working simple api bridge done
This commit is contained in:
TCUDIKEL
2
.gitignore
vendored
2
.gitignore
vendored
@@ -3,6 +3,6 @@ venv
|
||||
config.json
|
||||
mcphost
|
||||
mcphost_openai_api.log.*
|
||||
helpers/settings.py
|
||||
settings.py
|
||||
.idea
|
||||
__pycache__
|
||||
31
.idea/inspectionProfiles/Project_Default.xml
generated
31
.idea/inspectionProfiles/Project_Default.xml
generated
@@ -1,31 +0,0 @@
|
||||
<component name="InspectionProjectProfileManager">
|
||||
<profile version="1.0">
|
||||
<option name="myName" value="Project Default" />
|
||||
<inspection_tool class="PyPackageRequirementsInspection" enabled="true" level="WARNING" enabled_by_default="true">
|
||||
<option name="ignoredPackages">
|
||||
<value>
|
||||
<list size="18">
|
||||
<item index="0" class="java.lang.String" itemvalue="pandas" />
|
||||
<item index="1" class="java.lang.String" itemvalue="tiktoken" />
|
||||
<item index="2" class="java.lang.String" itemvalue="gensim" />
|
||||
<item index="3" class="java.lang.String" itemvalue="pydantic" />
|
||||
<item index="4" class="java.lang.String" itemvalue="configparser" />
|
||||
<item index="5" class="java.lang.String" itemvalue="future" />
|
||||
<item index="6" class="java.lang.String" itemvalue="tenacity" />
|
||||
<item index="7" class="java.lang.String" itemvalue="python-dotenv" />
|
||||
<item index="8" class="java.lang.String" itemvalue="aiohttp" />
|
||||
<item index="9" class="java.lang.String" itemvalue="pipmaster" />
|
||||
<item index="10" class="java.lang.String" itemvalue="setuptools" />
|
||||
<item index="11" class="java.lang.String" itemvalue="xlsxwriter" />
|
||||
<item index="12" class="java.lang.String" itemvalue="numpy" />
|
||||
<item index="13" class="java.lang.String" itemvalue="scipy" />
|
||||
<item index="14" class="java.lang.String" itemvalue="filelock" />
|
||||
<item index="15" class="java.lang.String" itemvalue="pycryptodomex" />
|
||||
<item index="16" class="java.lang.String" itemvalue="lxml" />
|
||||
<item index="17" class="java.lang.String" itemvalue="blobfile" />
|
||||
</list>
|
||||
</value>
|
||||
</option>
|
||||
</inspection_tool>
|
||||
</profile>
|
||||
</component>
|
||||
6
.idea/inspectionProfiles/profiles_settings.xml
generated
6
.idea/inspectionProfiles/profiles_settings.xml
generated
@@ -1,6 +0,0 @@
|
||||
<component name="InspectionProjectProfileManager">
|
||||
<settings>
|
||||
<option name="USE_PROJECT_PROFILE" value="false" />
|
||||
<version value="1.0" />
|
||||
</settings>
|
||||
</component>
|
||||
10
.idea/mcphost-api.iml
generated
10
.idea/mcphost-api.iml
generated
@@ -1,10 +0,0 @@
|
||||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<module type="PYTHON_MODULE" version="4">
|
||||
<component name="NewModuleRootManager">
|
||||
<content url="file://$MODULE_DIR$">
|
||||
<excludeFolder url="file://$MODULE_DIR$/.venv" />
|
||||
</content>
|
||||
<orderEntry type="jdk" jdkName="Python 3.12 (mcphost-api)" jdkType="Python SDK" />
|
||||
<orderEntry type="sourceFolder" forTests="false" />
|
||||
</component>
|
||||
</module>
|
||||
8
.idea/modules.xml
generated
8
.idea/modules.xml
generated
@@ -1,8 +0,0 @@
|
||||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<project version="4">
|
||||
<component name="ProjectModuleManager">
|
||||
<modules>
|
||||
<module fileurl="file://$PROJECT_DIR$/.idea/mcphost-api.iml" filepath="$PROJECT_DIR$/.idea/mcphost-api.iml" />
|
||||
</modules>
|
||||
</component>
|
||||
</project>
|
||||
6
.idea/vcs.xml
generated
6
.idea/vcs.xml
generated
@@ -1,6 +0,0 @@
|
||||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<project version="4">
|
||||
<component name="VcsDirectoryMappings">
|
||||
<mapping directory="$PROJECT_DIR$" vcs="Git" />
|
||||
</component>
|
||||
</project>
|
||||
102
.idea/workspace.xml
generated
102
.idea/workspace.xml
generated
@@ -1,102 +0,0 @@
|
||||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<project version="4">
|
||||
<component name="AutoImportSettings">
|
||||
<option name="autoReloadType" value="SELECTIVE" />
|
||||
</component>
|
||||
<component name="ChangeListManager">
|
||||
<list default="true" id="bbbc2257-1e9d-40b2-a56b-3daf4fc0a636" name="Changes" comment="">
|
||||
<change afterPath="$PROJECT_DIR$/.idea/inspectionProfiles/Project_Default.xml" afterDir="false" />
|
||||
<change afterPath="$PROJECT_DIR$/.idea/inspectionProfiles/profiles_settings.xml" afterDir="false" />
|
||||
<change afterPath="$PROJECT_DIR$/.idea/mcphost-api.iml" afterDir="false" />
|
||||
<change afterPath="$PROJECT_DIR$/.idea/modules.xml" afterDir="false" />
|
||||
<change afterPath="$PROJECT_DIR$/.idea/vcs.xml" afterDir="false" />
|
||||
<change afterPath="$PROJECT_DIR$/.idea/workspace.xml" afterDir="false" />
|
||||
<change afterPath="$PROJECT_DIR$/helpers/__init__.py" afterDir="false" />
|
||||
<change afterPath="$PROJECT_DIR$/test.sh" afterDir="false" />
