This is a variant of summing an arbitrary column which have been answered some times lately in this forum. See thread below for a comprehensive alternative list to this issue:
This variant doesn’t have a pure query to get the needed value, so in order to address the issue of matching against your particular setup I opted for using split() with a regex delimiter, where it stores the alternatives. This resulted in the following base query:
```dataview
TABLE WITHOUT ID proj[1] as Project, proj[2] as "Some number"
FLATTEN file.lists as item
WHERE item.Size
FLATTEN list(split(item.Size, "^(.*?) - ([^-]+)$")) as proj
WHERE file = this.file
```
The regex used splits the value into two segments, one everything else except the last dash followed by a text with no dashes (aka the number at the end). These two matching groups are stored into proj, where proj[0] is everything, proj[1] is the project name and proj[2] is the number.
Combined into a dataviewjs where the last part is more or less boilerplate from the thread above:
```dataviewjs
const result = await dv.query(`
TABLE WITHOUT ID proj[1] as Project, proj[2] as "Some number"
FLATTEN file.lists as item
WHERE item.Size
FLATTEN list(split(item.Size, "^(.*?) - ([^-]+)$")) as proj
WHERE file = this.file
`)
if ( result.successful ) {
const values = result.value.values
// Find the sum of all
const sum = values.map(m => parseInt(m[1])).reduce((a,c) => a + c, 0)
result.value.values.push(["<span style='float: right; font-weight: 750'>Total</span>", sum])
dv.table(result.value.headers, result.value.values)
} else
dv.paragraph("~~~~\n" + result.error + "\n~~~~")
```
Which based on (almost) your example set returns:
