From 70568520ab10be616f1ca7abaf6d1de11842c4ce Mon Sep 17 00:00:00 2001
From: Cole Schlesinger <cole@akitasoftware.com>
Date: Fri, 21 Oct 2022 12:20:11 -0700
Subject: [PATCH] Replace '\C' with '\\' instead of '\'

When preprocessing JSON to remove control characters, replace escaped control
characters with a backslash.  This ensures the JSON parser will interpret the
escaped control character as an escaped backslash, rather than escaping the
next character after the control character.
---
 learn/json_preprocessor.go      | 38 +++++++++++++++++-----
 learn/json_preprocessor_test.go | 57 +++++++++++++++++++++++----------
 2 files changed, 70 insertions(+), 25 deletions(-)

diff --git a/learn/json_preprocessor.go b/learn/json_preprocessor.go
index 9b137b10..737fff16 100644
--- a/learn/json_preprocessor.go
+++ b/learn/json_preprocessor.go
@@ -5,17 +5,23 @@ import (
 )
 
 type stripControlCharactersReader struct {
-	wrapped io.Reader
+	hasPrecedingBackslash bool
+	wrapped               io.Reader
 }
 
-func newStripControlCharactersReader(wrapped io.Reader) stripControlCharactersReader {
-	return stripControlCharactersReader{wrapped: wrapped}
+func newStripControlCharactersReader(wrapped io.Reader) *stripControlCharactersReader {
+	return &stripControlCharactersReader{wrapped: wrapped}
 }
 
 // Read up to len(p) bytes, removing any control characters found.
-// Removed characters do not count toward the total bytes read.  Returns the
-// number of bytes read.
-func (r stripControlCharactersReader) Read(p []byte) (n int, err error) {
+// Removed characters do not count toward the total bytes read.
+//
+// If the control character is preceded by a backslash, it is replaced with a
+// backslash rather than removed, e.g. "\<CR>" becomes "\\".  This prevents
+// the JSON parser from applying the backslash to escape the next character.
+//
+// Returns the number of bytes read.
+func (r *stripControlCharactersReader) Read(p []byte) (n int, err error) {
 	pIdx := 0
 	buf := make([]byte, len(p))
 
@@ -31,10 +37,26 @@ func (r stripControlCharactersReader) Read(p []byte) (n int, err error) {
 
 		// Copy from buf to p, skipping control characters.
 		for _, c := range bufSlice[:bufN] {
+			toWrite := c
+
 			if c <= 0x1f {
-				continue
+				if r.hasPrecedingBackslash {
+					// If the control character is escaped, replace it with a
+					// backslash.
+					toWrite = '\\'
+				} else {
+					// Otherwise, just remove it.
+					continue
+				}
+			}
+
+			if c == '\\' {
+				r.hasPrecedingBackslash = !r.hasPrecedingBackslash
+			} else {
+				r.hasPrecedingBackslash = false
 			}
-			p[pIdx] = c
+
+			p[pIdx] = toWrite
 			pIdx += 1
 		}
 
diff --git a/learn/json_preprocessor_test.go b/learn/json_preprocessor_test.go
index c8a60b70..449ed8e0 100644
--- a/learn/json_preprocessor_test.go
+++ b/learn/json_preprocessor_test.go
@@ -102,10 +102,9 @@ func TestSkipCarriageReturnReader(t *testing.T) {
 
 func TestSkipControlCharacterReader_JSON(t *testing.T) {
 	testCases := []struct {
-		Name          string
-		Input         string
-		Expected      interface{}
-		ExpectedError bool
+		Name     string
+		Input    string
+		Expected interface{}
 	}{
 		{
 			Name:     "no control characters",
@@ -119,11 +118,11 @@ func TestSkipControlCharacterReader_JSON(t *testing.T) {
 		},
 		{
 			Name:  "escaped control char",
-			Input: "{\"foo\\\r\\\n\": \"bar\"}",
+			Input: "{\"foo\\\r\": \"bar\"}",
 
 			// The Go JSON parser doesn't support escaping control characters.
 			// However, if someone were to try, the preprocessor would remove
-			// the control character but leave the backslash.
+			// the control character and replace it with a backslash.
 			Expected: map[string]interface{}{`foo\`: "bar"},
 		},
 		{
@@ -133,25 +132,49 @@ func TestSkipControlCharacterReader_JSON(t *testing.T) {
 ",
   "subject": "world"
 }`,
-			Expected: map[string]interface{}{"greeting": "hello", "subject": "world"},
-
-			// After the newline is removed, the JSON parser interprets the backslash
-			// as escaping the quote.
-			ExpectedError: true,
+			Expected: map[string]interface{}{"greeting": "hello \\", "subject": "world"},
 		},
 	}
 
 	for _, tc := range testCases {
 		// Parse JSON after removing control strings.
 		var parsed interface{}
+		var err error
+
+		// Try parsing using a JSON decoder.
 		decoder := json.NewDecoder(newStripControlCharactersReader(strings.NewReader(tc.Input)))
-		err := decoder.Decode(&parsed)
-		if tc.ExpectedError {
-			assert.Error(t, err, tc.Name)
-		} else {
-			assert.NoError(t, err, tc.Name)
-			assert.Equal(t, tc.Expected, parsed, tc.Name)
+		err = decoder.Decode(&parsed)
+		assert.NoError(t, err, "["+tc.Name+", decoder] error")
+		assert.Equal(t, tc.Expected, parsed, "["+tc.Name+", decoder] not equal")
+
+		// Try reading all bytes from the reader, one byte at a time, and then
+		// parsing.
+		bytesToRead := 1
+		reader := newStripControlCharactersReader(strings.NewReader(tc.Input))
+		read := make([]byte, 0, len([]byte(tc.Input)))
+		buf := make([]byte, bytesToRead)
+		for {
+			var n int
+			n, err = reader.Read(buf)
+
+			read = append(read, buf[:n]...)
+
+			if err != nil {
+				break
+			}
+
+			// Clear the buffer.
+			for i, _ := range buf {
+				buf[i] = '\x00'
+			}
+		}
+
+		// Parse.
+		if err == io.EOF {
+			err = json.Unmarshal(read, &parsed)
 		}
+		assert.NoError(t, err, "["+tc.Name+", byte-by-byte] error")
+		assert.Equal(t, tc.Expected, parsed, "["+tc.Name+", byte-by-byte] not equal")
 	}
 }