EntryParser.php 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300
  1. <?php
  2. declare(strict_types=1);
  3. namespace Dotenv\Parser;
  4. use Dotenv\Util\Regex;
  5. use Dotenv\Util\Str;
  6. use GrahamCampbell\ResultType\Error;
  7. use GrahamCampbell\ResultType\Result;
  8. use GrahamCampbell\ResultType\Success;
  9. final class EntryParser
  10. {
  11. private const INITIAL_STATE = 0;
  12. private const UNQUOTED_STATE = 1;
  13. private const SINGLE_QUOTED_STATE = 2;
  14. private const DOUBLE_QUOTED_STATE = 3;
  15. private const ESCAPE_SEQUENCE_STATE = 4;
  16. private const WHITESPACE_STATE = 5;
  17. private const COMMENT_STATE = 6;
  18. private const REJECT_STATES = [self::SINGLE_QUOTED_STATE, self::DOUBLE_QUOTED_STATE, self::ESCAPE_SEQUENCE_STATE];
  19. /**
  20. * This class is a singleton.
  21. *
  22. * @codeCoverageIgnore
  23. *
  24. * @return void
  25. */
  26. private function __construct()
  27. {
  28. //
  29. }
  30. /**
  31. * Parse a raw entry into a proper entry.
  32. *
  33. * That is, turn a raw environment variable entry into a name and possibly
  34. * a value. We wrap the answer in a result type.
  35. *
  36. * @param string $entry
  37. *
  38. * @return \GrahamCampbell\ResultType\Result<\Dotenv\Parser\Entry,string>
  39. */
  40. public static function parse(string $entry)
  41. {
  42. return self::splitStringIntoParts($entry)->flatMap(static function (array $parts) {
  43. [$name, $value] = $parts;
  44. return self::parseName($name)->flatMap(static function (string $name) use ($value) {
  45. /** @var Result<Value|null,string> */
  46. $parsedValue = $value === null ? Success::create(null) : self::parseValue($value);
  47. return $parsedValue->map(static function (?Value $value) use ($name) {
  48. return new Entry($name, $value);
  49. });
  50. });
  51. });
  52. }
  53. /**
  54. * Split the compound string into parts.
  55. *
  56. * @param string $line
  57. *
  58. * @return \GrahamCampbell\ResultType\Result<array{string,string|null},string>
  59. */
  60. private static function splitStringIntoParts(string $line)
  61. {
  62. /** @var array{string,string|null} */
  63. $result = Str::pos($line, '=')->map(static function () use ($line) {
  64. return \array_map('trim', \explode('=', $line, 2));
  65. })->getOrElse([$line, null]);
  66. if ($result[0] === '') {
  67. /** @var \GrahamCampbell\ResultType\Result<array{string,string|null},string> */
  68. return Error::create(self::getErrorMessage('an unexpected equals', $line));
  69. }
  70. /** @var \GrahamCampbell\ResultType\Result<array{string,string|null},string> */
  71. return Success::create($result);
  72. }
  73. /**
  74. * Parse the given variable name.
  75. *
  76. * That is, strip the optional quotes and leading "export" from the
  77. * variable name. We wrap the answer in a result type.
  78. *
  79. * @param string $name
  80. *
  81. * @return \GrahamCampbell\ResultType\Result<string,string>
  82. */
  83. private static function parseName(string $name)
  84. {
  85. if (Str::len($name) > 8 && Str::substr($name, 0, 6) === 'export' && \ctype_space(Str::substr($name, 6, 1))) {
  86. $name = \ltrim(Str::substr($name, 6));
  87. }
  88. if (self::isQuotedName($name)) {
  89. $name = Str::substr($name, 1, -1);
  90. }
  91. if (!self::isValidName($name)) {
  92. /** @var \GrahamCampbell\ResultType\Result<string,string> */
  93. return Error::create(self::getErrorMessage('an invalid name', $name));
  94. }
  95. /** @var \GrahamCampbell\ResultType\Result<string,string> */
  96. return Success::create($name);
  97. }
  98. /**
  99. * Is the given variable name quoted?
  100. *
  101. * @param string $name
  102. *
  103. * @return bool
  104. */
  105. private static function isQuotedName(string $name)
  106. {
  107. if (Str::len($name) < 3) {
  108. return false;
  109. }
  110. $first = Str::substr($name, 0, 1);
  111. $last = Str::substr($name, -1, 1);
  112. return ($first === '"' && $last === '"') || ($first === '\'' && $last === '\'');
  113. }
  114. /**
  115. * Is the given variable name valid?
  116. *
  117. * @param string $name
  118. *
  119. * @return bool
  120. */
  121. private static function isValidName(string $name)
  122. {
  123. return Regex::matches('~(*UTF8)\A[\p{Ll}\p{Lu}\p{M}\p{N}_.]+\z~', $name)->success()->getOrElse(false);
  124. }
  125. /**
  126. * Parse the given variable value.
  127. *
  128. * This has the effect of stripping quotes and comments, dealing with
  129. * special characters, and locating nested variables, but not resolving
  130. * them. Formally, we run a finite state automaton with an output tape: a
  131. * transducer. We wrap the answer in a result type.
