Refactor interactive mode in main.cpp

Author: thomasantony

main.cpp

@@ -51,6 +51,8 @@ void sigint_handler(int signo) {
 #endif
 
 
+void process_interactive_input(llama_context& ctx, const gpt_params& params);
+
 int main(int argc, char ** argv) {
     ggml_time_init();
     const int64_t t_main_start_us = ggml_time_us();
@@ -95,7 +97,7 @@ int main(int argc, char ** argv) {
     // tokenize the reverse prompt
     std::vector<gpt_vocab::id> antiprompt_inp = llama_tokenize_text(ctx, params.antiprompt);
 
-
+    // Setup interactive mode
     if (params.interactive) {
 #if defined (__unix__) || (defined (__APPLE__) && defined (__MACH__))
         struct sigaction sigint_action;
@@ -176,43 +178,16 @@ int main(int argc, char ** argv) {
 
         // in interactive mode, and not currently processing queued inputs;
         // check if we should prompt the user for more
-        if (params.interactive && !llama_has_unconsumed_input(ctx)) {
-            const std::vector<gpt_vocab::id>& last_n_tokens = llama_context_get_last_n_tokens(ctx);
+        if (params.interactive) {
             // check for reverse prompt
-            if (antiprompt_inp.size() && std::equal(antiprompt_inp.rbegin(), antiprompt_inp.rend(), last_n_tokens.rbegin())) {
+            if (antiprompt_inp.size() && llama_is_anti_prompt_present(ctx, antiprompt_inp)) {
                 // reverse prompt found
                 is_interacting = true;
             }
             if (is_interacting) {
                 // currently being interactive
-                bool another_line=true;
-                while (another_line) {
-                    fflush(stdout);
-                    char buf[256] = {0};
-                    int n_read;
-                    if(params.use_color) printf(ANSI_BOLD ANSI_COLOR_GREEN);
-                    if (scanf("%255[^\n]%n%*c", buf, &n_read) <= 0) {
-                        // presumable empty line, consume the newline
-                        scanf("%*c");
-                        n_read=0;
-                    }
-                    if(params.use_color) printf(ANSI_COLOR_RESET);
-
-                    if (n_read > 0 && buf[n_read-1]=='\\') {
-                        another_line = true;
-                        buf[n_read-1] = '\n';
-                        buf[n_read] = 0;
-                    } else {
-                        another_line = false;
-                        buf[n_read] = '\n';
-                        buf[n_read+1] = 0;
-                    }
-
-                    // Do not clear existing context in interactive mode
-                    llama_update_input(ctx, buf);
-                    input_noecho = true; // do not echo this again
-                }
-
+                process_interactive_input(ctx, params);
+                input_noecho = true; // do not echo this input again
                 is_interacting = false;
             }
         }
@@ -243,3 +218,33 @@ int main(int argc, char ** argv) {
     }
     return 0;
 }
+
+void process_interactive_input(llama_context& ctx, const gpt_params& params)
+{
+    bool another_line=true;
+    while (another_line) {
+        fflush(stdout);
+        char buf[256] = {0};
+        int n_read;
+        if(params.use_color) printf(ANSI_BOLD ANSI_COLOR_GREEN);
+        if (scanf("%255[^\n]%n%*c", buf, &n_read) <= 0) {
+            // presumable empty line, consume the newline
+            scanf("%*c");
+            n_read=0;
+        }
+        if(params.use_color) printf(ANSI_COLOR_RESET);
+
+        if (n_read > 0 && buf[n_read-1]=='\\') {
+            another_line = true;
+            buf[n_read-1] = '\n';
+            buf[n_read] = 0;
+        } else {
+            another_line = false;
+            buf[n_read] = '\n';
+            buf[n_read+1] = 0;
+        }
+
+        // Do not clear existing context in interactive mode
+        llama_update_input(ctx, buf);
+    }
+}