stringFuncs.decodeCSV - Code Metrics - Inspection of "build" - nightfury7/universalbasics.js - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 1709f1...334cbf )

by Night

created 2018-09-11 07:32 UTC

stringFuncs.decodeCSV F

↳ Parent: src/ub.files.csv.js

Complexity

Conditions	24
Paths	844

Size

Total Lines	133
Code Lines	61

Duplication

Lines	0
Ratio	0 %

Importance

Changes	3
Bugs	1	Features	1

Metric	Value
cc	24
eloc	61
nc	844
nop	4
dl	0
loc	133
rs	0
c	3
b	1
f	1

How to fix Long Method Complexity

/** global: UB */
//removeIf(nodejs)




var arrayFuncs = {


	/** Fast and simple CSV encoder.
		Supports multi-line values and values with double quotes.
		Automatic/smart escaping like Excel-generated CSVs. */
	encodeCSV: function(headers, trimValues, columnar, seperator = ",") {
		var linesData = this;
		
		if (columnar) {
			linesData = linesData.transpose();
		}
		
		var sb = [];
		
		// per line
		for (var l = (headers != null ? -1 : 0), ll = linesData.length - 1; l <= ll; l++) {
			
			// fetch header / line data words
			var words = l === -1 ? headers : linesData[l];
			
			// per word
			for (var v = 0, vl = words.length - 1; v <= vl; v++) {
				
				// if value given
				if (words[v] != null) {
					
					// convert value to text
					var word = words[v].toString();
					if (trimValues) {
						word = word.trim();
					}
					
					// add word
					if (word.indexOf('"') > -1) {
						
						// escape quotes, and enclose in quotes if word has a quote
						sb.push('"');
						sb.push(word.split("\"").join("\"\""));
						sb.push('"');
						
					} else if (word.indexOf(',') > -1) {
						
						// enclose in quotes if word has comma
						sb.push('"');
						sb.push(word);
						sb.push('"');
						
					} else {
						
						// add straightaway otherwise
						sb.push(word);
					}
				}
				
				// add seperator
				if (v < vl) {
					sb.push(seperator);
				}
			}
			
			// add newline
			if (l < ll) {
				sb.push("\r\n");
			}
		}
		
		return sb.join("");
	},
	

	/** Fast and simple TSV encoder.
		Supports multi-line values and values with double quotes.
		Automatic/smart escaping like Excel-generated TSV. */
	encodeTSV: function(headers, trimValues, columnar) {
		var linesData = this;
		return linesData.encodeCSV(headers, trimValues, columnar, "\t");
	},

    none:null,
}

// register funcs
UB.registerFuncs(Array.prototype, arrayFuncs);



var stringFuncs = {

	/** Decodes the given CSV file string and returns the cell data as an array.
	 *  Extremely robust and fast CSV parser. Only parser that works with all bizarre but valid test files.
	 * 
	 * 3 modes are available:
	 * if `headers` is null - All cells are returned as 2D array. (default)
	 * if `headers` is given and 0 length - The first row is stored in `headers`, remaining cells and returned as 2D array.
	 * if `headers` is given and >0 length - All rows are returned as objects, with the given headers treated as the prop names for the objects.
	 */
	decodeCSV: function (headers, trimValues, columnar = false, seperator = ",") {
		var csvString = this.toString();
	
		// cut String into lines
		var lines = csvString.trim().splitLines();
		var sep = seperator == "auto" ? UB.CSV_detectSeperator(csvString, lines.length) : seperator;
		
		// config
		var hasHeaders = headers != null;
		var returnAsObjs = hasHeaders && headers.exists();

		// status
		var inQuoted = false;
		
		// result
		var linesData = [];
		var word = [];
		var tempHeaders = [];
		var lineWords = [];

		// per line
		for (var l = 0, ll = lines.length; l < ll; l++) {
			var line = lines[l];
			var isHeader = (l === 0 && hasHeaders);
			
			// if we are in quoted text
			if (inQuoted) {

				
				// keep taking chars
				
			}else{
				
				// save words into headers array / new array
				lineWords = [];
				if (isHeader){
					if (returnAsObjs){
						lineWords = tempHeaders;
					}else{
						lineWords = headers;
					}
				}
				if (!isHeader) {
					linesData.push(lineWords);
				}
				
