Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
100.00% |
60 / 60 |
|
100.00% |
12 / 12 |
CRAP | |
100.00% |
1 / 1 |
| CsvParser | |
100.00% |
60 / 60 |
|
100.00% |
12 / 12 |
32 | |
100.00% |
1 / 1 |
| setColumnHeadings | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
6 | |||
| getColumnHeadings | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| isSingleVisibleCharacter | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
2 | |||
| setFieldDelimiterChar | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| getFieldDelimiterChar | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setFieldEnclosureChar | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| getFieldEnclosureChar | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setEscapeChar | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| getEscapeChar | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setFirstRowContainsColumnHeadings | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getFirstRowContainsColumnHeadings | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| parseValue | |
100.00% |
33 / 33 |
|
100.00% |
1 / 1 |
11 | |||
| setMsgContent | n/a |
0 / 0 |
n/a |
0 / 0 |
1 | |||||
| 1 | <?php |
| 2 | /** |
| 3 | * @author: Doug Wilbourne (dougwilbourne@gmail.com) |
| 4 | */ |
| 5 | |
| 6 | declare(strict_types=1); |
| 7 | |
| 8 | namespace pvc\parser\csv; |
| 9 | |
| 10 | use pvc\interfaces\msg\MsgInterface; |
| 11 | use pvc\parser\err\CsvParserException; |
| 12 | use pvc\parser\err\DuplicateColumnHeadingException; |
| 13 | use pvc\parser\err\OpenFileException; |
| 14 | use pvc\parser\err\InvalidColumnHeadingException; |
| 15 | use pvc\parser\err\InvalidEscapeCharacterException; |
| 16 | use pvc\parser\err\InvalidFieldDelimiterException; |
| 17 | use pvc\parser\err\InvalidFieldEnclosureCharException; |
| 18 | use pvc\parser\err\NonExistentColumnHeadingException; |
| 19 | use pvc\parser\err\NonExistentFilePathException; |
| 20 | use pvc\parser\Parser; |
| 21 | use Throwable; |
| 22 | |
| 23 | /** |
| 24 | * Class CsvParser. This class restricts record termination characters to be either LF or CRLF (windows). PHP is |
| 25 | * supposed to automatically detect line endings with its verbs that extract lines from a file. This class uses fgetscsv |
| 26 | * and detects the presence of byte order marks at the beginning of the file |
| 27 | * |
| 28 | * @extends Parser<array> |
| 29 | */ |
| 30 | class CsvParser extends Parser |
| 31 | { |
| 32 | /** |
| 33 | * @var string |
| 34 | */ |
| 35 | protected string $filePath; |
| 36 | |
| 37 | /** |
| 38 | * @var array <string> |
| 39 | */ |
| 40 | protected array $columnHeadings; |
| 41 | /** |
| 42 | * @var non-empty-string |
| 43 | */ |
| 44 | protected string $fieldDelimiterChar = ','; |
| 45 | |
| 46 | /** |
| 47 | * @var string |
| 48 | */ |
| 49 | protected string $fieldEnclosureChar = "\""; |
| 50 | |
| 51 | /** |
| 52 | * @var string |
| 53 | */ |
| 54 | protected string $escapeChar = "\\"; |
| 55 | |
| 56 | /** |
| 57 | * @var bool |
| 58 | */ |
| 59 | protected bool $firstRowContainsColumnHeadings = false; |
| 60 | |
| 61 | /** |
| 62 | * setColumnHeadings |
| 63 | * @param array<string> $columnHeadings |
| 64 | * @throws InvalidColumnHeadingException|DuplicateColumnHeadingException |
| 65 | */ |
| 66 | public function setColumnHeadings(array $columnHeadings) : void |
