Curl (#61)

.github/workflows/ci.yml

@@ -62,7 +62,8 @@ jobs:
         run: |
           sudo apt install -y ninja-build \
             nlohmann-json3-dev \
-            libwebkit2gtk-4.1-dev
+            libwebkit2gtk-4.1-dev \
+            libcurl4-openssl-dev
           # npx playwright install-deps
 
       - name: Configure (macOS)

CMakeLists.txt

@@ -16,14 +16,15 @@ endif()
 
 find_package(PkgConfig REQUIRED)
 pkg_check_modules(NlohmannJson REQUIRED IMPORTED_TARGET "nlohmann_json")
+find_package(CURL REQUIRED)
 
 if(LINUX)
     pkg_check_modules(webkit2gtk REQUIRED IMPORTED_TARGET "webkit2gtk-4.1")
-    set(LIBS PkgConfig::NlohmannJson PkgConfig::webkit2gtk)
+    set(LIBS PkgConfig::NlohmannJson PkgConfig::webkit2gtk CURL::libcurl)
 elseif(APPLE)
     find_library(COCOA_LIB Cocoa REQUIRED)
     find_library(WEBKIT_LIB WebKit REQUIRED)
-    set(LIBS PkgConfig::NlohmannJson ${COCOA_LIB} ${WEBKIT_LIB})
+    set(LIBS PkgConfig::NlohmannJson ${COCOA_LIB} ${WEBKIT_LIB} CURL::libcurl)
 endif()
 
 include_directories(pugixml/src)

docs/API.md

@@ -0,0 +1,54 @@
+# API
+
+## `curl` (async)
+
+With `curl`, you can interact with network resources directly.
+
+```ts
+async function curl(url: string, args: CurlArgs) => CurlResponse
+
+type CurlArgs = {
+    method?: "GET" | "POST" | "DELETE" | "HEAD" | "OPTIONS" | "PUT" | "PATCH",
+    headers?: object,
+    data?: string,    // ignored if `json` exists
+    json?: JSON,
+    binary?: bool,
+    timeout?: uint64  // milliseconds
+}
+
+type CurlResponse = {
+    status: number,
+    data: string,
+}
+```
+
+- If `args.binary` is `true`, then `response.data` will be a base64-encoded representation of the original data.
+- The precision of `args.timeout` is 50ms.
+
+**Example** POST w/ JSON:
+
+```js
+# Basic usage
+curl("https://httpbin.org/post", {
+    json: {arr: [1,2,3]}
+}).then(r => { 
+    console.log(r.status, r.status==200)
+    return JSON.parse(r.data)
+}).then(j => console.log(j.data))
+
+# OpenAI
+curl("https://api.openai.com/v1/chat/completions", {
+    headers: { "Authorization": "Bearer $OPENAI_API_KEY" },
+    json: {
+        "model": "gpt-4o-mini",
+        "messages": [{
+            "role": "system",
+            "content": "You are a poetic assistant, skilled in explaining complex programming concepts with creative flair."
+        }, {
+            "role": "user",
+            "content": "Compose a poem that explains the concept of recursion in programming."
+        }]
+    }
+}).then(r => JSON.parse(r.data))
+  .then(j => console.log(j))
+```

include/curl.hpp

@@ -0,0 +1,27 @@
+#pragma once
+
+#include <curl/curl.h>
+#include <functional>
+#include <shared_mutex>
+#include <thread>
+
+class CurlMultiManager {
+  public:
+    using Callback = std::function<void(CURLcode, CURL *, const std::string &)>;
+    using HandleData = std::pair<CURL *, Callback>;
+
+    static CurlMultiManager &shared();
+    CurlMultiManager();
+    ~CurlMultiManager();
+    void add(CURL *easy, CurlMultiManager::Callback cb);
+
+  private:
+    CURLM *multi;
+    std::thread worker_thread;
+    int controlfd[2];
+    std::shared_mutex m;
+    std::unordered_map<CURL *, std::string> buf;
+    std::unordered_map<CURL *, Callback> cb;
+
+    void run();
+};