|
||||
</list>
|
||||
<option name="SHOW_DIALOG" value="false" />
|
||||
<option name="HIGHLIGHT_CONFLICTS" value="true" />
|
||||
<option name="HIGHLIGHT_NON_ACTIVE_CHANGELIST" value="false" />
|
||||
<option name="LAST_RESOLUTION" value="IGNORE" />
|
||||
</component>
|
||||
<component name="FileTemplateManagerImpl">
|
||||
<option name="RECENT_TEMPLATES">
|
||||
<list>
|
||||
<option value="Python Script" />
|
||||
</list>
|
||||
</option>
|
||||
</component>
|
||||
<component name="Git.Settings">
|
||||
<option name="RECENT_GIT_ROOT_PATH" value="$PROJECT_DIR$" />
|
||||
</component>
|
||||
<component name="ProjectColorInfo"><![CDATA[{
|
||||
"associatedIndex": 5
|
||||
}]]></component>
|
||||
<component name="ProjectId" id="2wuM0HE6IiWx1ypaydX36jB6Bh3" />
|
||||
<component name="ProjectLevelVcsManager" settingsEditedManually="true" />
|
||||
<component name="ProjectViewState">
|
||||
<option name="hideEmptyMiddlePackages" value="true" />
|
||||
<option name="showLibraryContents" value="true" />
|
||||
</component>
|
||||
<component name="PropertiesComponent"><![CDATA[{
|
||||
"keyToString": {
|
||||
"Python.serve_openai_compatible.executor": "Debug",
|
||||
"RunOnceActivity.ShowReadmeOnStart": "true",
|
||||
"RunOnceActivity.git.unshallow": "true",
|
||||
"git-widget-placeholder": "main"
|
||||
}
|
||||
}]]></component>
|
||||
<component name="RecentsManager">
|
||||
<key name="MoveFile.RECENT_KEYS">
|
||||
<recent name="$PROJECT_DIR$/helpers" />
|
||||
<recent name="$PROJECT_DIR$/utils" />
|
||||
</key>
|
||||
</component>
|
||||
<component name="RunManager">
|
||||
<configuration name="serve_openai_compatible" type="PythonConfigurationType" factoryName="Python" temporary="true" nameIsGenerated="true">
|
||||
<module name="mcphost-api" />
|
||||
<option name="ENV_FILES" value="" />
|
||||
<option name="INTERPRETER_OPTIONS" value="" />
|
||||
<option name="PARENT_ENVS" value="true" />
|
||||
<envs>
|
||||
<env name="PYTHONUNBUFFERED" value="1" />
|
||||
</envs>
|
||||
<option name="SDK_HOME" value="" />
|
||||
<option name="WORKING_DIRECTORY" value="$PROJECT_DIR$" />
|
||||
<option name="IS_MODULE_SDK" value="true" />
|
||||
<option name="ADD_CONTENT_ROOTS" value="true" />
|
||||
<option name="ADD_SOURCE_ROOTS" value="true" />
|
||||
<option name="SCRIPT_NAME" value="$PROJECT_DIR$/serve_openai_compatible.py" />
|
||||
<option name="PARAMETERS" value="" />
|
||||
<option name="SHOW_COMMAND_LINE" value="false" />
|
||||
<option name="EMULATE_TERMINAL" value="false" />
|
||||
<option name="MODULE_MODE" value="false" />
|
||||
<option name="REDIRECT_INPUT" value="false" />
|
||||
<option name="INPUT_FILE" value="" />
|
||||
<method v="2" />
|
||||
</configuration>
|
||||
<recent_temporary>
|
||||
<list>
|
||||
<item itemvalue="Python.serve_openai_compatible" />
|
||||
</list>
|
||||
</recent_temporary>
|
||||
</component>
|
||||
<component name="SharedIndexes">
|
||||
<attachedChunks>
|
||||
<set>
|
||||
<option value="bundled-python-sdk-4f4e415b4190-aa17d162503b-com.jetbrains.pycharm.community.sharedIndexes.bundled-PC-243.26053.29" />
|
||||
</set>
|
||||
</attachedChunks>
|
||||
</component>
|
||||
<component name="SpellCheckerSettings" RuntimeDictionaries="0" Folders="0" CustomDictionaries="0" DefaultDictionary="application-level" UseSingleDictionary="true" transferred="true" />
|
||||
<component name="TaskManager">
|
||||
<task active="true" id="Default" summary="Default task">
|
||||
<changelist id="bbbc2257-1e9d-40b2-a56b-3daf4fc0a636" name="Changes" comment="" />
|
||||
<created>1746886781301</created>
|
||||
<option name="number" value="Default" />
|
||||
<option name="presentableId" value="Default" />
|
||||
<updated>1746886781301</updated>
|
||||
</task>
|
||||
<servers />
|
||||
</component>
|
||||
</project>
|
||||
@@ -1,15 +0,0 @@
|
||||
|
||||
|
||||
|
||||
class LoggerWriter:
|
||||
def __init__(self, logger_func):
|
||||
self.logger_func = logger_func
|
||||
|
||||
def write(self, message):
|
||||
# Remove trailing newlines for cleaner logs
|
||||
if message.strip():
|
||||
self.logger_func(message.rstrip())
|
||||
|
||||
def flush(self):
|
||||
# Required for file-like objects
|
||||
pass
|
||||
@@ -1,15 +0,0 @@
|
||||
from typing import Optional, List
|
||||
from pydantic import BaseModel
|
||||
|
||||
|
||||
# OpenAI-compatible models
|
||||
class ChatMessage(BaseModel):
|
||||
role: str
|
||||
content: str
|
||||
|
||||
class ChatCompletionRequest(BaseModel):
|
||||
model: str = "mcphost-model"
|
||||
messages: List[ChatMessage]
|
||||
max_tokens: Optional[int] = 512
|
||||
temperature: Optional[float] = 0.1
|
||||
stream: Optional[bool] = False
|
||||
425
serve.py
Normal file
425
serve.py
Normal file
@@ -0,0 +1,425 @@
|
||||
import json
|
||||
import re
|
||||
import sys
|
||||
import threading
|
||||
import time
|
||||
