dotenv: fix env-file parsing

Implementation of parsing is identical to dotenv npm package

Author: Luboš Matejčík

src/node_dotenv.cc

@@ -1,4 +1,5 @@
 #include "node_dotenv.h"
+#include <regex>
 #include <unordered_set>
 #include "env-inl.h"
 #include "node_file.h"
@@ -87,137 +88,58 @@ Local<Object> Dotenv::ToObject(Environment* env) const {
   return result;
 }
 
-std::string_view trim_spaces(std::string_view input) {
-  if (input.empty()) return "";
-  if (input.front() == ' ') {
-    input.remove_prefix(input.find_first_not_of(' '));
-  }
-  if (!input.empty() && input.back() == ' ') {
-    input = input.substr(0, input.find_last_not_of(' ') + 1);
-  }
-  return input;
+const std::regex whitespace_regex{
+    "(^\\s+|\\s+$)",
+    std::regex_constants::ECMAScript | std::regex_constants::multiline};
+
+std::string trim_spaces(const std::string& input) {
+  return std::regex_replace(input, whitespace_regex, "");
 }
 
+const std::regex env_regex{
+    R"((?:^|^)\s*(?:export\s+)?([\w.-]+)(?:\s*=\s*?|:\s+?)(\s*'(?:\\'|[^'])*'|\s*"(?:\\"|[^"])*"|\s*`(?:\\`|[^`])*`|[^#\r\n]+)?\s*(?:#.*)?(?:$|$))",
+    std::regex_constants::ECMAScript | std::regex_constants::multiline};
+
+const std::regex quotes_regex{
+    R"(^(['"`])([\s\S]*)\1$)",
+    std::regex_constants::ECMAScript | std::regex_constants::multiline};
+const std::regex newline_regex{
+    R"(\\n)",
+    std::regex_constants::ECMAScript | std::regex_constants::multiline};
+const std::regex cr_regex{
+    R"(\\r)",
+    std::regex_constants::ECMAScript | std::regex_constants::multiline};
+
 void Dotenv::ParseContent(const std::string_view input) {
   std::string lines(input);
 
   // Handle windows newlines "\r\n": remove "\r" and keep only "\n"
   lines.erase(std::remove(lines.begin(), lines.end(), '\r'), lines.end());
 
-  std::string_view content = lines;
-  content = trim_spaces(content);
-
-  std::string_view key;
-  std::string_view value;
+  std::string content = trim_spaces(lines);
 
-  while (!content.empty()) {
-    // Skip empty lines and comments
-    if (content.front() == '\n' || content.front() == '#') {
-      auto newline = content.find('\n');
-      if (newline != std::string_view::npos) {
-        content.remove_prefix(newline + 1);
-        continue;
-      }
-    }
+  std::string key;
+  std::string value;
 
-    // If there is no equal character, then ignore everything
-    auto equal = content.find('=');
-    if (equal == std::string_view::npos) {
-      break;
-    }
+  auto matches_begin =
+      std::sregex_iterator(content.cbegin(), content.cend(), env_regex);
+  auto matches_end = std::sregex_iterator();
 
-    key = content.substr(0, equal);
-    content.remove_prefix(equal + 1);
-    key = trim_spaces(key);
-    content = trim_spaces(content);
+  for (std::regex_iterator i = matches_begin; i != matches_end; ++i) {
+    key = i->str(1);
+    value = i->str(2);
+    value = trim_spaces(value);
 
-    if (key.empty()) {
-      break;
-    }
+    const auto has_opening_double_quote = value.size() > 0 && value[0] == '\"';
+    value = std::regex_replace(value, quotes_regex, "$2");
 
-    // Remove export prefix from key
-    if (key.starts_with("export ")) {
-      key.remove_prefix(7);
+    if (has_opening_double_quote) {
+      // replace escaped newlines and carriage returns
+      value = std::regex_replace(value, newline_regex, "\n");
+      value = std::regex_replace(value, cr_regex, "\r");
     }
 
-    // SAFETY: Content is guaranteed to have at least one character
-    if (content.empty()) {
-      // In case the last line is a single key without value
-      // Example: KEY= (without a newline at the EOF)
-      store_.insert_or_assign(std::string(key), "");
-      break;
-    }
-
-    // Expand new line if \n it's inside double quotes
-    // Example: EXPAND_NEWLINES = 'expand\nnew\nlines'
-    if (content.front() == '"') {
-      auto closing_quote = content.find(content.front(), 1);
-      if (closing_quote != std::string_view::npos) {
-        value = content.substr(1, closing_quote - 1);
-        std::string multi_line_value = std::string(value);
-
-        size_t pos = 0;
-        while ((pos = multi_line_value.find("\\n", pos)) !=
-               std::string_view::npos) {
-          multi_line_value.replace(pos, 2, "\n");
-          pos += 1;
-        }
-
-        store_.insert_or_assign(std::string(key), multi_line_value);
-        content.remove_prefix(content.find('\n', closing_quote + 1));
-        continue;
-      }
-    }
-
-    // Check if the value is wrapped in quotes, single quotes or backticks
-    if ((content.front() == '\'' || content.front() == '"' ||
-         content.front() == '`')) {
-      auto closing_quote = content.find(content.front(), 1);
-
-      // Check if the closing quote is not found
-      // Example: KEY="value
-      if (closing_quote == std::string_view::npos) {
-        // Check if newline exist. If it does, take the entire line as the value
-        // Example: KEY="value\nKEY2=value2
-        // The value pair should be `"value`
-        auto newline = content.find('\n');
-        if (newline != std::string_view::npos) {
-          value = content.substr(0, newline);
-          store_.insert_or_assign(std::string(key), value);
-          content.remove_prefix(newline);
-        }
-      } else {
-        // Example: KEY="value"
-        value = content.substr(1, closing_quote - 1);
-        store_.insert_or_assign(std::string(key), value);
-        // Select the first newline after the closing quotation mark
-        // since there could be newline characters inside the value.
-        content.remove_prefix(content.find('\n', closing_quote + 1));
-      }
-    } else {
-      // Regular key value pair.
-      // Example: `KEY=this is value`
-      auto newline = content.find('\n');
-
-      if (newline != std::string_view::npos) {
-        value = content.substr(0, newline);
-        auto hash_character = value.find('#');
-        // Check if there is a comment in the line
-        // Example: KEY=value # comment
-        // The value pair should be `value`
-        if (hash_character != std::string_view::npos) {
-          value = content.substr(0, hash_character);
-        }
-        content.remove_prefix(newline);
-      } else {
-        // In case the last line is a single key/value pair
-        // Example: KEY=VALUE (without a newline at the EOF)
-        value = content.substr(0);
-      }
-
-      value = trim_spaces(value);
-      store_.insert_or_assign(std::string(key), value);
-    }
+    store_.insert_or_assign(key, value);
   }
 }