From fa4caae02403f64aca99da275b6dd13a0164619f Mon Sep 17 00:00:00 2001
From: Paul Richmond <p.richmond@sheffield.ac.uk>
Date: Thu, 21 Nov 2024 15:56:11 +0000
Subject: [PATCH] Improvements to Telemetry (#1252)

Telemetry: Switch to ID per user rather than per build directory

This improves accuracy of unique user counting, especially for python usrs.

---------

Co-authored-by: Peter Heywood <p.heywood@sheffield.ac.uk>
---
 README.md                       |   2 +-
 include/flamegpu/io/Telemetry.h |  18 ++++++
 src/flamegpu/io/Telemetry.cpp   | 104 +++++++++++++++++++++++++++++++-
 3 files changed, 122 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index a25f4959d..d4edfd658 100644
--- a/README.md
+++ b/README.md
@@ -328,7 +328,7 @@ Information is collected when a simulation, ensemble or test suite run have comp
 The [TelemetryDeck](https://telemetrydeck.com/) service is used to store telemetry data. 
 All data is sent to their API endpoint of https://nom.telemetrydeck.com/v1/ via https. For more details please review the [TelmetryDeck privacy policy](https://telemetrydeck.com/privacy/).
 
-We do not collect any personal data such as usernames, email addresses or machine identifiers.
+We do not collect any personal data such as usernames, email addresses or hardware identifiers but we do generate a random user identifier. This identifier is salted and hashed by Telemetry deck.
 
 More information can be found in the [FLAMEGPU documentation](https://docs.flamegpu.com/guide/telemetry).
 
diff --git a/include/flamegpu/io/Telemetry.h b/include/flamegpu/io/Telemetry.h
index 1b6c87481..efefadf46 100644
--- a/include/flamegpu/io/Telemetry.h
+++ b/include/flamegpu/io/Telemetry.h
@@ -84,6 +84,24 @@ static void encourageUsage();
  * @return if telemetry is currently in test mode or not.
  */
 static bool isTestMode();
+
+/**
+ * Gets a cross platform location for storing user configuration data. Required to store a per user Id. On windows this uses the AppData folder (CSIDL_APPDATA) and in Linux this uses XDG_CONFIG_HOME.
+ * @return Root directory for storing configuration files
+ */
+static std::string getConfigDirectory();
+
+/**
+ * Generates a randomised 36 character alphanumeric string for use as a User Id. 
+ * @return A 36 character randomised alphanumeric string
+ */
+static std::string generateRandomId();
+
+/**
+ * Obtains a unique user Id. If a configuration file (i.e. ${XDG_CONFIG_HOME}/flamegpu/telemetry_user.cfg on linux) exists this will be loaded from disk otherwise it will be generated and stored in the configuration location. If the configuration location is not writeable a new user Id will be generated each time. The user Id will be further obfuscated by Telemetry Deck which will salt and hash the Id.
+ * @return A 36 character randomised alphanumeric string representing a unique user Id
+ */
+static std::string getUserId();
 };
 }  // namespace io
 }  // namespace flamegpu
diff --git a/src/flamegpu/io/Telemetry.cpp b/src/flamegpu/io/Telemetry.cpp
index e298cb136..cc2b292ee 100644
--- a/src/flamegpu/io/Telemetry.cpp
+++ b/src/flamegpu/io/Telemetry.cpp
@@ -6,11 +6,22 @@
 #include <memory>
 #include <array>
 #include <iostream>
+#include <fstream>
+#include <filesystem>
 #include <cstdio>
 #include <cctype>
 #include <sstream>
 #include <string>
 #include <map>
+#include <random>
+#ifdef _WIN32
+#include <windows.h>
+#include <shlobj.h>
+#else
+#include <sys/types.h>
+#include <unistd.h>
+#include <pwd.h>
+#endif
 
 #include "flamegpu/version.h"
 
@@ -159,7 +170,7 @@ std::string Telemetry::generateData(std::string event_name, std::map<std::string
     // check ENV for test variable FLAMEGPU_TEST_ENVIRONMENT
     std::string testmode = isTestMode() ? "true" : "false";
     std::string appID = TELEMETRY_APP_ID;
-    std::string telemetryRandomID = flamegpu::TELEMETRY_RANDOM_ID;
+    std::string telemetryRandomID = Telemetry::getUserId();  // Obtain a unique user ID
 
