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
|
|
|
* @version v0.2 |
10
|
|
|
* @copyright Copyright (c) 2016 Luke Visinoni <[email protected]> |
11
|
|
|
* @author Luke Visinoni <[email protected]> |
12
|
|
|
* @license https://github.com/deni-zen/csvelte/blob/master/LICENSE The MIT License (MIT) |
13
|
|
|
*/ |
14
|
|
|
namespace CSVelte; |
15
|
|
|
|
16
|
|
|
use \Closure; |
17
|
|
|
use \InvalidArgumentException; |
18
|
|
|
use \FilterIterator; |
19
|
|
|
use CSVelte\IO\Stream; |
20
|
|
|
use CSVelte\Contract\Readable; |
21
|
|
|
use CSVelte\Table\Row; |
22
|
|
|
use CSVelte\Table\HeaderRow; |
23
|
|
|
use CSVelte\Exception\EndOfFileException; |
24
|
|
|
use CSVelte\Reader\FilteredIterator as FilteredReader; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* CSV Reader |
28
|
|
|
* |
29
|
|
|
* Reads CSV data from any object that implements CSVelte\Contract\Readable. |
30
|
|
|
* |
31
|
|
|
* @package CSVelte |
32
|
|
|
* @subpackage Reader |
33
|
|
|
* @since v0.1 |
34
|
|
|
* @todo Also, is there any way to do some kind of caching or something? Probably |
35
|
|
|
* not but if you could that would be a cool feature... |
36
|
|
|
*/ |
37
|
|
|
class Reader implements \Iterator |
38
|
|
|
{ |
39
|
|
|
const PLACEHOLDER_DELIM = '[=[__DLIM__]=]'; |
40
|
|
|
const PLACEHOLDER_NEWLINE = '[=[__NWLN__]=]'; |
41
|
|
|
|
42
|
|
|
/** |
43
|
|
|
* This class supports any sources of input that implements this interface. |
44
|
|
|
* This way I can read from local files, streams, FTP, any class that implements |
45
|
|
|
* the "Readable" interface |
46
|
|
|
* @var \CSVelte\Contract\Readable |
47
|
|
|
*/ |
48
|
|
|
protected $source; |
49
|
|
|
|
50
|
|
|
/** |
51
|
|
|
* @var \CSVelte\Flavor The "flavor" or format of the CSV being read |
52
|
|
|
*/ |
53
|
|
|
protected $flavor; |
54
|
|
|
|
55
|
|
|
/** |
56
|
|
|
* @var \CSVelte\Table\Row|boolean Row currently loaded into memory |
57
|
|
|
*/ |
58
|
|
|
protected $current; |
59
|
|
|
|
60
|
|
|
/** |
61
|
|
|
* @var integer The current line being read (from input source) |
62
|
|
|
*/ |
63
|
|
|
protected $line = 0; |
64
|
|
|
|
65
|
|
|
/** |
66
|
|
|
* @var \CSVelte\Table\HeaderRow The header row (if any) |
67
|
|
|
*/ |
68
|
|
|
protected $header; |
69
|
|
|
|
70
|
|
|
/** |
71
|
|
|
* @var array An array of callback functions |
72
|
|
|
*/ |
73
|
|
|
protected $filters = array(); |
74
|
|
|
|
75
|
|
|
/** |
76
|
|
|
* @var bool True if current line ended while inside a quoted string |
77
|
|
|
*/ |
78
|
|
|
protected $open = false; |
79
|
|
|
|
80
|
|
|
/** |
81
|
|
|
* @var bool True if last character read was the escape character |
82
|
|
|
*/ |
83
|
|
|
protected $escape = false; |
84
|
|
|
|
85
|
|
|
/** |
86
|
|
|
* Reader Constructor. |
87
|
|
|
* Initializes a reader object using an input source and optionally a flavor |
88
|
|
|
* |
89
|
|
|
* @param \CSVelte\Contract\Readable $input The source of our CSV data |
90
|
|
|
* @param \CSVelte\Flavor $flavor The "flavor" or format specification object |
91
|
|
|
*/ |
92
|
22 |
|
public function __construct($input, $flavor = null) |
93
|
|
|
{ |
94
|
22 |
|
$this->setSource($input) |
95
|
22 |
|
->setFlavor($flavor) |
96
|
22 |
|
->rewind(); |
97
|
22 |
|
} |
98
|
|
|
