@@ -153,7 +153,7 @@ discard block |
||
| 153 | 153 | * @param string $data The data to analyze |
| 154 | 154 | * @param string $lineTerminator The line terminator char/sequence |
| 155 | 155 | * |
| 156 | - * @return array A two-row array containing quotechar, delimchar |
|
| 156 | + * @return string[] A two-row array containing quotechar, delimchar |
|
| 157 | 157 | */ |
| 158 | 158 | protected function sniffQuoteAndDelim($data, $lineTerminator) |
| 159 | 159 | { |
@@ -163,6 +163,8 @@ discard block |
||
| 163 | 163 | |
| 164 | 164 | /** |
| 165 | 165 | * @todo To make this class more oop and test-friendly, implement strategy pattern here with each delim sniffing method implemented in its own strategy class. |
| 166 | + * @param string|false $data |
|
| 167 | + * @param string $lineTerminator |
|
| 166 | 168 | */ |
| 167 | 169 | protected function sniffDelimiter($data, $lineTerminator) |
| 168 | 170 | { |
@@ -177,11 +179,17 @@ discard block |
||
| 177 | 179 | return current($winners); |
| 178 | 180 | } |
| 179 | 181 | |
| 182 | + /** |
|
| 183 | + * @param string $eols |
|
| 184 | + */ |
|
| 180 | 185 | protected function sniffQuotingStyle($delimiter, $eols) |
| 181 | 186 | { |
| 182 | 187 | return Dialect::QUOTE_MINIMAL; |
| 183 | 188 | } |
| 184 | 189 | |
| 190 | + /** |
|
| 191 | + * @param string $eols |
|
| 192 | + */ |
|
| 185 | 193 | protected function sniffHeader($delimiter, $eols) |
| 186 | 194 | { |
| 187 | 195 | return true; |
@@ -13,15 +13,12 @@ |
||
| 13 | 13 | namespace CSVelte; |
| 14 | 14 | |
| 15 | 15 | use CSVelte\Contract\Streamable; |
| 16 | - |
|
| 17 | 16 | use CSVelte\Exception\SnifferException; |
| 18 | 17 | use CSVelte\Sniffer\SniffDelimiterByConsistency; |
| 19 | 18 | use CSVelte\Sniffer\SniffDelimiterByDistribution; |
| 20 | 19 | use CSVelte\Sniffer\SniffLineTerminatorByCount; |
| 21 | 20 | use CSVelte\Sniffer\SniffQuoteAndDelimByAdjacency; |
| 22 | -use Noz\Collection\Collection; |
|
| 23 | 21 | use function Noz\to_array; |
| 24 | -use RuntimeException; |
|
| 25 | 22 | |
| 26 | 23 | use function Noz\collect; |
| 27 | 24 | use function Stringy\create as s; |
@@ -35,9 +35,9 @@ |
||
| 35 | 35 | // in this case we really only care about newlines so we pass in a comma as the delim |
| 36 | 36 | $str = $this->replaceQuotedSpecialChars($data, ','); |
| 37 | 37 | $eols = [ |
| 38 | - static::EOL_WINDOWS => "\r\n", // 0x0D - 0x0A - Windows, DOS OS/2 |
|
| 39 | - static::EOL_UNIX => "\n", // 0x0A - - Unix, OSX |
|
| 40 | - static::EOL_OTHER => "\r", // 0x0D - - Other |
|
| 38 | + static::EOL_WINDOWS => "\r\n", // 0x0D - 0x0A - Windows, DOS OS/2 |
|
| 39 | + static::EOL_UNIX => "\n", // 0x0A - - Unix, OSX |
|
| 40 | + static::EOL_OTHER => "\r", // 0x0D - - Other |
|
| 41 | 41 | ]; |
| 42 | 42 | |
| 43 | 43 | $curCount = 0; |
@@ -20,18 +20,18 @@ |
||
| 20 | 20 | |
| 21 | 21 | class SniffQuoteAndDelimByAdjacency extends AbstractSniffer |
| 22 | 22 | { |
| 23 | - /** |
|
| 24 | - * Guess quote and delimiter character(s) |
|
| 25 | - * |
|
| 26 | - * If there are quoted values within the data, it is often easiest to guess the quote and delimiter characters at |
|
| 27 | - * the same time by analyzing their adjacency to one-another. That is to say, in cases where certain values are |
|
| 28 | - * wrapped in quotes, it can often be determined what not only that quote character is, but also the delimiter |
|
| 29 | - * because it is often on either side of the quote character. |
|
| 30 | - * |
|
| 31 | - * @param string $data The data to analyze |
|
| 32 | - * |
|
| 33 | - * @return string[] |
|
| 34 | - */ |
|
| 23 | + /** |
|
| 24 | + * Guess quote and delimiter character(s) |
|
| 25 | + * |
|
| 26 | + * If there are quoted values within the data, it is often easiest to guess the quote and delimiter characters at |
|
| 27 | + * the same time by analyzing their adjacency to one-another. That is to say, in cases where certain values are |
|
| 28 | + * wrapped in quotes, it can often be determined what not only that quote character is, but also the delimiter |
|
| 29 | + * because it is often on either side of the quote character. |
|
| 30 | + * |
|
| 31 | + * @param string $data The data to analyze |
|
| 32 | + * |
|
| 33 | + * @return string[] |
|
| 34 | + */ |
|
| 35 | 35 | public function sniff($data) |
| 36 | 36 | { |
| 37 | 37 | /** |
@@ -68,7 +68,7 @@ |
||
| 68 | 68 | if (is_null($eol)) { |
| 69 | 69 | $eol = "\r\n|\r|\n"; |
| 70 | 70 | } |
| 71 | - return preg_replace_callback('/([\'"])(.*)\1/imsU', function ($matches) use ($delim, $eol) { |
|
| 71 | + return preg_replace_callback('/([\'"])(.*)\1/imsU', function($matches) use ($delim, $eol) { |
|
| 72 | 72 | $ret = preg_replace("/({$eol})/", self::PLACEHOLDER_NEWLINE, $matches[0]); |
| 73 | 73 | if (!is_null($delim)) { |
| 74 | 74 | $ret = str_replace($delim, self::PLACEHOLDER_DELIM, $ret); |