summaryrefslogtreecommitdiff
path: root/pdf
diff options
context:
space:
mode:
Diffstat (limited to 'pdf')
-rw-r--r--pdf/mupdf.h2
-rw-r--r--pdf/pdf_annot.c2
-rw-r--r--pdf/pdf_crypt.c125
-rw-r--r--pdf/pdf_form.c18
-rw-r--r--pdf/pdf_interpret.c15
-rw-r--r--pdf/pdf_nametree.c2
-rw-r--r--pdf/pdf_object.c8
-rw-r--r--pdf/pdf_parse.c12
-rw-r--r--pdf/pdf_write.c4
-rw-r--r--pdf/pdf_xobject.c8
10 files changed, 158 insertions, 38 deletions
diff --git a/pdf/mupdf.h b/pdf/mupdf.h
index f2d235b9..16b0f58b 100644
--- a/pdf/mupdf.h
+++ b/pdf/mupdf.h
@@ -17,7 +17,7 @@ pdf_obj *pdf_new_null(fz_context *ctx);
pdf_obj *pdf_new_bool(fz_context *ctx, int b);
pdf_obj *pdf_new_int(fz_context *ctx, int i);
pdf_obj *pdf_new_real(fz_context *ctx, float f);
-pdf_obj *fz_new_name(fz_context *ctx, char *str);
+pdf_obj *pdf_new_name(fz_context *ctx, char *str);
pdf_obj *pdf_new_string(fz_context *ctx, char *str, int len);
pdf_obj *pdf_new_indirect(fz_context *ctx, int num, int gen, void *doc);
pdf_obj *pdf_new_array(fz_context *ctx, int initialcap);
diff --git a/pdf/pdf_annot.c b/pdf/pdf_annot.c
index 5a463336..18a0c947 100644
--- a/pdf/pdf_annot.c
+++ b/pdf/pdf_annot.c
@@ -198,6 +198,8 @@ pdf_parse_action(pdf_document *xref, pdf_obj *action)
pdf_obj *obj, *dest;
fz_context *ctx = xref->ctx;
+ UNUSED(ctx);
+
ld.kind = FZ_LINK_NONE;
if (!action)
diff --git a/pdf/pdf_crypt.c b/pdf/pdf_crypt.c
index 5329e845..ec6f792b 100644
--- a/pdf/pdf_crypt.c
+++ b/pdf/pdf_crypt.c
@@ -103,8 +103,8 @@ pdf_new_crypt(fz_context *ctx, pdf_obj *dict, pdf_obj *id)
obj = pdf_dict_gets(dict, "O");
if (pdf_is_string(obj) && pdf_to_str_len(obj) == 32)
memcpy(crypt->o, pdf_to_str_buf(obj), 32);
- /* /O and /U are supposed to be 48 bytes long for revision 5, they're often longer, though */
- else if (crypt->r == 5 && pdf_is_string(obj) && pdf_to_str_len(obj) >= 48)
+ /* /O and /U are supposed to be 48 bytes long for revision 5 and 6, they're often longer, though */
+ else if (crypt->r >= 5 && pdf_is_string(obj) && pdf_to_str_len(obj) >= 48)
memcpy(crypt->o, pdf_to_str_buf(obj), 48);
else
{
@@ -115,8 +115,8 @@ pdf_new_crypt(fz_context *ctx, pdf_obj *dict, pdf_obj *id)
obj = pdf_dict_gets(dict, "U");
if (pdf_is_string(obj) && pdf_to_str_len(obj) == 32)
memcpy(crypt->u, pdf_to_str_buf(obj), 32);
- /* /O and /U are supposed to be 48 bytes long for revision 5, they're often longer, though */
- else if (crypt->r == 5 && pdf_is_string(obj) && pdf_to_str_len(obj) >= 48)
+ /* /O and /U are supposed to be 48 bytes long for revision 5 and 6, they're often longer, though */
+ else if (crypt->r >= 5 && pdf_is_string(obj) && pdf_to_str_len(obj) >= 48)
memcpy(crypt->u, pdf_to_str_buf(obj), 48);
else if (pdf_is_string(obj) && pdf_to_str_len(obj) < 32)
{
@@ -138,7 +138,7 @@ pdf_new_crypt(fz_context *ctx, pdf_obj *dict, pdf_obj *id)
crypt->p = 0xfffffffc;
}
- if (crypt->r == 5)
+ if (crypt->r == 5 || crypt->r == 6)
{
obj = pdf_dict_gets(dict, "OE");
if (!pdf_is_string(obj) || pdf_to_str_len(obj) != 32)
@@ -315,10 +315,9 @@ pdf_parse_crypt_filter(fz_context *ctx, pdf_crypt_filter *cf, pdf_crypt *crypt,
fz_throw(ctx, "invalid key length: %d", cf->length);
if ((crypt->r == 1 || crypt->r == 2 || crypt->r == 4) &&
- (cf->length < 0 || cf->length > 256))
+ (cf->length < 0 || cf->length > 128))
