Calendar Heatmap from Date List

Turns a column of dates into the GitHub-contribution-graph view of a year. Each line can be just a date (2026-03-15) for count 1, or a date with a number (2026-03-15 4) when you have weights. The grid auto-bins counts into five intensity levels relative to the year's peak, picks a palette, and surfaces the longest streak and current streak below. Useful for habit trackers, commit history copies, sales/error logs — any time-series where the rhythm matters more than the exact value.

Binning relative to the year's peak is what makes the picture readable and also what makes two charts incomparable. A single outlier day compresses everything else into the palest band, so a year with one enormous spike looks emptier than a steady year with a lower total. Two heatmaps side by side are showing different scales unless you say otherwise, and reading one against the other is the mistake the format invites.

The five levels mean the colour is an ordering rather than a quantity. You can see that Tuesdays are busier than Thursdays and that a habit lapsed in August; you cannot read a value off a square, and a step from one shade to the next is not a fixed number of units. That is the trade the format makes, and it is the right one when the question is about rhythm.

Streaks are counted in calendar days with no notion of a rest day, so a habit deliberately done five days a week shows a longest streak of five however consistent it is. A missing line is a zero rather than missing data, which matters when the source stops logging: a broken exporter and a genuinely idle month look identical here.

The grid is laid out in weeks, and which day the column starts on shifts the whole pattern. If you are comparing against a chart drawn elsewhere, check that both start the week the same way before reading anything into the shape.