word_parse_of_eng/paper_upload_parse/replace_pic_new.py

import re
from pprint import pprint
from util import del_outlier

pat1 = re.compile(r"_*\s*[(（]\s*(\d+)\s*[)）]\s*_*")
pat11 = re.compile(r"_*\s*[(（]?\s*(\d+)\s*[)）]?\s*[.、．]?\s*_+")
pat111 = re.compile(r"_+\s*[(（]?\s*(\d+)\s*[)）]?\s*_*")
pat1111 = re.compile(r"_+\s*[(（]?\s*(\d+)\s*[)）]?\s*[.、．]?\s*_+")

pat3 = re.compile(r"_*\s*[(（]\s*\d+\s*[)）]\s*_*")
pat33 = re.compile(r"_*\s*[(（]?\s*\d+\s*[)）]?\s*[.、．]?\s*_+")
pat333 = re.compile(r"_+\s*[(（]?\s*\d+\s*[)）]?\s*_*")
pat3333 = re.compile(r"_+\s*[(（]?\s*\d+\s*[)）]?\s*[.、．]?\s*_+")

pat2 = re.compile(r'<span style="text-decoration.*?>(\s*\d+)?|<span style="border-bottom:.*?>(\s*\d+)?')
pat4 = re.compile(
    r'(_*\s*<img.*?>\s*_*)|(_*\s*<span style="text-decoration.*?>.*?</span>\s*_*)|(_*\s*<span style="border-bottom:.*?>.*?</span>\s*_*)')
pat5 = re.compile(r"(<img.*?>)|[(（]?\s*\d+\s*[)）]?\s*[.、．]")
pat6 = re.compile(r"[(（]?\s*(\d+)\s*[)）]?\s*[.、．]")
pat7 = re.compile(r"[(（]?\s*\d+\s*[)）]?\s*[.、．]")


def pics(a, b, n=False):
    if n:
        t = "\n"
    else:
        t = ""
    replace = t + '''<img src="http://zsytk2.zhixinhuixue.com/static/images/{}.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/{}.gif" class="tiankong" data-num="{}">'''.format(
        a, a, b)
    return replace


def replace_pics1(content):
    if not content:
        return content
    old_content = content
    if "http://zsytk2.zhixinhuixue.com/static/images" in str(content):
        content = pat4.sub("#####", str(content))
        content = re.sub(r"_+\s*[(（]?\s*(\d+)\s*[)）]?\s*[.、．]?\s*_+", "#####", content)
        res = []
        for sss in range(0, 20):
            new_content1 = re.sub(r"#####", pics(sss + 1, sss), content, count=1, flags=re.S)
            content = new_content1
            res.append(new_content1)
            if "#####" not in new_content1:
                break
        if res:
            if "#####" in str(res[-1]):
                res_final = str(res[-1]).replace("#####", " ")
                return res_final
            else:
                return res[-1]
        else:
            return old_content
    else:
        if len(pat2.findall(content)):
            content = pat4.sub("#####", content)
            content = re.sub(r"_+\s*[(（]?\s*(\d+)\s*[)）]?\s*[.、．]?\s*_+", "#####", content)
            res = []
            for i in range(0, 20):
                new_content2 = re.sub(r"#####", pics(i + 1, i), content, count=1, flags=re.S)
                content = new_content2
                res.append(new_content2)
                if "#####" not in new_content2:
                    break
            if res:
                if "#####" in str(res[-1]):
                    res_final = str(res[-1]).replace("#####", " ")
                    return res_final
                else:
                    return res[-1]
            else:
                return old_content
        else:

            if "<br>" in old_content:
                old_content = str(old_content).split("<br>")
            elif "<br/>" in old_content:
                old_content = str(old_content).split("<br/>")
            elif "<br />" in old_content:
                old_content = str(old_content).split("<br />")
            else:
                old_content = str(old_content).split("\n")