     // Differentiate pyflamegpu in the payload via the SWIG compiler macro, which we only define when building for pyflamegpu.
     // A user could potentially static link against a build using that macro, but that's not a use-case we are currently concerned with.
@@ -178,6 +189,7 @@ std::string Telemetry::generateData(std::string event_name, std::map<std::string
     payload_items["appVersionPatch"] = std::to_string(flamegpu::VERSION_PATCH);
     payload_items["appVersionPreRelease"] = flamegpu::VERSION_PRERELEASE;
     payload_items["buildNumber"] = flamegpu::VERSION_BUILDMETADATA;  // e.g. '0553592f' (graphed in Telemetry deck)
+    payload_items["buildID"] = flamegpu::TELEMETRY_RANDOM_ID;  // e.g. 'e7e0fe30325c83a3ad52e2cb2180e50979d3e6bcb0692789977a06ad09889843' (ID generated a cmake time)
 
     // OS
 #ifdef _WIN32
@@ -291,5 +303,95 @@ void Telemetry::encourageUsage() {
     }
 }
 
+std::string Telemetry::getConfigDirectory() {
+#ifdef _WIN32
+    char path[MAX_PATH];
+    if (SUCCEEDED(SHGetFolderPathA(NULL, CSIDL_APPDATA, NULL, 0, path))) {
+        return std::string(path);
+    }
+    throw std::runtime_error("Unable to retrieve config directory on Windows");
+#else
+    const char* configHome = std::getenv("XDG_CONFIG_HOME");
+    if (configHome) {
+        return std::string(configHome);
+    }
+    const char* home = std::getenv("HOME");
+    if (home) {
+        return std::string(home) + "/.config";
+    }
+    // try and get the user directory if home is not set
+    struct passwd pwd;
+    struct passwd* result = nullptr;
+    char buffer[4096];
+    int ret = getpwuid_r(getuid(), &pwd, buffer, sizeof(buffer), &result);
+    if (ret == 0 && result != nullptr) {
+        return std::string(pwd.pw_dir) + "/.config";
+    }
+
+    throw std::runtime_error("Unable to retrieve config directory on Linux");
+#endif
+}
+
+std::string Telemetry::generateRandomId() {
+    const char charset[] =
+        "0123456789"
+        "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
+        "abcdefghijklmnopqrstuvwxyz";
+    const size_t length = 36;
+    std::random_device rd;
+    std::mt19937 generator(rd());
+    std::uniform_int_distribution<size_t> distribution(0, sizeof(charset) - 2);
+
+    std::string randomId;
+    for (size_t i = 0; i < length; ++i) {
+        randomId += charset[distribution(generator)];
+    }
+    return randomId;
+}
+
+std::string Telemetry::getUserId() {
+    // Generate and a new random 36-character user ID
+    std::string userId = Telemetry::generateRandomId();
+
+    // Try to load an existing ID from file if it exists
+    try {
+        // Determine config file location
+        std::string configDir = Telemetry::getConfigDirectory() + "/flamegpu";
+        std::filesystem::create_directories(configDir);  // Ensure the directory exists
+        std::string filePath = configDir + "/telemetry_user.cfg";
+
+        // Check if the file exists
+        if (std::filesystem::exists(filePath)) {
+            std::ifstream file(filePath, std::ios_base::binary);
+            if (file.is_open()) {
+                std::string cached_id;
+                std::getline(file, cached_id);  // overwrite existing Id
+                file.close();
+                // Config file and user id found so return it if not empty (either because file is externally modified or file is a directory)
+                if (!cached_id.empty())
+                    return cached_id;
+                else
+                    return userId;
+            }
+            else {
+                throw std::runtime_error("Unable to open user ID file for reading");
+            }
+        }
+
+        std::ofstream file(filePath, std::ios_base::binary);
+        if (file.is_open()) {
+            file << userId;
+            file.close();
+        }
+        else {
+            throw std::runtime_error("Unable to create user ID file");
+        }
+    } catch (const std::exception&) {
+        fprintf(stderr, "Warning: Telemetry User Id file is not read/writeable from config file. A new User Id will be used.\n");
+    }
+    return userId;
+}
+
+
 }  // namespace io
 }  // namespace flamegpu