Reader::parseLine() - Code Metrics - Inspection of "Release v0.3" - nozavroni/csvelte - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#183)

by Luke

created 2018-07-25 05:29 UTC

Reader::parseLine() A

↳ Parent: Reader

Complexity

Conditions	4
Paths	3

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	16
CRAP Score	4

Importance

Changes

Metric	Value
cc	4
nc	3
nop	1
dl	0
loc	19
ccs	16
cts	16
cp	1
crap	4
rs	9.6333
c	0
b	0
f	0

<?php
/**
 * CSVelte: Slender, elegant CSV for PHP
 *
 * Inspired by Python's CSV module and Frictionless Data and the W3C's CSV
 * standardization efforts, CSVelte was written in an effort to take all the
 * suck out of working with CSV.
 *
 * @copyright Copyright (c) 2018 Luke Visinoni
 * @author    Luke Visinoni <[email protected]>
 * @license   See LICENSE file (MIT license)
 */
namespace CSVelte;

use Iterator;
use Countable;
use CSVelte\Contract\Streamable;
use Noz\Collection\Collection;
use Stringy\Stringy;

use function Noz\collect;
use function Stringy\create as s;

class Reader implements Iterator, Countable
{
    /** @var Streamable The input stream to read from */
    protected $input;

    /** @var Dialect The *dialect* of CSV to read */
    protected $dialect;

    /** @var Collection The line currently sitting in memory */
    protected $current;

    /** @var Collection The header row */
    protected $header;

    /** @var int The current line number */
    protected $lineNo;

    /** @var bool Determines whether we've reached the end of the data */
    protected $valid;

    /**
     * Reader constructor.
     *
     * Although this is the constructor, I don't envision it being used much in userland. I think much more common
     * methods of creating readers will be available within CSVelte base class such as CSVelte::fromPath(),
     * CSVelte::fromString(), CSVelte::fromSplFileObject, CSVelte::toSplFileObject, CSVelte::toPath(), etc.
     *
     * @param Streamable $input The source being read from
     * @param Dialect $dialect The dialect being read
     */
    public function __construct(Streamable $input, Dialect $dialect = null)
    {
        if (is_null($dialect)) {
            $dialect = new Dialect;
        }
        $this->setInputStream($input)
            ->setDialect($dialect);
    }

    /**
     * Get csv data as a two-dimensional array
     *
     * @return array
     */
    public function toArray()
    {
        return iterator_to_array($this);
    }

    /**
     * Set the CSV dialect
     *
     * @param Dialect $dialect The *dialect* of CSV to read
     *
     * @return self
     */
    public function setDialect(Dialect $dialect)
    {
        $this->dialect = $dialect;
        // call rewind because new dialect needs to be used to re-read
        return $this->rewind();
    }

    /**
     * Get dialect
     *
     * @return Dialect
     */
    public function getDialect()
    {
        return $this->dialect;
    }

    /**
     * Get a single row
     *
     * Get the next row from the CSV data. If no more data available, returns false.
     *
     * @param int $offset An optional row offset (negative offsets not yet supported)
     *
     * @return array|false
     */
    public function getRow($offset = null)
    {
        if (!is_null($offset)) {
            if ($offset < 0) {
                return collect($this->toArray())->getValueAt($offset);
            }
            foreach ($this as $key => $row) {
                if ($key === $offset) {
                    return $row;
                }
            }
            return false;
        }
        if (!$this->valid() || $this->input->eof()) {
            return false;
        }
        $line = $this->current();
        $this->next();
        return $line;
    }

    /**
     * Get the column at given $index
     *
     * @param int|string $index The column index
     *
     * @return array
     */
    public function getColumn($index)
    {
        return collect($this->toArray())
            ->getColumn($index)
            ->toArray();
    }

    /**
     * Set input stream
     *
     * @param Streamable $stream The input stream to read from
     *
     * @return self
     */
    protected function setInputStream(Streamable $stream)
    {
        $this->input = $stream;
        return $this;
    }

