mirror of
https://github.com/gchq/CyberChef.git
synced 2024-11-02 14:11:02 +01:00
Tidied up 'To Table' operation, adding better CSV parsing support.
This commit is contained in:
parent
8fc5f59647
commit
8556bdcdeb
@ -737,37 +737,43 @@ const Utils = {
|
||||
* Parses CSV data and returns it as a two dimensional array or strings.
|
||||
*
|
||||
* @param {string} data
|
||||
* @param {string[]} [cellDelims=[","]]
|
||||
* @param {string[]} [lineDelims=["\n", "\r"]]
|
||||
* @returns {string[][]}
|
||||
*
|
||||
* @example
|
||||
* // returns [["head1", "head2"], ["data1", "data2"]]
|
||||
* Utils.parseCSV("head1,head2\ndata1,data2");
|
||||
*/
|
||||
parseCSV: function(data) {
|
||||
|
||||
parseCSV: function(data, cellDelims=[","], lineDelims=["\n", "\r"]) {
|
||||
let b,
|
||||
ignoreNext = false,
|
||||
next,
|
||||
renderNext = false,
|
||||
inString = false,
|
||||
cell = "",
|
||||
line = [],
|
||||
lines = [];
|
||||
|
||||
// Remove BOM, often present in Excel CSV files
|
||||
if (data.length && data[0] === "\uFEFF") data = data.substr(1);
|
||||
|
||||
for (let i = 0; i < data.length; i++) {
|
||||
b = data[i];
|
||||
if (ignoreNext) {
|
||||
next = data[i+1] || "";
|
||||
if (renderNext) {
|
||||
cell += b;
|
||||
ignoreNext = false;
|
||||
renderNext = false;
|
||||
} else if (b === "\\") {
|
||||
cell += b;
|
||||
ignoreNext = true;
|
||||
renderNext = true;
|
||||
} else if (b === "\"" && !inString) {
|
||||
inString = true;
|
||||
} else if (b === "\"" && inString) {
|
||||
inString = false;
|
||||
} else if (b === "," && !inString) {
|
||||
if (next === "\"") renderNext = true;
|
||||
else inString = false;
|
||||
} else if (!inString && cellDelims.indexOf(b) >= 0) {
|
||||
line.push(cell);
|
||||
cell = "";
|
||||
} else if ((b === "\n" || b === "\r") && !inString) {
|
||||
} else if (!inString && lineDelims.indexOf(b) >= 0) {
|
||||
line.push(cell);
|
||||
cell = "";
|
||||
lines.push(line);
|
||||
|
@ -67,7 +67,6 @@ const Categories = [
|
||||
"Encode text",
|
||||
"Decode text",
|
||||
"Swap endianness",
|
||||
"To Table",
|
||||
]
|
||||
},
|
||||
{
|
||||
@ -183,6 +182,7 @@ const Categories = [
|
||||
"To Lower case",
|
||||
"Add line numbers",
|
||||
"Remove line numbers",
|
||||
"To Table",
|
||||
"Reverse",
|
||||
"Sort",
|
||||
"Unique",
|
||||
|
@ -616,26 +616,22 @@ const OperationConfig = {
|
||||
},
|
||||
"To Table": {
|
||||
module: "Default",
|
||||
description: "Renders data as a table. Data can be split on different characters and output as a HTML or ASCII table with optional header row.",
|
||||
description: "Data can be split on different characters and rendered as an HTML or ASCII table with an optional header row.<br><br>Supports the CSV (Comma Separated Values) file format by default. Change the cell delimiter argument to <code>\\t</code> to support TSV (Tab Separated Values) or <code>|</code> for PSV (Pipe Separated Values).<br><br>You can enter as many delimiters as you like. Each character will be treat as a separate possible delimiter.",
|
||||
inputType: "string",
|
||||
outputType: "html",
|
||||
highlight: false,
|
||||
highlightReverse: false,
|
||||
manualBake: false,
|
||||
args: [
|
||||
{
|
||||
name: "Select separator",
|
||||
type: "populateOption",
|
||||
value: ToTable.SEPARATORS,
|
||||
target: 1
|
||||
},
|
||||
{
|
||||
name: "Separator",
|
||||
type: "string",
|
||||
name: "Cell delimiters",
|
||||
type: "binaryShortString",
|
||||
value: ","
|
||||
},
|
||||
{
|
||||
name: "First row header?",
|
||||
name: "Row delimiters",
|
||||
type: "binaryShortString",
|
||||
value: "\\n\\r"
|
||||
},
|
||||
{
|
||||
name: "Make first row header",
|
||||
type: "boolean",
|
||||
value: false
|
||||
},
|
||||
|
@ -2,19 +2,14 @@
|
||||
* ToTable operations.
|
||||
*
|
||||
* @author Mark Jones [github.com/justanothermark]
|
||||
* @copyright Crown Copyright 2018
|
||||
* @license Apache-2.0
|
||||
*
|
||||
* @namespace
|
||||
*/
|
||||
import Utils from "../Utils.js";
|
||||
|
||||
const ToTable = {
|
||||
/**
|
||||
* @constant
|
||||
* @default
|
||||
*/
|
||||
SEPARATORS: [
|
||||
{name: "Comma", value: ","},
|
||||
{name: "Tab", value: "\\t"},
|
||||
{name: "Pipe", value: "|"},
|
||||
{name: "Custom", value: ""}
|
||||
],
|
||||
|
||||
/**
|
||||
* @constant
|
||||
@ -25,6 +20,7 @@ const ToTable = {
|
||||
"HTML"
|
||||
],
|
||||
|
||||
|
||||
/**
|
||||
* To Table operation.
|
||||
*
|
||||
@ -33,42 +29,26 @@ const ToTable = {
|
||||
* @returns {html}
|
||||
*/
|
||||
runToTable: function (input, args) {
|
||||
let separator = args[1];
|
||||
let firstRowHeader = args[2];
|
||||
let format = args[3];
|
||||
let tableData = [];
|
||||
|
||||
// If the separator contains any tabs, convert them to tab characters.
|
||||
separator = separator.replace("\\t", "\t");
|
||||
const [cellDelims, rowDelims, firstRowHeader, format] = args;
|
||||
|
||||
// Process the input into a nested array of elements.
|
||||
let rows = input.split("\n");
|
||||
rows.forEach(function(element) {
|
||||
if (separator === "") {
|
||||
tableData.push([element]);
|
||||
} else {
|
||||
tableData.push(element.split(separator));
|
||||
}
|
||||
});
|
||||
const tableData = Utils.parseCSV(input, cellDelims.split(""), rowDelims.split(""));
|
||||
|
||||
if (!tableData.length) return "";
|
||||
|
||||
// Render the data in the requested format.
|
||||
let output = "";
|
||||
switch (format) {
|
||||
case "ASCII":
|
||||
output = asciiOutput(tableData);
|
||||
break;
|
||||
|
||||
return asciiOutput(tableData);
|
||||
case "HTML":
|
||||
default:
|
||||
output = htmlOutput(tableData);
|
||||
break;
|
||||
return htmlOutput(tableData);
|
||||
}
|
||||
|
||||
return output;
|
||||
|
||||
/**
|
||||
* Outputs an array of data as an ASCII table.
|
||||
*
|
||||
* @param {Array[]} tableData
|
||||
* @param {string[][]} tableData
|
||||
* @returns {string}
|
||||
*/
|
||||
function asciiOutput(tableData) {
|
||||
@ -137,6 +117,9 @@ const ToTable = {
|
||||
|
||||
/**
|
||||
* Outputs a table of data as a HTML table.
|
||||
*
|
||||
* @param {string[][]} tableData
|
||||
* @returns {string}
|
||||
*/
|
||||
function htmlOutput(tableData) {
|
||||
// Start the HTML output with suitable classes for styling.
|
||||
|
Loading…
Reference in New Issue
Block a user