#include "build-info.h"
#include "llama.h"
#include <cstdio>
#include <cstdlib>
#include <string>
#include <vector>
// embedded data generated by cmake
extern const char * LICENSES[];
// visible
int llama_server(int argc, char ** argv);
int llama_cli(int argc, char ** argv);
// hidden
int llama_completion(int argc, char ** argv);
int llama_bench(int argc, char ** argv);
int llama_batched_bench(int argc, char ** argv);
int llama_fit_params(int argc, char ** argv);
int llama_quantize(int argc, char ** argv);
int llama_perplexity(int argc, char ** argv);
int llama_download(int argc, char ** argv);
// Self-update is only supported for binaries built with llama-install.sh
static int llama_update(int argc, char ** argv) {
(void) argc;
(void) argv;
#ifdef LLAMA_INSTALL_BUILD
#if defined(_WIN32)
return system("powershell -NoProfile -ExecutionPolicy Bypass -Command \"irm https://llama.app/install.ps1 | iex\"");
#else
return system("curl -fsSL https://llama.app/install.sh | sh");
#endif
#else
printf("Updates are available only when installed from https://llama.app\n");
return 1;
#endif
}
static const char * progname;
static int help(int argc, char ** argv);
static int version(int argc, char ** argv);
static int licenses(int argc, char ** argv);
struct command {
const char * name;
const char * desc;
std::vector<std::string> aliases;
bool hidden;
int (*func)(int, char **);
bool flags = false; // allow --name
};
#ifdef LLAMA_INSTALL_BUILD
#define UPDATE_HIDDEN false
#else
#define UPDATE_HIDDEN true
#endif
static const command cmds[] = {
{"serve", "HTTP API server", {"server"}, false, llama_server },
{"cli", "Command-line interactive interface", {"client"}, false, llama_cli },
{"update", "Update llama to the latest release", {}, UPDATE_HIDDEN, llama_update },
{"download", "Download a model", {"get"}, false, llama_download },
{"completion", "Text completion", {"complete"}, true, llama_completion },
{"bench", "Benchmark prompt processing and text generation", {}, true, llama_bench },
{"batched-bench", "Benchmark batched decoding performance", {}, true, llama_batched_bench},
{"fit-params", "Compute parameters to fit a model in device memory", {}, true, llama_fit_params },
{"quantize", "Quantize a model", {}, true, llama_quantize },
{"perplexity", "Compute model perplexity and KL divergence", {}, true, llama_perplexity },
{"version", "Show version", {}, false, version, true },
{"licenses", "Show third-party licenses", {"credits"}, false, licenses, true },
{"help", "Show available commands", {}, false, help, true },
};
#undef UPDATE_HIDDEN
static int version(int /*argc*/, char ** /*argv*/) {
llama_print_build_info(llama_version());
return 0;
}
static int licenses(int /*argc*/, char ** /*argv*/) {
for (int i = 0; LICENSES[i]; ++i) {
printf("%s\n", LICENSES[i]);
}
return 0;
}
static int help(int argc, char ** argv) {
const bool show_all = argc >= 2 && std::string(argv[1]) == "all";
printf("Usage: %s <command> [options]\n\nAvailable commands:\n", progname);
for (const auto & cmd : cmds) {
if (show_all || !cmd.hidden) {
printf(" %-15s %s\n", cmd.name, cmd.desc);
}
}
printf("\n");
if (!show_all) {
printf("Run '%s help all' to show additional commands.\n", progname);
}
printf("Run '%s <command> --help' for command-specific usage.\n", progname);
return 0;
}
static bool matches(std::string arg, const command & cmd) {
if (cmd.flags && arg.size() > 2 && arg[0] == '-' && arg[1] == '-') {
arg.erase(0, 2);
}
if (arg == cmd.name) {
return true;
}
for (const auto & alias : cmd.aliases) {
if (arg == alias) {
return true;
}
}
return false;
}
int main(int argc, char ** argv) {
progname = argv[0];
const std::string arg = argc >= 2 ? argv[1] : "help";
for (const auto & cmd : cmds) {
if (matches(arg, cmd)) {
// keep cmd.name so the router's child processes re-invoke correctly
#ifdef _WIN32
_putenv_s("LLAMA_APP_CMD", cmd.name);
#else
setenv("LLAMA_APP_CMD", cmd.name, 1);
#endif
return cmd.func(argc - 1, argv + 1);
}
}
fprintf(stderr, "error: unknown command '%s'\n", arg.c_str());
return 1;
}