from http.server import HTTPServer, BaseHTTPRequestHandler
|
||||
from typing import Optional, List
|
||||
from functools import partial, wraps
|
||||
|
||||
import pexpect
|
||||
from loguru import logger
|
||||
|
||||
from settings import settings
|
||||
|
||||
|
||||
def log_performance(func):
|
||||
"""Decorator to log function performance"""
|
||||
|
||||
@wraps(func)
|
||||
def wrapper(*args, **kwargs):
|
||||
start_time = time.time()
|
||||
logger.debug("Starting {}", func.__name__)
|
||||
|
||||
try:
|
||||
result = func(*args, **kwargs)
|
||||
duration = time.time() - start_time
|
||||
logger.debug("{} completed in {:.2f}s", func.__name__, duration)
|
||||
return result
|
||||
except Exception as e:
|
||||
duration = time.time() - start_time
|
||||
logger.error("{} failed after {:.2f}s: {}", func.__name__, duration, str(e))
|
||||
raise
|
||||
|
||||
return wrapper
|
||||
|
||||
|
||||
# Configure loguru
|
||||
logger.remove() # Remove default handler
|
||||
|
||||
# Console handler only
|
||||
logger.add(
|
||||
sys.stderr,
|
||||
level="DEBUG" if settings.debug else "INFO",
|
||||
format="<green>{time:YYYY-MM-DD HH:mm:ss}</green> | <level>{level: <8}</level> | <cyan>{name}</cyan>:<cyan>{function}</cyan>:<cyan>{line}</cyan> - <level>{message}</level>",
|
||||
colorize=True,
|
||||
backtrace=True,
|
||||
diagnose=True
|
||||
)
|
||||
|
||||
|
||||
class Config:
|
||||
"""Configuration constants for the application"""
|
||||
SPAWN_TIMEOUT = 60
|
||||
ECHO_DELAY = 0.5
|
||||
READ_TIMEOUT = 0.1
|
||||
RESPONSE_WAIT_TIME = 2
|
||||
CHUNK_SIZE = 1000
|
||||
MAX_READ_SIZE = 10000
|
||||
PROMPT_INDICATOR = "Enter your prompt"
|
||||
ANSI_PATTERN = re.compile(r'\x1B(?:[@-Z\\-_]|\[[0-?]*[ -/]*[@-~])')
|
||||
TUI_BORDER = '┃'
|
||||
SKIP_PATTERNS = ['alt+enter', 'Enter your prompt']
|
||||
|
||||
|
||||
class ResponseCleaner:
|
||||
"""Handles cleaning and formatting of MCP responses"""
|
||||
|
||||
def __init__(self):
|
||||
self.ansi_pattern = Config.ANSI_PATTERN
|
||||
self.tui_border = Config.TUI_BORDER
|
||||
self.skip_patterns = Config.SKIP_PATTERNS
|
||||
|
||||
def clean(self, response: str, original_prompt: str) -> str:
|
||||
"""Clean response with clear steps"""
|
||||
if not response:
|
||||
return ""
|
||||
|
||||
response = self._remove_ansi(response)
|
||||
lines = self._extract_content_lines(response, original_prompt)
|
||||
return '\n'.join(lines)
|
||||
|
||||
def _remove_ansi(self, text: str) -> str:
|
||||
"""Remove ANSI escape sequences"""
|
||||
return self.ansi_pattern.sub('', text)
|
||||
|
||||
def _extract_content_lines(self, response: str, original_prompt: str) -> List[str]:
|
||||
"""Extract meaningful content lines from response"""
|
||||
lines = response.split('\n')
|
||||
cleaned_lines = []
|
||||
|
||||
for line in lines:
|
||||
cleaned_line = self._process_line(line, original_prompt)
|
||||
if cleaned_line is not None:
|
||||
cleaned_lines.append(cleaned_line)
|
||||
|
||||
return cleaned_lines
|
||||
|
||||
def _process_line(self, line: str, original_prompt: str) -> Optional[str]:
|
||||
"""Process a single line and return cleaned content or None to skip"""
|
||||
stripped = line.strip()
|
||||
|
||||
# Skip empty lines
|
||||
if not stripped:
|
||||
return None
|
||||
|
||||
# Skip the original prompt
|
||||
if stripped == original_prompt:
|
||||
return None
|
||||
|
||||
# Handle TUI decorations
|
||||
if stripped.startswith(self.tui_border):
|
||||
content = stripped.strip(self.tui_border).strip()
|
||||
if content and content != original_prompt:
|
||||
return content
|
||||
return None
|
||||
|
||||
# Skip navigation hints
|
||||
if any(pattern in line for pattern in self.skip_patterns):
|
||||
return None
|
||||
|
||||
# Return non-empty, non-decoration lines
|
||||
return stripped
|
||||
|
||||
|
||||
class ProcessManager:
|
||||
"""Manages the MCP process lifecycle"""
|
||||
|
||||
def __init__(self):
|
||||
self.child: Optional[pexpect.spawn] = None
|
||||
self.config = Config()
|
||||
|
||||
@log_performance
|
||||
def start(self) -> bool:
|
||||
"""Start the mcphost process"""
|
||||
command = self._build_command()
|
||||
logger.info("Starting mcphost: {}", ' '.join(command))
|
||||
|
||||
try:
|
||||
self.child = pexpect.spawn(
|
||||
' '.join(command),
|
||||
timeout=self.config.SPAWN_TIMEOUT,
|
||||
encoding='utf-8'
|
||||
)
|
||||
self._configure_process()
|
||||
return self._wait_for_ready()
|
||||
except Exception as e:
|
||||
logger.error("Error starting mcphost: {}", e)
|
||||
logger.exception("Full traceback:")
|
||||
return False
|
||||
|
||||
def is_alive(self) -> bool:
|
||||
"""Check if the process is running"""
|
||||
return self.child is not None and self.child.isalive()
|
||||
|
||||
def shutdown(self):
|
||||
"""Shutdown mcphost gracefully"""
|
||||
if self.is_alive():
|
||||
logger.info("Shutting down mcphost...")
