627b72e50a8e649c6329f1f1a8',
'file_type': 'pdf',
'file_name': '11.pdf',
'file_size': '3.99/KB'},
'pages': [ # 页面列表
{
'id': 1, # 页面 ID
'content': 'dsdddddddd', # 页面内容
'layout': [
{
'order_id': 0, # 阅读顺序
'label': 'text', # 布局类型
'box': [578, 1019, 972, 1095], # 布局框
'content': 'dsdddddddd' # 布局内容
}
],
'elapse_times': 2.4340291023254395 # 页面耗时
},
...
],
'content': 'dsdddddddd\n我也', # 文件内容
'elapse_times': 3.8336679935455322 # 文件耗时
},
'msg': 'success'
}
流式解析文件
import json
import httpx
args = {
"use_doc_cls": False,
"use_doc_rectifier": False,
"use_doc_layout": True,
"stream": True
}
url = f'http://your_ip:your_port/anyparse/invoke/v1'
headers = {
"Authorization": f"Bearer your_custom_secret_key",
}
files = {
'file': open('11.pdf','rb')
}
with httpx.Client(timeout=3600) as client:
with client.stream("POST", url, data=args, headers=headers, files = files) as response:
response.raise_for_status()
buffer = ""
for chunk_bytes in response.iter_bytes():
buffer += chunk_bytes.decode("utf-8")
lines = buffer.split("\n")
buffer = lines[-1] # 保留不完整行
for line in lines[:-1]:
line = line.strip()
if line and line.startswith("data:"):
payload = line[5:].strip()
if not payload:
continue
data = json.loads(payload)
text = data.get("content", "")
pageid = int(data.get("id", 1)) + 1
print(data)