            try:
                # start_index = [1 if len(re.findall(r'[\u4e00-\u9fa5]', con.strip())) > 5 else 0 for con in old_content[:6]].index(0)
                start_index = [
                    1 if len(re.findall(r'[\u4e00-\u9fa5]', con.strip())) / (len(con.strip()) + 0.0001) >= 0.5 else 0
                    for con in old_content[:3]].index(0)
            except:
                start_index = 0
            old_content = "\n".join(old_content[start_index:])

            replace_con = pat3.findall(old_content) + pat33.findall(old_content) + pat333.findall(
                old_content) + pat3333.findall(old_content)
            replace_con = list(set(replace_con))
            if replace_con:
                num_list = pat1.findall(old_content) + pat11.findall(old_content) + pat111.findall(
                    old_content) + pat1111.findall(old_content)
                new_num_list = del_outlier(num_list)
                new_replace_con_list = []
                for num in replace_con:
                    if re.search(pat1, str(num).strip()):
                        if re.search(pat1, str(num).strip()).group(1) in new_num_list:
                            new_replace_con_list.append(str(num).strip())
                            content = content.replace(str(num).strip(), "#####")

                    if re.search(pat11, str(num).strip()):
                        if re.search(pat11, str(num).strip()).group(1) in new_num_list:
                            new_replace_con_list.append(str(num).strip())
                            content = content.replace(str(num).strip(), "#####")

                    if re.search(pat111, str(num).strip()):
                        if re.search(pat111, str(num).strip()).group(1) in new_num_list:
                            new_replace_con_list.append(str(num).strip())
                            content = content.replace(str(num).strip(), "#####")

                    if re.search(pat1111, str(num).strip()):
                        if re.search(pat1111, str(num).strip()).group(1) in new_num_list:
                            new_replace_con_list.append(str(num).strip())
                            content = content.replace(str(num).strip(), "#####")

                res = []
                for i, new_replace_con in enumerate(new_replace_con_list):
                    new_content3 = re.sub(r"#####", pics(i + 1, i), content, count=1, flags=re.S)
                    content = new_content3
                    res.append(new_content3)
                if res:
                    return res[-1]
                else:
                    return old_content
            else:
                return old_content


def replace_pics2(content):
    if not content:
        return content
    old_content = content
    if "http://zsytk2.zhixinhuixue.com/static/images" in content:
        content = pat5.sub("#####", content)
        res = []
        for ii, sss in enumerate(range(len(pat5.findall(old_content)))):
            new_content1 = re.sub(r"#####", pics(sss + 1, sss, n=True), content, count=1, flags=re.S)
            content = new_content1
            res.append(new_content1)
        if res:
            return res[-1]
        else:
            return old_content
    else:
        replace_con = pat7.findall(old_content)
        num_list = pat6.findall(old_content)
        new_num_list = del_outlier(num_list)
        new_replace_con_list = []
        for num in replace_con:
            if pat6.search(str(num).strip()).group(1) in new_num_list:
                new_replace_con_list.append(str(num).strip())
                content = content.replace(str(num).strip(), "#####")
        res = []
        for i, new_replace_con in enumerate(new_replace_con_list):
            new_content2 = re.sub(r"#####", pics(i + 1, i), content, count=1, flags=re.S)
            content = new_content2
            res.append(new_content2)
        if res:
            return res[-1]
        else:
            return old_content


def content_replace_result(types, contents):
    content = contents.get("content", None)
    if types in ["完形填空", "任务型阅读", '七选五', "选词填空", '语法填空', '课文填空']:
        # if types in ["完形填空", "任务型阅读", '七选五', "选词填空", '语法填空']:
        return replace_pics1(content)
    elif types in ["句子翻译", "完成句子", "阅读表达"]:
        return replace_pics2(content)
    else:
        return content