|
99
|
|
|
/** |
100
|
|
|
* Set the flavor. |
101
|
|
|
* |
102
|
|
|
* Set the ``CSVelte\Flavor`` object, used to determine CSV format. |
103
|
|
|
* |
104
|
|
|
* @param \CSVelte\Flavor|array $flavor Either an array or a flavor object |
105
|
|
|
*/ |
106
|
20 |
|
protected function setFlavor($flavor = null) |
107
|
|
|
{ |
108
|
20 |
|
if (is_array($flavor)) $flavor = new Flavor($flavor); |
109
|
20 |
|
$taster = new Taster($this->source); |
110
|
|
|
// @todo put this inside a try/catch |
111
|
20 |
|
if (is_null($flavor)) { |
112
|
12 |
|
$flavor = $taster->lick(); |
113
|
12 |
|
} |
114
|
20 |
|
if (is_null($flavor->header)) { |
115
|
|
|
// Flavor is immutable, give me a new one with header set to lickHeader return val |
116
|
4 |
|
$flavor = $flavor->copy(['header' => $taster->lickHeader($flavor->delimiter, $flavor->lineTerminator)]); |
117
|
4 |
|
} |
118
|
20 |
|
$this->flavor = $flavor; |
119
|
20 |
|
return $this; |
120
|
|
|
} |
121
|
|
|
|
122
|
|
|
/** |
123
|
|
|
* Set the reader source. |
124
|
|
|
* |
125
|
|
|
* The reader can accept anything that implements Readable and is actually |
126
|
|
|
* readable (can be read). This will make sure that whatever is passed to |
127
|
|
|
* the reader meets these expectations and set $this->source. |
128
|
|
|
* |
129
|
|
|
* @param \CSVelte\Contract\Readable|object|string $input See description |
130
|
|
|
* @return $this |
131
|
|
|
*/ |
132
|
20 |
|
protected function setSource($input) |
133
|
|
|
{ |
134
|
20 |
|
if ($input instanceof Readable && $input->isReadable()) { |
135
|
9 |
|
$this->source = $input; |
136
|
9 |
|
} else { |
137
|
11 |
|
$this->source = Stream::streamize($input); |
138
|
|
|
} |
139
|
20 |
|
return $this; |
140
|
|
|
} |
141
|
|
|
|
142
|
|
|
/** |
143
|
|
|
* Load a line into memory |
144
|
|
|
* |
145
|
|
|
* @return void ($this?) |
146
|
|
|
* @access protected |
147
|
|
|
*/ |
148
|
20 |
|
protected function load() |
149
|
|
|
{ |
150
|
20 |
|
if (is_null($this->current)) { |
151
|
|
|
try { |
152
|
20 |
|
$line = $this->readLine(); |
153
|
20 |
|
$this->line++; |
154
|
20 |
|
$parsed = $this->parse($line); |
155
|
20 |
|
if ($this->hasHeader() && $this->line === 1) { |
156
|
14 |
|
$this->header = new HeaderRow($parsed); |
157
|
14 |
|
} else { |
158
|
20 |
|
$this->current = new Row($parsed); |
159
|
20 |
|
if ($this->header) $this->current->setHeaderRow($this->header); |
160
|
|
|
} |
161
|
20 |
|
} catch (EndOfFileException $e) { |
162
|
7 |
|
$this->current = false; |
163
|
|
|
} |
164
|
20 |
|
} |
165
|
20 |
|
} |
166
|
|
|
|
167
|
|
|
/** |
168
|
|
|
* Read single line from CSV data source (stream, file, etc.), taking into |
169
|
|
|
* account CSV's de-facto quoting rules with respect to designated line |
170
|
|
|
* terminator character when they fall within quoted strings. |
171
|
|
|
* |
172
|
|
|
* @return string A CSV row (could possibly span multiple lines depending on |
173
|
|
|
* quoting and escaping) |
174
|
|
|
* @throws \CSVelte\Exception\EndOfFileException when eof has been reached |
175
|
|
|
* and the read buffer has all been returned |
176
|
|
|
*/ |
177
|
20 |
|
protected function readLine() |
178
|
|
|
{ |
179
|
20 |
|
$f = $this->getFlavor(); |
180
|
20 |
|