| 67 | { |
| 68 | if (empty($columnHeadings)) { |
| 69 | throw new NonExistentColumnHeadingException(); |
| 70 | } |
| 71 | |
| 72 | /** |
| 73 | * re-initialize the attribute so that successive calls to the parser work properly |
| 74 | */ |
| 75 | $this->columnHeadings = []; |
| 76 | |
| 77 | foreach ($columnHeadings as $columnHeading) { |
| 78 | /** |
| 79 | * must be a string |
| 80 | */ |
| 81 | if (!is_string($columnHeading)) { |
| 82 | throw new InvalidColumnHeadingException(); |
| 83 | } |
| 84 | |
| 85 | /** |
| 86 | * characters in the column heading must all be visible |
| 87 | */ |
| 88 | if (!ctype_graph($columnHeading)) { |
| 89 | throw new InvalidColumnHeadingException(); |
| 90 | } |
| 91 | |
| 92 | /** |
| 93 | * no duplicate column headings since they become indices into an array |
| 94 | */ |
| 95 | if (in_array($columnHeading, $this->columnHeadings)) { |
| 96 | throw new DuplicateColumnHeadingException($columnHeading); |
| 97 | } |
| 98 | |
| 99 | $this->columnHeadings[] = $columnHeading; |
| 100 | } |
| 101 | } |
| 102 | |
| 103 | /** |
| 104 | * getColumnHeadings |
| 105 | * @return array<string> |
| 106 | */ |
| 107 | public function getColumnHeadings() : array |
| 108 | { |
| 109 | return $this->columnHeadings; |
| 110 | } |
| 111 | |
| 112 | protected function isSingleVisibleCharacter(string $char): bool |
| 113 | { |
| 114 | return (ctype_graph($char) && (strlen($char) == 1)); |
| 115 | } |
| 116 | |
| 117 | /** |
| 118 | * setFieldDelimiterChar |
| 119 | * @param non-empty-string $delimiterChar |
| 120 | * @throws InvalidFieldDelimiterException |
| 121 | */ |
| 122 | public function setFieldDelimiterChar(string $delimiterChar) : void |
| 123 | { |
| 124 | /** field delimiter must be a single visible character */ |
| 125 | if (!$this->isSingleVisibleCharacter($delimiterChar)) { |
| 126 | throw new InvalidFieldDelimiterException(); |
| 127 | } |
| 128 | $this->fieldDelimiterChar = $delimiterChar; |
| 129 | } |
| 130 | |
| 131 | /** |
| 132 | * getFieldDelimiterChar |
| 133 | * @return string |
| 134 | */ |
| 135 | public function getFieldDelimiterChar() : string |
| 136 | { |
| 137 | return $this->fieldDelimiterChar; |
| 138 | } |
| 139 | |
| 140 | /** |
| 141 | * setFieldEnclosureChar |
| 142 | * @param string $enclosureChar |
| 143 | * @throws InvalidFieldEnclosureCharException |
| 144 | */ |
| 145 | public function setFieldEnclosureChar(string $enclosureChar) : void |
| 146 | { |
| 147 | /** field enclosure must be a single visible character */ |
| 148 | if (!$this->isSingleVisibleCharacter($enclosureChar)) { |
| 149 | throw new InvalidFieldEnclosureCharException(); |
| 150 | } |
| 151 | $this->fieldEnclosureChar = $enclosureChar; |
| 152 | } |
| 153 | |
| 154 | /** |
| 155 | * getFieldEnclosureChar |
| 156 | * @return string |
| 157 | */ |
| 158 | public function getFieldEnclosureChar() : string |
| 159 | { |
| 160 | return $this->fieldEnclosureChar; |
| 161 | } |
| 162 | |
| 163 | /** |
| 164 | * setEscapeChar |
| 165 | * @param string $escapeChar |
| 166 | * @throws InvalidEscapeCharacterException |
| 167 | */ |
| 168 | public function setEscapeChar(string $escapeChar) : void |
| 169 | { |
| 170 | /** escape character must be a single visible character */ |
| 171 | if (!$this->isSingleVisibleCharacter($escapeChar)) { |
