Image pager

From Meta, a Wikimedia project coordination wiki

It would be nice to be able to dynamically link through the list of images in a category, rather than backing up to the category page itself. (This would make mediawiki a viable gallery program!)

Below are some implemenation ideas. No PHP work has been done on this topic, but a mediawiki template in conjunction with a bot can get a decent working version.

Example category[edit]

The category "foo" used in the examples on this page contains ten images (a, b, c, ... j) and three articles (1, 2, 3).

Example syntax[edit]

Like a namespace[edit]


This syntax allows parameters, so we could say [[Pager:foo|images]] to only page between images, and [[Pager:foo|articles]] for only articles (plus "all" and perhaps other restrictions on what gets paged). A big con is that it is not actually a namespace; it is one-to-one with the Category: namespace. A fix to this could be along the lines of [[Special:Pager]].

Like a template[edit]


This acts almost identically to the namespace suggestion, allowing further parameters.

Like a tag[edit]

This syntax would allow multiple categories to be stringed together in one pager:


This would create a simple pager which notes the order of the articles and images listed in the category view. It presents a link to the first, previous, current, next, and last article/image in the list, perhaps looking like this:

Example pager[edit]

This is the pager that Adam Katz uses; it is positioned at the top of this document.

Implementation w/out a patch[edit]

This is up live in a functional form; see It uses two templates; one for pagers overall and one for each category:


<!-- pager template syntax: {{pager|category|first|prev|next|last}} -->
<div style="background:white;color:black;border:dashed thin gray;padding:0 1em 0 1em;position:absolute;top:1em;right:1em">
[[:Category:{{{1}}}|<b>{{{1}}} pager</b>]]




With my hacks, you could call this with:


Creating the pagers w/out a patch[edit]

Creating the pager for each image is tedious; the Pywikipediabot can help, but even that is quite messy. Here's my implemenation:

echo "y" >yes     # (I don't like the 'yes' command)
export f=current
links -dump-width 300 -dump "" \
  |sed -e '/http:/!d' -e 's/^.*http.*wiki\///' |uniq >$f

# now manually edit the file
vim $f
# - remove all lines before the first entry and after the last entry
# - duplicate the first and last entries
#   (so {a, b, c} would be {a, a, b, c, c}, with each entry on its own line)

export n=2
# loop through each entry (excluding first and last)
while [ $n -lt `sed -n '$=' $f` ]
  do python \
    -page:`sed -n ${n}p $f` \
    -regex "(\]\])$" \
    "\1\n{{pager-foo|:`sed -n $((n-1))p $f`|:`sed -n $((n+1))p $f`}}" \
  # increment counter for loop
  export n=$((n+1))

The python pywikipediabot replace script:

  • -page... will modify the current page (line n of our list)
  • -regex... best way I know of to only replace once (how do we just match the EOF? That would be ideal.) ... This is the stupid part; it needs exactly one link at the end of a line in the entire description.
  • "\1\n{{pager-foo|... this inserts the template on a new line, pulling the previous and next lines from our list
  • <yes... we're prompted to accept the changes. I don't like the yes command, so we created a file to feed it earlier.