fz_throw(ctx, "invalid key length: %d", cf->length);
- if (crypt->r == 5 &&
- (cf->length != 128 && cf->length != 192 && cf->length == 256))
+ if ((crypt->r == 5 || crypt->r == 6) && cf->length != 256)
fz_throw(ctx, "invalid key length: %d", cf->length);
}
@@ -445,6 +444,101 @@ pdf_compute_encryption_key_r5(pdf_crypt *crypt, unsigned char *password, int pwl
}
/*
+ * Compute an encryption key (PDF 1.7 ExtensionLevel 8 algorithm)
+ *
+ * Adobe has not yet released the details, so the algorithm reference is:
+ * http://esec-lab.sogeti.com/post/The-undocumented-password-validation-algorithm-of-Adobe-Reader-X
+ */
+
+static void
+pdf_compute_hardened_hash_r6(unsigned char *password, int pwlen, unsigned char salt[16], unsigned char *ownerkey, unsigned char hash[32])
+{
+ unsigned char data[(128 + 64 + 48) * 64];
+ unsigned char block[64];
+ int block_size = 32;
+ int data_len = 0;
+ int i, j, sum;
+
+ fz_sha256 sha256;
+ fz_sha384 sha384;
+ fz_sha512 sha512;
+ fz_aes aes;
+
+ /* Step 1: calculate initial data block */
+ fz_sha256_init(&sha256);
+ fz_sha256_update(&sha256, password, pwlen);
+ fz_sha256_update(&sha256, salt, 8);
+ if (ownerkey)
+ fz_sha256_update(&sha256, ownerkey, 48);
+ fz_sha256_final(&sha256, block);
+
+ for (i = 0; i < 64 || i < data[data_len * 64 - 1] + 32; i++)
+ {
+ /* Step 2: repeat password and data block 64 times */
+ memcpy(data, password, pwlen);
+ memcpy(data + pwlen, block, block_size);
+ memcpy(data + pwlen + block_size, ownerkey, ownerkey ? 48 : 0);
+ data_len = pwlen + block_size + (ownerkey ? 48 : 0);
+ for (j = 1; j < 64; j++)
+ memcpy(data + j * data_len, data, data_len);
+
+ /* Step 3: encrypt data using data block as key and iv */
+ aes_setkey_enc(&aes, block, 128);
+ aes_crypt_cbc(&aes, AES_ENCRYPT, data_len * 64, block + 16, data, data);
+
+ /* Step 4: determine SHA-2 hash size for this round */
+ for (j = 0, sum = 0; j < 16; j++)
+ sum += data[j];
+
+ /* Step 5: calculate data block for next round */
+ block_size = 32 + (sum % 3) * 16;
+ switch (block_size)
+ {
+ case 32:
+ fz_sha256_init(&sha256);
+ fz_sha256_update(&sha256, data, data_len * 64);
+ fz_sha256_final(&sha256, block);
+ break;
+ case 48:
+ fz_sha384_init(&sha384);
+ fz_sha384_update(&sha384, data, data_len * 64);
+ fz_sha384_final(&sha384, block);
+ break;
+ case 64:
+ fz_sha512_init(&sha512);
+ fz_sha512_update(&sha512, data, data_len * 64);
+ fz_sha512_final(&sha512, block);
+ break;
+ }
+ }
+
+ memset(data, 0, sizeof(data));
+ memcpy(hash, block, 32);
+}
+
+static void
+pdf_compute_encryption_key_r6(pdf_crypt *crypt, unsigned char *password, int pwlen, int ownerkey, unsigned char *validationkey)
+{
+ unsigned char hash[32];
+ unsigned char iv[16];
+ fz_aes aes;
+
+ if (pwlen > 127)
+ pwlen = 127;
+
+ pdf_compute_hardened_hash_r6(password, pwlen,
+ (ownerkey ? crypt->o : crypt->u) + 32,
+ ownerkey ? crypt->u : NULL, validationkey);
+ pdf_compute_hardened_hash_r6(password, pwlen,
+ crypt->u + 40, NULL, hash);
+
+ memset(iv, 0, sizeof(iv));
+ aes_setkey_dec(&aes, hash, 256);
+ aes_crypt_cbc(&aes, AES_DECRYPT, 32, iv,
+ ownerkey ? crypt->oe : crypt->ue, crypt->key);
+}
+
+/*
* Computing the user password (PDF 1.7 algorithm 3.4 and 3.5)
* Also save the generated key for decrypting objects and streams in crypt->key.
