diff --git a/src/kernel_internals/parsing/logic_preprocess.mll b/src/kernel_internals/parsing/logic_preprocess.mll
index 34ea01b4d6bda1c8a16eb82d54425fc1b82b90ef..765d158daf4a966e8c609f7a6f0e56426b143a7e 100644
--- a/src/kernel_internals/parsing/logic_preprocess.mll
+++ b/src/kernel_internals/parsing/logic_preprocess.mll
@@ -62,9 +62,30 @@
let backslash = "__ANNOT_BACKSLASH__"
let annot_content = "__ANNOT_CONTENT__"
+ let utf8_prefix = "__FC_UTF8_"
+
+ let encode_utf8 c = utf8_prefix ^ (string_of_int (Char.code c))
let re_backslash = Str.regexp_string backslash
let re_annot_content = Str.regexp_string annot_content
+ let re_utf8 = Str.regexp (utf8_prefix ^ "\\([0-9]+\\)")
+
+ let decode_utf8 s =
+ let res = ref s in
+ let start = ref 0 in
+ try
+ while true do
+ let b = Str.search_forward re_utf8 !res !start in
+ let e = Str.match_end () in
+ let chr = Char.chr (int_of_string (Str.matched_group 1 !res)) in
+ let buf = Bytes.of_string !res in
+ Bytes.set buf b chr;
+ Bytes.blit buf e buf (b+1) (String.length !res - e);
+ res:= Bytes.sub_string buf 0 (String.length !res + 1 + b - e);
+ start := b+1;
+ done;
+ assert false;
+ with Not_found -> !res
(* Delimiters for the various annotations in the preprocessing buffer.
We have one delimiter for the beginning of an annotation (to discard
@@ -115,7 +136,7 @@
ignore_content ();
ignore (input_line file); (* ignore the #line directive *)
let with_nl, content = get_annot true in
- with_nl, replace_backslash content
+ with_nl, decode_utf8 @@ replace_backslash content
with End_of_file ->
Kernel.fatal
"too few annotations in result file while pre-processing annotations"
@@ -184,6 +205,8 @@
add_preprocess_line_info()
}
+let utf8 = ['\128'-'\255']
+
rule main = parse
| ("#define"|"#undef") [' ''\t']* ((['a'-'z''A'-'Z''0'-'9''_'])* as m)
{
@@ -388,6 +411,10 @@ and annot = parse
is_newline:=CHAR;
Buffer.add_char preprocess_buffer '"';
string annot lexbuf }
+ | utf8 as c {
+ Buffer.add_string preprocess_buffer (encode_utf8 c);
+ annot lexbuf
+ }
| _ as c { is_newline := CHAR;
Buffer.add_char preprocess_buffer c;
annot lexbuf }
diff --git a/src/plugins/markdown-report/tests/sarif/oracle/with-libc.sarif b/src/plugins/markdown-report/tests/sarif/oracle/with-libc.sarif
index 999ef8e57b7c0cf38f6383d4b549ebdd8f0b1b84..75cbdcd725df74bc9cf5732be0bcd534ed041e59 100644
--- a/src/plugins/markdown-report/tests/sarif/oracle/with-libc.sarif
+++ b/src/plugins/markdown-report/tests/sarif/oracle/with-libc.sarif
@@ -516,8 +516,8 @@
"startLine": 168,
"startColumn": 4,
"endLine": 170,
- "endColumn": 82,
- "byteLength": 146
+ "endColumn": 75,
+ "byteLength": 132
}
}
}
@@ -587,8 +587,8 @@
"startLine": 240,
"startColumn": 4,
"endLine": 244,
- "endColumn": 70,
- "byteLength": 180
+ "endColumn": 63,
+ "byteLength": 173
}
}
}
@@ -1151,7 +1151,7 @@
"startColumn": 4,
"endLine": 121,
"endColumn": 77,
- "byteLength": 155
+ "byteLength": 148
}
}
}
@@ -1266,7 +1266,7 @@
"startColumn": 4,
"endLine": 125,
"endColumn": 77,
- "byteLength": 156
+ "byteLength": 149
}
}
}
@@ -1455,7 +1455,7 @@
"startColumn": 4,
"endLine": 109,
"endColumn": 59,
- "byteLength": 125
+ "byteLength": 118
}
}
}
@@ -1992,8 +1992,8 @@
"startLine": 89,
"startColumn": 4,
"endLine": 90,
- "endColumn": 79,
- "byteLength": 108
+ "endColumn": 72,
+ "byteLength": 101
}
}
}
@@ -2202,8 +2202,8 @@
"startLine": 143,
"startColumn": 4,
"endLine": 147,
- "endColumn": 70,
- "byteLength": 177
+ "endColumn": 63,
+ "byteLength": 170
}
}
}
@@ -2439,7 +2439,7 @@
"startColumn": 4,
"endLine": 59,
"endColumn": 62,
- "byteLength": 148
+ "byteLength": 134
}
}
}
@@ -2788,7 +2788,7 @@
"startColumn": 4,
"endLine": 185,
"endColumn": 63,
- "byteLength": 150
