# -*- coding: utf-8 -*- """ MoinMoin - Macro for Attachment Administration Discover unused (not linked) and missing (not yet uploaded) attachments @copyright: 2015-2018, David Linke, MoinMoin:DavidLinke, https://github.com/dalito @license: GNU GPL, see COPYING for details. """ import os import sys from MoinMoin import config from MoinMoin import wikiutil from MoinMoin.PageEditor import PageEditor from MoinMoin.parser.text_moin_wiki import Parser from MoinMoin.action import AttachFile # Should only the superuser be allowed to use this macro ? SUPERUSER_ONLY = False def macro_StatusAttachments(macro): """ Purpose: Discover unused (not linked) attachments and missing attachment This macro finds and reports (i) attachments that are stored on pages put linked from nowhere (ii) attachment-links for which the attachment is not yet uploaded. The underlay directory is excluded. """ _ = macro.request.getText request = macro.request # do not show to users not in superuser list if SUPERUSER_ONLY and not request.user.isSuperUser(): error = _(u'Only superusers can use StatusAttachments macro.') # Send page with an error message request.theme.add_msg(error, 'info') return '' # do import from plugin "extensions" dir LinkParser = wikiutil.importWikiPlugin( request.cfg, 'parser', 'text_moin_wiki_links', 'LinkParser' ) # create instance of parser for link extraction lparser = LinkParser() # collect info per page in lists stored_attachments = {} attachments_no_ext = {} links_to_existing_atts = {} links_to_missing_atts = {} stored_non_linked_atts = {} # get list of all pagenames excluding underlay pages = request.rootpage.getPageDict(user=request.user, include_underlay=False) for pagename in pages: page = PageEditor(request, pagename) # (i) get linked attachments by parsing raw text raw = page.get_raw_body() atts = lparser.find_links(raw, link_filter=lparser.is_attachment) for att in atts: link = att.split(':', 1)[1] if '/' in link: storage_page, filename = link.rsplit('/', 1) else: storage_page, filename = pagename, link filename = wikiutil.taintfilename(filename) if AttachFile.exists(request, storage_page, filename): if storage_page in links_to_existing_atts: links_to_existing_atts[storage_page].append(filename) else: links_to_existing_atts[storage_page] = [filename] else: if storage_page in links_to_missing_atts: if filename not in links_to_missing_atts[storage_page]: links_to_missing_atts[storage_page].append(filename) else: links_to_missing_atts[storage_page] = [filename] # (ii) get stored attachments by looking at file system # & (iii) get stored attachments without any extension page_att_dir = AttachFile.getAttachDir(request, pagename) if os.path.isdir(page_att_dir): atts, atts_no_ext = [], [] for fname in os.listdir(page_att_dir): atts.append(fname.decode(config.charset)) if not os.path.splitext(fname)[1]: atts_no_ext.append(fname.decode(config.charset)) if atts: stored_attachments[pagename] = atts if atts_no_ext: attachments_no_ext[pagename] = atts_no_ext # when checking for availability of attachments we need to consider that # filenames are case insensitive on some platforms if sys.platform == 'win32': links_to_existing_atts_fs = {} for p in links_to_existing_atts: links_to_existing_atts_fs[p] = [os.path.normcase(att) for att in links_to_existing_atts[p]] else: links_to_existing_atts_fs = links_to_existing_atts # Find stored attachments that are not linked from anywhere for pagename in stored_attachments: for att in stored_attachments[pagename]: # look up in links_to_existing_atts if linked if (os.path.normcase(att) in links_to_existing_atts_fs.get(pagename, [])): continue # keep name of non-linked attachment if pagename in stored_non_linked_atts: stored_non_linked_atts[pagename].append(att) else: stored_non_linked_atts[pagename] = [att] # Created formatted output. We create wiki raw code that is then rendered. # Compare to using formatters this is much less verbose (but slower). # (i) attachments that are stored on pages put linked from nowhere ret = [] ret.append('<>') ret.append(u'== %s ==' % _(u'Attachments that are not linked from anywhere')) if not stored_non_linked_atts: ret.append(u" * None!") else: ret.append(u" * Found '''%i''' attachments that are not linked." % len(reduce(list.__add__, stored_non_linked_atts.values()))) for pagename in sorted(stored_non_linked_atts): ret.append(u" * '''[[%s||&action=AttachFile]]'''" % pagename) for att in sorted(stored_non_linked_atts[pagename]): ret.append(u" * `%s`" % att) # (ii) attachment-links for which the attachment is not yet uploaded. ret.append(u'== %s ==' % _(u'Attachments that are linked but missing')) if not links_to_missing_atts: ret.append(u" * None!") else: ret.append(u" * Found '''%i''' links to missing attachments." % len(reduce(list.__add__, links_to_missing_atts.values()))) for pagename in sorted(links_to_missing_atts): ret.append(u" * '''[[%s]]'''" % pagename) for att in sorted(links_to_missing_atts[pagename]): ret.append(u" * `%s`" % att) # (iii) List all stored attachments without any extension ret.append(u'== %s ==' % _(u'Attachments lacking a file extension')) if not attachments_no_ext: ret.append(u" * None!") else: ret.append(u" * Found '''%i''' attachments without extension." % len(reduce(list.__add__, attachments_no_ext.values()))) for pagename in sorted(attachments_no_ext): ret.append(u" * '''[[%s||&action=AttachFile]]'''" % pagename) for att in sorted(attachments_no_ext[pagename]): ret.append(u" * `%s`" % att) return wikiutil.renderText(request, Parser, '\n'.join(ret))