 |
handle unescaped string and unexpected EOF and improve handling surrogates - js… |
 |
git clone git://git.codemadness.org/json2tsv |
 |
Log |
|
Files |
|
Refs |
|
README |
|
LICENSE |
|
--- |
|
commit 9f4ab639718e4351b02b4bde6035cd588c32b169 |
|
parent f0b7f8935d41162e29c5a01f15273ba225909969 |
|
Author: Hiltjo Posthuma <[email protected]> |
|
Date: Tue, 15 Oct 2019 18:58:56 +0200 |
|
|
|
handle unescaped string and unexpected EOF and improve handling surrogates |
|
|
|
.... handle UTF-16 surrogate code-point errors and try to recover gracefully, j… |
|
output the raw bytes. |
|
|
|
Diffstat: |
|
M json2tsv.c | 45 ++++++++++++++++++++---------… |
|
|
|
1 file changed, 29 insertions(+), 16 deletions(-) |
|
--- |
|
diff --git a/json2tsv.c b/json2tsv.c |
|
@@ -126,7 +126,7 @@ parsejson(void (*cb)(struct json_node *, size_t, const char… |
|
nodes[depth].type = TYPE_PRIMITIVE; |
|
|
|
while ((c = GETNEXT()) != EOF) { |
|
- /* not whitespace or control-character */ |
|
+ /* not whitespace or control character */ |
|
if (c <= 0x20 || c == 0x7f) |
|
continue; |
|
|
|
@@ -149,14 +149,19 @@ parsejson(void (*cb)(struct json_node *, size_t, const ch… |
|
break; |
|
case '"': |
|
nodes[depth].type = TYPE_STRING; |
|
- for (escape = 0; (c = GETNEXT()) != EOF;) { |
|
- /* 0x7f is not defined as a control-character … |
|
- if (c < 0x20) |
|
- continue; |
|
+ escape = 0; |
|
+ for (;;) { |
|
+ c = GETNEXT(); |
|
+chr: |
|
+ if (c < 0x20) { |
|
+ /* EOF or control char: 0x7f is not de… |
|
+ *errstr = JSON_ERROR_INVALID_CHAR; |
|
+ goto end; |
|
+ } |
|
|
|
if (escape) { |
|
+escchr: |
|
escape = 0; |
|
- |
|
switch (c) { |
|
case '"': /* FALLTHROUGH */ |
|
case '\\': |
|
@@ -167,6 +172,8 @@ parsejson(void (*cb)(struct json_node *, size_t, const char… |
|
case 'r': c = '\r'; break; |
|
case 't': c = '\t'; break; |
|
case 'u': /* hex hex hex hex */ |
|
+ if (capacity(&value, &vz, v, 4… |
|
+ goto end; |
|
for (i = 12, cp = 0; i >= 0; i… |
|
if ((c = GETNEXT()) ==… |
|
*errstr = JSON… |
|
@@ -175,13 +182,17 @@ parsejson(void (*cb)(struct json_node *, size_t, const ch… |
|
cp |= (hexdigit(c) << … |
|
} |
|
/* See also: |
|
- * RFC7159 - 7. Strings and |
|
+ * RFC8259 - 7. Strings and |
|
* https://unicode.org/faq/utf… |
|
* 0xd800 - 0xdb7f - high surr… |
|
if (cp >= 0xd800 && cp <= 0xdb… |
|
- if (GETNEXT() != '\\' … |
|
- *errstr = JSON… |
|
- goto end; |
|
+ if ((c = GETNEXT()) !=… |
|
+ v += codepoint… |
|
+ goto chr; |
|
+ } |
|
+ if ((c = GETNEXT()) !=… |
|
+ v += codepoint… |
|
+ goto escchr; |
|
} |
|
for (hi = cp, i = 12, … |
|
if ((c = GETNE… |
|
@@ -191,14 +202,16 @@ parsejson(void (*cb)(struct json_node *, size_t, const ch… |
|
lo |= (hexdigi… |
|
} |
|
/* 0xdc00 - 0xdfff - l… |
|
- if (!(lo >= 0xdc00 && … |
|
- *errstr = JSON… |
|
- goto end; |
|
+ if (lo >= 0xdc00 && lo… |
|
+ cp = (hi << 10… |
|
+ } else { |
|
+ v += codepoint… |
|
+ if (capacity(&… |
|
+ goto e… |
|
+ v += codepoint… |
|
+ continue; |
|
} |
|
- cp = (hi << 10) + lo -… |
|
} |
|
- if (capacity(&value, &vz, v, 4… |
|
- goto end; |
|
v += codepointtoutf8(cp, &valu… |
|
continue; |
|
default: |
