Sane sanitization

src/mesh/TypeConversions.cpp

@@ -1,6 +1,7 @@
 #include "TypeConversions.h"
 #include "mesh/generated/meshtastic/deviceonly.pb.h"
 #include "mesh/generated/meshtastic/mesh.pb.h"
+#include "meshUtils.h"
 
 meshtastic_NodeInfo TypeConversions::ConvertToNodeInfo(const meshtastic_NodeInfoLite *lite)
 {
@@ -82,8 +83,10 @@ meshtastic_UserLite TypeConversions::ConvertToUserLite(meshtastic_User user)
 
     strncpy(lite.long_name, user.long_name, sizeof(lite.long_name));
     lite.long_name[sizeof(lite.long_name) - 1] = '\0';
+    sanitizeUtf8(lite.long_name, sizeof(lite.long_name));
     strncpy(lite.short_name, user.short_name, sizeof(lite.short_name));
     lite.short_name[sizeof(lite.short_name) - 1] = '\0';
+    sanitizeUtf8(lite.short_name, sizeof(lite.short_name));
     lite.hw_model = user.hw_model;
     lite.role = user.role;
     lite.is_licensed = user.is_licensed;
@@ -102,8 +105,10 @@ meshtastic_User TypeConversions::ConvertToUser(uint32_t nodeNum, meshtastic_User
     snprintf(user.id, sizeof(user.id), "!%08x", nodeNum);
     strncpy(user.long_name, lite.long_name, sizeof(user.long_name));
     user.long_name[sizeof(user.long_name) - 1] = '\0';
+    sanitizeUtf8(user.long_name, sizeof(user.long_name));
     strncpy(user.short_name, lite.short_name, sizeof(user.short_name));
     user.short_name[sizeof(user.short_name) - 1] = '\0';
+    sanitizeUtf8(user.short_name, sizeof(user.short_name));
     user.hw_model = lite.hw_model;
     user.role = lite.role;
     user.is_licensed = lite.is_licensed;

src/meshUtils.cpp

@@ -117,4 +117,93 @@ size_t pb_string_length(const char *str, size_t max_len)
         }
     }
     return len;
+}
+
+bool sanitizeUtf8(char *buf, size_t bufSize)
+{
+    if (!buf || bufSize == 0)
+        return false;
+
+    // Ensure null-terminated within buffer
+    buf[bufSize - 1] = '\0';
+
+    bool replaced = false;
+    size_t i = 0;
+    size_t len = strlen(buf);
+
+    while (i < len) {
+        uint8_t b = (uint8_t)buf[i];
+
+        // Determine expected sequence length from lead byte
+        size_t seqLen;
+        uint32_t minCodepoint;
+        if (b <= 0x7F) {
+            // ASCII — valid single byte
+            i++;
+            continue;
+        } else if ((b & 0xE0) == 0xC0) {
+            seqLen = 2;
+            minCodepoint = 0x80; // Reject overlong
+        } else if ((b & 0xF0) == 0xE0) {
+            seqLen = 3;
+            minCodepoint = 0x800;
+        } else if ((b & 0xF8) == 0xF0) {
+            seqLen = 4;
+            minCodepoint = 0x10000;
+        } else {
+            // Invalid lead byte (0x80-0xBF or 0xF8+)
+            buf[i] = '?';
+            replaced = true;
+            i++;
+            continue;
+        }
+
+        // Check that we have enough bytes remaining
+        if (i + seqLen > len) {
+            // Truncated sequence at end of string — replace remaining bytes
+            for (size_t j = i; j < len; j++) {
+                buf[j] = '?';
+            }
+            replaced = true;
+            break;
+        }
+
+        // Validate continuation bytes (must be 10xxxxxx)
+        bool valid = true;
+        for (size_t j = 1; j < seqLen; j++) {
+            if (((uint8_t)buf[i + j] & 0xC0) != 0x80) {
+                valid = false;
+                break;
+            }
+        }
+
+        if (valid) {
+            // Decode codepoint to check for overlong encodings and surrogates
+            uint32_t cp = 0;
+            if (seqLen == 2)
+                cp = b & 0x1F;
+            else if (seqLen == 3)
+                cp = b & 0x0F;
+            else
+                cp = b & 0x07;
+            for (size_t j = 1; j < seqLen; j++)
+                cp = (cp << 6) | ((uint8_t)buf[i + j] & 0x3F);
+
+            if (cp < minCodepoint || cp > 0x10FFFF || (cp >= 0xD800 && cp <= 0xDFFF)) {
+                // Overlong encoding, out of Unicode range, or surrogate half
+                valid = false;
+            }
+        }
+
+        if (valid) {
+            i += seqLen;
+        } else {
+            // Replace only the lead byte; continuation bytes will be caught on next iteration
+            buf[i] = '?';
+            replaced = true;
+            i++;
+        }
+    }
+
+    return replaced;
 }

src/meshUtils.h

@@ -38,6 +38,10 @@ const std::string vformat(const char *const zcFormat, ...);
 // Get actual string length for nanopb char array fields.
 size_t pb_string_length(const char *str, size_t max_len);
 
+// Sanitize a fixed-size char buffer in-place by replacing invalid UTF-8 sequences with '?'.
+// Ensures the result is null-terminated within bufSize. Returns true if any bytes were replaced.
+bool sanitizeUtf8(char *buf, size_t bufSize);
+
 /// Calculate 2^n without calling pow() - used for spreading factor and other calculations
 inline uint32_t pow_of_2(uint32_t n)
 {

src/modules/AdminModule.cpp

@@ -599,10 +599,14 @@ void AdminModule::handleSetOwner(const meshtastic_User &o)
     if (*o.long_name) {
         changed |= strcmp(owner.long_name, o.long_name);
         strncpy(owner.long_name, o.long_name, sizeof(owner.long_name));
+        owner.long_name[sizeof(owner.long_name) - 1] = '\0';
+        sanitizeUtf8(owner.long_name, sizeof(owner.long_name));
     }
     if (*o.short_name) {
         changed |= strcmp(owner.short_name, o.short_name);
         strncpy(owner.short_name, o.short_name, sizeof(owner.short_name));
+        owner.short_name[sizeof(owner.short_name) - 1] = '\0';
+        sanitizeUtf8(owner.short_name, sizeof(owner.short_name));
     }
     snprintf(owner.id, sizeof(owner.id), "!%08x", nodeDB->getNodeNum());
 
@@ -1400,7 +1404,11 @@ void AdminModule::handleSetHamMode(const meshtastic_HamParameters &p)
 
     // Set call sign and override lora limitations for licensed use
     strncpy(owner.long_name, p.call_sign, sizeof(owner.long_name));
+    owner.long_name[sizeof(owner.long_name) - 1] = '\0';
+    sanitizeUtf8(owner.long_name, sizeof(owner.long_name));
     strncpy(owner.short_name, p.short_name, sizeof(owner.short_name));
+    owner.short_name[sizeof(owner.short_name) - 1] = '\0';
+    sanitizeUtf8(owner.short_name, sizeof(owner.short_name));
     owner.is_licensed = true;
     config.lora.override_duty_cycle = true;
     config.lora.tx_power = p.tx_power;