$eol = $f->lineTerminator; |
181
|
|
|
try { |
182
|
|
|
do { |
183
|
20 |
|
if (!isset($lines)) $lines = array(); |
184
|
20 |
|
if (false === ($line = $this->source->readLine($eol))) { |
185
|
7 |
|
throw new EndOfFileException("End of file reached: " . $this->source->getName()); |
186
|
|
|
} |
187
|
20 |
|
array_push($lines, rtrim($line, $eol)); |
188
|
20 |
|
} while ($this->inQuotedString(end($lines), $f->quoteChar, $f->escapeChar)); |
189
|
20 |
|
} catch (EndOfFileException $e) { |
190
|
|
|
// only throw the exception if we don't already have lines in the buffer |
191
|
7 |
|
if (!count($lines)) throw $e; |
192
|
|
|
} |
193
|
20 |
|
return rtrim(implode($eol, $lines), $eol); |
194
|
|
|
} |
195
|
|
|
|
196
|
|
|
/** |
197
|
|
|
* Determine whether last line ended while a quoted string was still "open" |
198
|
|
|
* |
199
|
|
|
* This method is used in a loop to determine if each line being read ends |
200
|
|
|
* while a quoted string is still "open". |
201
|
|
|
* |
202
|
|
|
* @param string $line Line of csv to analyze |
203
|
|
|
* @param string $quoteChar The quote/enclosure character to use |
204
|
|
|
* @param string $escapeChar The escape char/sequence to use |
205
|
|
|
* @return bool True if currently within a quoted string |
206
|
|
|
*/ |
207
|
20 |
|
protected function inQuotedString($line, $quoteChar, $escapeChar) |
208
|
|
|
{ |
209
|
20 |
|
if (!empty($line)) { |
210
|
|
|
do { |
211
|
20 |
|
if (!isset($i)) $i = 0; |
212
|
20 |
|
$c = $line[$i++]; |
213
|
20 |
|
if ($this->escape) { |
214
|
|
|
$this->escape = false; |
215
|
|
|
continue; |
216
|
|
|
} |
217
|
20 |
|
$this->escape = ($c == $escapeChar); |
218
|
20 |
|
if ($c == $quoteChar) $this->open = !$this->open; |
219
|
20 |
|
} while ($i < strlen($line)); |
220
|
20 |
|
} |
221
|
20 |
|
return $this->open; |
222
|
|
|
} |
223
|
|
|
|
224
|
|
|
/** |
225
|
|
|
* Flavor Getter. |
226
|
|
|
* |
227
|
|
|
* Retreive the "flavor" object being used by the reader |
228
|
|
|
* |
229
|
|
|
* @return \CSVelte\Flavor |
230
|
|
|
* @access public |
231
|
|
|
*/ |
232
|
20 |
|
public function getFlavor() |
233
|
|
|
{ |
234
|
20 |
|
return $this->flavor; |
235
|
|
|
} |
236
|
|
|
|
237
|
|
|
/** |
238
|
|
|
* Check if flavor object defines header. |
239
|
|
|
* |
240
|
|
|
* Determine whether or not the input source's CSV data contains a header |
241
|
|
|
* row or not. Unless you explicitly specify so within your Flavor object, |
242
|
|
|
* this method is a logical best guess. The CSV format does not |
243
|
|
|
* provide metadata of any kind and therefor does not provide this info. |
244
|
|
|
* |
245
|
|
|
* @return boolean True if the input source has a header row (or, to be more ) |
246
|
|
|
* accurate, if the flavor SAYS it has a header row) |
247
|
|
|
* @todo Rather than always reading in Taster::SAMPLE_SIZE, read in ten lines at a time until |
248
|
|
|
* whatever method it is has enough data to make a reliable decision/guess |
249
|
|
|
*/ |
250
|
20 |
|
public function hasHeader() |
251
|
|
|
{ |
252
|
20 |
|
return $this->getFlavor()->header; |
253
|
|
|
} |
254
|
|
|
|
255
|
|
|
/** |
256
|
|
|
* Temporarily replace special characters within a quoted string |
257
|
|
|
* |
258
|
|
|