|
||||
self.child.sendcontrol('c')
|
||||
self.child.close()
|
||||
logger.info("MCPHost shutdown complete")
|
||||
|
||||
def _build_command(self) -> List[str]:
|
||||
"""Build the command to start mcphost"""
|
||||
command = [
|
||||
settings.mcphost_path,
|
||||
'--config', settings.mcphost_config,
|
||||
'--model', settings.mcphost_model,
|
||||
'--openai-url', settings.openai_url,
|
||||
'--openai-api-key', settings.openai_api_key
|
||||
]
|
||||
|
||||
if settings.debug:
|
||||
command.insert(1, '--debug')
|
||||
|
||||
return command
|
||||
|
||||
def _configure_process(self):
|
||||
"""Configure the spawned process"""
|
||||
self.child.setecho(False)
|
||||
|
||||
def _wait_for_ready(self) -> bool:
|
||||
"""Wait for the process to be ready"""
|
||||
try:
|
||||
self.child.expect(self.config.PROMPT_INDICATOR)
|
||||
logger.success("MCPHost started and ready")
|
||||
self._clear_buffer()
|
||||
return True
|
||||
except Exception as e:
|
||||
logger.error("Error waiting for prompt: {}", e)
|
||||
return False
|
||||
|
||||
def _clear_buffer(self):
|
||||
"""Clear any remaining output in the buffer"""
|
||||
time.sleep(self.config.ECHO_DELAY)
|
||||
try:
|
||||
self.child.read_nonblocking(
|
||||
size=self.config.MAX_READ_SIZE,
|
||||
timeout=self.config.READ_TIMEOUT
|
||||
)
|
||||
except:
|
||||
pass
|
||||
|
||||
|
||||
class MCPCommunicator:
|
||||
"""Handles communication with the MCP process"""
|
||||
|
||||
def __init__(self, process: ProcessManager, cleaner: ResponseCleaner):
|
||||
self.process = process
|
||||
self.cleaner = cleaner
|
||||
self.config = Config()
|
||||
|
||||
def send_prompt(self, prompt: str) -> str:
|
||||
"""Send a prompt and receive response"""
|
||||
if not self.process.is_alive():
|
||||
raise RuntimeError("MCP process is not running")
|
||||
|
||||
self._clear_pending_output()
|
||||
self._send_command(prompt)
|
||||
response = self._collect_response()
|
||||
return self.cleaner.clean(response, prompt)
|
||||
|
||||
def _clear_pending_output(self):
|
||||
"""Clear any pending output from the process"""
|
||||
try:
|
||||
self.process.child.read_nonblocking(
|
||||
size=self.config.MAX_READ_SIZE,
|
||||
timeout=self.config.READ_TIMEOUT
|
||||
)
|
||||
except:
|
||||
pass
|
||||
|
||||
def _send_command(self, prompt: str):
|
||||
"""Send a command to the process"""
|
||||
logger.debug("Sending prompt: {}", prompt)
|
||||
self.process.child.send(prompt)
|
||||
self.process.child.send('\r')
|
||||
|
||||
# Wait for the model to process
|
||||
time.sleep(self.config.RESPONSE_WAIT_TIME)
|
||||
|
||||
def _collect_response(self) -> str:
|
||||
"""Collect response from the process"""
|
||||
response = ""
|
||||
response_complete = False
|
||||
|
||||
with logger.catch(message="Error during response collection"):
|
||||
while not response_complete:
|
||||
try:
|
||||
chunk = self.process.child.read_nonblocking(
|
||||
size=self.config.CHUNK_SIZE,
|
||||
timeout=3
|
||||
)
|
||||
|
||||
if chunk:
|
||||
response += chunk
|
||||
logger.trace("Received chunk: {}", chunk[:50] + "..." if len(chunk) > 50 else chunk)
|
||||
if self.config.PROMPT_INDICATOR in chunk:
|
||||
response_complete = True
|
||||
else:
|
||||
break
|
||||
|
||||
except pexpect.TIMEOUT:
|
||||
if response and self.config.PROMPT_INDICATOR in response:
|
||||
response_complete = True
|
||||
elif response:
|
||||
logger.debug("Waiting for more response data...")
|
||||
time.sleep(1)
|
||||
continue
|
||||
else:
|
||||
break
|
||||
except Exception as e:
|
||||
logger.error("Error reading response: {}", e)
|
||||
break
|
||||
|
||||
logger.debug("Collected response length: {} characters", len(response))
|
||||
return response
|
||||
|
||||
|
||||
class MCPHostManager:
|
||||
"""Main manager that orchestrates process and communication"""
|
||||
|
||||
def __init__(self):
|
||||
self.process = ProcessManager()
|
||||
self.cleaner = ResponseCleaner()
|
||||
self.communicator = MCPCommunicator(self.process, self.cleaner)
|
||||
self.lock = threading.Lock()
|
||||
|
||||
def __enter__(self):
|
||||
self.start()
|
||||
return self
|
||||
|
||||
def __exit__(self, exc_type, exc_val, exc_tb):
|
||||
self.shutdown()
|
||||
|
||||
def start(self) -> bool:
|
||||
"""Start the MCP host"""
|
||||
return self.process.start()
|
||||
|
||||
@log_performance
|
||||
def send_prompt(self, prompt: str) -> str:
|
||||
"""Send a prompt to mcphost and get the response"""
|
||||
with self.lock:
|
||||
if not self.process.is_alive():
|
||||
logger.warning("MCPHost not running, attempting to restart...")