include/utility.hpp

@@ -50,3 +50,4 @@ template <typename T> function_traits<T> make_function_traits(const T &) {
 }
 
 std::string escape_html(const std::string &content);
+std::string base64(const std::string &s);

include/webview_candidate_window.hpp

@@ -9,6 +9,9 @@
 #include <sstream>
 
 namespace candidate_window {
+
+enum CustomAPI : uint64_t { kCurl = 1 };
+
 class WebviewCandidateWindow : public CandidateWindow {
   public:
     WebviewCandidateWindow();
@@ -33,6 +36,8 @@ class WebviewCandidateWindow : public CandidateWindow {
     void set_accent_color();
     void copy_html();
 
+    void set_api(uint64_t apis);
+
   private:
     std::shared_ptr<webview::webview> w_;
     double cursor_x_ = 0;
@@ -58,6 +63,10 @@ class WebviewCandidateWindow : public CandidateWindow {
     void *platform_data = nullptr;
     void platform_init();
 
+  private:
+    /* API */
+    void api_curl(std::string id, std::string req);
+
   private:
     /* Invoke a JavaScript function. */
     template <typename Ret = void, bool debug = false, typename... Args>

preview/CMakeLists.txt

@@ -1,9 +1,11 @@
-add_executable(preview preview.cpp)
-target_link_libraries(preview WebviewCandidateWindow)
-target_include_directories(preview PRIVATE "${PROJECT_SOURCE_DIR}/include")
-
 if(APPLE)
+    add_executable(preview MACOSX_BUNDLE preview.cpp)
     target_compile_options(preview PRIVATE "-Wno-auto-var-id" "-ObjC++")
+else()
+    add_executable(preview preview.cpp)
 endif()
 
+target_link_libraries(preview WebviewCandidateWindow)
+target_include_directories(preview PRIVATE "${PROJECT_SOURCE_DIR}/include")
+
 add_dependencies(preview GenerateHTML)

src/CMakeLists.txt

@@ -3,6 +3,7 @@ add_library(WebviewCandidateWindow
     "${PROJECT_SOURCE_DIR}/pugixml/src/pugixml.cpp"
     webview_candidate_window.cpp
     platform.cpp
+    curl.cpp
 )
 target_include_directories(WebviewCandidateWindow PUBLIC "${PROJECT_SOURCE_DIR}/include" "${PROJECT_SOURCE_DIR}/webview")
 target_link_libraries(WebviewCandidateWindow ${LIBS})

