非机构化解析【包含PDF、word、PPT】

此项目是针对PDF、docx、doc、PPT四种非结构化数据进行解析，识别里面的文本和图片。

代码结构

├── Dockerfile
├── requirements
├── resluts
├── test_data
│   ├── 20151202033304658.pdf
│   ├── 2020_World_Energy_Data.pdf
│   ├── 2022110404_pdf.docx
│   ├── 2022110404_pdf.pdf
│   ├── H3_AP201701200282787162_01.pdf
│   ├── H3_AP202205271568109307_1.pdf
│   ├── H3_AP202205271568109307_1.pptx
│   ├── test.pdf
│   ├── test.pptx
│   ├── test_table.pdf
│   └── test_word.docx
├── Unstr_ApiSever.py ###----------API服务
├── Unstructured_PDF_Operation_Code.py
├── Unstructured_PPT_Operation_Code.py
├── Unstructured_Word_Operation_Code.py

PDF操作

部分代码展示

import fitz,os
doc = fitz.open('./test_data/2022110404_pdf.pdf')
def func(doc):
    for i in range(len(doc)):
        imglist = doc.getPageImageList(i)
        for j, img in enumerate(imglist):
            xref = img[0]
            pix = fitz.Pixmap(doc, xref)  # make pixmap from image
            if pix.n - pix.alpha < 4:  # can be saved as PNG
                pix.writePNG("p%s-%s.png" % (i + 1, j + 1))
            else:  # CMYK: must convert first
                pix0 = fitz.Pixmap(fitz.csRGB, pix)
                pix0.writePNG("p%s-%s.png" % (i + 1, j + 1))
                pix0 = None  # free Pixmap resources
            pix = None  # free Pixmap resources
            
if __name__ == "__main__":
    func(doc=fitz.open('./test_data/2022110404_pdf.pdf'))         # input the path of pdf file
    func1('./test_data')  # input the path of pdf file            
	pdf_path = "./test_data/2022110404_pdf.pdf"
	doc = fitz.open(pdf_path)
	num_pages = doc.page_count
	# Text info of PDF
	for page_index in range(num_pages):
	    page = doc.load_page(page_index)
	    text = page.get_text()
	    print(f"第{page_index + 1}页的文本内容为：\n{text}\n")

结果如下：

word操作

import docx
import os, re
from docx import Document

class Word:
    """
    Word操作
    """
    def Word_get_pictures(self,infile):
        try:
            in_File = infile.split('/')[2][:-5]  ##---------Word名称
            new_filepath = os.path.join('%s/%s') % ('./resluts', in_File)
            doc = docx.Document(infile)
            dict_rel = doc.part._rels
            for rel in dict_rel:
                rel = dict_rel[rel]
                if "image" in rel.target_ref:
                    if not os.path.exists(new_filepath):
                        os.makedirs(new_filepath)
                    img_name = re.findall("/(.*)", rel.target_ref)[0]
                    word_name = os.path.splitext(new_filepath)[0]
                    if os.sep in word_name:
                        new_name = word_name.split('\\')[-1]
                    else:
                        new_name = word_name.split('/')[-1]
                    img_name = f'{new_name}-' + '-' + f'{img_name}'
                    with open(f'{new_filepath}/{img_name}', "wb") as f:
                        f.write(rel.target_part.blob)
        except:
            pass

    def Word_Get_txt(self,infile):
        in_File = infile.split('/')[2][:-5]  ##---------Word名称
        new_filepath = os.path.join('%s/%s') % ('./resluts', in_File)
        document = Document(infile)
        all_paragraphs = document.paragraphs
        all_tables = document.tables
        with open(os.path.join("%s/%s.txt") % (new_filepath, "resluts"), 'w', encoding='utf-8') as f:
            for paragraph in all_paragraphs:
                # print(paragraph.text.replace("   ", "").replace(" ", ""))
                f.write(paragraph.text.replace("   ", "").replace(" ", ""))
            for table in all_tables:
                for row in table.rows:
                    for cell in row.cells:
                        f.write(cell.text)
                        # print(cell.text)  # 打印
if __name__ == '__main__':
    # 获取文件夹下的word文档列表,路径自定义
    # os.chdir("./test_data/2022110404_pdf.docx")

    Word().Word_get_pictures("./test_data/2022110404_pdf.docx")
    Word().Word_Get_txt("./test_data/2022110404_pdf.docx")

结果如下：

PPT操作

import os
from zipfile import ZipFile
from pptx import Presentation
from docx import Document

class PPT:
    def PPT_get_pictrue(self,infile):
        in_File = infile.split('/')[2][:-5] 
        new_filepath = os.path.join('%s/%s') % ('./resluts', in_File)
        if not os.path.exists(new_filepath):
            os.makedirs(new_filepath)
        with ZipFile(infile) as f:
            for file in f.namelist():
                if file.startswith("ppt/media/"):
                    f.extract(file, path=new_filepath)

        return new_filepath

    def PPT_get_words_to_txt(self,inpath, outpath):
        m_ppt = Presentation(inpath)
        # print(len(m_ppt.slides))
        with open(os.path.join('%s/%s.txt') % (outpath, 'resluts'), 'w', encoding='utf-8') as f:
            for slide in m_ppt.slides: 
                for shape in slide.shapes:
                    if not shape.has_text_frame: 
                        continue
                    for paragraph in shape.text_frame.paragraphs: 
                        for content in paragraph.runs:
                            f.write(content.text + '\n')

    def PPT_get_words_to_docx(self,filepath,save_path):
        wordfile = Document()
        pptx = Presentation(filepath)
        for slide in pptx.slides:
            for shape in slide.shapes:
                if shape.has_text_frame:
                    text_frame = shape.text_frame
                    for paragraph in text_frame.paragraphs:
                        wordfile.add_paragraph(paragraph.text)

        wordfile.save(save_path)


if __name__ == "__main__":
    infile = "./test_data/OpenCV算法解析.pptx"
    new_infile=PPT().PPT_get_pictrue(infile)
    PPT().PPT_get_words_to_txt(infile,new_infile)

结果如下：