models.py 5.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173
  1. __package__ = 'archivebox.core'
  2. import uuid
  3. from django.db import models, transaction
  4. from django.utils.functional import cached_property
  5. from django.utils.text import slugify
  6. from ..util import parse_date
  7. from ..index.schema import Link
  8. from ..extractors import get_default_archive_methods
  9. EXTRACTORS = [(extractor[0], extractor[0]) for extractor in get_default_archive_methods()]
  10. STATUS_CHOICES = [
  11. ("succeeded", "succeeded"),
  12. ("failed", "failed"),
  13. ("skipped", "skipped")
  14. ]
  15. class Tag(models.Model):
  16. """
  17. Based on django-taggit model
  18. """
  19. name = models.CharField(verbose_name="name", unique=True, blank=False, max_length=100)
  20. slug = models.SlugField(verbose_name="slug", unique=True, max_length=100)
  21. class Meta:
  22. verbose_name = "Tag"
  23. verbose_name_plural = "Tags"
  24. def __str__(self):
  25. return self.name
  26. def slugify(self, tag, i=None):
  27. slug = slugify(tag)
  28. if i is not None:
  29. slug += "_%d" % i
  30. return slug
  31. def save(self, *args, **kwargs):
  32. if self._state.adding and not self.slug:
  33. self.slug = self.slugify(self.name)
  34. with transaction.atomic():
  35. slugs = set(
  36. type(self)
  37. ._default_manager.filter(slug__startswith=self.slug)
  38. .values_list("slug", flat=True)
  39. )
  40. i = None
  41. while True:
  42. slug = self.slugify(self.name, i)
  43. if slug not in slugs:
  44. self.slug = slug
  45. return super().save(*args, **kwargs)
  46. i = 1 if i is None else i+1
  47. else:
  48. return super().save(*args, **kwargs)
  49. class Snapshot(models.Model):
  50. id = models.UUIDField(primary_key=True, default=uuid.uuid4, editable=False)
  51. url = models.URLField(unique=True)
  52. timestamp = models.CharField(max_length=32, unique=True, db_index=True)
  53. title = models.CharField(max_length=128, null=True, blank=True, db_index=True)
  54. added = models.DateTimeField(auto_now_add=True, db_index=True)
  55. updated = models.DateTimeField(null=True, blank=True, db_index=True)
  56. tags = models.ManyToManyField(Tag)
  57. keys = ('url', 'timestamp', 'title', 'tags', 'updated')
  58. def __repr__(self) -> str:
  59. title = self.title or '-'
  60. return f'[{self.timestamp}] {self.url[:64]} ({title[:64]})'
  61. def __str__(self) -> str:
  62. title = self.title or '-'
  63. return f'[{self.timestamp}] {self.url[:64]} ({title[:64]})'
  64. @classmethod
  65. def from_json(cls, info: dict):
  66. info = {k: v for k, v in info.items() if k in cls.keys}
  67. return cls(**info)
  68. def as_json(self, *args) -> dict:
  69. args = args or self.keys
  70. return {
  71. key: getattr(self, key)
  72. if key != 'tags' else self.tags_str()
  73. for key in args
  74. }
  75. def as_link(self) -> Link:
  76. return Link.from_json(self.as_json())
  77. def as_link_with_details(self) -> Link:
  78. from ..index import load_link_details
  79. return load_link_details(self.as_link())
  80. def tags_str(self) -> str:
  81. return ','.join(self.tags.order_by('name').values_list('name', flat=True))
  82. @cached_property
  83. def bookmarked(self):
  84. return parse_date(self.timestamp)
  85. @cached_property
  86. def is_archived(self):
  87. return self.as_link().is_archived
  88. @cached_property
  89. def num_outputs(self):
  90. return self.as_link().num_outputs
  91. @cached_property
  92. def url_hash(self):
  93. return self.as_link().url_hash
  94. @cached_property
  95. def base_url(self):
  96. return self.as_link().base_url
  97. @cached_property
  98. def link_dir(self):
  99. return self.as_link().link_dir
  100. @cached_property
  101. def archive_path(self):
  102. return self.as_link().archive_path
  103. @cached_property
  104. def archive_size(self):
  105. return self.as_link().archive_size
  106. @cached_property
  107. def history(self):
  108. from ..index import load_link_details
  109. return load_link_details(self.as_link()).history
  110. @cached_property
  111. def latest_title(self):
  112. if ('title' in self.history
  113. and self.history['title']
  114. and (self.history['title'][-1].status == 'succeeded')
  115. and self.history['title'][-1].output.strip()):
  116. return self.history['title'][-1].output.strip()
  117. return None
  118. def save_tags(self, tags=[]):
  119. tags_id = []
  120. for tag in tags:
  121. tags_id.append(Tag.objects.get_or_create(name=tag)[0].id)
  122. self.tags.clear()
  123. self.tags.add(*tags_id)
  124. class ArchiveResult(models.Model):
  125. snapshot = models.ForeignKey(Snapshot, on_delete=models.CASCADE)
  126. cmd = models.JSONField()
  127. pwd = models.CharField(max_length=256)
  128. cmd_version = models.CharField(max_length=32)
  129. output = models.CharField(max_length=512)
  130. start_ts = models.DateTimeField()
  131. end_ts = models.DateTimeField()
  132. status = models.CharField(max_length=16, choices=STATUS_CHOICES)
  133. extractor = models.CharField(choices=EXTRACTORS, max_length=32)
  134. def __str__(self):
  135. return self.extractor