+ "byteLength": 143
}
}
}
@@ -3563,7 +3563,7 @@
"startColumn": 4,
"endLine": 202,
"endColumn": 22,
- "byteLength": 120
+ "byteLength": 113
}
}
}
@@ -3655,7 +3655,7 @@
"startColumn": 4,
"endLine": 269,
"endColumn": 29,
- "byteLength": 167
+ "byteLength": 153
}
}
}
@@ -3958,7 +3958,7 @@
"startColumn": 4,
"endLine": 97,
"endColumn": 58,
- "byteLength": 127
+ "byteLength": 120
}
}
}
@@ -4431,7 +4431,7 @@
"startColumn": 4,
"endLine": 59,
"endColumn": 62,
- "byteLength": 148
+ "byteLength": 134
}
}
}
@@ -4826,7 +4826,7 @@
"startColumn": 4,
"endLine": 105,
"endColumn": 51,
- "byteLength": 115
+ "byteLength": 108
}
}
}
@@ -4849,7 +4849,7 @@
"startColumn": 4,
"endLine": 135,
"endColumn": 38,
- "byteLength": 192
+ "byteLength": 185
}
}
}
@@ -5078,8 +5078,8 @@
"startLine": 68,
"startColumn": 4,
"endLine": 70,
- "endColumn": 70,
- "byteLength": 156
+ "endColumn": 63,
+ "byteLength": 135
}
}
}
@@ -5269,8 +5269,8 @@
"startLine": 252,
"startColumn": 4,
"endLine": 256,
- "endColumn": 60,
- "byteLength": 208
+ "endColumn": 53,
+ "byteLength": 194
}
}
}
@@ -5364,7 +5364,7 @@
"startColumn": 4,
"endLine": 269,
"endColumn": 29,
- "byteLength": 167
+ "byteLength": 153
}
}
}
@@ -5573,7 +5573,7 @@
"startColumn": 4,
"endLine": 101,
"endColumn": 59,
- "byteLength": 124
+ "byteLength": 117
}
}
}
@@ -5739,7 +5739,7 @@
"startColumn": 4,
"endLine": 113,
"endColumn": 51,
- "byteLength": 118
+ "byteLength": 111
}
}
}
@@ -5948,7 +5948,7 @@
"startColumn": 4,
"endLine": 117,
"endColumn": 62,
- "byteLength": 157
+ "byteLength": 143
}
}
}
@@ -6636,7 +6636,7 @@
"startColumn": 4,
"endLine": 130,
"endColumn": 38,
- "byteLength": 191
+ "byteLength": 184
}
}
}
@@ -6731,8 +6731,8 @@
"startLine": 39,
"startColumn": 4,
"endLine": 42,
- "endColumn": 77,
- "byteLength": 184
+ "endColumn": 70,
+ "byteLength": 170
}
}
}
@@ -6800,8 +6800,8 @@
"startLine": 155,
"startColumn": 4,
"endLine": 159,
- "endColumn": 60,
- "byteLength": 205
+ "endColumn": 53,
+ "byteLength": 191
}
}
}
@@ -7177,8 +7177,8 @@
"startLine": 143,
"startColumn": 4,
"endLine": 147,
- "endColumn": 70,
- "byteLength": 177
+ "endColumn": 63,
+ "byteLength": 170
}
}
}
@@ -7246,8 +7246,8 @@
"startLine": 240,
"startColumn": 4,
"endLine": 244,
- "endColumn": 70,
- "byteLength": 180
+ "endColumn": 63,
+ "byteLength": 173
}
}
}
@@ -7362,7 +7362,7 @@
"startColumn": 4,
"endLine": 82,
"endColumn": 40,
- "byteLength": 183
+ "byteLength": 169
}
}
}
@@ -7500,7 +7500,7 @@
"startColumn": 4,
"endLine": 135,
"endColumn": 38,
- "byteLength": 192
+ "byteLength": 185
}
}
}
@@ -7592,7 +7592,7 @@
"startColumn": 4,
"endLine": 197,
"endColumn": 41,
- "byteLength": 188
+ "byteLength": 174
}
}
}
@@ -7614,8 +7614,8 @@
"startLine": 155,
"startColumn": 4,
"endLine": 159,
- "endColumn": 60,
- "byteLength": 205
+ "endColumn": 53,
+ "byteLength": 191
}
}
}
@@ -8124,8 +8124,8 @@
"startLine": 252,
"startColumn": 4,
"endLine": 256,
- "endColumn": 60,
- "byteLength": 208
+ "endColumn": 53,
+ "byteLength": 194
}
}
}
@@ -8267,7 +8267,7 @@
"startColumn": 4,
"endLine": 185,
"endColumn": 63,
- "byteLength": 150
+ "byteLength": 143
}
}
}
@@ -8595,8 +8595,8 @@
"startLine": 39,
"startColumn": 4,
"endLine": 42,
- "endColumn": 77,
- "byteLength": 184
+ "endColumn": 70,
+ "byteLength": 170
}
}
}
@@ -8738,7 +8738,7 @@
"startColumn": 4,
"endLine": 87,
"endColumn": 22,
- "byteLength": 116
+ "byteLength": 109
}
}
}
@@ -8829,8 +8829,8 @@
"startLine": 68,
"startColumn": 4,
"endLine": 70,
- "endColumn": 70,
- "byteLength": 156
+ "endColumn": 63,
+ "byteLength": 135
}
}
}
@@ -9538,8 +9538,8 @@
"startLine": 168,
"startColumn": 4,
"endLine": 170,
- "endColumn": 82,
- "byteLength": 146
+ "endColumn": 75,
+ "byteLength": 132
}
}
}