src: remove regex usage for env file parsing

ilyasShabiCS · ilyasShabiCS · commit c42b6a0ff956 · 2024-04-09T11:26:46.000+02:00
diff --git a/src/node_dotenv.cc b/src/node_dotenv.cc
@@ -12,15 +12,6 @@ using v8::NewStringType;
 using v8::Object;
 using v8::String;
 
-/**
- * The inspiration for this implementation comes from the original dotenv code,
- * available at https://github.com/motdotla/dotenv
- */
-const std::regex LINE(
-    "\\s*(?:export\\s+)?([\\w.-]+)(?:\\s*=\\s*?|:\\s+?)(\\s*'(?:\\\\'|[^']"
-    ")*'|\\s*\"(?:\\\\\"|[^\"])*\"|\\s*`(?:\\\\`|[^`])*`|[^#\r\n]+)?\\s*(?"
-    ":#.*)?");  // NOLINT(whitespace/line_length)
-
 std::vector<std::string> Dotenv::GetPathFromArgs(
     const std::vector<std::string>& args) {
   const auto find_match = [](const std::string& arg) {
@@ -101,35 +92,120 @@ Local<Object> Dotenv::ToObject(Environment* env) {
   return result;
 }
 
-void Dotenv::ParseContent(const std::string_view content) {
-  std::string lines = std::string(content);
-  lines = std::regex_replace(lines, std::regex("\r\n?"), "\n");
+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;
+}
+
+void Dotenv::ParseContent(const std::string_view input) {
+  std::string_view content = input;
+
+  std::string_view key;
+  std::string_view value;
 
-  std::smatch match;
-  while (std::regex_search(lines, match, LINE)) {
-    const std::string key = match[1].str();
+  content = trim_spaces(content);
+
+  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;
+      }
+    }
+
+    // If there is no equal character, then ignore everything
+    auto equal = content.find('=');
+    if (equal == std::string_view::npos) {
+      break;
+    }
 
-    // Default undefined or null to an empty string
-    std::string value = match[2].str();
+    key = content.substr(0, equal);
+    content.remove_prefix(equal + 1);
+    key = trim_spaces(key);
 
-    // Remove leading whitespaces
-    value.erase(0, value.find_first_not_of(" \t"));
+    if (key.empty()) {
+      break;
+    }
 
-    // Remove trailing whitespaces
-    if (!value.empty()) {
-      value.erase(value.find_last_not_of(" \t") + 1);
+    // Remove export prefix from key
+    auto have_export = key.compare(0, 7, "export ") == 0;
+    if (have_export) {
+      key = key.substr(7);
     }
 
-    if (!value.empty() && value.front() == '"') {
-      value = std::regex_replace(value, std::regex("\\\\n"), "\n");
-      value = std::regex_replace(value, std::regex("\\\\r"), "\r");
+    // SAFETY: Content is guaranteed to have at least one character
+    if (content.empty()) {
+      break;
     }
 
-    // Remove surrounding quotes
-    value = trim_quotes(value);
+    // 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);
+      value = content.substr(1, closing_quote - 1);
+      if (closing_quote != std::string_view::npos) {
+        auto multi_line_value =
+            std::regex_replace(std::string(value), std::regex("\\\\n"), "\n");
+        store_.insert_or_assign(std::string(key), multi_line_value);
+        content.remove_prefix(content.find('\n', closing_quote + 1));
+        continue;
+      }
+    }
 
-    store_.insert_or_assign(std::string(key), value);
-    lines = match.suffix();
+    // 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);
+        }
+        value = trim_spaces(value);
+        content.remove_prefix(newline);
+        store_.insert_or_assign(std::string(key), value);
+      } else {
+        break;
+      }
+    }
   }
 }
 
@@ -179,13 +255,4 @@ void Dotenv::AssignNodeOptionsIfAvailable(std::string* node_options) {
   }
 }
 
-std::string_view Dotenv::trim_quotes(std::string_view str) {
-  static const std::unordered_set<char> quotes = {'"', '\'', '`'};
-  if (str.size() >= 2 && quotes.count(str.front()) &&
-      quotes.count(str.back())) {
-    str = str.substr(1, str.size() - 2);
-  }
-  return str;
-}
-
 }  // namespace node
diff --git a/src/node_dotenv.h b/src/node_dotenv.h
@@ -32,7 +32,6 @@ class Dotenv {
 
  private:
   std::map<std::string, std::string> store_;
-  std::string_view trim_quotes(std::string_view str);
 };
 
 }  // namespace node
diff --git a/test/fixtures/dotenv/valid.env b/test/fixtures/dotenv/valid.env
@@ -1,5 +1,9 @@
 BASIC=basic
 
