validate BSON document size against the bytes read (#5287)

.github/SECURITY.md

@@ -15,6 +15,14 @@ guidance.
 
 For vulnerabilities in third-party dependencies or modules, please report them directly to the respective maintainers.
 
+## Unofficial packages
+
+This project does not publish an official npm package. The npm package
+[`nlohmann-json`](https://www.npmjs.com/package/nlohmann-json) (or similarly named packages) is not maintained or
+endorsed by this project. See the
+[package managers documentation](https://json.nlohmann.me/integration/package_managers/#npm) for supported
+integration options.
+
 ## Additional Resources
 
 - Explore security-related topics and contribute to tools and projects through

docs/mkdocs/docs/integration/package_managers.md

@@ -930,6 +930,12 @@ If you are using [CocoaPods](https://cocoapods.org), you can use the library by
 to your podfile (see [an example](https://bitbucket.org/benman/nlohmann_json-cocoapod/src/master/)). Please file issues
 [here](https://bitbucket.org/benman/nlohmann_json-cocoapod/issues?status=new&status=open).
 
+## npm
+
+This project does not publish an official [npm](https://www.npmjs.com) package. The npm package
+[`nlohmann-json`](https://www.npmjs.com/package/nlohmann-json) (or similarly named packages) is not maintained or
+endorsed by this project. Use one of the package managers listed above, or integrate the single header directly.
+
 ## ESP-IDF and PlatformIO
 
 There is no official package published to the [ESP-IDF Component Registry](https://components.espressif.com) or the

include/nlohmann/detail/input/binary_reader.hpp

@@ -163,12 +163,39 @@ class binary_reader
     // BSON //
     //////////
 
+    /*!
+    @brief Validate a BSON document's declared size against the bytes read.
+
+    A BSON document starts with an int32 that counts its own total length in
+    bytes, including that prefix and the trailing 0x00. The reader is driven
+    by the terminator rather than the declared length, so without this check a
+    nested document could declare a length that disagrees with where its
+    terminator actually falls and quietly hand the bytes in between to the
+    enclosing document. A well-formed document is at least 5 bytes (the prefix
+    plus the terminator); the equality also rejects those impossible sizes,
+    since at least 5 bytes are always consumed.
+
+    @param[in] document_start  value of chars_read before the size prefix
+    @param[in] document_size   the declared document size
+    @return whether the declared size matches the number of bytes read
+    */
+    bool check_bson_document_size(const std::size_t document_start, const std::int32_t document_size)
+    {
+        if (JSON_HEDLEY_UNLIKELY(document_size < 0 || static_cast<std::size_t>(document_size) != chars_read - document_start))
+        {
+            return sax->parse_error(chars_read, get_token_string(), parse_error::create(112, chars_read,
+                                    exception_message(input_format_t::bson, concat("document size ", std::to_string(document_size), " does not match the number of bytes read (", std::to_string(chars_read - document_start), ")"), "document"), nullptr));
+        }
+        return true;
+    }
+
     /*!
     @brief Reads in a BSON-object and passes it to the SAX-parser.
     @return whether a valid BSON-value was passed to the SAX parser
     */
     bool parse_bson_internal()
     {
+        const std::size_t document_start = chars_read;
         std::int32_t document_size{};
         get_number<std::int32_t, true>(input_format_t::bson, document_size);
 
@@ -182,6 +209,11 @@ class binary_reader
             return false;
         }
 
+        if (JSON_HEDLEY_UNLIKELY(!check_bson_document_size(document_start, document_size)))
+        {
+            return false;
+        }
+
         return sax->end_object();
     }
 
@@ -397,6 +429,7 @@ class binary_reader
     */
     bool parse_bson_array()
     {
+        const std::size_t document_start = chars_read;
         std::int32_t document_size{};
         get_number<std::int32_t, true>(input_format_t::bson, document_size);
 
@@ -410,6 +443,11 @@ class binary_reader
             return false;
         }
 
+        if (JSON_HEDLEY_UNLIKELY(!check_bson_document_size(document_start, document_size)))
+        {
+            return false;
+        }
+
         return sax->end_array();
     }
 
@@ -1869,24 +1907,6 @@ class binary_reader
         return true;
     }
 
-    /*!
-    @brief create the error message for a missing/invalid length-type marker
-
-    Used both when reading a UBJSON/BJData string and when reading an optimized
-    container size. The accepted markers depend on the input format, and the
-    size variant appends " after '#'" via @a infix.
-
-    @param[in] last_token  the offending byte as returned by get_token_string()
-    @param[in] infix       extra context inserted after the marker list
-    @return the formatted error message
-    */
-    std::string unexpected_length_type_message(const std::string& last_token, const char* infix) const
-    {
-        return concat("expected length type specification (",
-                      input_format == input_format_t::bjdata ? "U, i, u, I, m, l, M, L" : "U, i, I, l, L",
-                      ")", infix, "; last byte: 0x", last_token);
-    }
-
     /*!
     @brief reads a UBJSON string
 