  132. *
  133. * @param string $value
  134. *
  135. * @return \GrahamCampbell\ResultType\Result<\Dotenv\Parser\Value,string>
  136. */
  137. private static function parseValue(string $value)
  138. {
  139. if (\trim($value) === '') {
  140. /** @var \GrahamCampbell\ResultType\Result<\Dotenv\Parser\Value,string> */
  141. return Success::create(Value::blank());
  142. }
  143. return \array_reduce(\iterator_to_array(Lexer::lex($value)), static function (Result $data, string $token) {
  144. return $data->flatMap(static function (array $data) use ($token) {
  145. return self::processToken($data[1], $token)->map(static function (array $val) use ($data) {
  146. return [$data[0]->append($val[0], $val[1]), $val[2]];
  147. });
  148. });
  149. }, Success::create([Value::blank(), self::INITIAL_STATE]))->flatMap(static function (array $result) {
  150. /** @psalm-suppress DocblockTypeContradiction */
  151. if (in_array($result[1], self::REJECT_STATES, true)) {
  152. /** @var \GrahamCampbell\ResultType\Result<\Dotenv\Parser\Value,string> */
  153. return Error::create('a missing closing quote');
  154. }
  155. /** @var \GrahamCampbell\ResultType\Result<\Dotenv\Parser\Value,string> */
  156. return Success::create($result[0]);
  157. })->mapError(static function (string $err) use ($value) {
  158. return self::getErrorMessage($err, $value);
  159. });
  160. }
  161. /**
  162. * Process the given token.
  163. *
  164. * @param int $state
  165. * @param string $token
  166. *
  167. * @return \GrahamCampbell\ResultType\Result<array{string,bool,int},string>
  168. */
  169. private static function processToken(int $state, string $token)
  170. {
  171. switch ($state) {
  172. case self::INITIAL_STATE:
  173. if ($token === '\'') {
  174. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  175. return Success::create(['', false, self::SINGLE_QUOTED_STATE]);
  176. } elseif ($token === '"') {
  177. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  178. return Success::create(['', false, self::DOUBLE_QUOTED_STATE]);
  179. } elseif ($token === '#') {
  180. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  181. return Success::create(['', false, self::COMMENT_STATE]);
  182. } elseif ($token === '$') {
  183. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  184. return Success::create([$token, true, self::UNQUOTED_STATE]);
  185. } else {
  186. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  187. return Success::create([$token, false, self::UNQUOTED_STATE]);
  188. }
  189. case self::UNQUOTED_STATE:
  190. if ($token === '#') {
  191. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  192. return Success::create(['', false, self::COMMENT_STATE]);
  193. } elseif (\ctype_space($token)) {
  194. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  195. return Success::create(['', false, self::WHITESPACE_STATE]);
  196. } elseif ($token === '$') {
  197. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  198. return Success::create([$token, true, self::UNQUOTED_STATE]);
  199. } else {
  200. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  201. return Success::create([$token, false, self::UNQUOTED_STATE]);
  202. }
  203. case self::SINGLE_QUOTED_STATE:
  204. if ($token === '\'') {
  205. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  206. return Success::create(['', false, self::WHITESPACE_STATE]);
  207. } else {
  208. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  209. return Success::create([$token, false, self::SINGLE_QUOTED_STATE]);
  210. }
  211. case self::DOUBLE_QUOTED_STATE:
  212. if ($token === '"') {
  213. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  214. return Success::create(['', false, self::WHITESPACE_STATE]);
  215. } elseif ($token === '\\') {
  216. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  217. return Success::create(['', false, self::ESCAPE_SEQUENCE_STATE]);
  218. } elseif ($token === '$') {
  219. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  220. return Success::create([$token, true, self::DOUBLE_QUOTED_STATE]);
  221. } else {
  222. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  223. return Success::create([$token, false, self::DOUBLE_QUOTED_STATE]);
  224. }
  225. case self::ESCAPE_SEQUENCE_STATE:
  226. if ($token === '"' || $token === '\\') {
  227. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  228. return Success::create([$token, false, self::DOUBLE_QUOTED_STATE]);
  229. } elseif ($token === '$') {
  230. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  231. return Success::create([$token, false, self::DOUBLE_QUOTED_STATE]);
  232. } else {
  233. $first = Str::substr($token, 0, 1);
  234. if (\in_array($first, ['f', 'n', 'r', 't', 'v'], true)) {
  235. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  236. return Success::create([\stripcslashes('\\'.$first).Str::substr($token, 1), false, self::DOUBLE_QUOTED_STATE]);
  237. } else {
  238. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  239. return Error::create('an unexpected escape sequence');
  240. }
  241. }
  242. case self::WHITESPACE_STATE:
  243. if ($token === '#') {
  244. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  245. return Success::create(['', false, self::COMMENT_STATE]);
  246. } elseif (!\ctype_space($token)) {
  247. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  248. return Error::create('unexpected whitespace');
  249. } else {
  250. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  251. return Success::create(['', false, self::WHITESPACE_STATE]);
  252. }
  253. case self::COMMENT_STATE:
  254. /** @var \GrahamCampbell\ResultType\Result<array{string,bool,int},string> */
  255. return Success::create(['', false, self::COMMENT_STATE]);
  256. default:
  257. throw new \Error('Parser entered invalid state.');
  258. }
  259. }
  260. /**
  261. * Generate a friendly error message.
  262. *
  263. * @param string $cause
  264. * @param string $subject
  265. *
  266. * @return string
  267. */
  268. private static function getErrorMessage(string $cause, string $subject)
  269. {
  270. return \sprintf(
  271. 'Encountered %s at [%s].',
  272. $cause,
  273. \strtok($subject, "\n")
  274. );
  275. }
  276. }