dataprep: Fix issue in uploading docx with embedding image (#561)

lianhao · XuhuiRen · web-flow · commit b873cf833a7a · 2024-09-04T20:57:48.000+08:00
Fix issue #407 Signed-off-by: Lianhao Lu <lianhao.lu@intel.com> Co-authored-by: XuhuiRen <44249229+XuhuiRen@users.noreply.github.com>
diff --git a/comps/dataprep/milvus/requirements.txt b/comps/dataprep/milvus/requirements.txt
@@ -25,5 +25,5 @@ python-pptx
 sentence_transformers
 shortuuid
 tiktoken
-unstructured[all-docs]==0.11.5
+unstructured[all-docs]==0.15.7
 uvicorn
diff --git a/comps/dataprep/pgvector/langchain/requirements.txt b/comps/dataprep/pgvector/langchain/requirements.txt
@@ -26,6 +26,6 @@ python-pptx
 sentence_transformers
 shortuuid
 tiktoken
-unstructured[all-docs]==0.11.5
+unstructured[all-docs]==0.15.7
 uvicorn
 
diff --git a/comps/dataprep/pinecone/requirements.txt b/comps/dataprep/pinecone/requirements.txt
@@ -26,5 +26,5 @@ python-docx
 python-pptx
 sentence_transformers
 shortuuid
-unstructured[all-docs]==0.11.5
+unstructured[all-docs]==0.15.7
 uvicorn
diff --git a/comps/dataprep/qdrant/requirements.txt b/comps/dataprep/qdrant/requirements.txt
@@ -23,5 +23,5 @@ python-pptx
 qdrant-client
 sentence_transformers
 shortuuid
-unstructured[all-docs]==0.11.5
+unstructured[all-docs]==0.15.7
 uvicorn
diff --git a/comps/dataprep/redis/langchain/docker/docker-compose-dataprep-redis.yaml b/comps/dataprep/redis/langchain/docker/docker-compose-dataprep-redis.yaml
@@ -32,8 +32,6 @@ services:
       no_proxy: ${no_proxy}
       http_proxy: ${http_proxy}
       https_proxy: ${https_proxy}
-      REDIS_HOST: ${REDIS_HOST}
-      REDIS_PORT: ${REDIS_PORT}
       REDIS_URL: ${REDIS_URL}
       INDEX_NAME: ${INDEX_NAME}
       TEI_ENDPOINT: ${TEI_ENDPOINT}
diff --git a/comps/dataprep/redis/langchain/requirements.txt b/comps/dataprep/redis/langchain/requirements.txt
@@ -25,5 +25,5 @@ python-pptx
 redis
 sentence_transformers
 shortuuid
-unstructured[all-docs]==0.11.5
+unstructured[all-docs]==0.15.7
 uvicorn
diff --git a/comps/dataprep/redis/langchain_ray/requirements.txt b/comps/dataprep/redis/langchain_ray/requirements.txt
@@ -24,5 +24,6 @@ ray
 redis
 sentence_transformers
 shortuuid
+unstructured[all-docs]==0.15.7
 uvicorn
 virtualenv
diff --git a/comps/dataprep/utils.py b/comps/dataprep/utils.py
@@ -12,6 +12,7 @@
 import shutil
 import signal
 import subprocess
+import tempfile
 import timeit
 import unicodedata
 import urllib.parse
@@ -192,8 +193,7 @@ def load_docx(docx_path):
         if isinstance(r._target, docx.parts.image.ImagePart):
             rid2img[r.rId] = os.path.basename(r._target.partname)
     if rid2img:
-        save_path = "./imgs/"
-        os.makedirs(save_path, exist_ok=True)
+        save_path = tempfile.mkdtemp()
         docx2txt.process(docx_path, save_path)
     for paragraph in doc.paragraphs:
         if hasattr(paragraph, "text"):
