'commit'

dsRag/Test/TestDocling.py

@@ -0,0 +1,45 @@
+# 参考文档
+# https://blog.csdn.net/make_progress/article/details/144490719
+
+# 魔搭社区下载DoclingModels
+# https://modelscope.cn/models/ds4sd/docling-models
+# pip install modelscope
+# modelscope download --model ds4sd/docling-models  --local_dir D:\Model\docling
+
+
+from docling.datamodel.base_models import InputFormat
+from docling.datamodel.pipeline_options import PdfPipelineOptions, EasyOcrOptions
+from docling.document_converter import PdfFormatOption, DocumentConverter, ImageFormatOption, PowerpointFormatOption, \
+    WordFormatOption, ExcelFormatOption, HTMLFormatOption
+
+# 参考接口地址
+# https://ds4sd.github.io/docling/v2/#access-document-structures
+
+# 配置OCR模型，设置EasyOCR模型的路径
+easyocr_model_storage_directory = "D:/Model/ocr/EasyOCR/model"
+easyocr_options = EasyOcrOptions()
+easyocr_options.model_storage_directory = easyocr_model_storage_directory
+
+# 配置pdf模型，设置Docling模型的路径
+pdf_artifacts_path = "D:/Model/docling"
+pdf_pipeline_options = PdfPipelineOptions(artifacts_path=pdf_artifacts_path)
+pdf_pipeline_options.ocr_options = easyocr_options
+
+# 转换模型
+converter = DocumentConverter(
+    format_options={
+        InputFormat.PDF: PdfFormatOption(pipeline_options=pdf_pipeline_options),
+        InputFormat.IMAGE: ImageFormatOption(pipeline_options=pdf_pipeline_options),
+        InputFormat.PPTX: PowerpointFormatOption(pipeline_options=pdf_pipeline_options),
+        InputFormat.DOCX: WordFormatOption(pipeline_options=pdf_pipeline_options),
+        InputFormat.XLSX: ExcelFormatOption(pipeline_options=pdf_pipeline_options),
+        InputFormat.HTML: HTMLFormatOption(pipeline_options=pdf_pipeline_options)
+    }
+)
+
+source = "E:/test/test.pdf"
+result = converter.convert(source)
+print(result.document.export_to_markdown())
+
+
+

dsRag/Test/TestEasyOCR.py

@@ -0,0 +1,24 @@
+"""
+conda activate rag
+
+# 安装docling
+pip install docling
+
+# 安装OCR工具
+pip install easyocr
+"""
+
+import easyocr
+ocr_model_path = "E:/model/ocr/EasyOCR/model"
+reader = easyocr.Reader(
+    # 设置识别图片中语言的模型
+    lang_list=['ch_sim', 'en'],
+    # 注意会自动下载模型到路径下，可以不指定模型目录，会自动下载到C盘下
+    # 包括三个模型
+    # 文本检测模型：CRAFT -- craft_mlt_25k.pth
+    # 中文简体模型：ch_sim -- zh_sim_g2.pth
+    # 英文模型：en -- latin_g2.pth
+    model_storage_directory=ocr_model_path
+)
+result = reader.readtext("E:/test/test.png")
+print(result)