Llama-quantize: Partial requant feature (#1313)

* Partial Requant feature for llama-quantize

- Inspired by the recently portcopied --dry-run feature.
- Allows to partially requantize a split quantized .gguf by requantizing only the missing splits in the destination directory.
- Works both for GGUF which are split tensors by tensors, or by group of several tensors (though this one is not very much tested beyond 2 tensors by split).
- Vibe coded.

* Create output directory if it doesn't exist in llama-quantize

* Create output directory if it doesn't exist in gguf-split

* Add exit when directory fails to be created on Windows

* Use std::filesystem

* cleanup

examples/gguf-split/gguf-split.cpp

@@ -7,6 +7,7 @@
 #include <fstream>
 #include <string>
 #include <vector>
+#include <filesystem>
 
 #include <stdio.h>
 #include <string.h>
@@ -190,6 +191,18 @@ static void zeros(std::ofstream & file, size_t n) {
     }
 }
 
+static void ensure_output_directory(const std::string & filepath) {
+    std::filesystem::path p(filepath);
+    if (p.has_parent_path()) {
+        std::error_code ec;
+        std::filesystem::create_directories(p.parent_path(), ec);
+        if (ec) {
+            fprintf(stderr, "Failed to create directory '%s': %s\n", p.parent_path().string().c_str(), ec.message().c_str());
+            exit(EXIT_FAILURE);
+        }
+    }
+}
+
 struct split_strategy {
     const split_params params;
     std::ifstream & f_input;
@@ -310,6 +323,8 @@ struct split_strategy {
             char split_path[PATH_MAX] = {0};
             llama_split_path(split_path, sizeof(split_path), params.output.c_str(), i_split, n_split);
 
+            ensure_output_directory(split_path);
+
             // open the output file
             printf("Writing file %s ... ", split_path);
             fflush(stdout);
@@ -401,6 +416,8 @@ static void gguf_merge(const split_params & split_params) {
     int n_split = 1;
     int total_tensors = 0;
 
+    ensure_output_directory(split_params.output);
+
     // avoid overwriting existing output file
     if (std::ifstream(split_params.output.c_str())) {
         fprintf(stderr, "%s: output file %s already exists\n", __func__, split_params.output.c_str());