Add function to grab UA details from useragentstring.com

4 files changed, 81 insertions, 1 deletions

diff --git a/src/Jamfile.jam b/src/Jamfile.jam
index e1d5ec5..9459dd8 100644
--- a/src/Jamfile.jam
+++ b/src/Jamfile.jam
@@ -1,4 +1,4 @@
-lib webstat : ingestor.cpp logTypes.cpp sql.cpp :
+lib webstat : [ glob *.cpp : *_main.cpp ] :
 	<include>.
 	<library>..//adhocutil
 	<library>..//dbppcore
diff --git a/src/uaLookup.cpp b/src/uaLookup.cpp
new file mode 100644
index 0000000..d62c313
--- /dev/null
+++ b/src/uaLookup.cpp
@@ -0,0 +1,50 @@
+#include "uaLookup.hpp"
+#include "util.hpp"
+#include <memory>
+#include <string_view>
+
+namespace WebStat {
+	namespace {
+		size_t
+		stringAppend(const char * ptr, size_t size, size_t nmemb, std::string * result)
+		{
+			result->append(ptr, nmemb * size);
+			return nmemb * size;
+		}
+
+		void
+		addForm(curl_mime * mime, const char * name, const std::string_view data)
+		{
+			auto part = curl_mime_addpart(mime);
+			curl_mime_data(part, data.data(), data.length());
+			curl_mime_name(part, name);
+		};
+	}
+
+	std::string
+	getUserAgentDetail(const std::string_view uas, const char * baseUrl)
+	{
+		std::string result;
+		std::array<char, CURL_ERROR_SIZE> err {};
+		std::unique_ptr<CURL, DeleteWith<&curl_easy_cleanup>> hnd {curl_easy_init()};
+		curl_easy_setopt(hnd.get(), CURLOPT_URL, baseUrl);
+		curl_easy_setopt(hnd.get(), CURLOPT_NOPROGRESS, 1L);
+		curl_easy_setopt(hnd.get(), CURLOPT_USERAGENT, "WebStat/0; +https://git.randomdan.homeip.net/repo/webstat/");
+		curl_easy_setopt(hnd.get(), CURLOPT_MAXREDIRS, 50L);
+		curl_easy_setopt(hnd.get(), CURLOPT_TCP_KEEPALIVE, 1L);
+		curl_easy_setopt(hnd.get(), CURLOPT_FAILONERROR, 1L);
+		curl_easy_setopt(hnd.get(), CURLOPT_ERRORBUFFER, err.data());
+		curl_easy_setopt(hnd.get(), CURLOPT_WRITEDATA, &result);
+		curl_easy_setopt(hnd.get(), CURLOPT_WRITEFUNCTION, &stringAppend);
+		std::unique_ptr<curl_mime, DeleteWith<&curl_mime_free>> mime {curl_mime_init(hnd.get())};
+		addForm(mime.get(), "uas", uas);
+		addForm(mime.get(), "getJSON", "all");
+		curl_easy_setopt(hnd.get(), CURLOPT_MIMEPOST, mime.get());
+
+		if (CURLcode ret = curl_easy_perform(hnd.get()); ret != CURLE_OK) {
+			throw CurlError {ret, err.data()};
+		}
+
+		return result;
+	}
+}
diff --git a/src/uaLookup.hpp b/src/uaLookup.hpp
new file mode 100644
index 0000000..5492f05
--- /dev/null
+++ b/src/uaLookup.hpp
@@ -0,0 +1,16 @@
+#pragma once
+
+#include <curl/curl.h>
+#include <stdexcept>
+#include <string>
+
+namespace WebStat {
+	class CurlError : public std::runtime_error {
+	public:
+		explicit CurlError(CURLcode code, const char * msg) : std::runtime_error {msg}, code(code) { }
+
+		CURLcode code;
+	};
+
+	std::string getUserAgentDetail(std::string_view uas, const char * baseUrl = "https://useragentstring.com");
+}
diff --git a/test/test-ingest.cpp b/test/test-ingest.cpp
index 5f29806..1a62b70 100644
--- a/test/test-ingest.cpp
+++ b/test/test-ingest.cpp
@@ -5,6 +5,7 @@
 #include "test-util.hpp"
 
 #include <ingestor.hpp>
+#include <uaLookup.hpp>
 
 namespace {
 	using namespace WebStat;
@@ -181,3 +182,16 @@ BOOST_DATA_TEST_CASE(StoreLogLine,
 	WebStat::Ingestor {WebStat::getTestUtsName("test-hostname"), std::make_shared<MockDBPool>("webstat")}.ingestLogLine(
 			DB::MockDatabase::openConnectionTo("webstat").get(), line);
 }
+
+BOOST_AUTO_TEST_CASE(FetchRealUserAgentDetail, *boost::unit_test::disabled())
+{
+	const auto uaDetail = WebStat::getUserAgentDetail(
+			R"(Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/139.0.0.0 Safari/537.36)");
+
+	BOOST_TEST_CONTEXT(uaDetail) {
+		BOOST_CHECK(uaDetail.starts_with("{"));
+		BOOST_CHECK(uaDetail.ends_with("}"));
+		BOOST_CHECK(uaDetail.contains(R"("agent_type":)"));
+		BOOST_CHECK(uaDetail.contains(R"("os_type":)"));
+	}
+}