|
||||
if not self.start():
|
||||
return "Error: Failed to restart MCPHost"
|
||||
|
||||
try:
|
||||
return self.communicator.send_prompt(prompt)
|
||||
except Exception as e:
|
||||
logger.exception("Exception in send_prompt: {}", str(e))
|
||||
return f"Error: {str(e)}"
|
||||
|
||||
def shutdown(self):
|
||||
"""Shutdown mcphost gracefully"""
|
||||
self.process.shutdown()
|
||||
|
||||
|
||||
class SubprocessHandler(BaseHTTPRequestHandler):
|
||||
"""HTTP request handler with dependency injection"""
|
||||
|
||||
def __init__(self, *args, manager: MCPHostManager, **kwargs):
|
||||
self.manager = manager
|
||||
super().__init__(*args, **kwargs)
|
||||
|
||||
@logger.catch
|
||||
def do_POST(self):
|
||||
"""Handle POST requests"""
|
||||
try:
|
||||
content_length = self._get_content_length()
|
||||
input_data = self._read_request_body(content_length)
|
||||
prompt = self._parse_prompt(input_data)
|
||||
|
||||
logger.info("Received prompt: {}", prompt)
|
||||
response = self.manager.send_prompt(prompt)
|
||||
logger.info("Response: {}", response)
|
||||
|
||||
self._send_success_response(response)
|
||||
except ValueError as e:
|
||||
logger.warning("Bad request: {}", str(e))
|
||||
self._send_error_response(400, str(e))
|
||||
except Exception as e:
|
||||
logger.exception("Unexpected error in POST handler")
|
||||
self._send_error_response(500, "Internal server error")
|
||||
|
||||
def do_GET(self):
|
||||
"""Handle GET requests (not supported)"""
|
||||
self._send_error_response(404, "Not Found")
|
||||
|
||||
def log_message(self, format, *args):
|
||||
"""Override to prevent default logging"""
|
||||
pass
|
||||
|
||||
def _get_content_length(self) -> int:
|
||||
"""Get content length from headers"""
|
||||
return int(self.headers.get('Content-Length', 0))
|
||||
|
||||
def _read_request_body(self, content_length: int) -> str:
|
||||
"""Read request body"""
|
||||
if content_length == 0:
|
||||
raise ValueError("Empty request body")
|
||||
return self.rfile.read(content_length).decode('utf-8')
|
||||
|
||||
def _parse_prompt(self, input_data: str) -> str:
|
||||
"""Parse prompt from input data"""
|
||||
try:
|
||||
data = json.loads(input_data)
|
||||
return data.get('prompt', input_data)
|
||||
except json.JSONDecodeError:
|
||||
return input_data
|
||||
|
||||
def _send_success_response(self, content: str):
|
||||
"""Send successful response"""
|
||||
self.send_response(200)
|
||||
self.send_header('Content-type', 'text/plain')
|
||||
self.end_headers()
|
||||
self.wfile.write(content.encode('utf-8'))
|
||||
|
||||
def _send_error_response(self, code: int, message: str):
|
||||
"""Send error response"""
|
||||
self.send_response(code)
|
||||
self.send_header('Content-type', 'text/plain')
|
||||
self.end_headers()
|
||||
self.wfile.write(message.encode('utf-8'))
|
||||
|
||||
|
||||
def create_server(manager: MCPHostManager) -> HTTPServer:
|
||||
"""Factory function to create HTTP server with manager"""
|
||||
handler = partial(SubprocessHandler, manager=manager)
|
||||
return HTTPServer((settings.host, settings.port), handler)
|
||||
|
||||
|
||||
@logger.catch
|
||||
def main():
|
||||
"""Main function with clean structure"""
|
||||
# Startup banner
|
||||
logger.info("=" * 50)
|
||||
logger.info("MCP Host Server v1.0")
|
||||
logger.info("Debug Mode: {}", "ON" if settings.debug else "OFF")
|
||||
logger.info("=" * 50)
|
||||
|
||||
logger.info("Initializing MCPHost...")
|
||||
|
||||
with MCPHostManager() as manager:
|
||||
server = create_server(manager)
|
||||
|
||||
try:
|
||||
logger.success("Server started at {}:{}", settings.host, settings.port)
|
||||
logger.info("Ready to accept requests.")
|
||||
logger.info("Press Ctrl+C to shutdown")
|
||||
server.serve_forever()
|
||||
except KeyboardInterrupt:
|
||||
logger.info("Received shutdown signal, shutting down gracefully...")
|
||||
except Exception as e:
|
||||
logger.error("Server error: {}", e)
|
||||
logger.exception("Full traceback:")
|
||||
finally:
|
||||
logger.info("Shutting down server...")
