在图像或 pdf 文档中将 pandas 数据帧保存为 table，具有良好的多索引显示

Question

我正在尝试在 pdf 格式的报告中包含一个具有多索引的数据框。我想要一个不错的 table 输出。

我找到了这两个解决方案：

pandas.df -> HTML -> pdf

    import pandas as pd
    from IPython.display import HTML
    import pdfkit

    # df generation
    df = pd.read_csv(path_to_csv, sep =',')
    groupeddf = df.groupby('Cluster')
    res = groupeddf.describe([0.05, 0.5, 0.95])
    res.index.rename(['Cluster', 'stats'], inplace=True)

    res['Cluster'] = res.index.get_level_values('Cluster')
    res['stats'] = res.index.get_level_values('stats')
    populations = (res.iloc[(res.index.get_level_values('stats') == 'count'), \
                                                            0].values).tolist()
    res['population'] = [populations[i] for i in res.index.labels[0].values()]
    total_pop = sum(populations)
    res['frequency'] =(res['population']/total_pop).round(3)
    res.set_index(['Cluster', 'population','frequency', 'stats'], inplace=True)



    res1 = res.iloc[(res.index.get_level_values('stats') == '5%') |
    (res.index.get_level_values('stats') == 'mean') |
    (res.index.get_level_values('stats') == '50%') |
    (res.index.get_level_values('stats') == '95%')]
    res1 = res1.round(2)
    # saving the df     
    h = HTML(res1.to_html())
    my_file = open('test.html', 'w')
    my_file.write(h.data)
    my_file.close()


    options = {
        'orientation': 'Landscape'
        }
    with open('test.html') as f:
        pdfkit.from_file(f, 'out.pdf', options=options)

但这对pdfkit有依赖性，这对我们来说很困难。这就是为什么我要尝试使用 pandas.df -> tex -> pdf（如 Export a Pandas dataframe as a table image 中所述）

    import pandas as pd
    import os
    # df generation              
    df = pd.read_csv(path_to_csv, sep =',')
    groupeddf = df.groupby('Cluster')
    res = groupeddf.describe([0.05, 0.5, 0.95])
    res.index.rename(['Cluster', 'stats'], inplace=True)

    res['Cluster'] = res.index.get_level_values('Cluster')
    res['stats'] = res.index.get_level_values('stats')
    populations = (res.iloc[(res.index.get_level_values('stats') == 'count'), \
                                                            0].values).tolist()
    res['population'] = [populations[i] for i in res.index.labels[0].values()]
    total_pop = sum(populations)
    res['frequency'] =(res['population']/total_pop).round(3)
    res.set_index(['Cluster', 'population','frequency', 'stats'], inplace=True)



    res1 = res.iloc[(res.index.get_level_values('stats') == '5%') |
    (res.index.get_level_values('stats') == 'mean') |
    (res.index.get_level_values('stats') == '50%') |
    (res.index.get_level_values('stats') == '95%')]
    res1 = res1.round(2)
    res1.rename(columns=lambda x: x.replace('_', ' '), inplace=True)    

    #latex
    template = r'''\documentclass[preview]{{standalone}}
    \usepackage{{booktabs}}
    \begin{{document}}
    {}
    \end{{document}}
    '''

    with open("outputfile.tex", "wb") as afile: 
        afile.write(template.format(res1.to_latex()))
    os.system("pdflatex outputfile.tex")

然而，我不熟悉乳胶，我得到这个错误：

  ! LaTeX Error: File `standalone.cls' not found.

 Type X to quit or <RETURN> to proceed,
 or enter a new name. (Default extension: cls)

关于错误或标准方法的任何想法 pandas.df -> pdf ？

Answer 1

好吧，一种方法是使用 markdown。您可以使用 df.to_html()。这会将数据帧转换为 html table。从那里您可以将生成的 html 放入 markdown 文件 (.md) 并使用包将 markdown 转换为 pdf。 https://www.npmjs.com/package/markdown-pdf

这是一个不错的选择吗？

Answer 2

适合我的解决方案： pandas >= 0.17 我安装了pdflatex。我复制了 booktabs.sty、geography.sty 和 pdflscape.sty

等乳胶包

import pandas as pd
import os
import math

def save_summary_table_as_pdf(path_to_csv, path_to_output_folder):
    pwd = os.getcwd()
    df = pd.read_csv(path_to_csv, sep =',')