*/
@@ -496,6 +590,11 @@ pdf_compute_user_password(pdf_crypt *crypt, unsigned char *password, int pwlen,
{
pdf_compute_encryption_key_r5(crypt, password, pwlen, 0, output);
}
+
+ if (crypt->r == 6)
+ {
+ pdf_compute_encryption_key_r6(crypt, password, pwlen, 0, output);
+ }
}
/*
@@ -510,7 +609,7 @@ pdf_authenticate_user_password(pdf_crypt *crypt, unsigned char *password, int pw
{
unsigned char output[32];
pdf_compute_user_password(crypt, password, pwlen, output);
- if (crypt->r == 2 || crypt->r == 5)
+ if (crypt->r == 2 || crypt->r == 5 || crypt->r == 6)
return memcmp(output, crypt->u, 32) == 0;
if (crypt->r == 3 || crypt->r == 4)
return memcmp(output, crypt->u, 16) == 0;
@@ -538,9 +637,13 @@ pdf_authenticate_owner_password(pdf_crypt *crypt, unsigned char *ownerpass, int
if (crypt->r == 5)
{
/* PDF 1.7 ExtensionLevel 3 algorithm 3.12 */
-
pdf_compute_encryption_key_r5(crypt, ownerpass, pwlen, 1, key);
-
+ return !memcmp(key, crypt->o, 32);
+ }
+ else if (crypt->r == 6)
+ {
+ /* PDF 1.7 ExtensionLevel 8 algorithm */
+ pdf_compute_encryption_key_r6(crypt, ownerpass, pwlen, 1, key);
return !memcmp(key, crypt->o, 32);
}
diff --git a/pdf/pdf_form.c b/pdf/pdf_form.c
index cb92956f..1a902df3 100644
--- a/pdf/pdf_form.c
+++ b/pdf/pdf_form.c
@@ -1452,7 +1452,7 @@ static void reset_field(pdf_document *doc, pdf_obj *obj)
fz_var(name);
fz_try(ctx)
{
- name = fz_new_name(ctx, "Off");
+ name = pdf_new_name(ctx, "Off");
pdf_dict_puts(obj, "AS", name);
}
fz_always(ctx)
@@ -1612,7 +1612,7 @@ static void check_off(fz_context *ctx, pdf_obj *obj)
fz_var(off);
fz_try(ctx);
{
- off = fz_new_name(ctx, "Off");
+ off = pdf_new_name(ctx, "Off");
pdf_dict_puts(obj, "AS", off);
}
fz_always(ctx)
@@ -1636,9 +1636,9 @@ static void set_check(fz_context *ctx, pdf_obj *chk, char *name)
/* If name is a possible value of this check
* box then use it, otherwise use "Off" */
if (pdf_dict_gets(n, name))
- val = fz_new_name(ctx, name);
+ val = pdf_new_name(ctx, name);
else
- val = fz_new_name(ctx, "Off");
+ val = pdf_new_name(ctx, "Off");
pdf_dict_puts(chk, "AS", val);
}
@@ -1947,15 +1947,15 @@ void pdf_field_setBorderStyle(pdf_document *doc, pdf_obj *field, char *text)
pdf_obj *val = NULL;
if (!strcmp(text, "Solid"))
- val = fz_new_name(ctx, "S");
+ val = pdf_new_name(ctx, "S");
else if (!strcmp(text, "Dashed"))
- val = fz_new_name(ctx, "D");
+ val = pdf_new_name(ctx, "D");
else if (!strcmp(text, "Beveled"))
- val = fz_new_name(ctx, "B");
+ val = pdf_new_name(ctx, "B");
else if (!strcmp(text, "Inset"))
- val = fz_new_name(ctx, "I");
+ val = pdf_new_name(ctx, "I");
else if (!strcmp(text, "Underline"))
- val = fz_new_name(ctx, "U");
+ val = pdf_new_name(ctx, "U");
else
return;
diff --git a/pdf/pdf_interpret.c b/pdf/pdf_interpret.c
index 704f6e78..f596b0ed 100644
--- a/pdf/pdf_interpret.c
