Refactored EnvFile & EnvLexer to support multi-line entries

wintercms · Mar 1, 2023 · 5d6a30d · 5d6a30d
1 parent 888053d
commit 5d6a30d
Show file tree

Hide file tree

Showing 4 changed files with 88 additions and 76 deletions.
diff --git a/src/Contracts/DataFileLexerInterface.php b/src/Contracts/DataFileLexerInterface.php
@@ -5,16 +5,16 @@
 interface DataFileLexerInterface
 {
  public const T_ENV = 'T_ENV';
- public const T_QUOTED_ENV = 'T_QUOTED_ENV';
- public const T_ENV_NO_VALUE = 'T_ENV_NO_VALUE';
+ public const T_VALUE = 'T_VALUE';
+ public const T_QUOTED_VALUE = 'T_QUOTED_VALUE';
  public const T_WHITESPACE = 'T_WHITESPACE';
  public const T_COMMENT = 'T_COMMENT';
 
  /**
  * Get the ast from array of src lines
  *
- * @param array<int, string> $src
+ * @param string $string
  * @return array<int, array>
  */
- public function parse(array $src): array;
+ public function parse(string $string): array;
 }
diff --git a/src/EnvFile.php b/src/EnvFile.php
@@ -11,7 +11,7 @@
 /**
  * Class EnvFile
  */
-class EnvFile implements DataFileInterface
+class EnvFile extends DataFile implements DataFileInterface
 {
  /**
  * Lines of env data
@@ -88,37 +88,50 @@ public function set($key, $value = null)
  }
 
  foreach ($this->ast as $index => $item) {
- if (
- !in_array($item['token'], [
- $this->lexer::T_ENV,
- $this->lexer::T_QUOTED_ENV,
- $this->lexer::T_ENV_NO_VALUE
- ])
- ) {
+ // Skip all but keys
+ if ($item['token'] !== $this->lexer::T_ENV) {
  continue;
  }
 
- if ($item['env']['key'] === $key) {
- $this->ast[$index]['env']['value'] = $this->castValue($value);
+ if ($item['value'] === $key) {
+ if (
+ !isset($this->ast[$index + 1])
+ || !in_array($this->ast[$index + 1]['token'], [$this->lexer::T_VALUE, $this->lexer::T_QUOTED_VALUE])
+ ) {
+ throw new \Exception('jack go fix');
+ }
+
+ $this->ast[$index + 1]['value'] = $this->castValue($value);
+
  // Reprocess the token type to ensure old casting rules are still applied
- $this->ast[$index]['token'] = (
- is_numeric($value)
- || is_bool($value)
- || is_null($value)
- || (is_string($value) && strpos($value, ' ') === false && $item['token'] === $this->lexer::T_ENV)
- ) ? $this->lexer::T_ENV : $this->lexer::T_QUOTED_ENV;
+ switch ($this->ast[$index + 1]['token']) {
+ case $this->lexer::T_VALUE:
+ if (
+ str_contains($this->ast[$index + 1]['value'], '"')
+ || str_contains($this->ast[$index + 1]['value'], '\'')
+ ) {
+ $this->ast[$index + 1]['token'] = $this->lexer::T_QUOTED_VALUE;
+ }
+ break;
+ case $this->lexer::T_QUOTED_VALUE:
+ if (is_null($value) || $value === true || $value === false) {
+ $this->ast[$index + 1]['token'] = $this->lexer::T_VALUE;
+ }
+ break;
+ }
 
  return $this;
  }
  }
 
  // We did not find the key in the AST, therefore we must create it
  $this->ast[] = [
- 'token' => (is_numeric($value) || is_bool($value)) ? $this->lexer::T_ENV : $this->lexer::T_QUOTED_ENV,
- 'env' => [
- 'key' => $key,
- 'value' => $this->castValue($value)
- ]
+ 'token' => $this->lexer::T_ENV,
+ 'value' => $key
+ ];
+ $this->ast[] = [
+ 'token' => (is_numeric($value) || is_bool($value)) ? $this->lexer::T_VALUE : $this->lexer::T_QUOTED_VALUE,
+ 'value' => $this->castValue($value)
  ];
 
  // Add a new line
@@ -171,7 +184,7 @@ protected function parse(string $filePath): array
  return [];
  }
 
- $contents = file($filePath);
+ $contents = file_get_contents($filePath);
 
  return $this->lexer->parse($contents);
  }
@@ -185,17 +198,19 @@ public function getVariables(): array
  {
  $env = [];
 
- foreach ($this->ast as $item) {
- if (
- !in_array($item['token'], [
- $this->lexer::T_ENV,
- $this->lexer::T_QUOTED_ENV
- ])
- ) {
+ foreach ($this->ast as $index => $item) {
+ if ($item['token'] !== $this->lexer::T_ENV) {
+ continue;
+ }
+
+ if (!(
+ isset($this->ast[$index + 1])
+ && in_array($this->ast[$index + 1]['token'], [$this->lexer::T_VALUE, $this->lexer::T_QUOTED_VALUE])
+ )) {
  continue;
  }
 
