|
| 1 | +#!/usr/bin/env python |
| 2 | +# Copyright The IETF Trust 2017-2019, All Rights Reserved |
| 3 | + |
| 4 | +import datetime |
| 5 | +import os |
| 6 | +import sys |
| 7 | +from pathlib2 import Path |
| 8 | +from contextlib import closing |
| 9 | + |
| 10 | +os.environ["DJANGO_SETTINGS_MODULE"] = "ietf.settings" |
| 11 | + |
| 12 | +import django |
| 13 | +django.setup() |
| 14 | + |
| 15 | +from django.conf import settings |
| 16 | +from django.core.validators import validate_email, ValidationError |
| 17 | +from ietf.utils.draft import Draft |
| 18 | +from ietf.submit.utils import update_authors |
| 19 | + |
| 20 | +import debug # pyflakes:ignore |
| 21 | + |
| 22 | +from ietf.doc.models import Document, NewRevisionDocEvent, DocEvent, State |
| 23 | +from ietf.person.models import Person |
| 24 | + |
| 25 | +system = Person.objects.get(name="(System)") |
| 26 | +expired = State.objects.get(type='draft',slug='expired') |
| 27 | + |
| 28 | +names = set() |
| 29 | +print 'collecting draft names ...' |
| 30 | +versions = 0 |
| 31 | +for p in Path(settings.INTERNET_DRAFT_PATH).glob('draft*.txt'): |
| 32 | + n = str(p).split('/')[-1].split('-') |
| 33 | + if n[-1][:2].isdigit(): |
| 34 | + name = '-'.join(n[:-1]) |
| 35 | + if '--' in name or '.txt' in name or '[' in name or '=' in name or '&' in name: |
| 36 | + continue |
| 37 | + if name.startswith('draft-draft-'): |
| 38 | + continue |
| 39 | + if name == 'draft-ietf-trade-iotp-v1_0-dsig': |
| 40 | + continue |
| 41 | + if len(n[-1]) != 6: |
| 42 | + continue |
| 43 | + if name.startswith('draft-mlee-'): |
| 44 | + continue |
| 45 | + names.add('-'.join(n[:-1])) |
| 46 | + |
| 47 | +count=0 |
| 48 | +print 'iterating through names ...' |
| 49 | +for name in sorted(names): |
| 50 | + if not Document.objects.filter(name=name).exists(): |
| 51 | + paths = list(Path(settings.INTERNET_DRAFT_PATH).glob('%s-??.txt'%name)) |
| 52 | + paths.sort() |
| 53 | + doc = None |
| 54 | + for p in paths: |
| 55 | + n = str(p).split('/')[-1].split('-') |
| 56 | + rev = n[-1][:2] |
| 57 | + with open(str(p)) as txt_file: |
| 58 | + raw = txt_file.read() |
| 59 | + try: |
| 60 | + text = raw.decode('utf8') |
| 61 | + except UnicodeDecodeError: |
| 62 | + text = raw.decode('latin1') |
| 63 | + try: |
| 64 | + draft = Draft(text, txt_file.name, name_from_source=True) |
| 65 | + except Exception as e: |
| 66 | + print name, rev, "Can't parse", p,":",e |
| 67 | + continue |
| 68 | + if draft.errors and draft.errors.keys()!=['draftname',]: |
| 69 | + print "Errors - could not process", name, rev, datetime.datetime.fromtimestamp(p.stat().st_mtime), draft.errors, draft.get_title().encode('utf8') |
| 70 | + else: |
| 71 | + time = datetime.datetime.fromtimestamp(p.stat().st_mtime) |
| 72 | + if not doc: |
| 73 | + doc = Document.objects.create(name=name, |
| 74 | + time=time, |
| 75 | + type_id='draft', |
| 76 | + title=draft.get_title(), |
| 77 | + abstract=draft.get_abstract(), |
| 78 | + rev = rev, |
| 79 | + pages=draft.get_pagecount(), |
| 80 | + words=draft.get_wordcount(), |
| 81 | + expires=time+datetime.timedelta(settings.INTERNET_DRAFT_DAYS_TO_EXPIRE), |
| 82 | + ) |
| 83 | + DocAlias.objects.create(name=doc.name).docs.add(doc) |
| 84 | + doc.states.add(expired) |
| 85 | + # update authors |
| 86 | + authors = [] |
| 87 | + for author in draft.get_author_list(): |
| 88 | + full_name, first_name, middle_initial, last_name, name_suffix, email, country, company = author |
| 89 | + |
| 90 | + author_name = full_name.replace("\n", "").replace("\r", "").replace("<", "").replace(">", "").strip() |
| 91 | + |
| 92 | + if email: |
| 93 | + try: |
| 94 | + validate_email(email) |
| 95 | + except ValidationError: |
| 96 | + email = "" |
| 97 | + |
| 98 | + def turn_into_unicode(s): |
| 99 | + if s is None: |
| 100 | + return u"" |
| 101 | + |
| 102 | + if isinstance(s, unicode): |
| 103 | + return s |
| 104 | + else: |
| 105 | + try: |
| 106 | + return s.decode("utf-8") |
| 107 | + except UnicodeDecodeError: |
| 108 | + try: |
| 109 | + return s.decode("latin-1") |
| 110 | + except UnicodeDecodeError: |
| 111 | + return "" |
| 112 | + |
| 113 | + author_name = turn_into_unicode(author_name) |
| 114 | + email = turn_into_unicode(email) |
| 115 | + company = turn_into_unicode(company) |
| 116 | + |
| 117 | + authors.append({ |
| 118 | + "name": author_name, |
| 119 | + "email": email, |
| 120 | + "affiliation": company, |
| 121 | + "country": country |
| 122 | + }) |
| 123 | + dummysubmission=type('', (), {})() #https://stackoverflow.com/questions/19476816/creating-an-empty-object-in-python |
| 124 | + dummysubmission.authors = authors |
| 125 | + update_authors(doc,dummysubmission) |
| 126 | + |
| 127 | + # add a docevent with words explaining where this came from |
| 128 | + events = [] |
| 129 | + e = NewRevisionDocEvent.objects.create( |
| 130 | + type="new_revision", |
| 131 | + doc=doc, |
| 132 | + rev=rev, |
| 133 | + by=system, |
| 134 | + desc="New version available: <b>%s-%s.txt</b>" % (doc.name, doc.rev), |
| 135 | + time=time, |
| 136 | + ) |
| 137 | + events.append(e) |
| 138 | + e = DocEvent.objects.create( |
| 139 | + type="comment", |
| 140 | + doc = doc, |
| 141 | + rev = rev, |
| 142 | + by = system, |
| 143 | + desc = "Revision added from id-archive on %s by %s"%(datetime.date.today(),sys.argv[0]), |
| 144 | + time=time, |
| 145 | + ) |
| 146 | + events.append(e) |
| 147 | + doc.time = time |
| 148 | + doc.rev = rev |
| 149 | + doc.save_with_history(events) |
| 150 | + print "Added",name, rev |
0 commit comments