From c69af34d976c5a47100e4f935081c1ce8a67494d Mon Sep 17 00:00:00 2001
From: "The gemma.cpp Authors" <no-reply@google.com>
Date: Fri, 28 Aug 2026 09:00:58 -0700
Subject: [PATCH] Internal Changes

PiperOrigin-RevId: 972619142
---
 deepseek/dsv4_tokenizer.cc | 16 ++++++++++++++--
 deepseek/dsv4_tokenizer.h  |  9 +++++++++
 2 files changed, 23 insertions(+), 2 deletions(-)

diff --git a/deepseek/dsv4_tokenizer.cc b/deepseek/dsv4_tokenizer.cc
index 15716e44..e339d809 100644
--- a/deepseek/dsv4_tokenizer.cc
+++ b/deepseek/dsv4_tokenizer.cc
@@ -22,8 +22,10 @@
 #include <limits>
 #include <sstream>
 #include <string>
+#include <string_view>
 #include <vector>
 
+
 #include "hwy/base.h"  // HWY_ABORT
 #include "nlohmann/json.hpp"
 
@@ -172,12 +174,22 @@ std::string ReadFileToStringOrAbort(const std::string& path) {
 
 Dsv4Tokenizer::Dsv4Tokenizer(const std::string& tokenizer_json_path) {
   const std::string contents = ReadFileToStringOrAbort(tokenizer_json_path);
-  json j = json::parse(contents, /*cb=*/nullptr, /*allow_exceptions=*/false);
+  Init(contents);
+}
+
+Dsv4Tokenizer::Dsv4Tokenizer(std::string_view json_content, bool /*is_content*/) {
+  Init(json_content);
+}
+
+void Dsv4Tokenizer::Init(std::string_view json_content) {
+  json j = json::parse(json_content.begin(), json_content.end(), /*cb=*/nullptr,
+                      /*allow_exceptions=*/false);
   if (j.is_discarded()) {
-    HWY_ABORT("Failed to parse tokenizer JSON %s", tokenizer_json_path.c_str());
+    HWY_ABORT("Failed to parse tokenizer JSON");
   }
 
   uint32_t byte_to_cp[256];
+
   BuildByteToCp(byte_to_cp);
   std::unordered_map<uint32_t, uint8_t> cp_to_byte;
   for (int b = 0; b < 256; ++b) {
diff --git a/deepseek/dsv4_tokenizer.h b/deepseek/dsv4_tokenizer.h
index 15bc9e8a..705e7b5e 100644
--- a/deepseek/dsv4_tokenizer.h
+++ b/deepseek/dsv4_tokenizer.h
@@ -25,6 +25,7 @@
 #include <stdint.h>
 
 #include <string>
+#include <string_view>
 #include <unordered_map>
 #include <vector>
 
@@ -35,6 +36,11 @@ class Dsv4Tokenizer {
   // Aborts on I/O or parse errors.
   explicit Dsv4Tokenizer(const std::string& tokenizer_json_path);
 
+  // Initializes tokenizer directly from JSON content in memory.
+  // 'is_content' is used to distinguish from path constructor.
+  Dsv4Tokenizer(std::string_view json_content, bool is_content);
+
+
   // Extracts added tokens (chat markers etc.), pre-tokenizes and BPE-encodes
   // everything in between. Equivalent to HF encode(add_special_tokens=false).
   std::vector<int> Encode(const std::string& text) const;
@@ -57,6 +63,9 @@ class Dsv4Tokenizer {
     int id;
   };
 
+  void Init(std::string_view json_content);
+
+
   // Splits `text` (a span with no added tokens) into pre-tokenization pieces
   // and BPE-encodes each, appending ids.
   void EncodeSegment(const char* bytes, size_t len,