quotes/massimport.py

   1 import re
   2 from django.db import DatabaseError, transaction
   3
   4 from quotes.models import QuoteTag, Quote, Work, Author
   5
   6 def get_or_create_author(authorname, resultcontext):
   7     authorname = authorname.strip()
   8     try:
   9         return Author.objects.get(name=authorname)
  10     except Author.DoesNotExist:
  11         a = Author.objects.create(name=authorname,
  12                                   pvt_notes="<p>Created during mass import</p>")
  13         resultcontext["created_authors"] += [a]
  14         return a
  15
  16 def get_or_create_work(workname, author, resultcontext):
  17     workname = workname.strip()
  18     try:
  19         return Work.objects.get(name=workname, author=author)
  20     except Work.DoesNotExist:
  21         w = Work.objects.create(name=workname,
  22                                 author=author,
  23                                 pvt_notes="<p>Created during mass import</p>")
  24         resultcontext["created_works"] += [w]
  25         return w
  26
  27 def get_or_create_tag(tagname, resultcontext):
  28     try:
  29         return QuoteTag.objects.get(tag=tagname)
  30     except QuoteTag.DoesNotExist:
  31         t = QuoteTag.objects.create(tag=tagname)
  32         resultcontext["created_tags"] += [t]
  33         return t
  34
  35 def add_tags_on_quote(quote, tagline, resultcontext):
  36     for tagname in tagline.split(","):
  37         tagname = tagname.strip()
  38         if not tagname:
  39             continue
  40         tag = get_or_create_tag(tagname, resultcontext)
  41         quote.tags.add(tag)
  42
  43 def paragraphize(text):
  44     paragraph = ""
  45     for line in text.splitlines():
  46         line = line.strip()
  47         if not line: continue
  48         paragraph += "<p>%s</p>" % line
  49         # rest of the HTML will be bleach.clean()'d
  50     return paragraph
  51
  52 def create_quote(quotetext, authorname, workname, tagline, resultcontext):
  53     author = get_or_create_author(authorname, resultcontext)
  54     work = get_or_create_work(workname, author, resultcontext)
  55     quotetext = paragraphize(quotetext)
  56     quote = Quote.objects.create(text=quotetext, work=work)
  57     add_tags_on_quote(quote, tagline, resultcontext)
  58     resultcontext["created_quotes"] += [quote]
  59
  60 def create_all_quotes(allquotes, resultcontext):
  61     quotesep = re.compile(r'^\s*===+.*$', re.MULTILINE)
  62     elemsep = re.compile(r'^\s*---+.*$', re.MULTILINE)
  63
  64     for fullquote in quotesep.split(allquotes):
  65         if not fullquote: continue
  66
  67         elements = elemsep.split(fullquote)
  68         if len(elements) not in [3, 4]:
  69             resultcontext["rejected"] += [fullquote]
  70             continue
  71         quotetext, authorname, workname = elements[0:3]
  72         if len(elements) == 4:
  73             tagline = elements[3]
  74         else:
  75             tagline = ''
  76         create_quote(quotetext, authorname, workname, tagline,
  77                      resultcontext)
  78
  79 def domassimport(allquotes):
  80     resultcontext = {}
  81     resultcontext["created_quotes"] = []
  82     resultcontext["created_tags"] = []
  83     resultcontext["created_works"] = []
  84     resultcontext["created_authors"] = []
  85     resultcontext["rejected"] = []
  86     resultcontext["fatal_error"] = False
  87
  88     try:
  89         with transaction.atomic():
  90             create_all_quotes(allquotes, resultcontext)
  91     except DatabaseError as e:
  92         resultcontext["fatal_error"] = True
  93         resultcontext["fatal_error_message"] = e.__cause__
  94
  95     return resultcontext