Use lua instead of curl

plugin/gemini/api.lua

@@ -13,7 +13,15 @@ local function make_request(prompt)
 		return nil
 	end
 
+	local socket = require("socket")
+	local ssl = require("ssl")
+	local https = require("socket.https")
+
 	local model = "gemini-2.0-flash" -- SPECIFY Gemini 2.0 Flash MODEL
+
+	local url =
+		string.format("https://generative-ai.googleapis.com/v1beta/models/%s:generateContent?key=%s", model, api_key)
+
 	-- Construct the JSON payload
 	local payload = vim.json.encode({
 		contents = {
@@ -27,25 +35,22 @@ local function make_request(prompt)
 		},
 	})
 
-	local command = string.format(
-		"curl -s -X POST "
-			.. "'https://generative-ai.googleapis.com/v1beta/models/%s:generateContent?key=%s' "
-			.. "-H 'Content-Type: application/json' "
-			.. "-d '%s'",
-		model,
-		api_key,
-		payload
-	)
+	local body, code, headers, status = https.request({
+		url = url,
+		method = "POST",
+		headers = {
+			["Content-Type"] = "application/json",
+		},
+		source = socket.source.string(payload),
+		protocol = "tlsv1_2", -- Require TLS 1.2
+	})
 
-	local result = vim.fn.system(command)
-
-	-- Check for errors during the curl execution
-	if vim.v.shell_error ~= 0 then
-		vim.notify("Error executing curl. Check your command and ensure curl is installed.", vim.log.levels.ERROR)
+	if code ~= 200 then
+		vim.notify(string.format("HTTP request failed with code %d: %s", code, status), vim.log.levels.ERROR)
 		return nil
 	end
 
-	local decoded_result = vim.json.decode(result)
+	local decoded_result = vim.json.decode(body)
 
 	return decoded_result
 end