models.py 5.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168
  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. class Tag(models.Model):
  11. """
  12. Based on django-taggit model
  13. """
  14. name = models.CharField(verbose_name="name", unique=True, blank=False, max_length=100)
  15. slug = models.SlugField(verbose_name="slug", unique=True, max_length=100)
  16. class Meta:
  17. verbose_name = "Tag"
  18. verbose_name_plural = "Tags"
  19. def __str__(self):
  20. return self.name
  21. def slugify(self, tag, i=None):
  22. slug = slugify(tag)
  23. if i is not None:
  24. slug += "_%d" % i
  25. return slug
  26. def save(self, *args, **kwargs):
  27. if self._state.adding and not self.slug:
  28. self.slug = self.slugify(self.name)
  29. with transaction.atomic():
  30. slugs = set(
  31. type(self)
  32. ._default_manager.filter(slug__startswith=self.slug)
  33. .values_list("slug", flat=True)
  34. )
  35. i = None
  36. while True:
  37. slug = self.slugify(self.name, i)
  38. if slug not in slugs:
  39. self.slug = slug
  40. return super().save(*args, **kwargs)
  41. i = 1 if i is None else i+1
  42. else:
  43. return super().save(*args, **kwargs)
  44. class Snapshot(models.Model):
  45. id = models.UUIDField(primary_key=True, default=uuid.uuid4, editable=False)
  46. url = models.URLField(unique=True)
  47. timestamp = models.CharField(max_length=32, unique=True, db_index=True)
  48. title = models.CharField(max_length=128, null=True, blank=True, db_index=True)
  49. added = models.DateTimeField(auto_now_add=True, db_index=True)
  50. updated = models.DateTimeField(null=True, blank=True, db_index=True)
  51. tags = models.ManyToManyField(Tag)
  52. keys = ('url', 'timestamp', 'title', 'tags', 'updated')
  53. def __repr__(self) -> str:
  54. title = self.title or '-'
  55. return f'[{self.timestamp}] {self.url[:64]} ({title[:64]})'
  56. def __str__(self) -> str:
  57. title = self.title or '-'
  58. return f'[{self.timestamp}] {self.url[:64]} ({title[:64]})'
  59. @classmethod
  60. def from_json(cls, info: dict):
  61. info = {k: v for k, v in info.items() if k in cls.keys}
  62. return cls(**info)
  63. def as_json(self, *args) -> dict:
  64. args = args or self.keys
  65. return {
  66. key: getattr(self, key)
  67. if key != 'tags' else self.tags_str()
  68. for key in args
  69. }
  70. def as_link(self) -> Link:
  71. return Link.from_json(self.as_json())
  72. def as_link_with_details(self) -> Link:
  73. from ..index import load_link_details
  74. return load_link_details(self.as_link())
  75. def tags_str(self) -> str:
  76. return ','.join(self.tags.order_by('name').values_list('name', flat=True))
  77. @cached_property
  78. def bookmarked(self):
  79. return parse_date(self.timestamp)
  80. @cached_property
  81. def is_archived(self):
  82. return self.as_link().is_archived
  83. @cached_property
  84. def num_outputs(self):
  85. return self.as_link().num_outputs
  86. @cached_property
  87. def url_hash(self):
  88. return self.as_link().url_hash
  89. @cached_property
  90. def base_url(self):
  91. return self.as_link().base_url
  92. @cached_property
  93. def link_dir(self):
  94. return self.as_link().link_dir
  95. @cached_property
  96. def archive_path(self):
  97. return self.as_link().archive_path
  98. @cached_property
  99. def archive_size(self):
  100. return self.as_link().archive_size
  101. @cached_property
  102. def history(self):
  103. from ..index import load_link_details
  104. return load_link_details(self.as_link()).history
  105. @cached_property
  106. def latest_title(self):
  107. if ('title' in self.history
  108. and self.history['title']
  109. and (self.history['title'][-1].status == 'succeeded')
  110. and self.history['title'][-1].output.strip()):
  111. return self.history['title'][-1].output.strip()
  112. return None
  113. def save_tags(self, tags=[]):
  114. tags_id = []
  115. for tag in tags:
  116. tags_id.append(Tag.objects.get_or_create(name=tag)[0].id)
  117. self.tags.clear()
  118. self.tags.add(*tags_id)
  119. class ArchiveResult(models.Model):
  120. snapshot = models.ForeignKey(Snapshot, on_delete=models.CASCADE)
  121. cmd = models.CharField(max_length=500, default="")
  122. pwd = models.CharField(max_length=200, default="")
  123. cmd_version = models.CharField(max_length=20, default="")
  124. output = models.CharField(max_length=500, default="")
  125. start_ts = models.DateTimeField()
  126. end_ts = models.DateTimeField()
  127. status = models.CharField(max_length=10)
  128. extractor = models.CharField(choices=EXTRACTORS, blank=False, max_length=20)
  129. def __str__(self):
  130. return self.extractor