|
||||
server.shutdown()
|
||||
logger.success("Server shutdown complete")
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
main()
|
||||
@@ -1,411 +0,0 @@
|
||||
#!/usr/bin/env python3
|
||||
"""
|
||||
OpenAI-compatible MCPhost API wrapper with enhanced debugging
|
||||
Usage: python mcphost_openai_api.py
|
||||
"""
|
||||
|
||||
import json
|
||||
import time
|
||||
import asyncio
|
||||
import os
|
||||
from typing import Optional
|
||||
from contextlib import asynccontextmanager
|
||||
import threading
|
||||
|
||||
from fastapi import FastAPI, HTTPException
|
||||
from starlette.responses import StreamingResponse
|
||||
import uvicorn
|
||||
from loguru import logger
|
||||
import pexpect
|
||||
import pyte
|
||||
|
||||
from helpers.settings import settings
|
||||
from helpers.models import ChatCompletionRequest
|
||||
from helpers.local_logger import LoggerWriter
|
||||
|
||||
# Global process variable
|
||||
mcp_process: Optional[pexpect.spawn] = None
|
||||
process_monitor_thread = None
|
||||
stop_monitoring = threading.Event()
|
||||
|
||||
|
||||
def monitor_process():
|
||||
"""Background thread to monitor MCPhost process status"""
|
||||
while not stop_monitoring.is_set():
|
||||
if mcp_process:
|
||||
is_alive = mcp_process.isalive()
|
||||
exit_status = mcp_process.exitstatus if not is_alive else None
|
||||
logger.info(f"MCPhost process status - Alive: {is_alive}, Exit status: {exit_status}")
|
||||
|
||||
if not is_alive and exit_status is not None:
|
||||
logger.error(f"MCPhost process died with exit code: {exit_status}")
|
||||
else:
|
||||
logger.warning("MCPhost process is None")
|
||||
|
||||
# Wait 5 seconds before next check
|
||||
for _ in range(5):
|
||||
if stop_monitoring.is_set():
|
||||
break
|
||||
time.sleep(1)
|
||||
|
||||
|
||||
@asynccontextmanager
|
||||
async def lifespan(app: FastAPI):
|
||||
# Startup: Start MCPhost subprocess
|
||||
global mcp_process, process_monitor_thread
|
||||
|
||||
try:
|
||||
# Build command
|
||||
cmd = [settings.mcphost_path]
|
||||
|
||||
if settings.debug:
|
||||
cmd.append("--debug")
|
||||
|
||||
cmd.extend([
|
||||
"--config", settings.mcphost_config,
|
||||
"--model", settings.mcphost_model,
|
||||
"--openai-url", settings.openai_url,
|
||||
"--openai-api-key", settings.openai_api_key
|
||||
])
|
||||
|
||||
# Set environment
|
||||
env = os.environ.copy()
|
||||
env["OLLAMA_NUM_CTX"] = str(settings.ollama_num_ctx)
|
||||
|
||||
logger.info(f"Starting MCPhost with command: {' '.join(cmd)}")
|
||||
logger.info(f"Environment: OLLAMA_NUM_CTX={settings.ollama_num_ctx}")
|
||||
|
||||
# Use pexpect to spawn MCPhost in interactive mode
|
||||
try:
|
||||
mcp_process = pexpect.spawn(" ".join(cmd), timeout=30, encoding='utf-8', env=env)
|
||||
logger.info(f"MCPhost process spawned with PID: {mcp_process.pid}")
|
||||
except Exception as spawn_error:
|
||||
logger.error(f"Failed to spawn MCPhost process: {spawn_error}")
|
||||
raise
|
||||
|
||||
# Enable detailed logging
|
||||
mcp_process.logfile_read = LoggerWriter(logger.debug)
|
||||
|
||||
# Wait for model to load
|
||||
logger.info("Waiting for model to load...")
|
||||
index = mcp_process.expect([
|
||||
"Model loaded",
|
||||
pexpect.TIMEOUT,
|
||||
pexpect.EOF
|
||||
], timeout=30)
|
||||
|
||||
if index == 0:
|
||||
logger.info("MCPhost model loaded successfully")
|
||||
elif index == 1:
|
||||
logger.error("Timeout waiting for model to load")
|
||||
logger.error(f"Output so far: {mcp_process.before}")
|
||||
raise RuntimeError("Timeout waiting for model to load")
|
||||
elif index == 2:
|
||||
logger.error("MCPhost process ended unexpectedly")
|
||||
logger.error(f"Output: {mcp_process.before}")
|
||||
raise RuntimeError("MCPhost process ended unexpectedly")
|
||||
|
||||
# Wait for prompt interface
|
||||
logger.info("Waiting for prompt...")
|
||||
|
||||
# Read output until we see the prompt interface
|
||||
seen_prompt = False
|
||||
start_time = time.time()
|
||||
|
||||
while time.time() - start_time < 10: # 10 second timeout
|
||||
try:
|
||||
output = mcp_process.read_nonblocking(size=1000, timeout=0.5)
|
||||
if "Enter your prompt" in output or "┃" in output:
|
||||
seen_prompt = True
|
||||
break
|
||||
except pexpect.TIMEOUT:
|
||||
continue
|
||||
except pexpect.EOF:
|
||||
logger.error("MCPhost process ended unexpectedly")
|
||||
raise RuntimeError("MCPhost process ended unexpectedly")
|
||||
|
||||
if seen_prompt:
|
||||
logger.info("MCPhost process started and ready")
|
||||
else:
|
||||
logger.error("Timeout waiting for prompt")
|
||||
raise RuntimeError("Timeout waiting for prompt")
|
||||
|
||||
# Start process monitoring thread
|
||||
stop_monitoring.clear()
|
||||
process_monitor_thread = threading.Thread(target=monitor_process, daemon=True)
|
||||
process_monitor_thread.start()
|
||||
logger.info("Started process monitoring thread")
|
||||
|
||||
# Final check
|
||||
if mcp_process.isalive():
|
||||
logger.info("MCPhost process is alive and ready to accept requests")
|
||||
else:
|
||||
logger.error("MCPhost process is not alive after startup")
|
||||
raise RuntimeError("MCPhost process died during startup")
|
||||
|
||||
yield
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"Failed to start MCPhost: {e}")
|
||||
if mcp_process:
|
||||
logger.error(f"MCPhost final output: {mcp_process.before}")
|
||||
logger.error(f"Process alive: {mcp_process.isalive()}")
|
||||
if not mcp_process.isalive():
|
||||
logger.error(f"Exit status: {mcp_process.exitstatus}")
|
||||
raise
|
||||
finally:
|
||||
# Shutdown: Clean up subprocess
|
||||
logger.info("Shutting down MCPhost...")
