提取進出ipython/jupyter筆記本的泡菜的方法

Question

我試圖總結一個數據分析項目，它運行許多ipython/jupyter筆記本，每個筆記本都相當長。有助於這個過程的事情之一是，如果我至少知道整個「投入」醬菜進入和「輸出」泡菜走出去。

什麼是最乾淨/最快/最有效的方式來做到這一點？

Answer 1

我不知道這是否是做的最好的方式，但它至少有一種方式......

def summerize_pickles(notebook_path): 
    from IPython.nbformat import current as nbformat 
    import re 

    with open(notebook_path) as fh: 
     nb = nbformat.reads_json(fh.read()) 

    list_of_input_pickles = [] 
    list_of_output_pickles = [] 

    for cell in nb["worksheets"][0]["cells"]: 
     # This confirms there is at least one pickle in it. 
     if cell["cell_type"] != "code" or cell["input"].find("pickle") == -1: # Skipping over those cells which aren't code or those cells with code but which don't reference "pickle 
      continue 

     # In case there are multiple lines, it iterates line by line. 
     for line in cell["input"].splitlines(): 
      if line.find("pickle") == -1: # Skips over lines w/ no mention of "pickle" to potentially reduce the number of times it's searched. 
       continue 
      ############################ ############################ ############################ ############################ 
      code_type = str() 
      if line.find("pickle.dump") != -1 or line.find(".to_pickle")!= -1: 
       code_type = "output"  
      elif line.find("pickle.load") != -1 or line.find(".read_pickle")!= -1: 
       code_type = "input" 
      else: 
       continue # This tells the code to skip over lines like "import cpickle as pickle" 
      ############################ ############################ ############################ ############################ 
      filename = re.findall(r'"(.*?)"', line) # This gets all the content between the quotes. See: http://stackoverflow.com/questions/171480/regex-grabbing-values-between-quotation-marks  
      ############################ ############################ ############################ ############################   
      if code_type == "input": 
       list_of_input_pickles.append(filename[0]) 
      elif code_type == "output": 
       list_of_output_pickles.append(filename[0]) 

    pickles_dict = {"input_pickles":list_of_input_pickles, 
        "output_pickles":list_of_output_pickles } 

    return pickles_dict