@@ -1979,8 +1999,17 @@ class binary_reader
                 break;
         }
         auto last_token = get_token_string();
-        return sax->parse_error(chars_read, last_token, parse_error::create(113, chars_read,
+        std::string message;
-                                exception_message(input_format, unexpected_length_type_message(last_token, ""), "string"), nullptr));
+
+        if (input_format != input_format_t::bjdata)
+        {
+            message = "expected length type specification (U, i, I, l, L); last byte: 0x" + last_token;
+        }
+        else
+        {
+            message = "expected length type specification (U, i, u, I, m, l, M, L); last byte: 0x" + last_token;
+        }
+        return sax->parse_error(chars_read, last_token, parse_error::create(113, chars_read, exception_message(input_format, message, "string"), nullptr));
     }
 
     /*!
@@ -2259,8 +2288,17 @@ class binary_reader
                 break;
         }
         auto last_token = get_token_string();
-        return sax->parse_error(chars_read, last_token, parse_error::create(113, chars_read,
+        std::string message;
-                                exception_message(input_format, unexpected_length_type_message(last_token, " after '#'"), "size"), nullptr));
+
+        if (input_format != input_format_t::bjdata)
+        {
+            message = "expected length type specification (U, i, I, l, L) after '#'; last byte: 0x" + last_token;
+        }
+        else
+        {
+            message = "expected length type specification (U, i, u, I, m, l, M, L) after '#'; last byte: 0x" + last_token;
+        }
+        return sax->parse_error(chars_read, last_token, parse_error::create(113, chars_read, exception_message(input_format, message, "size"), nullptr));
     }
 
     /*!

single_include/nlohmann/json.hpp

@@ -10712,12 +10712,39 @@ class binary_reader
     // BSON //
     //////////
 
+    /*!
+    @brief Validate a BSON document's declared size against the bytes read.
+
+    A BSON document starts with an int32 that counts its own total length in
+    bytes, including that prefix and the trailing 0x00. The reader is driven
+    by the terminator rather than the declared length, so without this check a
+    nested document could declare a length that disagrees with where its
+    terminator actually falls and quietly hand the bytes in between to the
+    enclosing document. A well-formed document is at least 5 bytes (the prefix
+    plus the terminator); the equality also rejects those impossible sizes,
+    since at least 5 bytes are always consumed.
+
+    @param[in] document_start  value of chars_read before the size prefix
+    @param[in] document_size   the declared document size
+    @return whether the declared size matches the number of bytes read
+    */
+    bool check_bson_document_size(const std::size_t document_start, const std::int32_t document_size)
+    {
+        if (JSON_HEDLEY_UNLIKELY(document_size < 0 || static_cast<std::size_t>(document_size) != chars_read - document_start))
+        {
+            return sax->parse_error(chars_read, get_token_string(), parse_error::create(112, chars_read,
+                                    exception_message(input_format_t::bson, concat("document size ", std::to_string(document_size), " does not match the number of bytes read (", std::to_string(chars_read - document_start), ")"), "document"), nullptr));
+        }
+        return true;
+    }
+
     /*!
     @brief Reads in a BSON-object and passes it to the SAX-parser.
     @return whether a valid BSON-value was passed to the SAX parser
     */
     bool parse_bson_internal()
     {
+        const std::size_t document_start = chars_read;
         std::int32_t document_size{};
         get_number<std::int32_t, true>(input_format_t::bson, document_size);
 
@@ -10731,6 +10758,11 @@ class binary_reader
             return false;
         }
 
+        if (JSON_HEDLEY_UNLIKELY(!check_bson_document_size(document_start, document_size)))
+        {
+            return false;
+        }
+
         return sax->end_object();
     }
 
@@ -10946,6 +10978,7 @@ class binary_reader
     */
     bool parse_bson_array()
     {
+        const std::size_t document_start = chars_read;
         std::int32_t document_size{};
         get_number<std::int32_t, true>(input_format_t::bson, document_size);
 
@@ -10959,6 +10992,11 @@ class binary_reader
             return false;
         }
 
+        if (JSON_HEDLEY_UNLIKELY(!check_bson_document_size(document_start, document_size)))
+        {
+            return false;
+        }
+
         return sax->end_array();
     }
 
@@ -12418,24 +12456,6 @@ class binary_reader
         return true;
     }
 
-    /*!
-    @brief create the error message for a missing/invalid length-type marker
-
-    Used both when reading a UBJSON/BJData string and when reading an optimized
-    container size. The accepted markers depend on the input format, and the
-    size variant appends " after '#'" via @a infix.
-
-    @param[in] last_token  the offending byte as returned by get_token_string()
-    @param[in] infix       extra context inserted after the marker list
-    @return the formatted error message
-    */
-    std::string unexpected_length_type_message(const std::string& last_token, const char* infix) const
-    {
-        return concat("expected length type specification (",
-                      input_format == input_format_t::bjdata ? "U, i, u, I, m, l, M, L" : "U, i, I, l, L",
-                      ")", infix, "; last byte: 0x", last_token);
-    }
-
     /*!
     @brief reads a UBJSON string
 