			}
			
			// per char
			for (var c = 0, clast = line.length - 1; c <= clast; c++) {
				var ch = line.charAt(c);
				
				// if we are in quoted text
				if (inQuoted) {
					
					// quotes..
					if (ch == "\"") {
						
						// quote may be escaped
						if (line.charAt(c + 1) == "\"") {
							c++;
							word.push("\"");
						}else {
							
							// quote means ending quoted text
							inQuoted = false;
						}
						
						continue;
					}
					
					// normal char
					word.push(ch);
					
					
				}else {
					
					// quote means beginning quoted text
					if (ch == "\""){
						inQuoted = true;
						continue;
					}
					
					// comma means end of word
					if (ch == sep) {
						lineWords.push(trimValues ? word.join("").trim() : word.join(""));
						word = [];
						continue;
					}
					
					// normal char
					word.push(ch);
					
					// newline means end of word
					if (c == clast) {
						lineWords.push(trimValues ? word.join("").trim() : word.join(""));
						word = [];
					}
					
				}
				
			}
			
			// at end of line take word
			if (!inQuoted && word.Length > 0) {
				lineWords.push(trimValues ? word.join("").trim() : word.join(""));
				word = [];
			}
		}

		// convert array to objs
		if (returnAsObjs){

			// go thru all rows
			for (var l = 0, ll = linesData.length; l < ll; l++) {

				var row = linesData[l];
				var obj = {};

				// convert all cells to obj props
				for (var h = 0, hl = headers.length; h < hl; h++) {
					var header = headers[h];
					obj[header] = row[h];
				}
				linesData[l] = obj;
			}
		}
		
		// convert 2D array to columnar
		if (columnar && !returnAsObjs) {
			linesData = linesData.transpose();
		}

		return linesData;
	},

	/** Decodes the given TSV file string and returns the cell data as an array.
	 *  Extremely robust and fast TSV parser. Only parser that works with all bizarre but valid test files.
	 * 
	 * 3 modes are available:
	 * if `headers` is null - All cells are returned as 2D array. (default)
	 * if `headers` is given and 0 length - The first row is stored in `headers`, remaining cells and returned as 2D array.
	 * if `headers` is given and >0 length - All rows are returned as objects, with the given headers treated as the prop names for the objects.
	 */
	decodeTSV: function (headers, trimValues, columnar = false) {
		var tsvString = this.toString();
		return tsvString.decodeCSV(headers, trimValues, columnar, "\t");
	},

    none:null,
}

// register funcs
UB.registerFuncs(String.prototype, stringFuncs);





/* File Utils - NodeJS only */

var fs = require('fs');
var pathUtil = require('path');

arrayFuncs = {

	saveToCSV: function(filePath, headers = null, trimValues = true, columnar = false, fixedSep = ",") {
		var data = this;

		// write CSV to string
		var str = data.encodeCSV(headers, trimValues, columnar, fixedSep);
		
		// save data as string via filestream
		str.saveToText(filePath);
	},

    none:null,
}

// register funcs
UB.registerFuncs(Array.prototype, arrayFuncs);


stringFuncs = {

	/**
	 * Loads the given CSV file, decoding the cell data and returning it as an array.
	 * Extremely robust and fast CSV parser. Only parser that works with all bizarre but valid test files.
	 * 
	 * 3 modes are available:
	 * if `headers` is null - All cells are returned as 2D array. (default)
	 * if `headers` is given and 0 length - The first row is stored in `headers`, remaining cells and returned as 2D array.
	 * if `headers` is given and >0 length - All rows are returned as objects, with the given headers treated as the prop names for the objects.
	 */
	loadCSV: function(encoding = "utf8", headers = null, trimValues = true, columnar = false, seperator = "auto") {

		// load text file
		var file = this.toString();
		var text = file.loadText(encoding);
		if (text == null) {
			return null;
		}
		
		// parse CSV string into Array
		return text.decodeCSV(headers, trimValues, columnar, seperator);
	},

    none:null,
}

// register funcs
UB.registerFuncs(String.prototype, stringFuncs);

//endRemoveIf(nodejs)





// UTILS
UB.CSV_seperators = [',', ';', ':', '\t'];

