From c3c0a3ee5cac4a77f038cd9490fe32909c3887f0 Mon Sep 17 00:00:00 2001
From: IlyasShabi <33763729+IlyasShabi@users.noreply.github.com>
Date: Wed, 31 Jan 2024 14:09:13 +0100
Subject: [PATCH] src: support multi-line values for .env file

PR-URL: https://github.com/nodejs/node/pull/51289
Reviewed-By: Yagiz Nizipli <yagiz.nizipli@sentry.io>
Reviewed-By: Zeyu "Alex" Yang <himself65@outlook.com>
Reviewed-By: Rafael Gonzaga <rafael.nunu@hotmail.com>
Reviewed-By: Franziska Hinkelmann <franziska.hinkelmann@gmail.com>
---
 doc/api/cli.md                  | 18 ++++++
 src/node_dotenv.cc              | 97 +++++++++++++++------------------
 src/node_dotenv.h               |  2 +-
 test/fixtures/dotenv/valid.env  | 25 +++++++++
 test/parallel/test-dotenv.js    | 14 +++++
 test/parallel/util-parse-env.js | 14 ++++-
 6 files changed, 114 insertions(+), 56 deletions(-)

diff --git a/doc/api/cli.md b/doc/api/cli.md
index 9d8d23a78ea734..f65a9f60a2e669 100644
--- a/doc/api/cli.md
+++ b/doc/api/cli.md
@@ -655,6 +655,10 @@ of `--enable-source-maps`.
 
 <!-- YAML
 added: v20.6.0
+changes:
+  - version: REPLACEME
+    pr-url: https://github.com/nodejs/node/pull/51289
+    description: Add support to multi-line values.
 -->
 
