q792602257/DataMate
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

bugfix: update values.yaml to enhance ray-cluster configuration with security context, environment variables, and resource limits (#172)

Browse Source 
* feature: unstructured支持简单pdf处理

* feature: update values.yaml to enhance ray-cluster configuration with security context, environment variables, and resource limits
This commit is contained in:
hhhhsc701
2025-12-17 10:41:13 +08:00
committed by GitHub
parent 082aca1597
commit 62b91b6deb
5 changed files with 62 additions and 51 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
runtime/python-executor/datamate/core/base_op.py
View File

@@ -146,10 +146,10 @@ class BaseOp:
def read_file(self, sample):
filepath = sample[self.filepath_key]
filetype = sample[self.filetype_key]
if filetype in ["ppt", "pptx", "docx", "doc", "xlsx"]:
if filetype in ["ppt", "pptx", "docx", "doc", "xlsx", "csv", "md", "pdf"]:
elements = partition(filename=filepath)
sample[self.text_key] = "\n\n".join([str(el) for el in elements])
elif filetype in ["txt", "md", "markdown", "xml", "html", "csv", "json", "jsonl"]:
elif filetype in ["txt", "md", "markdown", "xml", "html", "json", "jsonl"]:
with open(filepath, 'rb') as f:
content = f.read()
sample[self.text_key] = content.decode("utf-8-sig").replace("\r\n", "\n")

2
runtime/python-executor/pyproject.toml
View File

@@ -21,7 +21,7 @@ dependencies = [
"loguru>=0.7.3",
"opencv-python-headless>=4.12.0.88",
"ray[data,default]==2.52.1",
"unstructured[csv,docx,pptx,xlsx]==0.18.15",
"unstructured[csv,docx,pptx,xlsx,pdf,md]==0.18.15",
"uvicorn[standard]>=0.38.0",
]