    /**
     * Loads next line into memory
     *
     * Reads from input one character at a time until a newline is reached that isn't within quotes. Once a completed
     * line has been loaded, it is assigned to the `$this->current` property. Subsequent calls will continue to load
     * successive lines until the end of the input source is reached.
     *
     * @return self
     */
    protected function loadLine()
    {
        $d = $this->getDialect();
        $line = '';
        while ($str = $this->input->readLine($d->getLineTerminator())) {
            $line .= $str;
            if (count(s($line)->split($d->getQuoteChar())) % 2) {
                break;
            }
        }
        $this->current = $this->parseLine($line);
        return $this;
    }

    /**
     * Parse a line of CSV into individual fields
     *
     * Accepts a line (string) of CSV data that it then splits at the delimiter character. The method is smart, in that
     * it knows not to split at delimiters within quotes. Ultimately, fields are placed into a collection and returned.
     *
     * @param string $line A single line of CSV data to parse into individual fields
     *
     * @return Collection
     */
    protected function parseLine($line)
    {
        $d = $this->getDialect();
        $fields = collect(s($line)
            ->trimRight($d->getLineTerminator())
            ->split(" *{$d->getDelimiter()} *(?=([^\"]*\"[^\"]*\")*[^\"]*$)"));
        if (!is_null($this->header)) {
            // @todo there may be cases where this gives a false positive...
            if (count($fields) == count($this->header)) {
                $fields = $fields->rekey($this->header);
            }
        }
        return $fields->map(function(Stringy $field, $pos) use ($d) {

            if ($d->isDoubleQuote()) {
                $field = $field->replace('""', '"');
            }
            return (string) $field->trim($d->getQuoteChar());
        });
    }

    /** == BEGIN: SPL implementation methods == */

    /**
     * Get current row
     *
     * @return array
     */
    public function current()
    {
        return $this->current->toArray();
    }

    /**
     * Move pointer to beginning of the next line internally and then load the line
     *
     * @return self
     */
    public function next()
    {
        $this->loadLine();
        $this->lineNo++;
        return $this;
    }

    /**
     * Get current line number
     *
     * @return int
     */
    public function key()
    {
        return $this->lineNo;
    }

    /**
     * Have we reached the end of the CSV data?
     *
     * @return bool
     */
    public function valid()
    {
        if ($this->valid === false) {
            return false;
        }
        if ($this->input->tell() >= $this->input->getSize()) {
            $this->valid = false;
        }
        return true;
    }

    /**
     * Rewind to the beginning
     *
     * Rewinds the internal pointer to the beginning of the CSV data, load first line, and reset line number to 1. Also
     * loads the header (if one exists) and uses its values as indexes within rows.
     *
     * @return self
     */
    public function rewind()
    {
        $this->valid = null;
        $this->lineNo = 0;
        $this->input->rewind();
        if ($this->getDialect()->hasHeader()) {
            $this->loadLine();
            $this->header = $this->current();

        }
        $this->loadLine();
        return $this;
    }

    /**
     * Get number of lines in the CSV data (not including header)
     *
     * @return int
     */
    public function count()
    {
        return count($this->toArray());
    }

