feature: 更新算子名称;增加创建任务、模板校验 (#57)

* feature: 更新算子名称;增加创建任务、模板校验

* feature: 镜像构建增加缓存
This commit is contained in:
hhhhsc701
2025-11-05 17:38:03 +08:00
committed by GitHub
parent 6917ae5b30
commit 05b26a2981
16 changed files with 56 additions and 40 deletions

View File

@@ -21,7 +21,7 @@ def _import_operators():
from . import file_exporter
from . import slide_formatter
from . import unstructured_formatter
from . import external_pdf_formatter
from . import mineru_formatter
_import_operators()

View File

@@ -2,5 +2,5 @@
from datamate.core.base_op import OPERATORS
OPERATORS.register_module(module_name='ExternalPDFFormatter',
OPERATORS.register_module(module_name='MineruFormatter',
module_path="ops.formatter.external_pdf_formatter.process")

View File

@@ -1,10 +1,10 @@
name: 'MinerU PDF文本抽取'
name_en: 'External PDF Text Extraction'
name_en: 'MinerU PDF Text Extraction'
description: '基于MinerU API,抽取PDF中的文本。'
description_en: 'Extracts text from PDF files based on MinerU API.'
language: 'python'
vendor: 'huawei'
raw_id: 'ExternalPDFFormatter'
raw_id: 'MineruFormatter'
version: '1.0.0'
types:
- 'collect'

View File

@@ -15,11 +15,11 @@ from datamate.core.base_op import Mapper
from datamate.common.utils.rest_client import http_request
class ExternalPDFFormatter(Mapper):
class MineruFormatter(Mapper):
"""基于外部API,抽取PDF中的文本"""
def __init__(self, *args, **kwargs):
super(ExternalPDFFormatter, self).__init__(*args, **kwargs)
super(MineruFormatter, self).__init__(*args, **kwargs)
self.base_url = os.getenv("EXTERNAL_PDF_BASE_URL", "http://datamate-mineru:9001")
self.pdf_extract_url = f"{self.base_url}/api/pdf-extract"
@@ -31,8 +31,8 @@ class ExternalPDFFormatter(Mapper):
response = http_request(method="POST", url=self.pdf_extract_url, data=data)
sample[self.text_key] = json.loads(response.text).get("result")
logger.info(
f"fileName: {filename}, method: ExternalPDFFormatter costs {(time.time() - start):6f} s")
f"fileName: {filename}, method: MineruFormatter costs {(time.time() - start):6f} s")
except UnicodeDecodeError as err:
logger.exception(f"fileName: {filename}, method: ExternalPDFFormatter causes decode error: {err}")
logger.exception(f"fileName: {filename}, method: MineruFormatter causes decode error: {err}")
raise
return sample

View File

@@ -1,4 +1,4 @@
name: '非结构化文本抽取'
name: 'Unstructured文本抽取'
name_en: 'Unstructured Text Extraction'
description: '抽取非结构化文件的文本,目前支持PowerPoint演示文稿、Word文档以及Excel工作簿。'
description_en: 'Extracts text from Unstructured files, currently supporting PowerPoint presentations, Word documents and Excel spreadsheets files.'

View File

@@ -51,7 +51,7 @@ class LazyLoader(ModuleType):
def __init__(self,
package_name,
module_name=None,
whl_path="/dataset/ops_whl",
whl_path=None,
exact_version=None,
force_reinstall=False
):
@@ -72,7 +72,7 @@ class LazyLoader(ModuleType):
self._module_name = module_name if module_name else package_name
self._package_name = package_name
self.whl_path = Path(whl_path).resolve()
self.whl_path = whl_path
self.exact_version = exact_version
self.force_reinstall = force_reinstall
@@ -126,7 +126,10 @@ class LazyLoader(ModuleType):
need_install = True
if need_install:
self._pip_install_package(package_name)
if self.whl_path is None:
self._pip_install_package_pypi(package_name)
else:
self._pip_install_package_local(package_name)
module = importlib.import_module(module_name)
self._cached_module = module
self._register_alias(module)
@@ -168,13 +171,26 @@ class LazyLoader(ModuleType):
return line.split()[-1]
raise PackageNotFoundError()
def _pip_install_package(self, package_name: str):
def _pip_install_package_pypi(self, package_name: str):
if self.exact_version:
package_name += f"=={self.exact_version}"
try:
subprocess.check_call([
sys.executable, "-m", "pip", "install", str(package_name)
], stdout=subprocess.DEVNULL, stderr=subprocess.STDOUT)
logger.info(f"Successfully installed {package_name}")
except subprocess.CalledProcessError as e:
logger.error(f"Installation failed: {e}")
raise RuntimeError(f"Installation failed: {e}") from e
def _pip_install_package_local(self, package_name: str):
"""安装逻辑 """
if not self.whl_path.exists():
whl_path = Path(self.whl_path).resolve()
if not whl_path.exists():
raise FileNotFoundError(f"WHL directory not found: {self.whl_path}")
whl_files = list(self.whl_path.glob(f"{package_name}*.whl"))
whl_files = list(whl_path.glob(f"{package_name}*.whl"))
if not whl_files:
raise RuntimeError(f"No WHL files found for {package_name}")