Changeset 3227 for libcacaTweet

Timestamp:

Nov 2, 2008, 11:12:24 PM (15 years ago)

Author:

bsittler

Message:

some unicode characters could overflow the buffer; handle more control
characters and non-unicode codepoints when generating HTML.

File:

• libcaca/trunk/caca/codec/export.c (modified) (3 diffs)

libcaca/trunk/caca/codec/export.c

@@ -504 +504 @@
      * A glyph: up to 48 chars for "<td bgcolor=\"#xxxxxx\"><tt><font color=\"#xxxxxx\">"
      *          up to 36 chars for "<b><i><u><blink></blink></u></i></b>"
-     *          up to 9 chars for "&#xxxxxx;" (far less for pure ASCII)
+     *          up to 10 chars for "&#xxxxxxx;" (far less for pure ASCII)
      *          17 chars for "</font></tt></td>" */
-    *bytes = 1000 + cv->height * (10 + maxcols * (48 + 36 + 9 + 17));
+    *bytes = 1000 + cv->height * (10 + maxcols * (48 + 36 + 10 + 17));
     cur = data = malloc(*bytes);
 
@@ -584 +584 @@
                 if(linechar[x + i] == CACA_MAGIC_FULLWIDTH)
                     ;
-                else if(linechar[x + i] <= 0x00000020)
+                else if((linechar[x + i] <= 0x00000020)
+                        ||
+                        ((linechar[x + i] >= 0x0000007f)
+                         &&
+                         (linechar[x + i] <= 0x0000009f)))
                 {
                     /* Control characters and space converted to
@@ -614 +618 @@
                 else if(linechar[x + i] < 0x00000080)
                     cur += sprintf(cur, "%c", (uint8_t)linechar[x + i]);
+                else if((linechar[x + i] <= 0x0010fffd)
+                        &&
+                        ((linechar[x + i] & 0x0000fffe) != 0x0000fffe)
+                        &&
+                        ((linechar[x + i] < 0x0000d800)
+                         ||
+                         (linechar[x + i] > 0x0000dfff)))
+                    cur += sprintf(cur, "&#%i;", (unsigned int)linechar[x + i]);
                 else
-                    cur += sprintf(cur, "&#%i;", (unsigned int)linechar[x + i]);
+                    /* non-character codepoints become U+FFFD
+                     * REPLACEMENT CHARACTER */
+                    cur += sprintf(cur, "&#%i;", (unsigned int)0x0000fffd);
 
                 if (((i + 1) == len) || (lineattr[x + i + 1] != lineattr[x + i]))