    /** == END: SPL implementation methods == */
}

1		<?php
2		/**
3		* CSVelte: Slender, elegant CSV for PHP
4		*
5		* Inspired by Python's CSV module and Frictionless Data and the W3C's CSV
6		* standardization efforts, CSVelte was written in an effort to take all the
7		* suck out of working with CSV.
8		*
9		* @copyright Copyright (c) 2018 Luke Visinoni
10		* @author Luke Visinoni <[email protected]>
11		* @license See LICENSE file (MIT license)
12		*/
13		namespace CSVelte;
14
15		use Iterator;
16		use Countable;
17		use CSVelte\Contract\Streamable;
18		use Noz\Collection\Collection;
19		use Stringy\Stringy;
20
21		use function Noz\collect;
22		use function Stringy\create as s;
23
24		class Reader implements Iterator, Countable
25		{
26		/** @var Streamable The input stream to read from */
27		protected $input;
28
29		/** @var Dialect The dialect of CSV to read */
30		protected $dialect;
31
32		/** @var Collection The line currently sitting in memory */
33		protected $current;
34
35		/** @var Collection The header row */
36		protected $header;
37
38		/** @var int The current line number */
39		protected $lineNo;
40
41		/** @var bool Determines whether we've reached the end of the data */
42		protected $valid;
43
44		/**
45		* Reader constructor.
46		*
47		* Although this is the constructor, I don't envision it being used much in userland. I think much more common
48		* methods of creating readers will be available within CSVelte base class such as CSVelte::fromPath(),
49		* CSVelte::fromString(), CSVelte::fromSplFileObject, CSVelte::toSplFileObject, CSVelte::toPath(), etc.
50		*
51		* @param Streamable $input The source being read from
52		* @param Dialect $dialect The dialect being read
53		*/
54	18	public function __construct(Streamable $input, Dialect $dialect = null)
55		{
56	18	if (is_null($dialect)) {
57	11	$dialect = new Dialect;
58	11	}
59	18	$this->setInputStream($input)
60	18	->setDialect($dialect);
61	18	}
62
63		/**
64		* Get csv data as a two-dimensional array
65		*
66		* @return array
67		*/
68	6	public function toArray()
69		{
70	6	return iterator_to_array($this);
71		}
72
73		/**
74		* Set the CSV dialect
75		*
76		* @param Dialect $dialect The dialect of CSV to read
77		*
78		* @return self
79		*/
80	18	public function setDialect(Dialect $dialect)
81		{
82	18	$this->dialect = $dialect;
83		// call rewind because new dialect needs to be used to re-read
84	18	return $this->rewind();
85		}
86
87		/**
88		* Get dialect
89		*
90		* @return Dialect
91		*/
92	18	public function getDialect()
93		{
94	18	return $this->dialect;
95		}
96
97		/**
98		* Get a single row
99		*
100		* Get the next row from the CSV data. If no more data available, returns false.
101		*
102		* @param int $offset An optional row offset (negative offsets not yet supported)
103		*
104		* @return array\|false
105		*/
106	4	public function getRow($offset = null)
107		{
108	4	if (!is_null($offset)) {
109	2	if ($offset < 0) {
110	1	return collect($this->toArray())->getValueAt($offset);
111		}
112	1	foreach ($this as $key => $row) {
113	1	if ($key === $offset) {
114	1	return $row;
115		}
116	1	}
117		return false;
118		}
119	2	if (!$this->valid() \|\| $this->input->eof()) {
120	1	return false;
121		}
122	1	$line = $this->current();
123	1	$this->next();
124	1	return $line;
125		}
126
127		/**
128		* Get the column at given $index
129		*
130		* @param int\|string $index The column index
131		*
132		* @return array
133		*/
134	1	public function getColumn($index)
135		{
136	1	return collect($this->toArray())
137	1	->getColumn($index)
138	1	->toArray();
139		}
140
141		/**
142		* Set input stream
143		*
144		* @param Streamable $stream The input stream to read from
145		*
146		* @return self
147		*/
148	18	protected function setInputStream(Streamable $stream)
149		{
150	18	$this->input = $stream;
151	18	return $this;
152		}
153
154		/**
155		* Loads next line into memory
156		*
157		* Reads from input one character at a time until a newline is reached that isn't within quotes. Once a completed
158		* line has been loaded, it is assigned to the `$this->current` property. Subsequent calls will continue to load
159		* successive lines until the end of the input source is reached.
160		*
