Get only body of letters, emails from text files

问题

I want to remove all from, to, cc, subject sent tags from this text document and only keep the body of the mail so that I can use this to summarize content of the document. What is the best way to do this in python. I think it's better to first do the extraction and then use preprocessing for this case. Also attaching code here. So if anyone can suggest how to do this, would be really helpful. The payload and ismultipart part of the file is not done properly and there is where my doubt is and so have commented that part and require help there.

Attaching code and the .txt file below for reference.

import os, sys, csv
import glob
import re
import email
#from tika import parser
import warnings
warnings.filterwarnings(action='ignore', category=UserWarning, module='gensim')
from gensim.summarization import summarize, keywords

# Set path to directory where files are
dirs = 'C:\\Users\\Lenovo\\.spyder-py3\\Testing\\'
#os.chdir(dirs)
for filename in glob.glob(os.path.join(dirs, '*.txt')):
    try:
        for files in filename:
            file = open(filename, 'r', encoding ='utf-8')
            filecontents = file.read()
            filecontents = re.sub(r'\s+', ' ', filecontents)
            print(filecontents)
            filecontents = filecontents.strip('\n')
            b = email.message_from_string(filecontents)# NEED
            if b.is_multipart():#HELP
                for payload in b.get_payload():#HERE
                    # if payload.is_multipart(): ...#SO
                    print (payload.get_payload())#COMMENTED
            else:#
                print (b.get_payload())#
            summary = summarize(filecontents, ratio =0.10)
            print(summary)
            kw = keywords(filecontents, words=15)
            print(kw)
            break
            #writer.writerow([file, summary, kw])
    except Exception as e:
        pass

TEXT FILE

 Stephanie /ANN

From: Mr.A,  <.Mr.A@abc.com>
Sent: Wednesday, July 25, 2018 2:27 PM
To: , Tim /ANN; Abd, May /ANN
Cc: Mr.A, ; Theoder Jerry,
Subject: [EXTERNAL] RE:  Holdings: XXXX SPA – mfno.1322

Dear Dr. Tim A. , 

The option-2 is fine. By the way, we had received in the past Letter of Authorization for many companies other 
than Spa and I guess Xxxx does not do bANNiness with them either. If yes, then need to submit withdrawal 
of Letter of Authorization for those companies and send a Letter of Authorization for spa. stating for any 
applications submitted. We will send an administrative filing issue letter for both the holder and the agent.  

Thank you! 

Regards, 
 Mr.A 
PRODUCT Master File 
CDER 

Currently, there is no requirement to submit or resubmit NAs in any electronic format.  However, starting May 5, 2018, 
new NAs, as well as any submissions to the existing NAs mANNt be submitted electronically in legal (electronic Common 
Technical Document) format specified by GROUP A in the legal guidance. NA submissions that are not submitted in legal 
format after this date may be subject to rejection. For more information please check the NA website 
www.GROUP A.gov/abc/bca 

This communication is an informal communication consistent with which represents my best judgment 
at this time, but does not constitute an advisory opinion, does not necessarily represent the formal position of the 
GROUP A, and does not bind or otherwise obligate or commit the agency to the views expressed. This communication, 
including any attachments, is intended only for the person or entity to which it is addressed and may contain 
confidential material. Any review, retransmission, distribution or other ANNe of this information by persons or entities 
other than the intended recipient is prohibited. If you received this in error, please destroy any copies, contact the 
sender and delete the material from any computer. Thank you. 

From: Tim.@xxxx.com [mailto:Tim.@xxxx.com]  
Sent: Wednesday, July 25, 2018 2:10 PM 
To: Mr.A,  <.Mr.A@abc.com> 
Cc: May.Abd@xxxx.com 
Subject: RE: Holdings: XXXX SPA ‐ dm 013383 

Dear , 

XXXX

2

Thanks for your phone call to clarify your needs and to understand the situation. I have confirmed that Xxxx only does 
direct bANNiness for test  S intermediate with b. and not with the other companies (e, 
x, etc.) that are secondary companies. Based on our discANNsion, I believe that we do not need to 
provide QAs for these secondary companies or mention them in our NA file as they would be covered under a 
separate QA  S.p.A. to them. If this is correct, then I believe you mentioned that we have two options as 
described below: 

Option 1: We can issue a separate QA for each . NA to be specific on which NA is being cross‐referenced 
to our NA 13383. 

Option 2: We can do a single QA for  and mention that they can cross‐reference any of their NAs. This 
would allow them to cross‐reference any of their 

If I have misunderstood or am incorrect in my response and we need to discANNs further, please let me know. 

If not, when you issue your request, can you please send to me and May Abd by email? 

Kind regards. 

Tim 

Tim A. , BsC 
Director, YY SERVICES) 
Xxxx ANN 
Phone/FAX: 2312333 
Cell: 23312123131 
Email: tim.@xxxx.com 

From: , Tim /ANN  
Sent: Monday, July 23, 2018 7:05 AM 
To: 'Mr.A, ' 
Cc: Abd, May /ANN 
Subject: RE: [EXTERNAL] Holder: XXXX SPA - NA 013383 

Dear , 

May is now on vacation and I am covering for her during her absence. Is there a good time to call you today or later this 
week? Please let me know and we can schedule or please call my cell phone 21313131231 at your convenience. 

Kind regards. 

Tim 

Tim A. , MSC 
Director, PQR 
Xxxx 
Phone/FAX: 2312313313 
Cell: 3142342424 
Email: tim.@xxxx.com 

XXXX

3

‐‐‐‐‐‐‐‐‐‐ Forwarded message ‐‐‐‐‐‐‐‐‐‐ 
From: "Mr.A, " <.Mr.A@abc.com> 
Date: Jul 20, 2018 9:01 AM 
Subject: [EXTERNAL] Holder: XXXX SPA ‐ NA 013383 
To: "TRETE/ANN" <May.Abd@xxxx.com> 
Cc: "mno.com> 

Dear May Abd, 

. I need to talk to you on this.  

Thank you! 

Regards, 
 Mr.A 
PRODUCT Master File 
CDER 

Currently, there is no requirement to submit or resubmit NAs in any electronic format.   
format after this date may be subject to rejection. For more information please check the NA website 
www.GROUP A./cder/NA   

This communication is an informal communication  which represents my best judgment 
at this time, but does not constitute an advisory opinion, does not necessarily represent the formal position of the 
GROUP A, and does not bind or otherwise obligate or commit the agency to the views expressed. This communication, 
including any attachments, is intended only for the person or entity to which it is addressed and may contain 
confidential material. Any review, retransmission, distribution or other ANNe of this information by persons or entities 
other than the intended recipient is prohibited. If you received this in error, please destroy any copies, contact the 
sender and delete the material from any computer. Thank you. 

XXXX

回答1:

It's not really clear which part of the code you need help with, what you want it to do instead of what it currently does, or how to pass on the results for further processing correctly.

However, I will note that your code has a number of problems.

You cannot read an email message as UTF-8 text. Regardless of the file extension, an RFC822 message is simply a sequence of bytes. Traditional email could come in a large number of different encodings, and if you try to coerce it into UTF-8, you will run into UnicodeDecodeErrors and other snags.
As always, a blanket except Exception: is a major bug. Perhaps you only put this in for debugging, but it actually makes debugging harder.
Typical modern email messages come with somewhat complex MIME body structures which you have to analyze in context before you decide which one(s) you actually want to process. One common phenomenon is multipart/alternative where the same message is rendered in different formats so that recipients can decide whether they want to read it rendered as HTML, plain text, or, occasionally, perhaps PDF or RTF or a single image or whatever, depending on the application. Also, HTML structures often have multiple parts, because the main HTML wants to pull in small images which are supplied in the MIME structure as well (company logo, animated emojis, and other insults to the reader). Perhaps see also What are the "parts" in a multipart email?

Another complication for this answer is that Python's email library went through an overhaul relatively recently. The new functionality was introduced experimentally in Python 3.3, but only became the documented and default version in 3.6. Most of the code you will find out in the wild will be using the pre-3.6 facilities, but going forward, you will probably want to target the new and improved API.

With the legacy API your code might look something like

from email import message_from_binary_file

for filename in glob.glob(os.path.join(dirs, '*.txt')):
    # Not useful; we already have a filename
    #for files in filename:
    # Open in binary mode, don't try to guess encoding
    # Use a context manager so we don't leave the file open
    with open(filename, 'rb') as file:
        # Just let the email library take it from here
        #filecontents = file.read()
        #filecontents = re.sub(r'\s+', ' ', filecontents)
        #print(filecontents)
        #filecontents = filecontents.strip('\n')
        b = email.message_from_binary_file(file)
    if b.is_multipart():
        # There are a number of things you could do to pick out
        # one or more payloads for analysis, but let's just take
        # the first text/plain part and call it "main_part"
        for part in b.walk()
            if part.get_content_type() == 'text/plain':
                main_part = part.get_payload()
                break
    else:
        main_part = b.get_payload()
    summary = summarize(main_part, ratio =0.10)
    print(summary)
    kw = keywords(main_part, words=15)
    print(kw)

To use the new 3.6+ API you will need to adapt this to something like

from email.policy import default as default_email_policy
...
    b = email.message_from_binary_file(file, policy=default_email_policy)
    main_part = b.get_body(['related', 'plain', 'html'])

This will result in a new email.message.EmailMessage object which has some different methods and different behaviors than the legacy email.message.Message class. The documentation suggests that maybe one day the default policy will be passed in by default, at which point old code will switch to new behavior (but also probably some amount of unpleasant surprises and outright breakage).

Notice also the get_body() method which is new in 3.6 and which lets you easily pick out a "probable main part"; though if no text/plain part is available, the code above will fall back to HTML, which you will then need to process further to extract the actual text (look at Beautifulsoup maybe?)

There is no technical, robust, reliable way to separate boilerplate (headers, signatures, etc) from actual content in email. Some HTML email clients might provide hints in the generated message as to which <div> contains things the user typed in, but in the general case, you just have to wade up to your eyebrows in (frankly, hopeless) heuristics.

回答2:

If you want to only remove the From, Sent, To, Cc, Subject and Forwarded tags from the email you could use regex.

import re

with open('email_input.txt', 'r') as input:
   lines = input.readlines()
   no_new_lines = [i.strip() for i in lines]
   for line in no_new_lines:
      email_component = re.compile(r'((From:|Sent:|To:|Cc:|Subject:|Forwarded message).*)', re.IGNORECASE)
      remove_component = re.findall(email_component, line)
      if remove_component:
         print(line)

         # output
         ‐‐‐‐‐‐‐‐‐‐ Forwarded message ‐‐‐‐‐‐‐‐‐‐
         From: Mr.A,  <.Mr.A@abc.com>
         Sent: Wednesday, July 25, 2018 2:27 PM
         To: , Tim /ANN; Abd, May /ANN
         Cc: Mr.A, ; Theoder Jerry,
         Subject: [EXTERNAL] RE:  Holdings: XXXX SPA – mfno.1322

Concerning removing the content after 'regards'. I didn't add that to my regex, because emails can be signed severals ways. Here are some of the most common ways:

Best,
Best regards,
Best wishes,
Fond regards,
Kind regards,
Regards,
Sincerely,
Sincerely yours,
Thank you,
With appreciation,
With gratitude,
Yours sincerely,

UPDATED ANSWER ONE

The updated answer below cleans some more of your email input, but more cleaning is required.

import re

with open('email_input.txt', 'r') as input:
   lines = input.readlines()

   # Remove some of the extra lines
   no_new_lines = [i.strip() for i in lines]

   # regex to catch header lines
   email_component = re.compile(r'((From:|Sent:|To:|Cc:|Subject:|Date:|Forwarded message).*)', re.IGNORECASE)
   remove_headers = [x for x in no_new_lines if not email_component.findall(x)]

   # regex to catch greeting lines
   greeting_component = re.compile(r'(Dear.*)', re.IGNORECASE)
   remove_greeting = [x for x in remove_headers if not greeting_component.findall(x)]

   # regex to catch lines with contact details
   contact_component = re.compile(r'(Phone.*:)|(Cell:.*)|(Email:.*)', re.IGNORECASE)
   remove_contacts = [x for x in remove_greeting if not contact_component.findall(x)]

    # regex to catch lines with salutation
    email_salutation_component = re.compile(r'Best,(.*?)|Best regards,(.*?)|Best wishes,(.*?)|Fond regards,(.*?)|'
                                        r'Kind regards(.*?)|Regards,(.*?)|Sincerely,(.*?)|Sincerely yours,(.*?)|'
                                        r'Thank you,(.*?)|With appreciation,(.*?)|Yours sincerely,(.*?)', re.IGNORECASE)

    remove_salutations = [x for x in remove_contacts if not email_salutation_component.findall(x)]

    # do something else

UPDATED ANSWER TWO

The updated answer below uses the python email library. My input file was an original email message pulled from my email client. Using the code below, I was able to extract the body of every email message that I tried. I also tested the gensim module and it worked correctly.

import email
from gensim.summarization import summarize, keywords

with open('email_input.txt', 'r') as input:
  email_body = ''
  raw_message = input.read()

  # Return a message object structure from a string
  msg = email.message_from_string(raw_message)

  # iterate over all the parts and subparts of a message object tree
  for part in msg.walk():

    # Return the message’s content type.
    if part.get_content_type() == 'text/plain':
        email_body = part.get_payload()

  summary = summarize(email_body, ratio=0.10)
  print(summary)

  kw = keywords(email_body, words=15)
  print(kw)

FINAL ANSWER

This is my final answer to this question. Hopefully, one of these 4 answers meets your requirements.

You will have to do some small cleanup of the output, because I don't know all your requirements.

with open('email_input.txt') as infile:
  # Boolean state variable to keep track of whether we want to be printing lines or not
  lines_to_keep = False
  for line in infile:

    # Look for lines that start with a greeting
    if line.startswith("Dear"):
      # set lines_to_keep true and start capturing lines
      lines_to_keep = True

    # Look for lines that start with a salutation
    elif line.startswith("Regards") or line.startswith("Kind regards"):
        # set lines_to_keep false and stop capturing lines
        lines_to_keep = False


    if lines_to_keep:
        greeting_component = re.compile(r'(Dear.*)', re.IGNORECASE)
        remove_greeting = re.match(greeting_component, line)
        if not remove_greeting:
            print (line.rstrip('\n'))
            # output 
            The option-2 is fine. By the way, we had received in the past Letter of Authorization for many companies other than Spa and I guess Xxxx does not do bANNiness with them either. If yes, then need to submit withdrawal of Letter of Authorization for those companies and send a Letter of Authorization for spa. stating for any applications submitted. We will send an administrative filing issue letter for both the holder and the agent.  

           more here....

来源：https://stackoverflow.com/questions/54760414/get-only-body-of-letters-emails-from-text-files

标签

python

summarization

document-body