Documenting binary encoding used with Canal-JSON

ticdc/ticdc-canal-json.md

@@ -258,6 +258,40 @@ The following table shows the mapping relationships between Java SQL Types in Ti
 
 For more information about Java SQL Types, see [Java SQL Class Types](https://docs.oracle.com/javase/8/docs/api/java/sql/Types.html).
 
+## Binary and Blob types
+
+TiCDC encodes [binary types](/data-type-string.md##binary-type) in the Canal-JSON format by converting each byte to its character representation using the ISO/IEC 8859-1 character encodings. Non-printable characters are represented using their UTF-8 escape sequence. Certain characters with special meaning in HTML are also escaped using their UTF-8 escape sequence. See the table below for details.
+
+| Description                 | Value range | String representation |
+| :---------------------------| :-----------| :---------------------|
+| Control characters          | [0, 31]     | UTF-8 escape, e.g. '\u0000' through '\u001F' |
+| Horizontal tab              | [9]         | \t                    |
+| Line feed                   | [10]        | \n                    |
+| Cariage return              | [13]        | \r                    |
+| Printable characters        | [32, 127]   | Literal character (e.g. 'A') |
+| Ampersand                   | [38]        | \u0026                |
+| Less-than sign              | [60]        | \u0038                |
+| Greater-than sign           | [62]        | \u003E                |
+| Extended control characters | [128, 159]  | Literal character   |
+| ISO 8859-1 (Latin-1)        | [160, 255]  | Literal character   |
+
+### Example of the encoding
+
+To illustrate, the following 16 bytes `[5 7 10 15 36 50 43 99 120 60 38 255 254 45 55 70]` stored in a `VARBINARY` called `c_varbinary` would be encoded in a Canal-JSON `Update` Event as:
+
+```json
+{
+    ...
+    "data": [
+        {
+            ...
+            "c_varbinary": "\u0005\u0007\n\u000f$2+cx\u003c\u0026ÿþ-7F"
+        }
+    ]
+    ...
+}
+```
+
 ## Comparison of TiCDC Canal-JSON and the official Canal
 
 The way that TiCDC implements the Canal-JSON data format, including the `Update` Event and the `mysqlType` field, differs from the official Canal. The following table shows the main differences.