@@ -76,7 +76,7 @@ char *pdf_convert_utf(char *begin, size_t sz)

     char *res=NULL;

 #if HAVE_ICONV

     char *buf, *outbuf, *p1, *p2;

-    size_t inlen, outlen, i;

+    size_t sz2, inlen, outlen, i;

     char *encodings[] = {

         "UTF-16",

         NULL

@@ -87,10 +87,63 @@ char *pdf_convert_utf(char *begin, size_t sz)

     if (!(buf))

         return NULL;

-    memcpy(buf, begin, sz);

+    /* convert PDF specific escape sequences, like octal sequences */

+    sz2 = 0;

+    for (i = 0; i < sz; ++i) {

+        if ((i+1 < sz) && begin[i] == '\\') {

+            if ((i+3 < sz) &&

+                (isdigit(begin[i+1]) && isdigit(begin[i+2]) && isdigit(begin[i+3]))) {

+                /* octal sequence */

+                char octal[4], *check;

+                unsigned long value;

+

+                memcpy(octal, &begin[i+1], 3);

+                octal[3] = '\0';

+

+                value = (char)strtoul(octal, &check, 8);

+                /* check if all characters were converted */

+                if (check == &octal[3])

+                    buf[sz2++] = value;

+                i += 3;

+            } else {

+                /* other sequences */

+                switch(begin[i+1]) {

+                case 'n':

+                    buf[sz2++] = 0x0a;

+                    break;

+                case 'r':

+                    buf[sz2++] = 0x0d;

+                    break;

+                case 't':

+                    buf[sz2++] = 0x09;

+                    break;

+                case 'b':

+                    buf[sz2++] = 0x08;

+                    break;

+                case 'f':

+                    buf[sz2++] = 0x0c;

+                    break;

+                case '(':

+                    buf[sz2++] = 0x28;

+                    break;

+                case ')':

+                    buf[sz2++] = 0x29;

+                    break;

+                case '\\':

+                    buf[sz2++] = 0x5c;

+                    break;

+                default:

+                    /* IGNORE THE REVERSE SOLIDUS - PDF3000-2008 */

+                    break;

+                }

+            }

+        } else

+            buf[sz2++] = begin[i]; 

+    }

+    //memcpy(buf, begin, sz);

     p1 = buf;

-    p2 = outbuf = cli_calloc(1, sz+1);

+    p2 = outbuf = cli_calloc(1, sz2+1);

     if (!(outbuf)) {

         free(buf);

         return NULL;

@@ -99,7 +152,7 @@ char *pdf_convert_utf(char *begin, size_t sz)

     for (i=0; encodings[i] != NULL; i++) {

         p1 = buf;

         p2 = outbuf;

-        inlen = outlen = sz;

+        inlen = outlen = sz2;

         cd = iconv_open("UTF-8", encodings[i]);

         if (cd == (iconv_t)(-1)) {

@@ -109,13 +162,13 @@ char *pdf_convert_utf(char *begin, size_t sz)

         iconv(cd, (char **)(&p1), &inlen, &p2, &outlen);

-        if (outlen == sz) {

+        if (outlen == sz2) {

             /* Decoding unsuccessful right from the start */

             iconv_close(cd);

             continue;

         }

-        outbuf[sz - outlen] = '\0';

+        outbuf[sz2 - outlen] = '\0';

         res = strdup(outbuf);

         iconv_close(cd);

@@ -277,10 +330,11 @@ char *pdf_parse_string(struct pdf_struct *pdf, struct pdf_obj *obj, const char *

     p2 = (char *)(q + objsize);

     if (is_object_reference(p1, &p2, &objid)) {

         struct pdf_obj *newobj;

-        char *begin;

+        char *begin, *p3;

         STATBUF sb;

         uint32_t objflags;

         int fd;

+        size_t objsize2;

         newobj = find_obj(pdf, obj, objid);

         if (!(newobj))

@@ -343,25 +397,32 @@ char *pdf_parse_string(struct pdf_struct *pdf, struct pdf_obj *obj, const char *

                 return NULL;

             }

-            switch (begin[0]) {

+            p3 = begin;

+            objsize2 = sb.st_size;

+            while ((size_t)(p3 - begin) < objsize2 && isspace(p3[0])) {

+                p3++;

+                objsize2--;

+            }

+

+            switch (*p3) {

                 case '(':

                 case '<':

-                    res = pdf_parse_string(pdf, obj, begin, sb.st_size, NULL, NULL);

+                    res = pdf_parse_string(pdf, obj, p3, objsize2, NULL, NULL);

                     free(begin);

                     break;

                 default:

-                    for (i=0; i < sb.st_size; i++) {

-                        if (begin[i] >= 0x7f) {

+                    for (i=0; i < objsize2; i++) {

+                        if (p3[i] >= 0x7f) {

                             likelyutf=1;

                             break;

                         }

                     }

-                    res = likelyutf ? pdf_convert_utf(begin, sb.st_size) : NULL;

+                    res = likelyutf ? pdf_convert_utf(p3, objsize2) : NULL;

                     if (!(res)) {

                         res = begin;

-                        res[sb.st_size] = '\0';

+                        res[objsize2] = '\0';

                     } else {

                         free(begin);

                     }