* Replace all instances of newlines and whatever character you specify (as |
259
|
|
|
* the delimiter) that are contained within quoted text. The replacements are |
260
|
|
|
* simply a special placeholder string. This is done so that I can use the |
261
|
|
|
* very unsmart "explode" function and not have to worry about it exploding |
262
|
|
|
* on delimiters or newlines within quotes. Once I have exploded, I typically |
263
|
|
|
* sub back in the real characters before doing anything else. |
264
|
|
|
* |
265
|
|
|
* @param string $data The string to do the replacements on |
266
|
|
|
* @param string $delim The delimiter character to replace |
267
|
|
|
* @param string $quo The quote character |
268
|
|
|
* @param string $eol Line terminator character/sequence |
269
|
|
|
* @return string The data with replacements performed |
270
|
|
|
* @access protected |
271
|
|
|
* @internal |
272
|
|
|
* @todo I could probably pass in (maybe optionally) the newline character I |
273
|
|
|
* want to replace as well. I'll do that if I need to. |
274
|
|
|
* @todo Create a regex class so you can do $regex->escape() rather than |
275
|
|
|
* preg_quote |
276
|
|
|
*/ |
277
|
20 |
|
protected function replaceQuotedSpecialChars($data, $delim, $quo, $eol) |
278
|
|
|
{ |
279
|
|
|
return preg_replace_callback('/(['. preg_quote($quo, '/') . '])(.*)\1/imsU', function($matches) use ($delim, $eol) { |
280
|
11 |
|
$ret = str_replace($eol, self::PLACEHOLDER_NEWLINE, $matches[0]); |
281
|
11 |
|
$ret = str_replace($delim, self::PLACEHOLDER_DELIM, $ret); |
282
|
11 |
|
return $ret; |
283
|
20 |
|
}, $data); |
284
|
|
|
} |
285
|
|
|
|
286
|
|
|
/** |
287
|
|
|
* Undo temporary special char replacements |
288
|
|
|
* |
289
|
|
|
* Replace the special character placeholders with the characters they |
290
|
|
|
* originally substituted. |
291
|
|
|
* |
292
|
|
|
* @param string $data The data to undo replacements in |
293
|
|
|
* @param string $delim The delimiter character |
294
|
|
|
* @param string $eol The character or string of characters used to terminate lines |
295
|
|
|
* @return string The data with placeholders replaced with original characters |
296
|
|
|
* @internal |
297
|
|
|
*/ |
298
|
20 |
|
protected function undoReplaceQuotedSpecialChars($data, $delim, $eol) |
299
|
|
|
{ |
300
|
20 |
|
$replacements = array(self::PLACEHOLDER_DELIM => $delim, self::PLACEHOLDER_NEWLINE => $eol); |
301
|
|
|
if (array_walk($replacements, function($replacement, $placeholder) use (&$data) { |
302
|
20 |
|
$data = str_replace($placeholder, $replacement, $data); |
303
|
20 |
|
})) { |
304
|
20 |
|
return $data; |
305
|
|
|
} |
306
|
|
|
} |
307
|
|
|
|
308
|
|
|
/** |
309
|
|
|
* Remove quotes wrapping text. |
310
|
|
|
* |
311
|
|
|
* @param string $data The data to unquote |
312
|
|
|
* @return string The data with quotes stripped from the outside of it |
313
|
|
|
* @internal |
314
|
|
|
*/ |
315
|
20 |
|
protected function unQuote($data) |
316
|
|
|
{ |
317
|
20 |
|
$escapeChar = $this->getFlavor()->doubleQuote ? $this->getFlavor()->quoteChar : $this->getFlavor()->escapeChar; |
318
|
20 |
|
$quoteChar = $this->getFlavor()->quoteChar; |
319
|
20 |
|
$data = $this->unEscape($data, $escapeChar, $quoteChar); |
320
|
20 |
|
