From 9c0d859e91ad1e9b1f1a2bd79d302bb9d71a53f0 Mon Sep 17 00:00:00 2001
From: hhhhsc <1710496817@qq.com>
Date: Tue, 25 Aug 2026 11:36:14 +0800
Subject: [PATCH 1/2] =?UTF-8?q?=E9=A2=84=E5=88=B6tiktoken?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 scripts/images/backend-python/Dockerfile | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/scripts/images/backend-python/Dockerfile b/scripts/images/backend-python/Dockerfile
index cf24083cd..e06b808bc 100644
--- a/scripts/images/backend-python/Dockerfile
+++ b/scripts/images/backend-python/Dockerfile
@@ -26,7 +26,8 @@ ENV PYTHONDONTWRITEBYTECODE=1 \
     POETRY_VERSION=2.2.1 \
     POETRY_NO_INTERACTION=1 \
     POETRY_VIRTUALENVS_CREATE=false \
-    POETRY_CACHE_DIR=/tmp/poetry_cache
+    POETRY_CACHE_DIR=/tmp/poetry_cache \
+    TIKTOKEN_CACHE_DIR=/opt/tiktoken-cache
 
 ENV JAVA_HOME=/usr/lib/jvm/java-21-openjdk
 
@@ -49,6 +50,10 @@ COPY runtime/datamate-python/pyproject.toml runtime/datamate-python/poetry.lock*
 RUN --mount=type=cache,target=$POETRY_CACHE_DIR \
     poetry install --no-root --only main
 
+# Preload the tokenizer vocabulary so knowledge-base operations work offline.
+RUN mkdir -p "$TIKTOKEN_CACHE_DIR" \
+    && python -c "import tiktoken; tiktoken.get_encoding('cl100k_base')"
+
 # Download NLTK data
 RUN python -c "import nltk; nltk.download(['punkt_tab','averaged_perceptron_tagger_eng'], download_dir='/usr/local/nltk_data')"
 ENV NLTK_DATA=/usr/local/nltk_data

From a2e57f3d96670c90f79764caa92eef8a676e0f2e Mon Sep 17 00:00:00 2001
From: hhhhsc <1710496817@qq.com>
Date: Tue, 25 Aug 2026 11:41:51 +0800
Subject: [PATCH 2/2] =?UTF-8?q?=E9=A2=84=E5=88=B6tiktoken?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 scripts/images/backend-python/Dockerfile | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/scripts/images/backend-python/Dockerfile b/scripts/images/backend-python/Dockerfile
index e06b808bc..7819f92cc 100644
--- a/scripts/images/backend-python/Dockerfile
+++ b/scripts/images/backend-python/Dockerfile
@@ -52,7 +52,7 @@ RUN --mount=type=cache,target=$POETRY_CACHE_DIR \
 
 # Preload the tokenizer vocabulary so knowledge-base operations work offline.
 RUN mkdir -p "$TIKTOKEN_CACHE_DIR" \
-    && python -c "import tiktoken; tiktoken.get_encoding('cl100k_base')"
+    && python -c "import tiktoken; [tiktoken.get_encoding(name) for name in ('cl100k_base', 'o200k_base')]"
 
 # Download NLTK data
 RUN python -c "import nltk; nltk.download(['punkt_tab','averaged_perceptron_tagger_eng'], download_dir='/usr/local/nltk_data')"