|
||||
stop_monitoring.set()
|
||||
|
||||
if process_monitor_thread:
|
||||
process_monitor_thread.join(timeout=2)
|
||||
|
||||
if mcp_process:
|
||||
if mcp_process.isalive():
|
||||
logger.info("Terminating MCPhost process...")
|
||||
mcp_process.terminate()
|
||||
mcp_process.wait()
|
||||
logger.info("MCPhost process stopped")
|
||||
|
||||
|
||||
# Create FastAPI app
|
||||
app = FastAPI(title="OpenAI-compatible MCPhost API", lifespan=lifespan)
|
||||
|
||||
|
||||
async def _resp_async_generator(response_text: str, model: str):
|
||||
"""Generator for streaming responses in OpenAI format"""
|
||||
tokens = response_text.split(" ")
|
||||
|
||||
for i, token in enumerate(tokens):
|
||||
chunk = {
|
||||
"id": f"chatcmpl-{int(time.time())}-{i}",
|
||||
"object": "chat.completion.chunk",
|
||||
"created": int(time.time()),
|
||||
"model": model,
|
||||
"choices": [{
|
||||
"index": 0,
|
||||
"delta": {"content": token + " "},
|
||||
"finish_reason": None
|
||||
}]
|
||||
}
|
||||
yield f"data: {json.dumps(chunk)}\n\n"
|
||||
await asyncio.sleep(0.01)
|
||||
|
||||
final_chunk = {
|
||||
"id": f"chatcmpl-{int(time.time())}-final",
|
||||
"object": "chat.completion.chunk",
|
||||
"created": int(time.time()),
|
||||
"model": model,
|
||||
"choices": [{
|
||||
"index": 0,
|
||||
"delta": {},
|
||||
"finish_reason": "stop"
|
||||
}]
|
||||
}
|
||||
yield f"data: {json.dumps(final_chunk)}\n\n"
|
||||
yield "data: [DONE]\n\n"
|
||||
|
||||
|
||||
@app.post("/v1/chat/completions")
|
||||
@app.post("/chat/completions")
|
||||
async def chat_completions(request: ChatCompletionRequest):
|
||||
"""OpenAI-compatible chat completions endpoint"""
|
||||
if not mcp_process:
|
||||
logger.error("MCPhost process object is None")
|
||||
raise HTTPException(status_code=500, detail="MCPhost process not initialized")
|
||||
|
||||
if not mcp_process.isalive():
|
||||
logger.error(f"MCPhost process not running. Exit status: {mcp_process.exitstatus}")
|
||||
raise HTTPException(status_code=500, detail="MCPhost process not running")
|
||||
|
||||
try:
|
||||
# Extract the last user message
|
||||
user_message = ""
|
||||
for message in reversed(request.messages):
|
||||
if message.role == "user":
|
||||
user_message = message.content
|
||||
break
|
||||
|
||||
if not user_message:
|
||||
user_message = request.messages[-1].content if request.messages else ""
|
||||
|
||||
logger.debug(f"Sending to MCPhost: {user_message}")
|
||||
|
||||
# Clear any pending output (non-blocking read)
|
||||
try:
|
||||
mcp_process.read_nonblocking(size=1000, timeout=0.1)
|
||||
except:
|
||||
pass
|
||||
|
||||
# Send message to MCPhost
|
||||
mcp_process.sendline(user_message)
|
||||
|
||||
# Create virtual terminal to parse output
|
||||
screen = pyte.Screen(80, 24)
|
||||
stream = pyte.ByteStream(screen)
|
||||
|
||||
# Wait for the response
|
||||
response_text = ""
|
||||
last_screen_content = ""
|
||||
|
||||
try:
|
||||
start_time = time.time()
|
||||
no_change_count = 0
|
||||
|
||||
while time.time() - start_time < 30: # 30 second timeout
|
||||
try:
|
||||
# Read available data
|
||||
data = mcp_process.read_nonblocking(size=1024, timeout=0.1)
|
||||
|
||||
# Feed data to virtual terminal
|
||||
stream.feed(data.encode('utf-8'))
|
||||
|
||||
# Get current screen content
|
||||
current_screen = '\n'.join(screen.display).strip()
|
||||
|
||||
# Check if screen content has changed
|
||||
if current_screen == last_screen_content:
|
||||
no_change_count += 1
|
||||
if no_change_count > 10: # No change for 1 second
|
||||
break
|
||||
else:
|
||||
no_change_count = 0
|
||||
last_screen_content = current_screen
|
||||
|
||||
# Extract response text from screen
|
||||
lines = current_screen.split('\n')
|
||||
response_lines = []
|
||||
|
||||
for line in lines:
|
||||
# Skip empty lines and prompt-related lines
|
||||
if line.strip() and not "Enter your prompt" in line and not line.strip().startswith(
|
||||
"alt+enter"):
|
||||
# Remove the prompt box character if present
|
||||
if line.startswith("┃"):
|
||||
clean_line = line[1:].strip()
|
||||
if clean_line and not "Enter your prompt" in clean_line:
|
||||
response_lines.append(clean_line)
|
||||
else:
|
||||
response_lines.append(line.strip())
|
||||
|
||||
# Update response text
|
||||
response_text = ' '.join(response_lines)
|
||||
|
||||
except pexpect.TIMEOUT:
|
||||
# If we haven't received anything yet, continue waiting
|
||||
if not response_text:
|
||||
continue
|
||||
else:
|
||||
no_change_count += 1
|
||||
if no_change_count > 10:
|
||||
break
|
||||
except pexpect.EOF:
|
||||
logger.error("MCPhost process ended unexpectedly")
|
||||
break
|
||||
|
||||
except Exception as e:
|
||||
logger.warning(f"Error reading response: {e}")
|
||||
|
||||
response_text = response_text.strip()
|
||||
|
||||
# Clean up the response text - remove any duplicates or artifacts
|
||||
if response_text:
|
||||
# Sometimes the terminal might show duplicated content
|
||||
words = response_text.split()
|
||||
clean_words = []
|
||||
for i, word in enumerate(words):
|
||||
if i == 0 or word != words[i - 1]:
|
||||
clean_words.append(word)
|
||||
response_text = ' '.join(clean_words)
|
||||
|
||||
if not response_text:
|
||||
response_text = "No response received from MCPhost"
|
||||
|
||||
logger.debug(f"MCPhost response: {response_text}")
|
||||
|
||||
# Handle streaming response
|
||||
if request.stream:
|
||||
return StreamingResponse(
|
||||
_resp_async_generator(response_text, request.model),
|
||||
media_type="text/event-stream"
|
||||
)
|
||||
|
||||
# Non-streaming response
|
||||
return {
|
||||
"id": f"chatcmpl-{int(time.time())}",
|
||||
"object": "chat.completion",
|
||||
"created": int(time.time()),
|
||||
"model": request.model,
|
||||
"choices": [{
|
||||
"index": 0,
|
||||
"message": {
|
||||
"role": "assistant",
|
||||
"content": response_text
|
||||
},
|
||||
"finish_reason": "stop"
|
||||
}],
|
||||
"usage": {
|
||||
"prompt_tokens": len(user_message.split()),
|
||||
"completion_tokens": len(response_text.split()),
|
||||
"total_tokens": len(user_message.split()) + len(response_text.split())
|
||||
}
|
||||
}
|
||||
|
||||
except Exception as e:
|
||||
logger.exception(f"Error in chat completion")
|
||||
raise HTTPException(status_code=500, detail=str(e))
|
||||
|
||||
|
||||
@app.get("/v1/models")
|
||||
@app.get("/models")
|
||||
async def list_models():
|
||||
"""List available models (OpenAI-compatible endpoint)"""
|
||||
return {
|
||||
"object": "list",
|
||||
"data": [{
|
||||
"id": settings.mcphost_model,
|
||||
"object": "model",
|
||||
"created": int(time.time()),
|
||||
"owned_by": "mcphost",
|
||||
"permission": [],
|
||||
"root": settings.mcphost_model,
|
||||
"parent": None
|
||||
}]
|
||||
}
|
||||
|
||||
|
||||
@app.get("/health")
|
||||
async def health_check():
|
||||
"""Health check endpoint"""
|
||||
if not mcp_process:
|
||||
logger.warning("Health check: MCPhost process is None")
|
||||
return {"status": "unhealthy", "detail": "Process not initialized"}
|
||||
|
||||
is_healthy = mcp_process.isalive()
|
||||
status = "healthy" if is_healthy else "unhealthy"
|
||||
detail = {"pid": mcp_process.pid if mcp_process else None}
|
||||
|
||||
if not is_healthy and mcp_process:
|
||||
detail["exit_status"] = mcp_process.exitstatus
|
||||
|
||||
logger.info(f"Health check: {status}, details: {detail}")
|
||||
return {"status": status, "detail": detail}
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
logger.add(
|
||||
"mcphost_openai_api.log",
|
||||
rotation="10 MB",
|
||||
retention="10 days",
|
||||
level="DEBUG"
|
||||
)
|
||||
|
||||
logger.info("Starting OpenAI-compatible MCPhost API server...")
|
||||
logger.info(f"Configuration:")
|
||||
logger.info(f" MCPhost Path: {settings.mcphost_path}")
|
||||
logger.info(f" Model: {settings.mcphost_model}")
|
||||
logger.info(f" OpenAI URL: {settings.openai_url}")
|
||||
logger.info(f" Debug: {settings.debug}")
|
||||
|
||||
uvicorn.run(app, host=settings.host, port=settings.port, log_config=None)
|
||||
23
test.sh
23
test.sh
@@ -1,10 +1,17 @@
|
||||
#!/bin/bash
|
||||
|
||||
curl -X POST http://0.0.0.0:8000/v1/chat/completions -H "Content-Type: application/json" -H "Authorization: Bearer fake-api-key" -d '{
|
||||
"model": "mcphost-model",
|
||||
"messages": [
|
||||
{"role": "system", "content": "You are a helpful assistant."},
|
||||
{"role": "user", "content": "Tell me a joke."}
|
||||
],
|
||||
"temperature": 0.7
|
||||
}'
|
||||
|
||||
#clear
|
||||
curl -X POST \
|
||||
-H "Content-Type: plain/text" \
|
||||
-d "When is your knowledge cut off? /no_think" \
|
||||
http://localhost:8000
|
||||
|
||||
#curl -X POST http://0.0.0.0:8000/v1/chat/completions -H "Content-Type: application/json" -H "Authorization: Bearer fake-api-key" -d '{
|
||||
# "model": "mcphost-model",
|
||||
# "messages": [
|
||||
# {"role": "system", "content": "You are a helpful assistant."},
|
||||
# {"role": "user", "content": "Tell me a joke."}
|
||||
# ],
|
||||
# "temperature": 0.7
|
||||
# }'
|
||||
|
||||
Reference in New Issue
Block a user