src: support multi-line values for .env file

Author: ilyasShabiCS

src/node_dotenv.cc

@@ -1,4 +1,6 @@
 #include "node_dotenv.h"
+#include <regex>  // NOLINT(build/c++11)
+#include <unordered_set>
 #include "env-inl.h"
 #include "node_file.h"
 #include "uv.h"
@@ -8,6 +10,17 @@ namespace node {
 using v8::NewStringType;
 using v8::String;
 
+/**
+ * The inspiration for this implementation comes from the original dotenv code,
+ * available at https://github.com/motdotla/dotenv
+ */
+
+std::regex LINE(
+    "(?:^|^)\\s*(?:export\\s+)?([\\w.-]+)(?:\\s*=\\s*?|:\\s+?)(\\s*'(?:\\\\'|"
+    "[^'])*'|\\s*\"(?:\\\\\"|[^\"])*\"|\\s*`(?:\\\\`|[^`])*`|[^#\\r\\n]+)?"
+    "\\s*(?:#.*)?(?:$|$)",
+    std::regex_constants::multiline);
+
 std::vector<std::string> Dotenv::GetPathFromArgs(
     const std::vector<std::string>& args) {
   const auto find_match = [](const std::string& arg) {
@@ -81,7 +94,7 @@ bool Dotenv::ParsePath(const std::string_view path) {
     uv_fs_req_cleanup(&close_req);
   });
 
-  std::string result{};
+  std::string lines{};
   char buffer[8192];
   uv_buf_t buf = uv_buf_init(buffer, sizeof(buffer));
 
@@ -95,15 +108,32 @@ bool Dotenv::ParsePath(const std::string_view path) {
     if (r <= 0) {
       break;
     }
-    result.append(buf.base, r);
+    lines.append(buf.base, r);
   }
 
-  using std::string_view_literals::operator""sv;
-  auto lines = SplitString(result, "\n"sv);
+  // Convert line breaks to the same format
+  std::regex_replace(lines, std::regex("\r\n?"), "\n");
+
+  std::smatch match;
+  while (std::regex_search(lines, match, LINE)) {
+    const std::string key = match[1].str();
+
+    // Default undefined or null to an empty string
+    std::string value = match[2].str();
+
+    // Remove leading whitespaces
+    value.erase(0, value.find_first_not_of(" \t"));
+
+    // Remove trailing whitespaces
+    value.erase(value.find_last_not_of(" \t") + 1);
+
+    // Remove surrounding quotes
+    value = trim_quotes(value);
 
-  for (const auto& line : lines) {
-    ParseLine(line);
+    store_.insert_or_assign(std::string(key), value);
+    lines = match.suffix();
   }
+
   return true;
 }
 
@@ -115,56 +145,12 @@ void Dotenv::AssignNodeOptionsIfAvailable(std::string* node_options) {
   }
 }
 
-void Dotenv::ParseLine(const std::string_view line) {
-  auto equal_index = line.find('=');
-
-  if (equal_index == std::string_view::npos) {
-    return;
+std::string Dotenv::trim_quotes(std::string str) {
+  static const std::unordered_set<char> quotes = {'"', '\'', '`'};
+  if (str.size() >= 2 && quotes.count(str[0]) && quotes.count(str.back())) {
+    str = str.substr(1, str.size() - 2);
   }
-
-  auto key = line.substr(0, equal_index);
-
-  // Remove leading and trailing space characters from key.
-  while (!key.empty() && std::isspace(key.front())) key.remove_prefix(1);
-  while (!key.empty() && std::isspace(key.back())) key.remove_suffix(1);
-
-  // Omit lines with comments
-  if (key.front() == '#' || key.empty()) {
-    return;
-  }
-
-  auto value = std::string(line.substr(equal_index + 1));
-
-  // Might start and end with `"' characters.
-  auto quotation_index = value.find_first_of("`\"'");
-
-  if (quotation_index == 0) {
-    auto quote_character = value[quotation_index];
-    value.erase(0, 1);
-
-    auto end_quotation_index = value.find_last_of(quote_character);
-
-    // We couldn't find the closing quotation character. Terminate.
-    if (end_quotation_index == std::string::npos) {
-      return;
-    }
-
-    value.erase(end_quotation_index);
-  } else {
-    auto hash_index = value.find('#');
-
-    // Remove any inline comments
-    if (hash_index != std::string::npos) {
-      value.erase(hash_index);
-    }
-
-    // Remove any leading/trailing spaces from unquoted values.
-    while (!value.empty() && std::isspace(value.front())) value.erase(0, 1);
-    while (!value.empty() && std::isspace(value.back()))
-      value.erase(value.size() - 1);
-  }
-
-  store_.insert_or_assign(std::string(key), value);
+  return str;
 }
 
 }  // namespace node

src/node_dotenv.h

@@ -26,8 +26,8 @@ class Dotenv {
       const std::vector<std::string>& args);
 
  private:
-  void ParseLine(const std::string_view line);
   std::map<std::string, std::string> store_;
+  std::string trim_quotes(std::string str);
 };
 
 }  // namespace node

test/fixtures/dotenv/valid.env

@@ -33,3 +33,24 @@ RETAIN_INNER_QUOTES_AS_BACKTICKS=`{"foo": "bar's"}`
 TRIM_SPACE_FROM_UNQUOTED=    some spaced out string
 EMAIL=[email protected]
     SPACED_KEY = parsed
+
+MULTI_DOUBLE_QUOTED="THIS
+IS
+A
+MULTILINE
+STRING"
+
+MULTI_SINGLE_QUOTED='THIS
+IS
+A
+MULTILINE
+STRING'
+
+MULTI_BACKTICKED=`THIS
+IS
+A
+"MULTILINE'S"
+STRING`
+MULTI_NOT_VALID_QUOTE="
+MULTI_NOT_VALID=THIS
+IS NOT MULTILINE

test/parallel/test-dotenv.js

@@ -68,3 +68,9 @@ assert.strictEqual(process.env.TRIM_SPACE_FROM_UNQUOTED, 'some spaced out string
 assert.strictEqual(process.env.EMAIL, '[email protected]');
 // Parses keys and values surrounded by spaces
 assert.strictEqual(process.env.SPACED_KEY, 'parsed');
+// Test multiple-line value
+assert.strictEqual(process.env.MULTI_DOUBLE_QUOTED, 'THIS\nIS\nA\nMULTILINE\nSTRING');
+assert.strictEqual(process.env.MULTI_SINGLE_QUOTED, 'THIS\nIS\nA\nMULTILINE\nSTRING');
+assert.strictEqual(process.env.MULTI_BACKTICKED, 'THIS\nIS\nA\n"MULTILINE\'S"\nSTRING');
+assert.strictEqual(process.env.MULTI_NOT_VALID_QUOTE, '"');
+assert.strictEqual(process.env.MULTI_NOT_VALID, 'THIS');