| 172 | throw new InvalidEscapeCharacterException(); |
| 173 | } |
| 174 | $this->escapeChar = $escapeChar; |
| 175 | } |
| 176 | |
| 177 | /** |
| 178 | * getEscapeChar |
| 179 | * @return string |
| 180 | */ |
| 181 | public function getEscapeChar() : string |
| 182 | { |
| 183 | return $this->escapeChar; |
| 184 | } |
| 185 | |
| 186 | public function setFirstRowContainsColumnHeadings(bool $value) : void |
| 187 | { |
| 188 | $this->firstRowContainsColumnHeadings = $value; |
| 189 | } |
| 190 | |
| 191 | public function getFirstRowContainsColumnHeadings() : bool |
| 192 | { |
| 193 | return $this->firstRowContainsColumnHeadings; |
| 194 | } |
| 195 | |
| 196 | /** |
| 197 | * parse |
| 198 | * @param string $data |
| 199 | * @return bool |
| 200 | * @throws NonExistentColumnHeadingException |
| 201 | */ |
| 202 | protected function parseValue(string $data): bool |
| 203 | { |
| 204 | try { |
| 205 | $handle = fopen($data, 'r'); |
| 206 | } catch (Throwable $e) { |
| 207 | throw new OpenFileException($data, $e); |
| 208 | } |
| 209 | $this->filePath = $data; |
| 210 | |
| 211 | $bom = "\xef\xbb\xbf"; |
| 212 | if (fgets($handle, 4) !== $bom) { |
| 213 | // BOM not found - rewind pointer to start of file. |
| 214 | rewind($handle); |
| 215 | } |
| 216 | |
| 217 | $rows = []; |
| 218 | while (false !== ($line = (fgetcsv( |
| 219 | $handle, |
| 220 | null, |
| 221 | $this->getFieldDelimiterChar(), |
| 222 | $this->getFieldEnclosureChar(), |
| 223 | $this->getEscapeChar() |
| 224 | )))) { |
| 225 | /** |
| 226 | * fgetcsv returns an array with a single element consisting of a null value if the line is empty |
| 227 | */ |
| 228 | if (!is_null($line[0])) { |
| 229 | $rows[] = $line; |
| 230 | } |
| 231 | } |
| 232 | |
| 233 | /** |
| 234 | * if we are not at the end of the file then fgetcsv returned false because it could not parse a line. |
| 235 | */ |
| 236 | if (!feof($handle)) { |
| 237 | throw new CsvParserException($data); |
| 238 | } |
| 239 | fclose($handle); |
| 240 | |
| 241 | /** |
| 242 | * array_combine would automatically convert invalid array keys to strings, but it will not check |
| 243 | * for duplicate column names or verify that all the characters are graphic (e.g. visible), so the |
| 244 | * setColumnHeadings method ensures those things. Also, array_combine fails if the number of headings does |
| 245 | * not match the number of elements in each and every row of data. It is certainly possible to ensure the |
| 246 | * shapes match and reshape as necessary, but it's about as much trouble as handling each row manually...... |
| 247 | */ |
| 248 | |
| 249 | if ($this->getFirstRowContainsColumnHeadings()) { |
| 250 | $firstRow = array_shift($rows); |
| 251 | if ($firstRow) { |
| 252 | $this->setColumnHeadings($firstRow); |
| 253 | } |
| 254 | |
| 255 | foreach ($rows as $row) { |
| 256 | $newRow = []; |
| 257 | foreach ($row as $index => $element) { |
| 258 | if (isset($this->columnHeadings[$index])) { |
| 259 | $newRow[$this->columnHeadings[$index]] = $element; |
| 260 | } else { |
| 261 | $newRow[$index] = $element; |
| 262 | } |
| 263 | } |
| 264 | $this->parsedValue[] = $newRow; |
| 265 | } |
| 266 | } else { |
| 267 | $this->parsedValue = $rows; |
| 268 | } |
| 269 | |
| 270 | return true; |
| 271 | } |
| 272 | |
| 273 | public function setMsgContent(MsgInterface $msg): void {} |
| 274 | } |