Relate elements in table form from Json file with jq

问题

I'm new to jq and I have the following code to obtain tabulated the values for each element called Abc:

["Abc"], ( .. | objects | select(has("Abc")) | [.["Abc"]] ) | @tsv

This is the current output I get:

"Abc"
"4"
"2"
"1"
"9"
"3"
"2"
"4"
"9"

I would like to add 4 columns to the left to show for each Abc value the corresponding page, row and column. Additionally if possible as first column add a counter from 1 to number of "Abc" elements.

Below I show the current output, compared with the desired output and the structure of the Json file in order to clarify:

The input Json file is below:

{
  "document": {
    "page": [
      {
        "@index": "0",
        "image": {
          "Abc": "4"
        }
      },
      {
        "@index": "1",
        "row": [
          {
            "column": [
              {
                "text": {
                  "Abc": "2"
                }
              }
            ]
          },
          {
            "column": [
              {
                "text": {
                  "Abc": "1"
                }
              },
              {
                "text": {
                  "Abc": "9"
                }
              }
            ]
          },
          {
            "column": [
              {
                "text": {
                  "Abc": "3"
                }
              }
            ]
          }
        ]
      },
      {
        "@index": "2",
        "row": [
          {
            "column": [
              {
                "text": {
                  "Abc": "2"
                }
              }
            ]
          },
          {
            "column": [
              {
                "text": {
                  "Abc": "4"
                }
              },
              {
                "text": {
                  "Abc": "9"
                }
              }
            ]
          }
        ]
      }
    ]
  }
}

I hope someone could help me. Thanks in advance.

回答1:

The following solution uses paths and has several advantages, including brevity, simplicity, and that it can easily be adapted to a handle data that is in a different format.

For clarity, we begin by defining a function that adds the row numbers:

# add a sequential id, starting at 1
def tsvRows(s):
  foreach s as $s (0; .+1; [.] + $s)
  | @tsv;

(["counter", "page", "row", "column", "Abc"] | @tsv),
tsvRows(paths as $p
  | select($p[-1] == "Abc")
  | getpath($p) as $v
  | $p
  | .[2] as $page
  | (if .[3] == "row" then .[4] else null end) as $row
  | (if .[5] == "column" then .[6] else null end) as $column
  | [$page, $row, $column, $v] )

回答2:

The irregularity of the input data makes the requirements a little opaque, but the following produces the desired output.

["counter", "page", "row", "column", "Abc"],
(foreach (.document.page[] | objects) as $page ({page: -1, counter: 0};
  .page += 1
  | if ($page | (has("image") and (.image|has("Abc"))))
    then
      .counter +=1
      | .out = [.counter, .page, null, null, ($page|.image.Abc)]
    else foreach ($page | .row[]?) as $row (.row=-1;
      .row += 1
      | foreach ($row | .column[]) as $column (.column=-1;
          .column +=1
          | foreach ($column | .text | objects) as $x (.;
              .counter += 1
              | .out = [.counter, .page, .row, .column, $x["Abc"]]
              ; . )
           ; . )
      ; . )
    end
    ; .out )
)
| @tsv

Output

Specifically, with the -r command-line option, the output produced from the given input is as follows (tabs included):

counter page    row column  Abc
1   0           4
2   1   0   0   2
3   1   1   0   1
4   1   1   1   9
5   1   2   0   3
6   2   0   0   2
7   2   1   0   4
8   2   1   1   9

回答3:

even though the question was for jq, for those who interested in the alternative solutions, here's one based on a walk-path unix utility jtc:

bash $ <file.json jtc -w' ' -T0 -w'[0][page][0]<p>k<Abc>l' -T'"{p}\t\t\t{}"' -w'[0][page][1:]<p>k[row][:]<r>k[column][:]<c>k<Abc>l' -T'"{p}\t{r}\t{c}\t{}"' -j | jtc -qqw' ' -T'"#\tpage\trow\tcolumn\tAbc"' -w'[1:]<#>k<v>v' -T'"{#}\t{v}"'
#       page    row     column  Abc
1       0                       4
2       1       0       0       2
3       1       1       0       1
4       1       1       1       9
5       1       2       0       3
6       2       0       0       2
7       2       1       0       4
8       2       1       1       9
bash $

in the first jtc call, first dummy walk-path (-w' ') is need only for purposes of offsetting start index to 1 (otherwise column # would be displayed starting from idx 0)

second walk processes the first page (0) only,

third walk-path processes the rest of JSON.

PS> Disclosure: I'm the creator of the jtc - shell cli tool for JSON operations

来源：https://stackoverflow.com/questions/56386521/relate-elements-in-table-form-from-json-file-with-jq

标签

json

tabulate