161		* @return self
162		*/
163	18	protected function loadLine()
164		{
165	18	$d = $this->getDialect();
166	18	$line = '';
167	18	while ($str = $this->input->readLine($d->getLineTerminator())) {
168	18	$line .= $str;
169	18	if (count(s($line)->split($d->getQuoteChar())) % 2) {
170	18	break;
171		}
172	11	}
173	18	$this->current = $this->parseLine($line);
174	18	return $this;
175		}
176
177		/**
178		* Parse a line of CSV into individual fields
179		*
180		* Accepts a line (string) of CSV data that it then splits at the delimiter character. The method is smart, in that
181		* it knows not to split at delimiters within quotes. Ultimately, fields are placed into a collection and returned.
182		*
183		* @param string $line A single line of CSV data to parse into individual fields
184		*
185		* @return Collection
186		*/
187	18	protected function parseLine($line)
188		{
189	18	$d = $this->getDialect();
190	18	$fields = collect(s($line)
191	18	->trimRight($d->getLineTerminator())
192	18	->split(" {$d->getDelimiter()} (?=([^\"]\"[^\"]\")[^\"]$)"));
193	18	if (!is_null($this->header)) {
194		// @todo there may be cases where this gives a false positive...
195	13	if (count($fields) == count($this->header)) {
196	13	$fields = $fields->rekey($this->header);
197	13	}
198	13	}
199	18	return $fields->map(function(Stringy $field, $pos) use ($d) {
		0 ignored issues – show Unused Code introduced 2018-07-20 01:01 UTC by Report Bug Copy Issue Report The parameter `$pos` is not used and could be removed. This check looks from parameters that have been defined for a function or method, but which are not used in the method body. Loading history...
200	18	if ($d->isDoubleQuote()) {
201	18	$field = $field->replace('""', '"');
202	18	}
203	18	return (string) $field->trim($d->getQuoteChar());
204	18	});
205		}
206
207		/** == BEGIN: SPL implementation methods == */
208
209		/**
210		* Get current row
211		*
212		* @return array
213		*/
214	17	public function current()
215		{
216	17	return $this->current->toArray();
217		}
218
219		/**
220		* Move pointer to beginning of the next line internally and then load the line
221		*
222		* @return self
223		*/
224	10	public function next()
225		{
226	10	$this->loadLine();
227	10	$this->lineNo++;
228	10	return $this;
229		}
230
231		/**
232		* Get current line number
233		*
234		* @return int
235		*/
236	10	public function key()
237		{
238	10	return $this->lineNo;
239		}
240
241		/**
242		* Have we reached the end of the CSV data?
243		*
244		* @return bool
245		*/
246	9	public function valid()
247		{
248	9	if ($this->valid === false) {
249	6	return false;
250		}
251	9	if ($this->input->tell() >= $this->input->getSize()) {
252	7	$this->valid = false;
253	7	}
254	9	return true;
255		}
256
257		/**
258		* Rewind to the beginning
259		*
260		* Rewinds the internal pointer to the beginning of the CSV data, load first line, and reset line number to 1. Also
261		* loads the header (if one exists) and uses its values as indexes within rows.
262		*
263		* @return self
264		*/
265	18	public function rewind()
266		{
267	18	$this->valid = null;
268	18	$this->lineNo = 0;
269	18	$this->input->rewind();
270	18	if ($this->getDialect()->hasHeader()) {
271	13	$this->loadLine();
272	13	$this->header = $this->current();
		0 ignored issues – show Documentation Bug introduced 2018-07-20 01:01 UTC by Report Bug Copy Issue Report It seems like `$this->current()` of type `array` is incompatible with the declared type `object<Noz\Collection\Collection>` of property `$header`. Our type inference engine has found an assignment to a property that is incompatible with the declared type of that property. Either this assignment is in error or the assigned type should be added to the documentation/type hint for that property.. Loading history...
273	13	}
274	18	$this->loadLine();
275	18	return $this;
276		}
277
278		/**
279		* Get number of lines in the CSV data (not including header)
280		*
281		* @return int
282		*/
283	1	public function count()
284		{
285	1	return count($this->toArray());
286		}
287
288		/** == END: SPL implementation methods == */
289		}

nozavroni / csvelte

Pull Request — master (#183)

Reader::parseLine() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like