noah/rules/wuppertal/nbn.json

[
  {
    "op": "core/column-addition-by-fetching-urls",
    "engineConfig": {
      "facets": [
        {
          "type": "list",
          "name": "id",
          "expression": "isBlank(value)",
          "columnName": "urn",
          "invert": false,
          "omitBlank": false,
          "omitError": false,
          "selection": [
            {
              "v": {
                "v": false,
                "l": "false"
              }
            }
          ],
          "selectBlank": false,
          "selectError": false
        }
      ],
      "mode": "row-based"
    },
    "baseColumnName": "urn",
    "urlExpression": "grel:'https://nbn-resolving.org/process-urn-form?identifier=' + value + '&verb=FULL&xml=on'",
    "onError": "set-to-blank",
    "newColumnName": "nbn-resolving",
    "columnInsertIndex": 2,
    "delay": 0,
    "cacheResponses": true,
    "httpHeadersJson": [
      {
        "name": "authorization",
        "value": ""
      },
      {
        "name": "user-agent",
        "value": "OpenRefine 3.4 [6443506]"
      },
      {
        "name": "accept",
        "value": "*/*"
      }
    ],
    "description": "Create column nbn-resolving at index 2 by fetching URLs based on column urn using expression grel:'https://nbn-resolving.org/process-urn-form?identifier=' + value + '&verb=FULL&xml=on'"
  },
  {
    "op": "core/column-addition",
    "engineConfig": {
      "facets": [
        {
          "type": "list",
          "name": "urn",
          "expression": "isBlank(value)",
          "columnName": "urn",
          "invert": false,
          "omitBlank": false,
          "omitError": false,
          "selection": [
            {
              "v": {
                "v": false,
                "l": "false"
              }
            }
          ],
          "selectBlank": false,
          "selectError": false
        }
      ],
      "mode": "row-based"
    },
    "baseColumnName": "nbn-resolving",
    "expression": "grel:value.parseXml().select('pidef|pidef pidef|data pidef|resolving_information pidef|url_info pidef|url')[0].htmlText()",
    "onError": "set-to-blank",
    "newColumnName": "url",
    "columnInsertIndex": 3,
    "description": "Create column url at index 3 based on column nbn-resolving using expression grel:value.parseXml().select('pidef|pidef pidef|data pidef|resolving_information pidef|url_info pidef|url')[0].htmlText()"
  },
  {
    "op": "core/column-removal",
    "columnName": "nbn-resolving",
    "description": "Remove column nbn-resolving"
  },
  {
    "op": "core/text-transform",
    "engineConfig": {
      "facets": [],
      "mode": "row-based"
    },
    "columnName": "url",
    "expression": "grel:if(isNonBlank(cells['id'].value),row.record.cells[columnName].value.join('␞'),null)",
    "onError": "keep-original",
    "repeat": false,
    "repeatCount": 10,
    "description": "Text transform on cells in column url using expression grel:if(isNonBlank(cells['id'].value),row.record.cells[columnName].value.join('␞'),null)"
  },
  {
    "op": "core/text-transform",
    "engineConfig": {
      "facets": [],
      "mode": "row-based"
    },
    "columnName": "pdf",
    "expression": "grel:if(isNonBlank(cells['id'].value),row.record.cells[columnName].value.join('␞'),null)",
    "onError": "keep-original",
    "repeat": false,
    "repeatCount": 10,
    "description": "Text transform on cells in column pdf using expression grel:if(isNonBlank(cells['id'].value),row.record.cells[columnName].value.join('␞'),null)"
  },
  {
    "op": "core/text-transform",
    "engineConfig": {
      "facets": [
        {
          "type": "list",
          "name": "id",
          "expression": "isBlank(value)",
          "columnName": "id",
          "invert": false,
          "omitBlank": false,
          "omitError": false,
          "selection": [
            {
              "v": {
                "v": false,
                "l": "false"
              }
            }
          ],
          "selectBlank": false,
          "selectError": false
        }
      ],
      "mode": "row-based"
    },
    "columnName": "url",
    "expression": "grel:if(value.contains('.pdf'),value,cells['pdf'].value)",
    "onError": "keep-original",
    "repeat": false,
    "repeatCount": 10,
    "description": "Text transform on cells in column url using expression grel:if(value.contains('.pdf'),value,cells['pdf'].value)"
  },
  {
    "op": "core/column-removal",
    "columnName": "pdf",
    "description": "Remove column pdf"
  },
  {
    "op": "core/row-removal",
    "engineConfig": {
      "facets": [
        {
          "type": "list",
          "name": "url",
          "expression": "grel:row.record.cells['url'].value.join('').toLowercase().contains('.pdf')",
          "columnName": "url",
          "invert": false,
          "omitBlank": false,
          "omitError": false,
          "selection": [
            {
              "v": {
                "v": false,
                "l": "false"
              }
            }
          ],
          "selectBlank": false,
          "selectError": false
        }
      ],
      "mode": "record-based"
    },
    "description": "Remove rows"
  }
]