- $env[$item['env']['key']] = trim($item['env']['value']);
+ $env[$item['value']] = trim($this->ast[$index + 1]['value']);
  }
 
  return $env;

diff --git a/src/Parser/EnvLexer.php b/src/Parser/EnvLexer.php
@@ -8,37 +8,49 @@
 class EnvLexer implements DataFileLexerInterface
 {
  protected $tokenMap = [
- '/^([\w]*)="?(.+)["|$]/' => self::T_QUOTED_ENV,
- '/^([\w]*)="?(.*)(?:"|$)/' => self::T_ENV,
- '/^(\w+)/' => self::T_ENV_NO_VALUE,
- '/^(\s+)/' => self::T_WHITESPACE,
- '/(#[\w\s]).*/' => self::T_COMMENT,
+ '/^(\s+)/'  => self::T_WHITESPACE,
+ '/^(#.*)/'  => self::T_COMMENT,
+ '/^(\w+)/s'  => self::T_ENV,
+ '/^="([^"\\\]*(?:\\\.[^"\\\]*)*)"/s' => self::T_QUOTED_VALUE,
+ '/^\=(.*)/'  => self::T_VALUE,
  ];
 
- public function parse(array $src): array
+ /**
+ * Parses an array of lines into an AST
+ *
+ * @param string $string
+ * @return array|array[]
+ * @throws EnvParserException
+ */
+ public function parse(string $string): array
  {
  $tokens = [];
+ $offset = 0;
+ do {
+ $result = $this->match($string, $offset);
 
- foreach ($src as $line => $str) {
- $read = 0;
- do {
- $result = $this->match($str, $line, $read);
-
- if (is_null($result)) {
- throw new EnvParserException("Unable to parse line " . ($line + 1) . ".");
- }
+ if (is_null($result)) {
+ throw new EnvParserException("Unable to parse file, failed at: " . $offset . ".");
+ }
 
-  $tokens[] = $result;
+ $tokens[] = $result;
 
- $read += strlen($result['match']);
- } while ($read < strlen($str));
- }
+ $offset += strlen($result['match']);
+ } while ($offset < strlen($string));
 
  return $tokens;
  }
 
- public function match(string $str, int $line, int $offset): ?array
+ /**
+ * Parse a string against our token map and return a node
+ *
+ * @param string $str
+ * @param int $offset
+ * @return array|null
+ */
+ public function match(string $str, int $offset): ?array
  {
+ $source = $str;
  $str = substr($str, $offset);
 
  foreach ($this->tokenMap as $pattern => $name) {
@@ -48,37 +60,22 @@ public function match(string $str, int $line, int $offset): ?array
 
  switch ($name) {
  case static::T_ENV:
- case static::T_QUOTED_ENV:
- return [
- 'match' => $matches[0],
- 'env' => [
- 'key' => $matches[1],
- 'value' => $matches[2],
- ],
- 'token' => $name,
- 'line' => $line + 1
- ];
- case static::T_ENV_NO_VALUE:
+ case static::T_VALUE:
+ case static::T_QUOTED_VALUE:
  return [
  'match' => $matches[0],
- 'env' => [
- 'key' => $matches[1],
- 'value' => '',
- ],
+ 'value' => $matches[1] ?? '',
  'token' => $name,
- 'line' => $line + 1
  ];
  case static::T_COMMENT:
  return [
  'match' => $matches[0],
  'token' => $name,
- 'line' => $line + 1
  ];
  case static::T_WHITESPACE:
  return [
  'match' => $matches[1],
  'token' => $name,
- 'line' => $line + 1
  ];
  }
  }

diff --git a/src/Printer/EnvPrinter.php b/src/Printer/EnvPrinter.php
@@ -14,13 +14,13 @@ public function render(array $ast): string
  foreach ($ast as $item) {
  switch ($item['token']) {
  case EnvLexer::T_ENV:
- $output .= sprintf('%s=%s', $item['env']['key'], $item['env']['value']);
+ $output .= $item['value'];
  break;
- case EnvLexer::T_QUOTED_ENV:
- $output .= sprintf('%s="%s"', $item['env']['key'], $item['env']['value']);
+ case EnvLexer::T_VALUE:
+ $output .= sprintf('=%s', $item['value']);
  break;
- case EnvLexer::T_ENV_NO_VALUE:
- $output .= $item['env']['key'];
+ case EnvLexer::T_QUOTED_VALUE:
+ $output .= sprintf('="%s"', $item['value']);
  break;
  case EnvLexer::T_COMMENT:
  case EnvLexer::T_WHITESPACE: