admin.py 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361
  1. __package__ = 'archivebox.core'
  2. from io import StringIO
  3. from pathlib import Path
  4. from contextlib import redirect_stdout
  5. from django.contrib import admin
  6. from django.urls import path
  7. from django.utils.html import format_html
  8. from django.utils.safestring import mark_safe
  9. from django.shortcuts import render, redirect
  10. from django.contrib.auth import get_user_model
  11. from django import forms
  12. from ..util import htmldecode, urldecode, ansi_to_html
  13. from core.models import Snapshot, ArchiveResult, Tag
  14. from core.forms import AddLinkForm
  15. from core.mixins import SearchResultsAdminMixin
  16. from index.html import snapshot_icons
  17. from logging_util import printable_filesize
  18. from main import add, remove
  19. from config import OUTPUT_DIR, SNAPSHOTS_PER_PAGE
  20. from extractors import archive_links
  21. # Admin URLs
  22. # /admin/
  23. # /admin/login/
  24. # /admin/core/
  25. # /admin/core/snapshot/
  26. # /admin/core/snapshot/:uuid/
  27. # /admin/core/tag/
  28. # /admin/core/tag/:uuid/
  29. # TODO: https://stackoverflow.com/questions/40760880/add-custom-button-to-django-admin-panel
  30. class ArchiveResultInline(admin.TabularInline):
  31. model = ArchiveResult
  32. class TagInline(admin.TabularInline):
  33. model = Snapshot.tags.through
  34. from django.contrib.admin.helpers import ActionForm
  35. class SnapshotActionForm(ActionForm):
  36. tag = forms.ModelChoiceField(queryset=Tag.objects.all(), required=False)
  37. # pass
  38. class SnapshotAdmin(SearchResultsAdminMixin, admin.ModelAdmin):
  39. list_display = ('added', 'title_str', 'url_str', 'files', 'size')
  40. sort_fields = ('title_str', 'url_str', 'added')
  41. readonly_fields = ('uuid', 'num_outputs', 'is_archived', 'url_hash', 'added', 'updated')
  42. search_fields = ['url__icontains', 'timestamp', 'title', 'tags__name']
  43. fields = ('timestamp', 'url', 'title', 'tags', *readonly_fields)
  44. list_filter = ('added', 'updated', 'tags')
  45. ordering = ['-added']
  46. actions = ['delete_snapshots', 'overwrite_snapshots', 'update_snapshots', 'update_titles', 'verify_snapshots', 'add_tag', 'remove_tag']
  47. autocomplete_fields = ['tags']
  48. inlines = [ArchiveResultInline]
  49. list_per_page = SNAPSHOTS_PER_PAGE
  50. action_form = SnapshotActionForm
  51. def get_urls(self):
  52. urls = super().get_urls()
  53. custom_urls = [
  54. path('grid/', self.admin_site.admin_view(self.grid_view),name='grid')
  55. ]
  56. return custom_urls + urls
  57. def get_queryset(self, request):
  58. self.request = request
  59. return super().get_queryset(request).prefetch_related('tags')
  60. def tag_list(self, obj):
  61. return ', '.join(obj.tags.values_list('name', flat=True))
  62. # TODO: figure out a different way to do this, you cant nest forms so this doenst work
  63. # def action(self, obj):
  64. # # csrfmiddlewaretoken: Wa8UcQ4fD3FJibzxqHN3IYrrjLo4VguWynmbzzcPYoebfVUnDovon7GEMYFRgsh0
  65. # # action: update_snapshots
  66. # # select_across: 0
  67. # # _selected_action: 76d29b26-2a88-439e-877c-a7cca1b72bb3
  68. # return format_html(
  69. # '''
  70. # <form action="/admin/core/snapshot/" method="post" onsubmit="e => e.stopPropagation()">
  71. # <input type="hidden" name="csrfmiddlewaretoken" value="{}">
  72. # <input type="hidden" name="_selected_action" value="{}">
  73. # <button name="update_snapshots">Check</button>
  74. # <button name="update_titles">Pull title + favicon</button>
  75. # <button name="update_snapshots">Update</button>
  76. # <button name="overwrite_snapshots">Re-Archive (overwrite)</button>
  77. # <button name="delete_snapshots">Permanently delete</button>
  78. # </form>
  79. # ''',
  80. # csrf.get_token(self.request),
  81. # obj.id,
  82. # )
  83. def uuid(self, obj):
  84. return format_html(
  85. '<code style="font-size: 10px">{}</code><br/><a href="/archive/{}">View index ➡️</a> &nbsp; &nbsp; <a href="/admin/core/snapshot/?id__exact={}">View actions ⚙️</a>',
  86. obj.id,
  87. obj.timestamp,
  88. obj.id,
  89. )
  90. def title_str(self, obj):
  91. canon = obj.as_link().canonical_outputs()
  92. tags = ''.join(
  93. format_html('<a href="/admin/core/snapshot/?id__startswith={}"><span class="tag">{}</span></a> ', tag.id, tag)
  94. for tag in obj.tags.all()
  95. if str(tag).strip()
  96. )
  97. return format_html(
  98. '<a href="/{}">'
  99. '<img src="/{}/{}" class="favicon" onerror="this.remove()">'
  100. '</a>'
  101. '<a href="/{}/index.html">'
  102. '<b class="status-{}">{}</b>'
  103. '</a>',
  104. obj.archive_path,
  105. obj.archive_path, canon['favicon_path'],
  106. obj.archive_path,
  107. 'fetched' if obj.latest_title or obj.title else 'pending',
  108. urldecode(htmldecode(obj.latest_title or obj.title or ''))[:128] or 'Pending...'
  109. ) + mark_safe(f' <span class="tags">{tags}</span>')
  110. def files(self, obj):
  111. return snapshot_icons(obj)
  112. def size(self, obj):
  113. archive_size = (Path(obj.link_dir) / 'index.html').exists() and obj.archive_size
  114. if archive_size:
  115. size_txt = printable_filesize(archive_size)
  116. if archive_size > 52428800:
  117. size_txt = mark_safe(f'<b>{size_txt}</b>')
  118. else:
  119. size_txt = mark_safe('<span style="opacity: 0.3">...</span>')
  120. return format_html(
  121. '<a href="/{}" title="View all files">{}</a>',
  122. obj.archive_path,
  123. size_txt,
  124. )
  125. def url_str(self, obj):
  126. return format_html(
  127. '<a href="{}"><code>{}</code></a>',
  128. obj.url,
  129. obj.url.split('://www.', 1)[-1].split('://', 1)[-1][:64],
  130. )
  131. def grid_view(self, request):
  132. # cl = self.get_changelist_instance(request)
  133. # Save before monkey patching to restore for changelist list view
  134. saved_change_list_template = self.change_list_template
  135. saved_list_per_page = self.list_per_page
  136. saved_list_max_show_all = self.list_max_show_all
  137. # Monkey patch here plus core_tags.py
  138. self.change_list_template = 'private_index_grid.html'
  139. self.list_per_page = 20
  140. self.list_max_show_all = self.list_per_page
  141. # Call monkey patched view
  142. rendered_response = self.changelist_view(request)
  143. # Restore values
  144. self.change_list_template = saved_change_list_template
  145. self.list_per_page = saved_list_per_page
  146. self.list_max_show_all = saved_list_max_show_all
  147. return rendered_response
  148. def update_snapshots(self, request, queryset):
  149. archive_links([
  150. snapshot.as_link()
  151. for snapshot in queryset
  152. ], out_dir=OUTPUT_DIR)
  153. update_snapshots.short_description = "Archive"
  154. def update_titles(self, request, queryset):
  155. archive_links([
  156. snapshot.as_link()
  157. for snapshot in queryset
  158. ], overwrite=True, methods=('title','favicon'), out_dir=OUTPUT_DIR)
  159. update_titles.short_description = "Pull title"
  160. def overwrite_snapshots(self, request, queryset):
  161. archive_links([
  162. snapshot.as_link()
  163. for snapshot in queryset
  164. ], overwrite=True, out_dir=OUTPUT_DIR)
  165. overwrite_snapshots.short_description = "Re-archive (overwrite)"
  166. def verify_snapshots(self, request, queryset):
  167. for snapshot in queryset:
  168. print(snapshot.timestamp, snapshot.url, snapshot.is_archived, snapshot.archive_size, len(snapshot.history))
  169. verify_snapshots.short_description = "Check"
  170. def delete_snapshots(self, request, queryset):
  171. remove(snapshots=queryset, yes=True, delete=True, out_dir=OUTPUT_DIR)
  172. delete_snapshots.short_description = "Delete"
  173. def add_tag(self, request, queryset):
  174. tag = request.POST['tag']
  175. for obj in queryset:
  176. obj.tags.add(tag)
  177. add_tag.short_description = "Add tag"
  178. def remove_tag(self, request, queryset):
  179. tag = request.POST['tag']
  180. for obj in queryset:
  181. obj.tags.remove(tag)
  182. remove_tag.short_description = "Remove tag"
  183. title_str.short_description = 'Title'
  184. url_str.short_description = 'Original URL'
  185. title_str.admin_order_field = 'title'
  186. url_str.admin_order_field = 'url'
  187. class TagAdmin(admin.ModelAdmin):
  188. list_display = ('slug', 'name', 'num_snapshots', 'snapshots', 'id')
  189. sort_fields = ('id', 'name', 'slug')
  190. readonly_fields = ('id', 'num_snapshots', 'snapshots')
  191. search_fields = ('id', 'name', 'slug')
  192. fields = (*readonly_fields, 'name', 'slug')
  193. actions = ['delete_selected']
  194. ordering = ['-id']
  195. def num_snapshots(self, obj):
  196. return format_html(
  197. '<a href="/admin/core/snapshot/?tags__id__exact={}">{} total</a>',
