pip install fitz

pip install PyMuPDF
pip install easygui  # 用來(lái)彈出文件選擇框的，thinker的話會(huì)彈出兩個(gè)窗口怪怪的

# -*- coding:utf-8 -*-
# author: peng
# file: mypdf.py
# time: 2021/9/8 17:47
# desc:壓縮PDF，對(duì)純圖片的PDF效果效果較好，有文字內(nèi)容的可能會(huì)比較模糊，推薦高質(zhì)量的壓縮

import fitz
from PIL import Image
import os
from shutil import copyfile, rmtree
from math import ceil

from time import strftime, localtime, time
import easygui as g

from functools import wraps
# 時(shí)間計(jì)數(shù)裝飾器，func如果有return值，必須返回才能有值
def runtime(func):
    @wraps(func)
    def wrapper(*args, **kwargs):
        print(strftime("%Y-%m-%d %H:%M:%S", localtime()))
        start = time()
        func_return = func(*args, **kwargs)
        end = time()
        print(func.__name__, args[-1], args[-2], " spend time ", end - start, " sec")
        return func_return

    return wrapper


class Luban(object):

    def __init__(self, quality, ignoreBy=102400):
        self.ignoreBy = ignoreBy
        self.quality = quality

    def setPath(self, path):
        self.path = path

    def setTargetDir(self, foldername="target"):
        self.dir, self.filename = os.path.split(self.path)
        self.targetDir = os.path.join(self.dir, foldername)

        if not os.path.exists(self.targetDir):
            os.makedirs(self.targetDir)

        self.targetPath = os.path.join(self.targetDir, "c_" + self.filename)

    def load(self):
        self.img = Image.open(self.path)

        if self.img.mode == "RGB":
            self.type = "JPEG"
        elif self.img.mode == "RGBA":
            self.type = "PNG"
        else:  # 其他的圖片就轉(zhuǎn)成JPEG
            self.img = self.img.convert("RGB")
            self.type = "JPEG"

    def computeScale(self):
        # 計(jì)算縮小的倍數(shù)

        srcWidth, srcHeight = self.img.size

        srcWidth = srcWidth + 1 if srcWidth % 2 == 1 else srcWidth
        srcHeight = srcHeight + 1 if srcHeight % 2 == 1 else srcHeight

        longSide = max(srcWidth, srcHeight)
        shortSide = min(srcWidth, srcHeight)

        scale = shortSide / longSide
        if (scale <= 1 and scale > 0.5625):
            if (longSide < 1664):
                return 1
            elif (longSide < 4990):
                return 2
            elif (longSide > 4990 and longSide < 10240):
                return 4
            else:
                return max(1, longSide // 1280)

        elif (scale <= 0.5625 and scale > 0.5):
            return max(1, longSide // 1280)

        else:
            return ceil(longSide / (1280.0 / scale))

    def compress(self):
        self.setTargetDir()
        # 先調(diào)整大小，再調(diào)整品質(zhì)
        if os.path.getsize(self.path) <= self.ignoreBy:
            copyfile(self.path, self.targetPath)
        else:
            self.load()

            scale = self.computeScale()
            srcWidth, srcHeight = self.img.size
            cache = self.img.resize((srcWidth // scale, srcHeight // scale),
                                    Image.ANTIALIAS)

            cache.save(self.targetPath, self.type, quality=self.quality)


# 提取成圖片
def covert2pic(doc, totaling, zooms=None):
    '''
    :param totaling: pdf的頁(yè)數(shù)
    :param zooms: 值越大，分辨率越高，文件越清晰，列表內(nèi)兩個(gè)浮點(diǎn)數(shù)，每個(gè)尺寸的縮放系數(shù)，默認(rèn)為分辨率的2倍
    :return:
    '''
    if zooms is None:
        zooms = [2.0, 2.0]
    if os.path.exists('.pdf'):  # 臨時(shí)文件，需為空
        rmtree('.pdf')
    os.mkdir('.pdf')
    print(f"pdf頁(yè)數(shù)為 {totaling} \n創(chuàng)建臨時(shí)文件夾.....")
    for pg in range(totaling):
        page = doc[pg]
        print(f"\r{page}", end="")
        trans = fitz.Matrix(*zooms).preRotate(0)  # 0為旋轉(zhuǎn)角度
        pm = page.getPixmap(matrix=trans, alpha=False)
        lurl = '.pdf/%s.jpg' % str(pg + 1)
        pm.writePNG(lurl) #保存
    doc.close()

