From 8f46aebdd1333ed6d017472114d0cef04ce70e8a Mon Sep 17 00:00:00 2001
From: Jean Machuca <correojean@gmail.com>
Date: Wed, 11 Nov 2026 05:38:21 +0000
Subject: [PATCH] fix: build aports buffers for tokenizer in cgo bridge

---
 internal/llm/bridge.go        | 26 +++++++++++++++++---------
 internal/server/middleware.go | 13 +++++++++++--
 2 files changed, 28 insertions(+), 11 deletions(-)

diff --git a/internal/llm/bridge.go b/internal/llm/bridge.go
index 34c619c..d4b513f 100644
--- a/internal/llm/bridge.go
+++ b/internal/llm/bridge.go
@@ -38,6 +38,7 @@ func bridgeLoadModel(modelPath string, nCtx, nGPULayers, nThreads int) (*cgoMode
 	if model == nil {
 		return nil, fmt.Errorf("E_MODEL_LOAD_FAILED: llama_model_load_from_file returned null for %q", modelPath)
 	}
+	fmt.Printf("DEBUG: load model=%p, vocab=%p, path=%q\n", model, C.llama_model_get_vocab(model), modelPath)
 
 	cparams := C.llama_context_default_params()
 	cparams.n_ctx = C.uint32_t(nCtx)
@@ -90,7 +91,9 @@ func bridgeLoadAdapter(cm *cgoModel, adapterPath string) error {
 }
 
 func (m *cgoModel) vocab() *C.struct_llama_vocab {
-	return C.llama_model_get_vocab(m.model)
+	v := C.llama_model_get_vocab(m.model)
+	fmt.Printf("DEBUG: vocab() m.model=%p, vocab=%p\n", m.model, v)
+	return v
 }
 
 func (m *cgoModel) nCtx() int {
@@ -118,15 +121,20 @@ func bridgeTokenize(vocab *C.struct_llama_vocab, text string, addSpecial bool) (
 	ctext := C.CString(text)
 	defer C.free(unsafe.Pointer(ctext))
 
-	n := C.llama_tokenize(vocab, ctext, C.int32_t(len(text)), nil, 0, C.bool(addSpecial), false)
-	if n <= 0 {
-		return nil, fmt.Errorf("tokenize failed: text too long or empty")
-	}
-
-	ctokens := make([]C.llama_token, n)
-	n = C.llama_tokenize(vocab, ctext, C.int32_t(len(text)), &ctokens[0], C.int32_t(len(ctokens)), C.bool(addSpecial), false)
+	textLen := C.int32_t(len(text))
+	// First call: estimate buffer size. Some llama.cpp versions don't support
+	// NULL tokens for counting, so start with a reasonable estimate.
+	nTokens := C.int32_t(len(text)*2 + 16)
+	ctokens := make([]C.llama_token, nTokens)
+	n := C.llama_tokenize(vocab, ctext, textLen, &ctokens[0], nTokens, C.bool(addSpecial), false)
 	if n < 0 {
-		return nil, fmt.Errorf("tokenize failed on second pass")
+		// Buffer too small, retry with exact size returned (negative = required size)
+		nTokens = C.int32_t(-n)
+		ctokens = make([]C.llama_token, nTokens)
+		n = C.llama_tokenize(vocab, ctext, textLen, &ctokens[0], nTokens, C.bool(addSpecial), false)
+		if n < 0 {
+			return nil, fmt.Errorf("tokenize failed: %d", n)
+		}
 	}
 
 	tokens := make([]int32, n)
diff --git a/internal/server/middleware.go b/internal/server/middleware.go
index fe890b1..c10ae33 100644
--- a/internal/server/middleware.go
+++ b/internal/server/middleware.go
@@ -1,7 +1,9 @@
 package server
 
 import (
+	"bytes"
 	"encoding/json"
+	"io"
 	"net/http"
 	"sync"
 	"time"
@@ -48,15 +50,22 @@ func (rl *rateLimiter) allow(ip string) bool {
 }
 
 // validateJSONBody checks that the request body is valid JSON for POST requests.
+// It reads the body into memory, validates JSON, then replaces the body reader
+// so downstream handlers can still read it.
 func validateJSONBody(next http.HandlerFunc) http.HandlerFunc {
 	return func(w http.ResponseWriter, r *http.Request) {
 		if r.Method == http.MethodPost && r.Body != nil {
+			body, err := io.ReadAll(r.Body)
+			if err != nil {
+				http.Error(w, `{"error":"invalid JSON body","code":"E_INVALID_JSON"}`, http.StatusBadRequest)
+				return
+			}
 			var v interface{}
-			if err := json.NewDecoder(r.Body).Decode(&v); err != nil {
+			if err := json.Unmarshal(body, &v); err != nil {
 				http.Error(w, `{"error":"invalid JSON body","code":"E_INVALID_JSON"}`, http.StatusBadRequest)
 				return
 			}
-			_ = v
+			r.Body = io.NopCloser(bytes.NewReader(body))
 		}
 		next.ServeHTTP(w, r)
 	}