  198. obj.id,
  199. obj.snapshot_set.count(),
  200. )
  201. def snapshots(self, obj):
  202. total_count = obj.snapshot_set.count()
  203. return mark_safe('<br/>'.join(
  204. format_html(
  205. '{} <code><a href="/admin/core/snapshot/{}/change"><b>[{}]</b></a> {}</code>',
  206. snap.updated.strftime('%Y-%m-%d %H:%M'),
  207. snap.id,
  208. snap.timestamp,
  209. snap.url,
  210. )
  211. for snap in obj.snapshot_set.order_by('-updated')[:10]
  212. ) + (f'<br/><a href="/admin/core/snapshot/?tags__id__exact={obj.id}">and {total_count-10} more...<a>' if obj.snapshot_set.count() > 10 else ''))
  213. class ArchiveResultAdmin(admin.ModelAdmin):
  214. list_display = ('id', 'start_ts', 'extractor', 'snapshot_str', 'cmd_str', 'status', 'output_str')
  215. sort_fields = ('start_ts', 'extractor', 'status')
  216. readonly_fields = ('id', 'uuid', 'snapshot_str')
  217. search_fields = ('id', 'uuid', 'snapshot__url', 'extractor', 'output', 'cmd_version', 'cmd', 'snapshot__timestamp')
  218. fields = (*readonly_fields, 'snapshot', 'snapshot__tags', 'extractor', 'status', 'start_ts', 'end_ts', 'pwd', 'cmd', 'cmd_version', 'output')
  219. autocomplete_fields = ['snapshot']
  220. list_filter = ('status', 'extractor', 'start_ts', 'cmd_version')
  221. ordering = ['-start_ts']
  222. list_per_page = SNAPSHOTS_PER_PAGE
  223. def snapshot_str(self, obj):
  224. return format_html(
  225. '<a href="/archive/{}/index.html"><b><code>[{}]</code></b></a><br/>'
  226. '<small>{}</small>',
  227. obj.snapshot.timestamp,
  228. obj.snapshot.timestamp,
  229. obj.snapshot.url[:128],
  230. )
  231. def cmd_str(self, obj):
  232. return format_html(
  233. '<pre>{}</pre>',
  234. ' '.join(obj.cmd) if isinstance(obj.cmd, list) else str(obj.cmd),
  235. )
  236. def output_str(self, obj):
  237. return format_html(
  238. '<a href="/archive/{}/{}" class="output-link">↗️</a><pre>{}</pre>',
  239. obj.snapshot.timestamp,
  240. obj.output if (obj.status == 'succeeded') and obj.extractor not in ('title', 'archive_org') else 'index.html',
  241. obj.output,
  242. )
  243. snapshot_str.short_description = 'snapshot'
  244. class ArchiveBoxAdmin(admin.AdminSite):
  245. site_header = 'ArchiveBox'
  246. index_title = 'Links'
  247. site_title = 'Index'
  248. def get_urls(self):
  249. return [
  250. path('core/snapshot/add/', self.add_view, name='Add'),
  251. ] + super().get_urls()
  252. def add_view(self, request):
  253. if not request.user.is_authenticated:
  254. return redirect(f'/admin/login/?next={request.path}')
  255. request.current_app = self.name
  256. context = {
  257. **self.each_context(request),
  258. 'title': 'Add URLs',
  259. }
  260. if request.method == 'GET':
  261. context['form'] = AddLinkForm()
  262. elif request.method == 'POST':
  263. form = AddLinkForm(request.POST)
  264. if form.is_valid():
  265. url = form.cleaned_data["url"]
  266. print(f'[+] Adding URL: {url}')
  267. depth = 0 if form.cleaned_data["depth"] == "0" else 1
  268. input_kwargs = {
  269. "urls": url,
  270. "depth": depth,
  271. "update_all": False,
  272. "out_dir": OUTPUT_DIR,
  273. }
  274. add_stdout = StringIO()
  275. with redirect_stdout(add_stdout):
  276. add(**input_kwargs)
  277. print(add_stdout.getvalue())
  278. context.update({
  279. "stdout": ansi_to_html(add_stdout.getvalue().strip()),
  280. "form": AddLinkForm()
  281. })
  282. else:
  283. context["form"] = form
  284. return render(template_name='add.html', request=request, context=context)
  285. admin.site = ArchiveBoxAdmin()
  286. admin.site.register(get_user_model())
  287. admin.site.register(Snapshot, SnapshotAdmin)
  288. admin.site.register(Tag, TagAdmin)
  289. admin.site.register(ArchiveResult, ArchiveResultAdmin)
  290. admin.site.disable_action('delete_selected')