gguf : add string length accessor

Add gguf_get_arr_str_len() to obtain the length of a string stored in a
GGUF string array without scanning for the terminator. This allows
callers to pre-compute the exact size of a result buffer before
formatting string arrays (ref:
https://github.com/ggml-org/llama.cpp/pull/30679#discussion_r4324797291).

Also bounds-check the index in gguf_get_arr_str().

Assisted-by: pi:llama.cpp/ggml-org/Qwen3.8-Flash-Next-GGUF (RTX 5090 32GB, DDR5 128GB)

ggml/include/gguf.h

@@ -125,6 +125,10 @@ extern "C" {
     // get ith C string from array with given key_id
     GGML_API const char * gguf_get_arr_str (const struct gguf_context * ctx, int64_t key_id, size_t i);
 
+    // get length of the ith C string from array with given key_id, excluding the terminating NUL
+    // for a non-array string KV pair, use i = 0
+    GGML_API size_t       gguf_get_arr_str_len(const struct gguf_context * ctx, int64_t key_id, size_t i);
+
     GGML_API int64_t         gguf_get_n_tensors    (const struct gguf_context * ctx);
     GGML_API int64_t         gguf_find_tensor      (const struct gguf_context * ctx, const char * name); // returns -1 if the tensor is not found
     GGML_API size_t          gguf_get_tensor_offset(const struct gguf_context * ctx, int64_t tensor_id);

ggml/src/gguf.cpp

@@ -1096,9 +1096,17 @@ const void * gguf_get_arr_data(const struct gguf_context * ctx, int64_t key_id)
 const char * gguf_get_arr_str(const struct gguf_context * ctx, int64_t key_id, size_t i) {
     GGML_ASSERT(key_id >= 0 && key_id < gguf_get_n_kv(ctx));
     GGML_ASSERT(ctx->kv[key_id].get_type() == GGUF_TYPE_STRING);
+    GGML_ASSERT(i < ctx->kv[key_id].data_string.size());
     return ctx->kv[key_id].data_string[i].c_str();
 }
 
+size_t gguf_get_arr_str_len(const struct gguf_context * ctx, int64_t key_id, size_t i) {
+    GGML_ASSERT(key_id >= 0 && key_id < gguf_get_n_kv(ctx));
+    GGML_ASSERT(ctx->kv[key_id].get_type() == GGUF_TYPE_STRING);
+    GGML_ASSERT(i < ctx->kv[key_id].data_string.size());
+    return ctx->kv[key_id].data_string[i].size();
+}
+
 size_t gguf_get_arr_n(const struct gguf_context * ctx, int64_t key_id) {
     GGML_ASSERT(key_id >= 0 && key_id < gguf_get_n_kv(ctx));
 

tests/test-gguf.cpp

@@ -577,6 +577,10 @@ static bool handcrafted_check_kv(const gguf_context * gguf_ctx, const unsigned i
                 ok = false;
                 continue;
             }
+            if (gguf_get_arr_str_len(gguf_ctx, id, 0) != n) { // a string KV pair has a single element
+                ok = false;
+                continue;
+            }
             if (!std::equal(str, str + n, data8)) {
                 ok = false;
             }
@@ -602,6 +606,10 @@ static bool handcrafted_check_kv(const gguf_context * gguf_ctx, const unsigned i
                         ok = false;
                         continue;
                     }
+                    if (gguf_get_arr_str_len(gguf_ctx, id, istr) != n) {
+                        ok = false;
+                        continue;
+                    }
                     const char * str_expected = reinterpret_cast<const char *>(&data[istr]);
                     if (strncmp(str, str_expected, n) != 0) {
                         ok = false;