检查word里参考文献引用依出现先后顺序排列（半自动化）

1、word导出为txt

另存为纯文本txt格式，命名为paper_old.txt，选择这个配置

举例，得到如下的txt内容：

　　xxxxxxxxxxxxxxxxx[5]xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
　　xxxxxxx[3-4]xxxx
　　
　　xxxxxxxxxxxasdasdasdxxx[1]
　　
参考文献开始
[1] Lee B J, Lee J, Kim K E. Representation Balancing Offline Model-based Reinforcement Learning[C]//9th International Conference on Learning Representations, 2021, Virtual Event, Austria, May 3-7, 2021. 2021.
[2] 鲍军威. 基于知识的自动问答与问题生成的研究[D]. 哈尔滨工业大学, 2019.
[3] 乔振浩, 车万翔. 基于知识图谱的问答系统研究与应用[D]. 哈尔滨工业大学, 2020.
[4] 向伟富, 段贵多. 面向问答系统的知识图谱推理算法的研究与实现[D]. 电子科技大学, 2020.
[5] 刘良. 基于领域知识图谱的智能问答关键技术研究[D]. 电子科技大学, 2020.

注：“参考文献开始”那一行是手动加上去的，以区分前文的引用和后文的引文。

可见编号有两种类型：[x]或者[x-y]。

2、引文顺序检查

python文件citation_sort.py

输出：

1. quotation_old2new_id.txt：旧引文编号->新引文编号，末尾再跟上未引用的旧引文编号
2. quotation_new.txt：更新后的带编号新引文序列
3. paper_new.txt：更新后的带引用正文+新版参考文献

注：paper_old.txt、citation_sort.py、quotation_old2new_id.txt、quotation_new.txt均在相同目录

import re
from ordered_set import OrderedSet
import collections

old_filename = 'paper_old.txt'
quotation_old2new_id_filename = 'quotation_old2new_id.txt'
quotation_new_filename = 'quotation_new.txt'
new_filename = 'paper_new.txt'

with open(old_filename, 'r', encoding='utf-8') as f:
    file_lines = f.readlines()

quotation_start = False
cite_history = []
quotation_dict = collections.OrderedDict()

for line in file_lines:
    line = line.strip('\n')
    if line == '参考文献开始':
        quotation_start = True
    else:
        if quotation_start == False:
            # (\[\d+\]|\[\d+\-\d+\])
            matchCite = re.findall(r'(\[\d+\]|\[\d+\-\d+\])', line)
            if matchCite:
                # print(matchCite)
                for cite in matchCite:
                    cite = cite.replace('[', '').replace(']', '')
                    if '-' in cite:
                        start, end = cite.split('-')
                        start, end = int(start), int(end)
                        cite_temp = list(range(start, end+1))
                        cite_history.extend(cite_temp)
                    else:
                        cite = int(cite)
                        cite_history.append(cite)
        else:
            matchQuotation = re.search(r'(\[\d+\])\s(.+)', line)
            id_old = int(matchQuotation.group(
                1).replace('[', '').replace(']', ''))
            quotation = matchQuotation.group(2)
            quotation_dict[id_old] = quotation

# print(cite_history)
cite_history_OrderedSet = OrderedSet(cite_history)
# print(cite_history_OrderedSet)
# print(quotation_dict)

quotation_full = OrderedSet(quotation_dict.keys())
quotation_useless = quotation_full-cite_history_OrderedSet
print('未使用的参考文献:', list(quotation_useless))

# print()
quotation_old2new = collections.OrderedDict(
    zip(range(1, quotation_full[-1]+1), [-1]*5))
# print(quotation_old2new)
count = 1
for id_old in cite_history_OrderedSet:
    quotation_old2new[id_old] = count
    count += 1
for id_old in quotation_useless:
    quotation_old2new[id_old] = count
    count += 1

# quotation_old2new_id_filename = 'quotation_old2new_id.txt'
# quotation_new_filename = 'quotation_new.txt'

with open(quotation_old2new_id_filename, 'w+', encoding='utf-8') as f:
    for old_id, new_id in quotation_old2new.items():
        f.write('%d->%d\n' % (old_id, new_id))
    f.write('未使用的参考文献:' + str(list(quotation_useless)))

# quotation_new2old_temp = dict(
#     zip(quotation_old2new.values(), quotation_old2new.keys()))
quotation_new2old = collections.OrderedDict()

for key, value in quotation_old2new.items():
    quotation_new2old[value] = key
# print(quotation_new2old)

with open(quotation_new_filename, 'w+', encoding='utf-8') as f:
    count = 1
    for old_id, new_id in quotation_new2old.items():
        if count == quotation_old2new[quotation_useless[0]]:
            f.write('*******以下为未引用文献********\n')
        f.write('[%d]\t%s\n' %
                (count, quotation_dict[quotation_new2old[new_id]]))
        count += 1


