tha80
983ca8992e
* This commit enables the router to forward form-data to model server. Fixes #22044 (enabling to use the /v1/audio/transcriptions in router mode) * * Applied the suggestion from Copilots first comment: using the non-throwing json::parse overload. * Addressed Copilots third comment by extending the files representation to also include filename and content-type * Addressed Copilots fourth comment by making the RNG thread_local * Changed variable body from std::string to std::ostringstream in build_multipart_body as suggested by ngxson in https://github.com/ggml-org/llama.cpp/pull/22118#discussion_r3127099053 * Added sanitize_field lambda in build_multipart_body for key, filename and content_type as suggested by ngxson in https://github.com/ggml-org/llama.cpp/pull/22118#discussion_r3127104647 * explicitly checking if value/item is string before calling value/item.get<std::string>() as requested by ngxson in https://github.com/ggml-org/llama.cpp/pull/22118#discussion_r3127111279 * Added double quote to the sanitize lambda and throw on json parse failure --------- Co-authored-by: Ralph Paßgang <ralph@trust-it.de>
90 lines
2.6 KiB
C++
90 lines
2.6 KiB
C++
#pragma once
|
|
|
|
#include <atomic>
|
|
#include <functional>
|
|
#include <map>
|
|
#include <string>
|
|
#include <thread>
|
|
#include <vector>
|
|
#include <cstdint>
|
|
|
|
struct common_params;
|
|
|
|
// generator-like API for HTTP response generation
|
|
// this object response with one of the 2 modes:
|
|
// 1) normal response: `data` contains the full response body
|
|
// 2) streaming response: each call to next(output) generates the next chunk
|
|
// when next(output) returns false, no more data after the current chunk
|
|
// note: some chunks can be empty, in which case no data is sent for that chunk
|
|
struct server_http_res {
|
|
std::string content_type = "application/json; charset=utf-8";
|
|
int status = 200;
|
|
std::string data;
|
|
std::map<std::string, std::string> headers;
|
|
|
|
// TODO: move this to a virtual function once we have proper polymorphism support
|
|
std::function<bool(std::string &)> next = nullptr;
|
|
bool is_stream() const {
|
|
return next != nullptr;
|
|
}
|
|
|
|
virtual ~server_http_res() = default;
|
|
};
|
|
|
|
// unique pointer, used by set_chunked_content_provider
|
|
// httplib requires the stream provider to be stored in heap
|
|
using server_http_res_ptr = std::unique_ptr<server_http_res>;
|
|
using raw_buffer = std::vector<uint8_t>;
|
|
|
|
struct uploaded_file {
|
|
raw_buffer data;
|
|
std::string filename;
|
|
std::string content_type;
|
|
};
|
|
|
|
struct server_http_req {
|
|
std::map<std::string, std::string> params; // path_params + query_params
|
|
std::map<std::string, std::string> headers; // used by MCP proxy
|
|
std::string path;
|
|
std::string query_string; // query parameters string (e.g. "action=save")
|
|
std::string body;
|
|
std::map<std::string, uploaded_file> files; // used for file uploads (form data)
|
|
const std::function<bool()> & should_stop;
|
|
|
|
std::string get_param(const std::string & key, const std::string & def = "") const {
|
|
auto it = params.find(key);
|
|
if (it != params.end()) {
|
|
return it->second;
|
|
}
|
|
return def;
|
|
}
|
|
};
|
|
|
|
struct server_http_context {
|
|
class Impl;
|
|
std::unique_ptr<Impl> pimpl;
|
|
|
|
std::thread thread; // server thread
|
|
std::atomic<bool> is_ready = false;
|
|
|
|
std::string path_prefix;
|
|
std::string hostname;
|
|
int port;
|
|
|
|
server_http_context();
|
|
~server_http_context();
|
|
|
|
bool init(const common_params & params);
|
|
bool start();
|
|
void stop() const;
|
|
|
|
// note: the handler should never throw exceptions
|
|
using handler_t = std::function<server_http_res_ptr(const server_http_req & req)>;
|
|
|
|
void get(const std::string & path, const handler_t & handler) const;
|
|
void post(const std::string & path, const handler_t & handler) const;
|
|
|
|
// for debugging
|
|
std::string listening_address;
|
|
};
|