Handle cells with rowspan when hiding table rows
algorithm, html, javascript, jquery
Solution
OK I really spent a hell of a long time over this question, so here goes...
For those of you who just want to see the working solution, click here
Update: I've changed the visual columns calculation method to iterate over the table and create a 2-dimensional array, to see the old method which used the jQuery `offset()` method, click here. The code is shorter, but more time costly.
The problem exists because when we hide a row, whilst we want all the cells to be hidden, we want the pseudo-cells — that is, the cells that appear to be in the following rows due to the cells `rowspan` attribute — to persist. To get around this, whenever we come across a hidden cell with a `rowspan`, we try to move it down the the next visible row (decrementing it's `rowspan` value as we go). With either our original cell or it's clone, we then iterate down the table once more for every row that would contain a pseudo-cell, and if the row is hidden we decrement the `rowspan` again. (To understand why, look at the working example, and note that when the blue row is hidden, red cell 9's rowspan must be reduced from 2 to 1, else it would push green 9 right).
With that in mind, we must apply the following function whenever rows are shown/hidden:
function calculate_rowspans() {
// Remove all temporary cells
$(".tmp").remove();
// We don't care about the last row
// If it's hidden, it's cells can't go anywhere else
$("tr").not(":last").each(function() {
var $tr = $(this);
// Iterate over all non-tmp cells with a rowspan
$("td[rowspan]:not(.tmp)", $tr).each(function() {
$td = $(this);
var $rows_down = $tr;
var new_rowspan = 1;
// If the cell is visible then we don't need to create a copy
if($td.is(":visible")) {
// Traverse down the table given the rowspan
for(var i = 0; i < $td.data("rowspan") - 1; i ++) {
$rows_down = $rows_down.next();
// If our cell's row is visible then it can have a rowspan
if($rows_down.is(":visible")) {
new_rowspan ++;
}
}
// Set our rowspan value
$td.attr("rowspan", new_rowspan);
}
else {
// We'll normally create a copy, unless all of the rows
// that the cell would cover are hidden
var $copy = false;
// Iterate down over all rows the cell would normally cover
for(var i = 0; i < $td.data("rowspan") - 1; i ++) {
$rows_down = $rows_down.next();
// We only consider visible rows
if($rows_down.is(":visible")) {
// If first visible row, create a copy
if(!$copy) {
$copy = $td.clone(true).addClass("tmp");
// You could do this 1000 better ways, using classes e.g
$copy.css({
"background-color": $td.parent().css("background-color")
});
// Insert the copy where the original would normally be
// by positioning it relative to it's columns data value
var $before = $("td", $rows_down).filter(function() {
return $(this).data("column") > $copy.data("column");
});
if($before.length) $before.eq(0).before($copy);
else $(".delete-cell", $rows_down).before($copy);
}
// For all other visible rows, increment the rowspan
else new_rowspan ++;
}
}
// If we made a copy then set the rowspan value
if(copy) copy.attr("rowspan", new_rowspan);
}
});
});
}
The next, really difficult part of the question is calculating at which index to place the copies of the cells within the row. Note in the example, blue cell 2 has an actual index within its row of 0, i.e. it's the first actual cell within the row, however we can see that visually it lies in column 2 (0-indexed).
I took the approach of calculating this only once, as soon as the document is loaded. I then store this value as a data attribute of the cell, so that I can position a copy of it in the right place (I've had many Eureka moments on this one, and made many pages of notes!). To do this calculation, I ended up constructing a 2-dimensional Array `matrix` which keeps track of all of the used-visual columns. At the same time, I store the cells original `rowspan` value, as this will change with hiding/showing rows:
function get_cell_data() {
var matrix = [];
$("tr").each(function(i) {
var $cells_in_row = $("td", this);
// If doesn't exist, create array for row
if(!matrix[i]) matrix[i] = [];
$cells_in_row.each(function(j) {
// CALCULATE VISUAL COLUMN
// Store progress in matrix
var column = next_column(matrix[i]);
// Store it in data to use later
$(this).data("column", column);
// Consume this space
matrix[i][column] = "x";
// If the cell has a rowspan, consume space across
// Other rows by iterating down
if($(this).attr("rowspan")) {
// Store rowspan in data, so it's not lost
var rowspan = parseInt($(this).attr("rowspan"));
$(this).data("rowspan", rowspan);
for(var x = 1; x < rowspan; x++) {
// If this row doesn't yet exist, create it
if(!matrix[i+x]) matrix[i+x] = [];
matrix[i+x][column] = "x";
}
}
});
});
// Calculate the next empty column in our array
// Note that our array will be sparse at times, and
// so we need to fill the first empty index or push to end
function next_column(ar) {
for(var next = 0; next < ar.length; next ++) {
if(!ar[next]) return next;
}
return next;
}
}
Then simply apply this on page load:
$(document).ready(function() {
get_cell_data();
});
(Note: whilst the code here is longer than my jQuery `.offset()` alternative, it's probably quicker to calculate. Please correct me if I'm wrong).
Problem
I have a table containing cells with rowspan attributes, I would like to: - Whenever a `tr` is hidden, the table will rearrange itself correctly - Whenever a `tr` is shown again, it will be restored to original state So if you have a table like this clicking on `X` shouldn't destroy the layout. and click a `come back` button, should restore the original layout. (try removing all rows from bottom-up, and than restoring them from right-to-left, this is a desired flow) I had some semi-solutions, but all seem too complicated, and i'm sure there is a nice way to handle this.