# 圖片合成pdf
def pic2pdf(obj, ratio, totaling):
    doc = fitz.open()
    compressor = Luban(quality=ratio)
    for pg in range(totaling):
        path = '.pdf/%s.jpg' % str(pg + 1)
        compressor.setPath(path)
        compressor.compress()
        print(f"\r 插入圖片 {pg + 1}/{totaling} 中......", end="")
        img = '.pdf/target/c_%s.jpg' % str(pg + 1)
        imgdoc = fitz.open(img)  # 打開(kāi)圖片
        pdfbytes = imgdoc.convertToPDF()  # 使用圖片創(chuàng)建單頁(yè)的 PDF
        os.remove(img)
        imgpdf = fitz.open("pdf", pdfbytes)
        doc.insertPDF(imgpdf)  # 將當(dāng)前頁(yè)插入文檔
    if os.path.exists(obj):  # 若pdf文件存在先刪除
        os.remove(obj)
    doc.save(obj)  # 保存pdf文件
    doc.close()


@runtime
def pdfz(doc, obj, ratio, totaling):
    covert2pic(doc, totaling)
    pic2pdf(obj, ratio, totaling)


def pic_quality():
    print("輸入壓縮等級(jí)1~3：")
    comp_level = input("壓縮等級(jí)（1=高畫(huà)質(zhì)50%，2=中畫(huà)質(zhì)70%，3=低畫(huà)質(zhì)80%）：（輸入數(shù)字并按回車(chē)鍵）")
    # 用字典模擬Switch分支，注意輸入的值是str類型
    ratio = {'1': 40, '2': 20, '3': 10}
    # 字典中沒(méi)有則默認(rèn) 低畫(huà)質(zhì)壓縮
    return ratio.get(comp_level, 10)


if __name__ == "__main__":
    print("請(qǐng)選擇需要壓縮的PDF文件")
    while True:
        '''打開(kāi)選擇文件夾對(duì)話框'''
        filepath = g.fileopenbox(title=u"選擇PDF", filetypes=['*.pdf'])
        if filepath == None:
            input("還未選擇文件，輸入任意鍵繼續(xù).......")
            continue
        else:
            filedir, filename = os.path.split(filepath)
            print(u'已選中文件【%s】' % (filename))
            if filename.endswith(".pdf") == False:
                input("選擇的文件類型不對(duì)，輸入任意鍵繼續(xù).......")
                continue
            ratio = pic_quality()
            obj = "new_" + filename
            doc = fitz.open(filepath)
            totaling = doc.pageCount

            pdfz(doc, obj, ratio, totaling)
            rmtree('.pdf')
            oldsize = os.stat(filepath).st_size
    
            newsize = os.stat(obj).st_size
            print('壓縮結(jié)果 %.2f M >>>> %.2f M'%(oldsize/(1024 * 1024),newsize/(1024 * 1024)))
            input(f"壓縮已完成，文件保存在改程序目錄下{filedir}，如需繼續(xù)壓縮請(qǐng)按任意鍵")

from multiprocessing import Process, Queue

# 提取成圖片
def covert2pic(filepath, qpaper, zooms=None):
    '''
    :param filepath: pdf文件的位置
    :param qpaper: 數(shù)據(jù)頁(yè)的隊(duì)列
    :param zooms: 值越大，分辨率越高，文件越清晰，列表內(nèi)兩個(gè)浮點(diǎn)數(shù)，每個(gè)尺寸的縮放系數(shù)，默認(rèn)為分辨率的2倍
    :return:
    '''
    doc = fitz.open(filepath)
    totaling = doc.pageCount
    if zooms is None:
        zooms = [2.0, 2.0]
    if path.exists('.pdf'):  # 臨時(shí)文件，需為空
        rmtree('.pdf')
    mkdir('.pdf')
    print(f"pdf頁(yè)數(shù)為 {totaling} \n創(chuàng)建臨時(shí)文件夾.....")
    for pg in range(totaling):
        page = doc[pg]
        print(f"\r{page}", end="")
        trans = fitz.Matrix(*zooms).preRotate(0)  # 0為旋轉(zhuǎn)角度
        pm = page.getPixmap(matrix=trans, alpha=False)
        lurl = '.pdf/%s.jpg' % str(pg + 1)
        pm.writePNG(lurl)  # 保存
        qpaper.put(pg)
    doc.close()