 Loads environment variables from a file relative to the current directory,
@@ -691,6 +695,20 @@ They are omitted from the values.
 USERNAME="nodejs" # will result in `nodejs` as the value.
 ```
 
+Multi-line values are supported:
+
+```text
+MULTI_LINE="THIS IS
+A MULTILINE"
+# will result in `THIS IS\nA MULTILINE` as the value.
+```
+
+Export keyword before a key is ignored:
+
+```text
+export USERNAME="nodejs" # will result in `nodejs` as the value.
+```
+
 ### `-e`, `--eval "script"`
 
 <!-- YAML
diff --git a/src/node_dotenv.cc b/src/node_dotenv.cc
index 718e5407040505..53faf461bb42f2 100644
--- a/src/node_dotenv.cc
+++ b/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"
@@ -10,6 +12,15 @@ 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) {
@@ -91,11 +102,34 @@ Local<Object> Dotenv::ToObject(Environment* env) {
 }
 
 void Dotenv::ParseContent(const std::string_view content) {
-  using std::string_view_literals::operator""sv;
-  auto lines = SplitString(content, "\n"sv);
+  std::string lines = std::string(content);
+  lines = 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);
+
+    const char maybeQuote = value.front();
+
+    if (maybeQuote == '"') {
+      value = std::regex_replace(value, std::regex("\\\\n"), "\n");
+      value = std::regex_replace(value, std::regex("\\\\r"), "\r");
+    }
+
+    // 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();
   }
 }
 
@@ -145,56 +179,13 @@ 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;
-  }
-
-  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(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);
+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);
   }
-
-  store_.insert_or_assign(std::string(key), value);
+  return str;
 }
 
 }  // namespace node
diff --git a/src/node_dotenv.h b/src/node_dotenv.h
index f2a9ce56f41570..5872627f0c25fe 100644
--- a/src/node_dotenv.h
+++ b/src/node_dotenv.h
@@ -31,8 +31,8 @@ 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_view trim_quotes(std::string_view str);
 };
 
 }  // namespace node
diff --git a/test/fixtures/dotenv/valid.env b/test/fixtures/dotenv/valid.env
index 980d3621b0c4df..ccffa779cb0d86 100644
--- a/test/fixtures/dotenv/valid.env
+++ b/test/fixtures/dotenv/valid.env
@@ -20,6 +20,9 @@ BACKTICKS_SPACED=`    backticks    `
 DOUBLE_QUOTES_INSIDE_BACKTICKS=`double "quotes" work inside backticks`
 SINGLE_QUOTES_INSIDE_BACKTICKS=`single 'quotes' work inside backticks`
 DOUBLE_AND_SINGLE_QUOTES_INSIDE_BACKTICKS=`double "quotes" and single 'quotes' work inside backticks`
+EXPAND_NEWLINES="expand\nnew\nlines"
+DONT_EXPAND_UNQUOTED=dontexpand\nnewlines
+DONT_EXPAND_SQUOTED='dontexpand\nnewlines'
 # COMMENTS=work
 INLINE_COMMENTS=inline comments # work #very #well
 INLINE_COMMENTS_SINGLE_QUOTES='inline comments outside of #singlequotes' # work
@@ -34,3 +37,25 @@ TRIM_SPACE_FROM_UNQUOTED=    some spaced out string
 EMAIL=therealnerdybeast@example.tld
     SPACED_KEY = parsed
 EDGE_CASE_INLINE_COMMENTS="VALUE1" # or "VALUE2" or "VALUE3"
+
+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
+export EXAMPLE = ignore export
diff --git a/test/parallel/test-dotenv.js b/test/parallel/test-dotenv.js
index efc5a164b39334..45b89db468510d 100644
--- a/test/parallel/test-dotenv.js
+++ b/test/parallel/test-dotenv.js
@@ -52,6 +52,8 @@ assert.strictEqual(process.env.INLINE_COMMENTS_DOUBLE_QUOTES, 'inline comments o
 assert.strictEqual(process.env.INLINE_COMMENTS_BACKTICKS, 'inline comments outside of #backticks');
 // Treats # character as start of comment
 assert.strictEqual(process.env.INLINE_COMMENTS_SPACE, 'inline comments start with a');
+// ignore comment
+assert.strictEqual(process.env.COMMENTS, undefined);
 // Respects equals signs in values
 assert.strictEqual(process.env.EQUAL_SIGNS, 'equals==');
 // Retains inner quotes
@@ -70,3 +72,15 @@ assert.strictEqual(process.env.EMAIL, 'therealnerdybeast@example.tld');
 assert.strictEqual(process.env.SPACED_KEY, 'parsed');
 // Parse inline comments correctly when multiple quotes
 assert.strictEqual(process.env.EDGE_CASE_INLINE_COMMENTS, 'VALUE1');
+// Test multi-line values with line breaks
+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');
+// Test that \n is expanded to a newline in double-quoted string
+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');
diff --git a/test/parallel/util-parse-env.js b/test/parallel/util-parse-env.js
index 54488e50da91c8..9e0ab4f6e2970e 100644
--- a/test/parallel/util-parse-env.js
+++ b/test/parallel/util-parse-env.js
@@ -17,23 +17,33 @@ const fs = require('node:fs');
     BACKTICKS_INSIDE_SINGLE: '`backticks` work inside single quotes',
     BACKTICKS_SPACED: '    backticks    ',
     BASIC: 'basic',
-    DOUBLE_AND_SINGLE_QUOTES_INSIDE_BACKTICKS: 'double "quotes" and single \'quotes\' work inside backticks',
+    DONT_EXPAND_SQUOTED: 'dontexpand\\nnewlines',
+    DONT_EXPAND_UNQUOTED: 'dontexpand\\nnewlines',
+    DOUBLE_AND_SINGLE_QUOTES_INSIDE_BACKTICKS: "double \"quotes\" and single 'quotes' work inside backticks",
     DOUBLE_QUOTES: 'double_quotes',
     DOUBLE_QUOTES_INSIDE_BACKTICKS: 'double "quotes" work inside backticks',
     DOUBLE_QUOTES_INSIDE_SINGLE: 'double "quotes" work inside single quotes',
     DOUBLE_QUOTES_SPACED: '    double quotes    ',
     DOUBLE_QUOTES_WITH_NO_SPACE_BRACKET: '{ port: $MONGOLAB_PORT}',
+    EDGE_CASE_INLINE_COMMENTS: 'VALUE1',
     EMAIL: 'therealnerdybeast@example.tld',
     EMPTY: '',
     EMPTY_BACKTICKS: '',
     EMPTY_DOUBLE_QUOTES: '',
     EMPTY_SINGLE_QUOTES: '',
     EQUAL_SIGNS: 'equals==',
+    EXAMPLE: 'ignore export',
+    EXPAND_NEWLINES: 'expand\nnew\nlines',
     INLINE_COMMENTS: 'inline comments',
     INLINE_COMMENTS_BACKTICKS: 'inline comments outside of #backticks',
     INLINE_COMMENTS_DOUBLE_QUOTES: 'inline comments outside of #doublequotes',
     INLINE_COMMENTS_SINGLE_QUOTES: 'inline comments outside of #singlequotes',
     INLINE_COMMENTS_SPACE: 'inline comments start with a',
+    MULTI_BACKTICKED: 'THIS\nIS\nA\n"MULTILINE\'S"\nSTRING',
+    MULTI_DOUBLE_QUOTED: 'THIS\nIS\nA\nMULTILINE\nSTRING',
+    MULTI_NOT_VALID: 'THIS',
+    MULTI_NOT_VALID_QUOTE: '"',
+    MULTI_SINGLE_QUOTED: 'THIS\nIS\nA\nMULTILINE\nSTRING',
     RETAIN_INNER_QUOTES: '{"foo": "bar"}',
     RETAIN_INNER_QUOTES_AS_BACKTICKS: '{"foo": "bar\'s"}',
     RETAIN_INNER_QUOTES_AS_STRING: '{"foo": "bar"}',
@@ -42,7 +52,7 @@ const fs = require('node:fs');
     SINGLE_QUOTES_INSIDE_DOUBLE: "single 'quotes' work inside double quotes",
     SINGLE_QUOTES_SPACED: '    single quotes    ',
     SPACED_KEY: 'parsed',
-    TRIM_SPACE_FROM_UNQUOTED: 'some spaced out string'
+    TRIM_SPACE_FROM_UNQUOTED: 'some spaced out string',
   });
 }