'''
Created on Apr 16, 2014
    refactor 5-17-17
@author: Brett Paufler
Copyright Brett Paufler

For all .doc files in ./input, outputs
    .txt
    .html
    
Works on Microsoft Word 2003
    On upgrade, may fail
    On upgrade, pdf conversion may be available

NOTE: the html output sucks
    so if using, create utility to clean it up
'''

from win32com.client.gencache import EnsureDispatch
from win32com.client  import constants
from os import listdir
from os.path import abspath



def all_docs_to_txt_and_html(txt=True, html=True):
    '''Creates new .txt & .html files in ./input
    for every .doc file in ./input.'''  

    WORD = EnsureDispatch('Word.Application')

    dirs_in = './input/'
    docs = [abspath(dirs_in + doc) for doc
        in listdir(dirs_in)
        if doc.endswith('doc')]
  
    for doc_file in docs:
        
        print 'CONVERTING'
        print doc_file
        
        doc = WORD.Documents.Open(doc_file)
        
        if txt:
            txt_file = doc_file.replace('.doc', '.txt')
            print txt_file
            txt_file_magic_number = 7
            doc.SaveAs(txt_file, txt_file_magic_number)
        
        if html:
            html_file = doc_file.replace('.doc', '.html')
            print html_file
            doc.SaveAs(html_file, constants.wdFormatHTML)

        #if upgrade to better office: pdf magic number below
        ##wdFormatPDF=17
        
        doc.Close()



if __name__ == '__main__':
    print 'Starting'
    all_docs_to_txt_and_html()
    print 'Finished'