+# COMMENTS=work
+#BASIC=basic2
+#BASIC=basic3
+
 # previous line intentionally left blank
 AFTER_LINE=after_line
 EMPTY=
@@ -55,7 +59,16 @@ IS
 A
 "MULTILINE'S"
 STRING`
+export EXPORT_EXAMPLE = ignore export
+
+MULTI_PEM_DOUBLE_QUOTED="-----BEGIN PUBLIC KEY-----
+MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAnNl1tL3QjKp3DZWM0T3u
+LgGJQwu9WqyzHKZ6WIA5T+7zPjO1L8l3S8k8YzBrfH4mqWOD1GBI8Yjq2L1ac3Y/
+bTdfHN8CmQr2iDJC0C6zY8YV93oZB3x0zC/LPbRYpF8f6OqX1lZj5vo2zJZy4fI/
+kKcI5jHYc8VJq+KCuRZrvn+3V+KuL9tF9v8ZgjF2PZbU+LsCy5Yqg1M8f5Jp5f6V
+u4QuUoobAgMBAAE=
+-----END PUBLIC KEY-----"
+
 MULTI_NOT_VALID_QUOTE="
 MULTI_NOT_VALID=THIS
 IS NOT MULTILINE
-export EXAMPLE = ignore export
diff --git a/test/parallel/test-dotenv.js b/test/parallel/test-dotenv.js
@@ -58,10 +58,6 @@ assert.strictEqual(process.env.COMMENTS, undefined);
 assert.strictEqual(process.env.EQUAL_SIGNS, 'equals==');
 // Retains inner quotes
 assert.strictEqual(process.env.RETAIN_INNER_QUOTES, '{"foo": "bar"}');
-// Respects equals signs in values
-assert.strictEqual(process.env.EQUAL_SIGNS, 'equals==');
-// Retains inner quotes
-assert.strictEqual(process.env.RETAIN_INNER_QUOTES, '{"foo": "bar"}');
 assert.strictEqual(process.env.RETAIN_INNER_QUOTES_AS_STRING, '{"foo": "bar"}');
 assert.strictEqual(process.env.RETAIN_INNER_QUOTES_AS_BACKTICKS, '{"foo": "bar\'s"}');
 // Retains spaces in string
@@ -83,4 +79,13 @@ assert.strictEqual(process.env.EXPAND_NEWLINES, 'expand\nnew\nlines');
 assert.strictEqual(process.env.DONT_EXPAND_UNQUOTED, 'dontexpand\\nnewlines');
 assert.strictEqual(process.env.DONT_EXPAND_SQUOTED, 'dontexpand\\nnewlines');
 // Ignore export before key
-assert.strictEqual(process.env.EXAMPLE, 'ignore export');
+assert.strictEqual(process.env.EXPORT_EXAMPLE, 'ignore export');
+
+const multiPem = `-----BEGIN PUBLIC KEY-----
+MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAnNl1tL3QjKp3DZWM0T3u
+LgGJQwu9WqyzHKZ6WIA5T+7zPjO1L8l3S8k8YzBrfH4mqWOD1GBI8Yjq2L1ac3Y/
+bTdfHN8CmQr2iDJC0C6zY8YV93oZB3x0zC/LPbRYpF8f6OqX1lZj5vo2zJZy4fI/
+kKcI5jHYc8VJq+KCuRZrvn+3V+KuL9tF9v8ZgjF2PZbU+LsCy5Yqg1M8f5Jp5f6V
+u4QuUoobAgMBAAE=
+-----END PUBLIC KEY-----`;
+assert.strictEqual(process.env.MULTI_PEM_DOUBLE_QUOTED, multiPem);
diff --git a/test/parallel/util-parse-env.js b/test/parallel/util-parse-env.js
@@ -6,6 +6,14 @@ const assert = require('node:assert');
 const util = require('node:util');
 const fs = require('node:fs');
 
+const multiPem = `-----BEGIN PUBLIC KEY-----
+MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAnNl1tL3QjKp3DZWM0T3u
+LgGJQwu9WqyzHKZ6WIA5T+7zPjO1L8l3S8k8YzBrfH4mqWOD1GBI8Yjq2L1ac3Y/
+bTdfHN8CmQr2iDJC0C6zY8YV93oZB3x0zC/LPbRYpF8f6OqX1lZj5vo2zJZy4fI/
+kKcI5jHYc8VJq+KCuRZrvn+3V+KuL9tF9v8ZgjF2PZbU+LsCy5Yqg1M8f5Jp5f6V
+u4QuUoobAgMBAAE=
+-----END PUBLIC KEY-----`;
+
 {
   const validEnvFilePath = fixtures.path('dotenv/valid.env');
   const validContent = fs.readFileSync(validEnvFilePath, 'utf8');
@@ -32,7 +40,7 @@ const fs = require('node:fs');
     EMPTY_DOUBLE_QUOTES: '',
     EMPTY_SINGLE_QUOTES: '',
     EQUAL_SIGNS: 'equals==',
-    EXAMPLE: 'ignore export',
+    EXPORT_EXAMPLE: 'ignore export',
     EXPAND_NEWLINES: 'expand\nnew\nlines',
     INLINE_COMMENTS: 'inline comments',
     INLINE_COMMENTS_BACKTICKS: 'inline comments outside of #backticks',
@@ -53,6 +61,7 @@ const fs = require('node:fs');
     SINGLE_QUOTES_SPACED: '    single quotes    ',
     SPACED_KEY: 'parsed',
     TRIM_SPACE_FROM_UNQUOTED: 'some spaced out string',
+    MULTI_PEM_DOUBLE_QUOTED: multiPem,
   });
 }
 
