[main-] handle position args as +sheet:subsheet:col:row

[midichef]

Two bug fixes: 3ea655baacc1ebe7ac2c4488d938aadb26dd8b9e - fixes setting column position in sheet Currently the starting column cannot be set, only the row. repro command: vd sample_data/sample.tsv sample_data/StatusPR.csv +2:2 This occurs because the column position changes too early, before the sheet is loaded.

1aa2d9808d79d89810bf1837d39b5f89b71d83f5 - fixes setting column position in subsheet A similar problem, changing position before the subsheet is loaded. repro: vd ~vdsrc/sample_data/employees.sqlite +:emp:5:2

And then some functionality changes: d16743d96114949aafb6f63b507ed01125f31406 - change arg parsing, from row:col to col:row Strings of the form a:b are parsed as row:column. This commit changes it to column:row. Previously, I discussed some pros and cons of this change. This is technically a breaking change, but as I noted in that comment, the form +a:b has not worked since 2.9 so it's not likely that anyone is currently using that syntax. However, if they are using +:a:b with the extra colon, this will indeed be a breaking change for them. repro: vd ~vdsrc/sample_data/sample.tsv +:1:4 (note the : before the 1. It's needed (on develop) to demonstrate the column-positioning bug, until the 2 commits above are applied.)

0a8f6ee3ca4b1b62fb514fbec7eb29e4c3b9b20b - handling args of form +a:b vs +:a:b. Comments in the code suggest that '+col:row' should apply to only the last sheet on the list, vs. adding a colon, '+:col:row', which should apply to all sheets: https://github.com/saulpw/visidata/blob/c01c2b5fe506632590c8d7f5bcf23b1afe9c1985/visidata/main.py#L90 (The suggestion comes from the phrase Empty sheetstr in startsheets. An empty sheetstr happens only when arg.split(':') has more than 2 items and pos[:-2] is '', which happens when arg follows the pattern +:a:b: https://github.com/saulpw/visidata/blob/c01c2b5fe506632590c8d7f5bcf23b1afe9c1985/visidata/main.py#L105 Currently the behavior is the reverse: +col:row applies to all sheets. repro: vd sample_data/sample.tsv sample_data/StatusPR.csv +:2:2 and look at the cursor position on each sheet. This commit changes +:a:b to apply to all sheets, and +a:b to apply only to the last sheet.

da52ac3b7b72243e25bfd4cb8e85bb95dd4eb9be - improve error for invalid sheets Attempting to index a sheet that does not exist, as with: repro: vd sample_data/sample.tsv +10:0:0 gives a traceback with IndexError: list index out of range. This commit handles the error and shows a status message of no sheet "10".

a0d4913a94d7147975c28eb3ec3ceb8014addc66 - handling subsheet names that are integers For consistency with sheets/rows/cols, this commit allows indexing subsheets with integers, like: numeric-subsheet-names.vds.txt repro: vd -f vds numeric-subsheet-names.vds.txt +0:3:0:0 This is also a change that breaks current behavior, for a subsheet that happens to have a name that is an integer. Such a subsheet would become unable to opened by using its name, as with sheets 5, 6, 7, and 8 in numeric-subsheet-names.vds.txt; +0:5:0:0 currently works, but after this commit only +0:0:0:0 would work.

Thoughts? I'm not so worried about breaking changes, because I don't think the vd + feature is heavily used.

Just to write out all the cases, here's what the args would mean after all these commits:

    +2         = row 2
    +1:        = last sheet; col 1
    +:1:       = all sheets; col 1
    +:2        = last sheet; row 2
    +1:2       = last sheet; col 1, row 2
    +:1:2      = all sheets; col 1, row 2
    +0:1:2     = sheet 0; col 1, row 2
    +:a:1:2    = last sheet, subsheet a; col 1, row 2
    +0:a:1:2   = sheet 0, subsheet a; col 1, row 2
    +0:a:b:1:2 = sheet 0, subsheet a from sheet 0, subsheet b from sheet a; col 1, row 2
    +0:0:1:2   = sheet 0, first subsheet of sheet 0, col 1, row 2
    +1::       = 2nd sheet, col 0, row 0
    +-2::      = 2nd sheet from last, col 0, row 0