UB.CSV_detectSeperator = function(csvString, rowCount) {

	var sepCount = UB.newArray(0, UB.CSV_seperators.length);

	var character;

	var quoted = false;
	var firstChar = true;
	var foundAny = false;

	var c = 0;
	var cl = csvString.length - 50;  // skip last few chars
	while (c < cl) {
		character = csvString.charAt(c);
		c++;

		switch (character) {
			case '"':
				if (quoted) {
					if (csvString.charAt(c + 1) != '"') {  // Value is quoted and current character is " and next character is not ".
						quoted = false;
					} else {
						c++;  // Value is quoted and current and next characters are "" - read (skip) peeked qoute.
					}
				} else {
					if (firstChar) {  // Set value as quoted only if this quote is the first char in the value.
						quoted = true;
					}
				}
				break;
			case '\r':
			case '\n':
				if (!quoted) {
					firstChar = true;
					continue;
				}
				break;
			default:
				if (!quoted) {
					var index = UB.CSV_seperators.indexOf(character);
					if (index != -1) {
						sepCount[index]++;
						firstChar = true;
						foundAny = true;
						continue;
					}
				}
				break;
		}

		if (firstChar) {
			firstChar = false;
		}
	}

	return !foundAny ? ',' : UB.CSV_seperators[sepCount.indexOfMax()];
}




1			/** global: UB */
2			//removeIf(nodejs)
3
4
5
6
7			var arrayFuncs = {
8
9
10			/** Fast and simple CSV encoder.
11			Supports multi-line values and values with double quotes.
12			Automatic/smart escaping like Excel-generated CSVs. */
13			encodeCSV: function(headers, trimValues, columnar, seperator = ",") {
14			var linesData = this;
15
16			if (columnar) {
17			linesData = linesData.transpose();
18			}
19
20			var sb = [];
21
22			// per line
23			for (var l = (headers != null ? -1 : 0), ll = linesData.length - 1; l <= ll; l++) {
24
25			// fetch header / line data words
26			var words = l === -1 ? headers : linesData[l];
27
28			// per word
29			for (var v = 0, vl = words.length - 1; v <= vl; v++) {
30
31			// if value given
32			if (words[v] != null) {
33
34			// convert value to text
35			var word = words[v].toString();
36			if (trimValues) {
37			word = word.trim();
38			}
39
40			// add word
41			if (word.indexOf('"') > -1) {
42
43			// escape quotes, and enclose in quotes if word has a quote
44			sb.push('"');
45			sb.push(word.split("\"").join("\"\""));
46			sb.push('"');
47
48			} else if (word.indexOf(',') > -1) {
49
50			// enclose in quotes if word has comma
51			sb.push('"');
52			sb.push(word);
53			sb.push('"');
54
55			} else {
56
57			// add straightaway otherwise
58			sb.push(word);
59			}
60			}
61
62			// add seperator
63			if (v < vl) {
64			sb.push(seperator);
65			}
66			}
67
68			// add newline
69			if (l < ll) {
70			sb.push("\r\n");
71			}
72			}
73
74			return sb.join("");
75			},
76
77
78			/** Fast and simple TSV encoder.
79			Supports multi-line values and values with double quotes.
80			Automatic/smart escaping like Excel-generated TSV. */
81			encodeTSV: function(headers, trimValues, columnar) {
82			var linesData = this;
83			return linesData.encodeCSV(headers, trimValues, columnar, "\t");
84			},
85
86			none:null,
87			}
88
89			// register funcs
90			UB.registerFuncs(Array.prototype, arrayFuncs);
91
92
93
94			var stringFuncs = {
95
96			/** Decodes the given CSV file string and returns the cell data as an array.
97			* Extremely robust and fast CSV parser. Only parser that works with all bizarre but valid test files.
98			*
99			* 3 modes are available:
100			* if `headers` is null - All cells are returned as 2D array. (default)
101			* if `headers` is given and 0 length - The first row is stored in `headers`, remaining cells and returned as 2D array.
102			* if `headers` is given and >0 length - All rows are returned as objects, with the given headers treated as the prop names for the objects.
103			*/
104			decodeCSV: function (headers, trimValues, columnar = false, seperator = ",") {
105			var csvString = this.toString();
106
107			// cut String into lines
108			var lines = csvString.trim().splitLines();
109			var sep = seperator == "auto" ? UB.CSV_detectSeperator(csvString, lines.length) : seperator;
110
111			// config
112			var hasHeaders = headers != null;
113			var returnAsObjs = hasHeaders && headers.exists();
114
115			// status
116			var inQuoted = false;
117
118			// result
119			var linesData = [];
120			var word = [];
121			var tempHeaders = [];
122			var lineWords = [];
123
124			// per line
125			for (var l = 0, ll = lines.length; l < ll; l++) {
126			var line = lines[l];
127			var isHeader = (l === 0 && hasHeaders);
128
129			// if we are in quoted text
130			if (inQuoted) {
			0 ignored issues – show Comprehensibility Documentation Best Practice introduced 2018-07-12 06:57 UTC by Report Bug Copy Issue Report This code block is empty. Consider removing it or adding a comment to explain. Loading history...
131
132			// keep taking chars
133
134			}else{
135
136			// save words into headers array / new array
137			lineWords = [];
138			if (isHeader){
139			if (returnAsObjs){
140			lineWords = tempHeaders;
141			}else{
142			lineWords = headers;
143			}
144			}
145			if (!isHeader) {
146			linesData.push(lineWords);
147			}
148
149			}
150
151			// per char
152			for (var c = 0, clast = line.length - 1; c <= clast; c++) {
153			var ch = line.charAt(c);
154
155			// if we are in quoted text
156			if (inQuoted) {
157
158			// quotes..
159			if (ch == "\"") {
160
161			// quote may be escaped
162			if (line.charAt(c + 1) == "\"") {
163			c++;
164			word.push("\"");
165			}else {
166
167			// quote means ending quoted text
168			inQuoted = false;
169			}
170
171			continue;
172			}
173
174			// normal char
175			word.push(ch);
176
177
178			}else {
179
180			// quote means beginning quoted text
181			if (ch == "\""){
182			inQuoted = true;
183			continue;
184			}
185
186			// comma means end of word
187			if (ch == sep) {
188			lineWords.push(trimValues ? word.join("").trim() : word.join(""));
189			word = [];
190			continue;
191			}
192
193			// normal char
194			word.push(ch);
195
196			// newline means end of word
197			if (c == clast) {
198			lineWords.push(trimValues ? word.join("").trim() : word.join(""));
199			word = [];
200			}
201
202			}
203
204			}
205
206			// at end of line take word
207			if (!inQuoted && word.Length > 0) {
208			lineWords.push(trimValues ? word.join("").trim() : word.join(""));
209			word = [];
210			}
211			}
212
213			// convert array to objs
214			if (returnAsObjs){
215
216			// go thru all rows
217			for (var l = 0, ll = linesData.length; l < ll; l++) {
			0 ignored issues – show Comprehensibility Naming Best Practice introduced 2018-07-12 12:22 UTC by Report Bug Copy Issue Report The variable `ll` already seems to be declared on line `125`. Consider using another variable name or omitting the `var` keyword. This check looks for variables that are declared in multiple lines. There may be several reasons for this. In the simplest case the variable name was reused by mistake. This may lead to very hard to locate bugs. If you want to reuse a variable for another purpose, consider declaring it at or near the top of your function and just assigning to it subsequently so it is always declared. Loading history... Comprehensibility Naming Best Practice introduced 2018-07-12 12:22 UTC by Report Bug Copy Issue Report The variable `l` already seems to be declared on line `125`. Consider using another variable name or omitting the `var` keyword. This check looks for variables that are declared in multiple lines. There may be several reasons for this. In the simplest case the variable name was reused by mistake. This may lead to very hard to locate bugs. If you want to reuse a variable for another purpose, consider declaring it at or near the top of your function and just assigning to it subsequently so it is always declared. Loading history...
218			var row = linesData[l];
219			var obj = {};
220
221			// convert all cells to obj props
222			for (var h = 0, hl = headers.length; h < hl; h++) {
223			var header = headers[h];
224			obj[header] = row[h];
225			}
226			linesData[l] = obj;
227			}
228			}
229
230			// convert 2D array to columnar
231			if (columnar && !returnAsObjs) {
232			linesData = linesData.transpose();
233			}
234
235			return linesData;
236			},
237
238			/** Decodes the given TSV file string and returns the cell data as an array.
239			* Extremely robust and fast TSV parser. Only parser that works with all bizarre but valid test files.
240			*
241			* 3 modes are available:
242			* if `headers` is null - All cells are returned as 2D array. (default)
243			* if `headers` is given and 0 length - The first row is stored in `headers`, remaining cells and returned as 2D array.
244			* if `headers` is given and >0 length - All rows are returned as objects, with the given headers treated as the prop names for the objects.
245			*/
246			decodeTSV: function (headers, trimValues, columnar = false) {
247			var tsvString = this.toString();
248			return tsvString.decodeCSV(headers, trimValues, columnar, "\t");
249			},
250
251			none:null,
252			}
253
254			// register funcs
255			UB.registerFuncs(String.prototype, stringFuncs);
256
257
258
259
260
261			/* File Utils - NodeJS only */
262
263			var fs = require('fs');
264			var pathUtil = require('path');
265
266			arrayFuncs = {
267
268			saveToCSV: function(filePath, headers = null, trimValues = true, columnar = false, fixedSep = ",") {
269			var data = this;
270
271			// write CSV to string
272			var str = data.encodeCSV(headers, trimValues, columnar, fixedSep);
273
274			// save data as string via filestream
275			str.saveToText(filePath);
276			},
277
278			none:null,
279			}
280
281			// register funcs
282			UB.registerFuncs(Array.prototype, arrayFuncs);
283
284
285			stringFuncs = {
286
287			/**
288			* Loads the given CSV file, decoding the cell data and returning it as an array.
289			* Extremely robust and fast CSV parser. Only parser that works with all bizarre but valid test files.
290			*
291			* 3 modes are available:
292			* if `headers` is null - All cells are returned as 2D array. (default)
293			* if `headers` is given and 0 length - The first row is stored in `headers`, remaining cells and returned as 2D array.
294			* if `headers` is given and >0 length - All rows are returned as objects, with the given headers treated as the prop names for the objects.
295			*/
296			loadCSV: function(encoding = "utf8", headers = null, trimValues = true, columnar = false, seperator = "auto") {
297
298			// load text file
299			var file = this.toString();
300			var text = file.loadText(encoding);
301			if (text == null) {
302			return null;
303			}
304
305			// parse CSV string into Array
306			return text.decodeCSV(headers, trimValues, columnar, seperator);
307			},
308
309			none:null,
310			}
311
312			// register funcs
313			UB.registerFuncs(String.prototype, stringFuncs);
314
315			//endRemoveIf(nodejs)
316
317
318
319
320
321			// UTILS
322			UB.CSV_seperators = [',', ';', ':', '\t'];
323
324			UB.CSV_detectSeperator = function(csvString, rowCount) {
			0 ignored issues – show Unused Code introduced 2018-07-12 06:57 UTC by Report Bug Copy Issue Report The parameter `rowCount` is not used and could be removed. This check looks for parameters in functions that are not used in the function body and are not followed by other parameters which are used inside the function. Loading history...
325			var sepCount = UB.newArray(0, UB.CSV_seperators.length);
326
327			var character;
328
329			var quoted = false;
330			var firstChar = true;
331			var foundAny = false;
332
333			var c = 0;
334			var cl = csvString.length - 50; // skip last few chars
335			while (c < cl) {
336			character = csvString.charAt(c);
337			c++;
338
339			switch (character) {
340			case '"':
341			if (quoted) {
342			if (csvString.charAt(c + 1) != '"') { // Value is quoted and current character is " and next character is not ".
343			quoted = false;
344			} else {
345			c++; // Value is quoted and current and next characters are "" - read (skip) peeked qoute.
346			}
347			} else {
348			if (firstChar) { // Set value as quoted only if this quote is the first char in the value.
349			quoted = true;
350			}
351			}
352			break;
353			case '\r':
354			case '\n':
355			if (!quoted) {
356			firstChar = true;
357			continue;
358			}
359			break;
360			default:
361			if (!quoted) {
362			var index = UB.CSV_seperators.indexOf(character);
363			if (index != -1) {
364			sepCount[index]++;
365			firstChar = true;
366			foundAny = true;
367			continue;
368			}
369			}
370			break;
371			}
372
373			if (firstChar) {
374			firstChar = false;
375			}
376			}
377
378			return !foundAny ? ',' : UB.CSV_seperators[sepCount.indexOfMax()];
379			}
380
381
382

nightfury7 / universalbasics.js

Push — master ( 1709f1...334cbf )

stringFuncs.decodeCSV F

Complexity

Size

Duplication

Importance

How to fix Long Method Complexity

Long Method

Complexity

Duplication Side-by-Side

Filter issues like