    #data preparation
    groupeddf = df.groupby('Cluster')
    res = groupeddf.describe([0.05, 0.5, 0.95])
    res.index.rename(['Cluster', 'Stats'], inplace=True)

    res['cluster'] = res.index.get_level_values('Cluster')
    res['stats'] = res.index.get_level_values('Stats')
    populations = (res.iloc[(res.index.get_level_values('Stats') == 'count'), \
                                                            0].values).tolist()
    res['population'] = [populations[i] for i in res.index.labels[0].values()]
    total_pop = sum(populations)
    res['frequency'] =(res['population']/total_pop).round(3)
    res.set_index(['cluster', 'population','frequency', 'stats'], inplace=True)
    res1 = res.iloc[(res.index.get_level_values('stats') == '5%') |
    (res.index.get_level_values('stats') == 'mean') |
    (res.index.get_level_values('stats') == '50%') |
    (res.index.get_level_values('stats') == '95%')]
    res1 = res1.round(2)
    res1.rename(columns=lambda x: x.replace('_', ' '), inplace=True)  

    #latex
    nbpages = int(math.ceil(res1.shape[0]*1.0/40))

    templatetop = r'''\documentclass[a3paper, 5pt]{article}
    \usepackage{booktabs}
    \usepackage{pdflscape}
    \usepackage[a4paper,bindingoffset=0.2in,%
            left=0.25in,right=0.25in,top=1in,bottom=1in,%
            footskip=.25in]{geometry}
    \begin{document}
    \begin{landscape}
    \pagenumbering{gobble}
    \oddsidemargin = 0pt
    \hoffset = -0.25in
    \topmargin = 1pt
    \headheight = 0pt
    \headsep = 0pt
    '''
    templatebottom = '''
    \end{landscape}
    \end{document}
    '''
    output_folder_path_abs = path_to_output_folder
    output_tex = os.path.join(output_folder_path_abs, 
    "clustering_summary_table.tex")

    with open(output_tex, "wb") as afile: 
        afile.write(templatetop +'\n')
        for i in range(0, nbpages):
            afile.write(res1.iloc[(i*40):((i+1)*40), :].to_latex() +'\n' + 
                                                """\pagenumbering{gobble}""")
        afile.write(templatebottom +'\n')
    os.chdir(output_folder_path_abs)
    os.system('pdflatex clustering_summary_table.tex')
    os.chdir(pwd)
    os.remove(output_tex)
    os.remove(os.path.join(path_to_output_folder, 
                                           'clustering_summary_table.aux'))
    os.remove(os.path.join(path_to_output_folder, 
                                           'clustering_summary_table.log'))

if __name__ == "__main__":
    print 'begin generate pdf table about clustering'
    import argparse
    parser = argparse.ArgumentParser()
    parser.add_argument("path_to_csv")
    parser.add_argument("outputfolder")
    args = vars(parser.parse_args())
    filedir = os.path.abspath(os.path.dirname(__file__))
    output_folder_path_abs = os.path.abspath(args['outputfolder'])
    input_folder_path_abs = os.path.abspath(args['path_to_csv'])
    # copy the user package latex to the folder
    os.system('scp '
    +os.path.abspath(os.path.join(filedir, 'userpackagelatex/booktabs.sty'))+
    ' ' +output_folder_path_abs)
    os.system('scp '
    +os.path.abspath(os.path.join(filedir, 'userpackagelatex/geography.sty'))+
    ' ' +output_folder_path_abs)
    os.system('scp '
    +os.path.abspath(os.path.join(filedir, 'userpackagelatex/pdflscape.sty'))+
    ' ' +output_folder_path_abs)
    save_summary_table_as_pdf(input_folder_path_abs, output_folder_path_abs)
    os.remove(os.path.join(output_folder_path_abs, 'booktabs.sty'))
    os.remove(os.path.join(output_folder_path_abs, 'geography.sty'))
    os.remove(os.path.join(output_folder_path_abs, 'pdflscape.sty'))

在图像或 pdf 文档中将 pandas 数据帧保存为 table，具有良好的多索引显示

Save a pandas dataframe as table in Image or pdf document with nice multi index display

python

pdf

latex

pdflatex

pandas