@@ -12528,8 +12548,17 @@ class binary_reader
                 break;
         }
         auto last_token = get_token_string();
-        return sax->parse_error(chars_read, last_token, parse_error::create(113, chars_read,
+        std::string message;
-                                exception_message(input_format, unexpected_length_type_message(last_token, ""), "string"), nullptr));
+
+        if (input_format != input_format_t::bjdata)
+        {
+            message = "expected length type specification (U, i, I, l, L); last byte: 0x" + last_token;
+        }
+        else
+        {
+            message = "expected length type specification (U, i, u, I, m, l, M, L); last byte: 0x" + last_token;
+        }
+        return sax->parse_error(chars_read, last_token, parse_error::create(113, chars_read, exception_message(input_format, message, "string"), nullptr));
     }
 
     /*!
@@ -12808,8 +12837,17 @@ class binary_reader
                 break;
         }
         auto last_token = get_token_string();
-        return sax->parse_error(chars_read, last_token, parse_error::create(113, chars_read,
+        std::string message;
-                                exception_message(input_format, unexpected_length_type_message(last_token, " after '#'"), "size"), nullptr));
+
+        if (input_format != input_format_t::bjdata)
+        {
+            message = "expected length type specification (U, i, I, l, L) after '#'; last byte: 0x" + last_token;
+        }
+        else
+        {
+            message = "expected length type specification (U, i, u, I, m, l, M, L) after '#'; last byte: 0x" + last_token;
+        }
+        return sax->parse_error(chars_read, last_token, parse_error::create(113, chars_read, exception_message(input_format, message, "size"), nullptr));
     }
 
     /*!

tests/src/unit-bson.cpp

@@ -854,6 +854,62 @@ TEST_CASE("Unsupported BSON input")
     CHECK(!json::sax_parse(bson, &scp, json::input_format_t::bson));
 }
 
+TEST_CASE("BSON document size mismatch")
+{
+    json _;
+
+    SECTION("top-level document declaring more bytes than it contains")
+    {
+        // empty object, but the length prefix claims 6 bytes instead of 5
+        std::vector<std::uint8_t> const input = {0x06, 0x00, 0x00, 0x00, 0x00};
+        CHECK_THROWS_WITH_AS(_ = json::from_bson(input), "[json.exception.parse_error.112] parse error at byte 5: syntax error while parsing BSON document: document size 6 does not match the number of bytes read (5)", json::parse_error&);
+        CHECK(json::from_bson(input, true, false).is_discarded());
+    }
+
+    SECTION("top-level document with a negative size")
+    {
+        std::vector<std::uint8_t> const input = {0xFF, 0xFF, 0xFF, 0xFF, 0x00};
+        CHECK_THROWS_WITH_AS(_ = json::from_bson(input), "[json.exception.parse_error.112] parse error at byte 5: syntax error while parsing BSON document: document size -1 does not match the number of bytes read (5)", json::parse_error&);
+        CHECK(json::from_bson(input, true, false).is_discarded());
+    }
+
+    SECTION("embedded document whose size disagrees with its terminator")
+    {
+        // the embedded document "d" declares 0x7FFFFFFF bytes but its 0x00
+        // terminator falls right after {"a":null}; the length prefix would
+        // otherwise let the following "h" element be read as a member of the
+        // enclosing document instead of "d"
+        std::vector<std::uint8_t> const input =
+        {
+            0x00, 0x00, 0x00, 0x00, // outer size
+            0x03, 'd', 0x00,        // entry: embedded document "d"
+            0xFF, 0xFF, 0xFF, 0x7F, //   embedded size 0x7FFFFFFF
+            0x0A, 'a', 0x00,        //   entry: null "a"
+            0x00,                   //   embedded end marker
+            0x08, 'h', 0x00, 0x01,  // entry: bool "h" = true
+            0x00                    // outer end marker
+        };
+        CHECK_THROWS_WITH_AS(_ = json::from_bson(input), "[json.exception.parse_error.112] parse error at byte 15: syntax error while parsing BSON document: document size 2147483647 does not match the number of bytes read (8)", json::parse_error&);
+        CHECK(json::from_bson(input, true, false).is_discarded());
+    }
+
+    SECTION("embedded array whose size disagrees with its terminator")
+    {
+        // array [42] is 12 bytes, but the length prefix claims 13
+        std::vector<std::uint8_t> const input =
+        {
+            0x00, 0x00, 0x00, 0x00, // outer size
+            0x04, 'a', 0x00,        // entry: array "a"
+            0x0D, 0x00, 0x00, 0x00, //   array size 13 (real is 12)
+            0x10, '0', 0x00, 0x2A, 0x00, 0x00, 0x00, //   entry: int32 "0" = 42
+            0x00,                   //   array end marker
+            0x00                    // outer end marker
+        };
+        CHECK_THROWS_WITH_AS(_ = json::from_bson(input), "[json.exception.parse_error.112] parse error at byte 19: syntax error while parsing BSON document: document size 13 does not match the number of bytes read (12)", json::parse_error&);
+        CHECK(json::from_bson(input, true, false).is_discarded());
+    }
+}
+
 TEST_CASE("BSON numerical data")
 {
     SECTION("number")