# 重新编排原文的正文引用
quotation_start = False
with open(new_filename, 'w+', encoding='utf-8') as f:
    for line in file_lines:
        # line = line.strip('\n')
        if line == '参考文献开始\n':
            quotation_start = True
            f.write(line)
        else:
            if quotation_start == False:
                # (\[\d+\]|\[\d+\-\d+\])
                matchCite = re.findall(r'(\[\d+\]|\[\d+\-\d+\])', line)
                if matchCite:
                    # print(matchCite)
                    for cite in matchCite:
                        cite = cite.replace('[', '').replace(']', '')
                        if '-' in cite:
                            start, end = cite.split('-')
                            start, end = int(start), int(end)
                            start_new, end_new = quotation_old2new[start], quotation_old2new[end]
                            line = line.replace('[%d-%d]' % (start, end),
                                                '[%d-%d]' % (start_new, end_new))
                            # cite_temp = list(range(start, end+1))
                            # cite_history.extend(cite_temp)
                        else:
                            cite = int(cite)
                            cite_new = quotation_old2new[cite]
                            line = line.replace('[%d]' %
                                                (cite), '[%d]' % (cite_new))
                            # cite_history.append(cite)
                f.write(line)
            else:
                break
                # matchQuotation = re.search(r'(\[\d+\])\s(.+)', line)
                # id_old = int(matchQuotation.group(
                #     1).replace('[', '').replace(']', ''))
                # id_new = quotation_old2new[id_old]
                # line=line.replace('[%d]' % (id_new), '[%d]' % (id_old))

                # quotation = matchQuotation.group(2)
                # quotation_dict[id_old] = quotation
    count = 1
    for old_id, new_id in quotation_new2old.items():
        if count == quotation_old2new[quotation_useless[0]]:
            f.write('*******以下为未引用文献********\n')
        f.write('[%d]\t%s\n' %
                (count, quotation_dict[quotation_new2old[new_id]]))
        count += 1

执行

python citation_sort.py

得到quotation_old2new_id.txt：旧引文编号->新引文编号，末尾再跟上未引用的旧引文编号

1->4
2->5
3->2
4->3
5->1
未使用的参考文献:[2]

以及quotation_new.txt：更新后的带编号新引文序列

[1] 刘良. 基于领域知识图谱的智能问答关键技术研究[D]. 电子科技大学, 2020.
[2] 乔振浩, 车万翔. 基于知识图谱的问答系统研究与应用[D]. 哈尔滨工业大学, 2020.
[3] 向伟富, 段贵多. 面向问答系统的知识图谱推理算法的研究与实现[D]. 电子科技大学, 2020.
[4] Lee B J, Lee J, Kim K E. Representation Balancing Offline Model-based Reinforcement Learning[C]//9th International Conference on Learning Representations, 2021, Virtual Event, Austria, May 3-7, 2021. 2021.
*******以下为未引用文献********
[5] 鲍军威. 基于知识的自动问答与问题生成的研究[D]. 哈尔滨工业大学, 2019.

注释：“*******以下为未引用文献********”行为自动添加的额外行，为区分已用和未用的参考文献。

正文paper_new.txt长这样：

xxxxxxxxxxxxxxxxx[1]xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
　　xxxxxxx[2-3]xxxx
　　
　　xxxxxxxxxxxasdasdasdxxx[4]
　　
参考文献开始
[1]	刘良. 基于领域知识图谱的智能问答关键技术研究[D]. 电子科技大学, 2020.
[2]	乔振浩, 车万翔. 基于知识图谱的问答系统研究与应用[D]. 哈尔滨工业大学, 2020.
[3]	向伟富, 段贵多. 面向问答系统的知识图谱推理算法的研究与实现[D]. 电子科技大学, 2020.
[4]	Lee B J, Lee J, Kim K E. Representation Balancing Offline Model-based Reinforcement Learning[C]//9th International Conference on Learning Representations, 2021, Virtual Event, Austria, May 3-7, 2021. 2021.
*******以下为未引用文献********
[5]	鲍军威. 基于知识的自动问答与问题生成的研究[D]. 哈尔滨工业大学, 2019.

可以作为修改为，检查引用和引文顺序是否修改正确的，辅助对照工具。

3、修改word的引文

方法：依照quotation_old2new_id.txt的方式，将箭头左边的编号，完整ctrl x剪贴后，放在右边对应位置。再ctrl a全选，f9刷新全文的引用编号。如图

注：如果要在第一个前面插文献x，建议先在1下面回车新建个空的2，把x先剪切到2，再把1剪切到2的下面，就行了。不然直接在[1]后面回车，会把原来的引文1干成两段，f9之后会出错的——原来指向文献x的就指向了文献1。。。。

最后，复制引文部分，到文件quotation_new_self.txt。

再和quotation_new.txt校验一遍是否相同即可（以下为未引用文献 行就不管了），比如用vscode打开文件夹，先右键quotation_new_self.txt“选择以进行比较”，再右键quotation_new.txt“与已选择的进行比较”。

除了我们自己加的“*******以下为未引用文献********”行，其他的都完全一致。

说明我们都改对了，游戏结束。