# 圖片合成pdf
def pic2pdf(obj, ratio, qpaper, totaling):
    doc2 = fitz.open()
    compressor = Luban(quality=ratio)
    for pg in range(totaling):
        picpath = '.pdf/%s.jpg' % str(pg + 1)
        compressor.setPath(picpath)
        while qpaper.empty():
            # 如果隊(duì)列為空，則循環(huán)等待
            pass
        qpaper.get()
        compressor.compress()
        print(f"\r 插入圖片 {pg + 1}/{totaling} 中......", end="")
        img = '.pdf/target/c_%s.jpg' % str(pg + 1)
        imgdoc = fitz.open(img)  # 打開(kāi)圖片
        pdfbytes = imgdoc.convertToPDF()  # 使用圖片創(chuàng)建單頁(yè)的 PDF
        remove(img)
        imgpdf = fitz.open("pdf", pdfbytes)
        doc2.insertPDF(imgpdf)  # 將當(dāng)前頁(yè)插入文檔
    if path.exists(obj):  # 若pdf文件存在先刪除
        remove(obj)
    doc2.save(obj)  # 保存pdf文件
    doc2.close()


@runtime
def pdfz(filepath, obj, ratio, totaling):
# 參數(shù)傳遞變?yōu)閒ilepath
    qpaper = Queue() # 創(chuàng)建隊(duì)列
    threads = []
    #read_thread = threading.Thread(target=covert2pic, args=(doc, totaling, qpaper))
    read_thread = Process(target=covert2pic, args=(filepath, qpaper))
    '''
    多進(jìn)程這里傳參數(shù)不一定成功，參數(shù)需要可以序列化才行，這里如果傳doc的變量，會(huì)報(bào)錯(cuò)WeakValueDictionary.__init__.<locals>.remove
    '''
    threads.append(read_thread)
    #write_thread = threading.Thread(target=pic2pdf, args=(obj, ratio, totaling, qpaper))
    write_thread = Process(target=pic2pdf, args=(obj, ratio, qpaper, totaling))
    threads.append(write_thread)

    for th in threads:
        th.start()  # 開(kāi)始執(zhí)行線程

    for th in threads:
        th.join()
    print("結(jié)束")

pip install PyMuPDF

import fitz
import os


def covert2pic(zoom):
    if os.path.exists('.pdf'):       # 臨時(shí)文件，需為空
         os.removedirs('.pdf')
    os.mkdir('.pdf')
    for pg in range(totaling):
        page = doc[pg]
        zoom = int(zoom)            #值越大，分辨率越高，文件越清晰
        rotate = int(0)
        print(page)
        trans = fitz.Matrix(zoom / 100.0, zoom / 100.0).preRotate(rotate)
        pm = page.getPixmap(matrix=trans, alpha=False)
      
        lurl='.pdf/%s.jpg' % str(pg+1)
        pm.writePNG(lurl)
    doc.close()

def pic2pdf(obj):
    doc = fitz.open()
    for pg in range(totaling):
        img = '.pdf/%s.jpg' % str(pg+1)
        imgdoc = fitz.open(img)                 # 打開(kāi)圖片
        pdfbytes = imgdoc.convertToPDF()        # 使用圖片創(chuàng)建單頁(yè)的 PDF
        os.remove(img)  
        imgpdf = fitz.open("pdf", pdfbytes)
        doc.insertPDF(imgpdf)                   # 將當(dāng)前頁(yè)插入文檔
    if os.path.exists(obj):         # 若文件存在先刪除
        os.remove(obj)
    doc.save(obj)                   # 保存pdf文件
    doc.close()


def pdfz(sor, obj, zoom):    
    covert2pic(zoom)
    pic2pdf(obj)
    
if __name__  == "__main__":

    sor = "source.pdf"              # 需要壓縮的PDF文件
    obj = "new" + sor
    doc = fitz.open(sor) 
    totaling = doc.pageCount
    
    zoom = 200                     # 清晰度調(diào)節(jié)，縮放比率
    pdfz(sor, obj, zoom)
    os.removedirs('.pdf')

pip install fitz
pip install pymupdf

lenXREF = doc.xref_length()
text = doc.xref_object(i)  # 定義對(duì)象字符串

im = im.resize((1376, y_s), Image.ANTIALIAS)
im.save(pic_path_d, quality=comp_ratio)

import fitz
import re
import os
from PIL import Image
from tkinter import filedialog


def pdf2pic(path, pic_path, comp_ratio):
    checkXO = r"/Type(?= */XObject)"  # 使用正則表達(dá)式來(lái)查找圖片
    checkIM = r"/Subtype(?= */Image)"
    doc = fitz.open(path)  # 打開(kāi)pdf文件
    imgcount = 0  # 圖片計(jì)數(shù)
    lenXREF = doc.xref_length()    # 獲取對(duì)象數(shù)量長(zhǎng)度