return preg_replace('/^(["\'])(.*)\1$/ms', '\2', $data); |
321
|
|
|
} |
322
|
|
|
|
323
|
|
|
/** |
324
|
|
|
* @internal |
325
|
|
|
* @todo This actually shouldn't even be necessary. Characters should be read |
326
|
|
|
* in one at a time and a quote that follows another should just be ignored |
327
|
|
|
* deeming this unnecessary. |
328
|
|
|
*/ |
329
|
20 |
|
protected function unEscape($str, $esc, $quo) |
330
|
|
|
{ |
331
|
20 |
|
return str_replace($esc . $quo, $quo, $str); |
332
|
|
|
} |
333
|
|
|
|
334
|
|
|
/** |
335
|
|
|
* Parse a line of CSV data into an array of columns |
336
|
|
|
* |
337
|
|
|
* @param string A line of CSV data to parse |
338
|
|
|
* @return array An array of columns |
339
|
|
|
* @access protected |
340
|
|
|
* @internal |
341
|
|
|
*/ |
342
|
20 |
|
protected function parse($line) |
343
|
|
|
{ |
344
|
20 |
|
$f = $this->getFlavor(); |
345
|
20 |
|
$replaced = $this->replaceQuotedSpecialChars($line, $f->delimiter, $f->quoteChar, $f->lineTerminator); |
346
|
20 |
|
$columns = explode($f->delimiter, $replaced); |
347
|
20 |
|
$that = $this; |
348
|
|
|
return array_map(function($val) use ($that, $f) { |
349
|
20 |
|
$undone = $that->undoReplaceQuotedSpecialChars($val, $f->delimiter, $f->lineTerminator); |
350
|
20 |
|
return $this->unQuote($undone); |
351
|
20 |
|
}, $columns); |
352
|
|
|
} |
353
|
|
|
|
354
|
20 |
|
public function current() |
355
|
|
|
{ |
356
|
20 |
|
return $this->current; |
357
|
|
|
} |
358
|
|
|
|
359
|
15 |
|
public function next() |
360
|
|
|
{ |
361
|
|
|
|
362
|
15 |
|
$this->current = null; |
363
|
15 |
|
$this->load(); |
364
|
15 |
|
return $this->current; |
365
|
|
|
} |
366
|
|
|
|
367
|
8 |
|
public function valid() |
368
|
|
|
{ |
369
|
8 |
|
return (bool) $this->current; |
370
|
|
|
} |
371
|
|
|
|
372
|
5 |
|
public function key() |
373
|
|
|
{ |
374
|
5 |
|
return $this->line; |
375
|
|
|
} |
376
|
|
|
|
377
|
20 |
|
public function rewind() |
378
|
|
|
{ |
379
|
20 |
|
$this->line = 0; |
380
|
20 |
|
$this->source->rewind(); |
381
|
20 |
|
$this->current = null; |
382
|
20 |
|
$this->load(); |
383
|
20 |
|
if ($this->hasHeader()) { |
384
|
14 |
|
$this->next(); |
385
|
14 |
|
} |
386
|
20 |
|
return $this->current(); |
387
|
|
|
} |
388
|
|
|
|
389
|
2 |
|
public function header() |
390
|
|
|
{ |
391
|
2 |
|
return $this->header; |
392
|
|
|
} |
393
|
|
|
|
394
|
|
|
/** |
395
|
|
|
* @todo Closure should be changed to "Callable" (php5.4+) |
396
|
|
|
*/ |
397
|
3 |
|
public function addFilter(Closure $filter) |
398
|
|
|
{ |
399
|
3 |
|
array_push($this->filters, $filter); |
400
|
3 |
|
return $this; |
401
|
|
|
} |
402
|
|
|
|
403
|
1 |
|
public function addFilters(array $filters) |
404
|
|
|
{ |
405
|
1 |
|
foreach ($filters as $filter) { |
406
|
1 |
|
$this->addFilter($filter); |
407
|
1 |
|
} |
408
|
1 |
|
return $this; |
409
|
|
|
} |
410
|
|
|
|
411
|
3 |
|
public function filter() |
412
|
|
|
{ |
413
|
3 |
|
return new FilteredReader($this, $this->filters); |
414
|
|
|
} |
415
|
|
|
|
416
|
|
|
public function toArray() |
417
|
|
|
{ |
418
|
1 |
|
return array_map(function($row){ |
419
|
1 |
|
return $row->toArray(); |
420
|
1 |
|
}, iterator_to_array($this)); |
421
|
|
|
} |
422
|
|
|
|
423
|
|
|
} |
424
|
|
|
|