if __name__ == '__main__':
    # a = ['阅读下面短文,\xa0从短文后各题所给的四个选项(A、B、C\xa0和\xa0D)中,\xa0选出可以填人\xa0空白处的最佳选项，并在答题卡上将该项涂黑。\nThe new science of spending comes to a surprising conclusion. How we use\xa0our money may \n<img src="http://zsytk2.zhixinhuixue.com/static/images/1.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/1.gif" class="tiankong" data-num="0"/>21 as much or more than how much we’ve got it. Money spent\xa0on experiences, \xa0\xa0\n<img src="http://zsytk2.zhixinhuixue.com/static/images/2.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/2.gif" class="tiankong" data-num="1"/>22 \xa0\xa0\xa0material goods, brings about more\xa0happiness.\nImagine that you wake up tomorrow morning to \xa0\n<img src="http://zsytk2.zhixinhuixue.com/static/images/3.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/3.gif" class="tiankong" data-num="2"/>23 \xa0＄1 million under\xa0your\nmattress. What would you do \n<img src="http://zsytk2.zhixinhuixue.com/static/images/4.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/4.gif" class="tiankong" data-num="3"/>24 that cash? The new-found wealth will probably\xa0make\xa0you think about one thing \n<img src="http://zsytk2.zhixinhuixue.com/static/images/5.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/5.gif" class="tiankong" data-num="4"/>25 all else—yourself. A growing body of research\xa0reveals\xa0that money \n<img src="http://zsytk2.zhixinhuixue.com/static/images/6.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/6.gif" class="tiankong" data-num="5"/>26 our selfish sides, focusing us on \n<img src="http://zsytk2.zhixinhuixue.com/static/images/7.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/7.gif" class="tiankong" data-num="6"/>27 that money can do for us, and\xa0us\xa0alone. Perhaps you are imagining buying new \xa0\xa0\n<img src="http://zsytk2.zhixinhuixue.com/static/images/8.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/8.gif" class="tiankong" data-num="7"/>28 \xa0\xa0: a faster car, or even a\xa0larger\xa0house.\nHowever, studies show that material goods often \xa0\n<img src="http://zsytk2.zhixinhuixue.com/static/images/9.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/9.gif" class="tiankong" data-num="8"/>29 \xa0deliver lasting\xa0happiness.\nFortunately, our ongoing research \n<img src="http://zsytk2.zhixinhuixue.com/static/images/10.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/10.gif" class="tiankong" data-num="9"/>30 various ways to get more happiness from\xa0every dollar you spend. Changing how you spend money can lead to measurable increases\xa0in your happiness. But making these \n<img src="http://zsytk2.zhixinhuixue.com/static/images/11.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/11.gif" class="tiankong" data-num="10"/>31 requires challenging some of our\xa0fundamental\xa0assumptions about spending, since it’s\xa0\n<img src="http://zsytk2.zhixinhuixue.com/static/images/12.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/12.gif" class="tiankong" data-num="11"/>32 not to view buying a house as a\xa0wise\xa0investment. But new research shows it has little impact on your overall happiness.\xa0A\xa0study in the United States found that homeowners, on average, were no happier\xa0than\n\n<img src="http://zsytk2.zhixinhuixue.com/static/images/13.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/13.gif" class="tiankong" data-num="12"/>\xa0\xa0\n<img src="http://zsytk2.zhixinhuixue.com/static/images/14.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/14.gif" class="tiankong" data-num="13"/>33 \xa0\xa0.\nThus, diligently saving up for a down payment might not be such a good idea\xa0\n<img src="http://zsytk2.zhixinhuixue.com/static/images/15.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/15.gif" class="tiankong" data-num="14"/>34 \xa0it means skipping after-work beers with friends or your annual celebration at a\xa0favorite\xa0restaurant. Dozens of studies show that people get more happiness from buying\xa0\n<img src="http://zsytk2.zhixinhuixue.com/static/images/16.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/16.gif" class="tiankong" data-num="15"/>35 \xa0than material things. Experiential purchases—such as trips, concerts and special\xa0meals\n—are specially \xa0\n<img src="http://zsytk2.zhixinhuixue.com/static/images/17.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/17.gif" class="tiankong" data-num="16"/>36 \xa0to our sense of self, making us who we\xa0are.\nExperiences come with one more \n<img src="http://zsytk2.zhixinhuixue.com/static/images/18.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/18.gif" class="tiankong" data-num="17"/>37 : They are likely to make us \n<img src="http://zsytk2.zhixinhuixue.com/static/images/19.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/19.gif" class="tiankong" data-num="18"/>38 to\xa0other\xa0people, while more often material things are enjoyed \n<img src="http://zsytk2.zhixinhuixue.com/static/images/20.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/20.gif" class="tiankong" data-num="19"/>39 . The importance of\xa0social\xa0contact for \xa0\xa0\n<img src="http://zsytk2.zhixinhuixue.com/static/images/21.gif" _src="http://zsytk2.zhixinhuixue.com/static/images/21.gif" class="tiankong" data-num="20"/>40 \xa0\xa0mental and physical health has been stressed for\xa0decades.\n', '1.\nA.present\nB.matter\nC.appear\nD.equal\n', '2.\nA.better\xa0than\nB.other\xa0than\nC.rather\xa0than\nD.more\xa0than\n', '3.\nA.keep\nB.own\nC.discover\nD.accept\n', '4.\nA.to\nB.with\nC.about\nD.for\n', '5.\nA.above\nB.beyond\nC.except\nD.after\n', '6.\nA.draws\xa0out\nB.takes\xa0in\nC.stands\xa0for\nD.makes\xa0up\n', '7.\nA.who\nB.why\nC.what\nD.how\n', '8.\nA.life\nB.equipment\nC.vehicles\nD.possessions\n', '9.\nA.manage\xa0to\nB.fail\xa0to\nC.tend\xa0to\nD.expect\xa0to\n', '10.\nA.invites\nB.offers\nC.prefers\nD.follows\n', '11.\nA.changes\nB.plans\nC.decisions\nD.mistakes\n', '12.\nA.easy\nB.hard\nC.foolish\nD.sensible\n', '13.\nA.buyers\nB.sellers\nC.builders\nD.renters\n', '14.\nA.if\nB.until\nC.unless\nD.though\n', '15.\nA.experiences\nB.goods\nC.cars\nD.houses\n', '16.\nA.exposed\nB.added\nC.applied\nD.related\n', '17.\nA.disadvantage\nB.conclusion\nC.purpose\nD.benefit\n', '18.\nA.grateful\nB.close\nC.similar\nD.equal\n', '19.\nA.together\nB.worldwide\nC.long\nD.alone\n', '20.\nA.increasing\nB.damaging\nC.improving\nD.recovering\n']

    a = ['''It is always interesting and revealing (发人深省的) to observe human behavior. In the past years, we have always held a party for Christmas gift __16___ at our house with at least ten members of our __17__ . Each year, we all buy gifts rapped (包裹) in Christmas paper. When everyone is sitting around a circle at home, all the gifts are __18__ in the middle of the floor.



People choose a __19__ from a bowl that tells him or her when it is their __20__ to choose a gift. The person with the number one can choose any gift and then, each person __21__ his or her turn. Before they choose a gift, they have the___22___to take the gift from the former person. __23__ when all people have chosen a gift, the person __24__ the first number can then take gift away from any other person.



What is always interesting is __25__ each person as they choose a gift. What happens next usually happens more often than not. The biggest gift __26__ gets chosen first. It is never the __27__ gift, nor is it the most expensive one. What does this tell us about most human __28__ ?



Is the __29__ always the best or worth the most? Do we __30__ that something small is not worth as much or cannot live up to what we expect?



Human beings, in fact, __31__ all sorts of __32__ and packagings. Some are bigger and some are smaller; some are dressed __33__ and some are dressed poorly. Behind these lie __34__ characteristics, talents, desires, hurts and pains and dreams.



Terry Josephson, a motivational (励志的) speaker says, “ __35__ thinking in terms of limitations and start thinking in terms of possibilities.”''']

    # pprint(a)
    # b = replace_pics1(",".join(a))
    b = replace_pics1(a[0])
    print(b)