    # 打印PDF的信息
    print("文件名:{}, 頁(yè)數(shù): {}, 對(duì)象: {}".format(path, len(doc), lenXREF - 1))

    # 遍歷每一個(gè)對(duì)象
    for i in range(1, lenXREF):
        text = doc.xref_object(i)  # 定義對(duì)象字符串
        isXObject = re.search(checkXO, text)  # 使用正則表達(dá)式查看是否是對(duì)象
        isImage = re.search(checkIM, text)  # 使用正則表達(dá)式查看是否是圖片
        if not isXObject or not isImage:  # 如果不是對(duì)象也不是圖片，則continue
            continue
        imgcount += 1
        pix = fitz.Pixmap(doc, i)  # 生成圖像對(duì)象
        new_name = "pic{}.jpg".format(imgcount)  # 生成圖片的名稱
        print(new_name)
        if pix.n < 5:  # 如果pix.n<5,可以直接存為PNG
            pic_path_d = os.path.join(pic_path, new_name)
            pix.writeImage(os.path.join(pic_path, new_name))
            im = Image.open(pic_path_d)
            x, y = im.size
            if x > 1376:
                y_s = int(y * 1376 / x)
                im = im.resize((1376, y_s), Image.ANTIALIAS)

            im.save(pic_path_d, quality=comp_ratio)
        else:  # 否則先轉(zhuǎn)換CMYK
            pix0 = fitz.Pixmap(fitz.csRGB, pix)
            pix0.writeImage(os.path.join(pic_path, new_name))
            pix0 = None

        pix = None  # 釋放資源
        print("提取了{(lán)}張圖片".format(imgcount))

    os.startfile(pic_path)

def rea(path, pdf_name):
    file_list = os.listdir(path)
    pic_name = []
    im_list = []
    for x in file_list:
        if "jpg" in x or 'png' in x or 'jpeg' in x:
            pic_name.append(x)

    pic_name.sort()
    new_pic = []

    for x in pic_name:
        if "jpg" in x:
            new_pic.append(x)

    for x in pic_name:
        if "png" in x:
            new_pic.append(x)

    print("hec", new_pic)

    im1 = Image.open(os.path.join(path, new_pic[0]))
    new_pic.pop(0)
    for i in new_pic:
        img = Image.open(os.path.join(path, i))
        # im_list.append(Image.open(i))
        if img.mode == "RGBA":
            img = img.convert('RGB')
            im_list.append(img)
        else:
            im_list.append(img)
    im1.save(pdf_name, "PDF", resolution=100.0, save_all=True, append_images=im_list)
    print("輸出文件名稱：", pdf_name)


def pdf_out():
    print('功能完善中')

if __name__ == '__main__':
    print("Hello world!請(qǐng)先輸入壓縮等級(jí)1~3，然后在彈出的對(duì)話框中選擇需要壓縮的文件")
    comp_level = input("壓縮等級(jí)（1=高畫(huà)質(zhì)，2=中畫(huà)質(zhì)，3=低畫(huà)質(zhì)）：（輸入數(shù)字并按回車(chē)鍵）")
    ratio = 10
    if comp_level == "1":
        ratio = 20
    elif comp_level == "2":
        ratio = 10
    elif comp_level == "3":
        ratio = 5
    '''打開(kāi)選擇文件夾對(duì)話框'''
    filepath = filedialog.askopenfilename()  # 獲得選擇好的文件
    print('選擇的PDF地址:', filepath)
    if os.path.exists("./pdf_output"):
        pass
    else:
        os.mkdir("./pdf_output")
    pic_path = str(os.getcwd()) + "\pdf_output"
    print('提取圖片的輸出地址：', pic_path )
    pdf2pic(filepath, pic_path, comp_ratio=ratio)

    pdf_name = 'Compressed.pdf'
    if ".pdf" in pdf_name:
        rea(pic_path, pdf_name=pdf_name)
    else:
        rea(pic_path, pdf_name="{}.pdf".format(pdf_name))

    print("壓縮完成，請(qǐng)關(guān)閉窗口。若壓縮等級(jí)不合適，請(qǐng)先刪除圖片和文件并重新打開(kāi)程序。")