src/curl.cpp

@@ -0,0 +1,134 @@
+#include "curl.hpp"
+#include <atomic>
+#include <cassert>
+#include <errno.h>
+#include <mutex>
+#include <stdexcept>
+#include <thread>
+#include <unistd.h>
+
+static size_t _on_data_cb(char *data, size_t size, size_t nmemb,
+                          std::string *outbuf);
+static bool write_char_strong(int fd, char c);
+static bool read_char_strong(int fd, char *c);
+std::mutex m;
+std::atomic<bool> running;
+
+CurlMultiManager &CurlMultiManager::shared() {
+    static CurlMultiManager instance;
+    return instance;
+}
+
+CurlMultiManager::CurlMultiManager() {
+    bool expected = false;
+    if (!running.compare_exchange_strong(expected, true)) {
+        throw std::runtime_error("should only run one curl manager");
+    }
+    if (pipe(controlfd) < 0) {
+        throw std::runtime_error("failed to create curl control pipe");
+    }
+    curl_global_init(CURL_GLOBAL_ALL);
+    multi = curl_multi_init();
+    worker_thread = std::thread(&CurlMultiManager::run, this);
+}
+
+CurlMultiManager::~CurlMultiManager() {
+    write_char_strong(controlfd[1], 'q');
+    if (worker_thread.joinable()) {
+        worker_thread.join();
+    }
+    curl_multi_cleanup(multi);
+    curl_global_cleanup();
+    close(controlfd[0]);
+    close(controlfd[1]);
+    running.store(false);
+}
+
+void CurlMultiManager::add(CURL *easy, CurlMultiManager::Callback callback) {
+    {
+        std::unique_lock g(m);
+        buf[easy] = "";
+        cb[easy] = callback;
+    }
+    curl_easy_setopt(easy, CURLOPT_WRITEFUNCTION, _on_data_cb);
+    curl_easy_setopt(easy, CURLOPT_WRITEDATA, &buf[easy]);
+    curl_multi_add_handle(multi, easy);
+    std::atomic_thread_fence(std::memory_order_release);
+    write_char_strong(controlfd[1], 'a');
+}
+
+void CurlMultiManager::run() {
+    while (true) {
+        int still_running = 0;
+        int ret;
+        ret = curl_multi_perform(multi, &still_running);
+        int numfds;
+        struct curl_waitfd wfd;
+        wfd.fd = controlfd[0];
+        wfd.events = CURL_WAIT_POLLIN;
+        wfd.revents = 0;
+        // NOTE: poll does not return when any of the easy handle's
+        // timeout expires.  By setting poll's timeout to 50ms, we
+        // effectively set the timeout precision to 50ms.
+        curl_multi_poll(multi, &wfd, 1, 50, &numfds);
+        std::atomic_thread_fence(std::memory_order_acquire);
+        if (wfd.revents) {
+            char cmd;
+            read_char_strong(controlfd[0], &cmd);
+            switch (cmd) {
+            case 'q': // quit
+                return;
+            case 'a': // added a new handle
+                break;
+            default:
+                assert(false && "unreachable");
+            }
+        }
+        CURLMsg *msg;
+        int msgs_left;
+        while ((msg = curl_multi_info_read(multi, &msgs_left))) {
+            if (msg->msg == CURLMSG_DONE) {
+                CURL *easy = msg->easy_handle;
+                CURLcode res = msg->data.result;
+                {
+                    // do this because the cb might be slow
+                    std::shared_lock g(m);
+                    try {
+                        cb[easy](res, easy, buf[easy]);
+                    } catch (...) {
+                        assert(false && "curl callback must not throw!");
+                    }
+                }
+                curl_multi_remove_handle(multi, easy);
+                curl_easy_cleanup(easy);
+                {
+                    std::unique_lock g(m);
+                    cb.erase(easy);
+                    buf.erase(easy);
+                }
+            }
+        }
+    }
+}
+
+size_t _on_data_cb(char *data, size_t size, size_t nmemb, std::string *outbuf) {
+    size_t realsize = size * nmemb;
+    outbuf->append(data, realsize);
+    return realsize;
+}
+
+static bool write_char_strong(int fd, char c) {
+    ssize_t ret;
+    do {
+        ret = write(fd, &c, 1);
+    } while (ret == -1 && errno == EINTR);
+    return ret == 1;
+}
+
+static bool read_char_strong(int fd, char *c) {
+    ssize_t ret;
+    do {
+        ret = read(fd, c, 1);
+    } while (ret == -1 && errno == EINTR);
+    return ret == 1;
+}

src/utility.cpp

@@ -9,3 +9,25 @@ std::string escape_html(const std::string &content) {
     doc.print(ss, "", pugi::format_raw);
     return ss.str();
 }
+
+std::string base64(const std::string &s) {
+    static const char *chars =
+        "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
+    std::string ret;
+    ret.reserve((s.size() + 2) / 3 * 4);
+    unsigned int w = 0;
+    int b = -6;
+    for (unsigned char c : s) {
+        w = (w << 8) + c;
+        b += 8;
+        while (b >= 0) {
+            ret += chars[(w >> b) & 0x3F];
+            b -= 6;
+        }
+    }
+    if (b > -6)
+        ret += chars[((w << 8) >> (b + 8)) & 0x3F];
+    while (ret.size() % 4)
+        ret += '=';
+    return ret;
+}