+++ b/pdf/pdf_interpret.c
@@ -1538,7 +1538,7 @@ pdf_run_extgstate(pdf_csi *csi, pdf_obj *rdb, pdf_obj *extgstate)
if (pdf_is_dict(val))
{
pdf_xobject *xobj;
- pdf_obj *group, *luminosity, *bc;
+ pdf_obj *group, *luminosity, *bc, *tr;
if (gstate->softmask)
{
@@ -1572,6 +1572,19 @@ pdf_run_extgstate(pdf_csi *csi, pdf_obj *rdb, pdf_obj *extgstate)
gstate->luminosity = 1;
else
gstate->luminosity = 0;
+
+ tr = pdf_dict_gets(val, "TR2");
+ if (tr)
+ {
+ if (strcmp(pdf_to_name(tr), "Identity") && strcmp(pdf_to_name(tr), "Default"))
+ fz_warn(ctx, "ignoring transfer function");
+ }
+ else
+ {
+ tr = pdf_dict_gets(val, "TR");
+ if (strcmp(pdf_to_name(tr), "Identity"))
+ fz_warn(ctx, "ignoring transfer function");
+ }
}
else if (pdf_is_name(val) && !strcmp(pdf_to_name(val), "None"))
{
diff --git a/pdf/pdf_nametree.c b/pdf/pdf_nametree.c
index 25fced52..ab3244d3 100644
--- a/pdf/pdf_nametree.c
+++ b/pdf/pdf_nametree.c
@@ -118,6 +118,8 @@ pdf_load_name_tree_imp(pdf_obj *dict, pdf_document *xref, pdf_obj *node)
pdf_obj *names = pdf_dict_gets(node, "Names");
int i;
+ UNUSED(ctx);
+
if (kids && !pdf_dict_mark(node))
{
int len = pdf_array_len(kids);
diff --git a/pdf/pdf_object.c b/pdf/pdf_object.c
index fb53d9b8..3a1e7655 100644
--- a/pdf/pdf_object.c
+++ b/pdf/pdf_object.c
@@ -117,7 +117,7 @@ pdf_new_string(fz_context *ctx, char *str, int len)
}
pdf_obj *
-fz_new_name(fz_context *ctx, char *str)
+pdf_new_name(fz_context *ctx, char *str)
{
pdf_obj *obj;
obj = Memento_label(fz_malloc(ctx, offsetof(pdf_obj, u.n) + strlen(str) + 1), "pdf_obj(name)");
@@ -588,12 +588,12 @@ pdf_obj *pdf_new_rect(fz_context *ctx, fz_rect *rect)
pdf_drop_obj(item);
item = NULL;
- item = pdf_new_real(ctx, rect->x1 - rect->x0);
+ item = pdf_new_real(ctx, rect->x1);
pdf_array_push(arr, item);
pdf_drop_obj(item);
item = NULL;
- item = pdf_new_real(ctx, rect->y1 - rect->y0);
+ item = pdf_new_real(ctx, rect->y1);
pdf_array_push(arr, item);
pdf_drop_obj(item);
item = NULL;
@@ -945,7 +945,7 @@ pdf_dict_put(pdf_obj *obj, pdf_obj *key, pdf_obj *val)
void
pdf_dict_puts(pdf_obj *obj, char *key, pdf_obj *val)
{
- pdf_obj *keyobj = fz_new_name(obj->ctx, key);
+ pdf_obj *keyobj = pdf_new_name(obj->ctx, key);
pdf_dict_put(obj, keyobj, val);
pdf_drop_obj(keyobj);
}
diff --git a/pdf/pdf_parse.c b/pdf/pdf_parse.c
index 4a7e421f..ed7889ed 100644
--- a/pdf/pdf_parse.c
+++ b/pdf/pdf_parse.c
@@ -196,7 +196,7 @@ pdf_obj *
pdf_to_utf8_name(pdf_document *xref, pdf_obj *src)
{
char *buf = pdf_to_utf8(xref, src);
- pdf_obj *dst = fz_new_name(xref->ctx, buf);
+ pdf_obj *dst = pdf_new_name(xref->ctx, buf);
fz_free(xref->ctx, buf);
return dst;
}
@@ -289,7 +289,7 @@ pdf_parse_array(pdf_document *xref, fz_stream *file, pdf_lexbuf *buf)
break;
case PDF_TOK_NAME:
- obj = fz_new_name(ctx, buf->scratch);
+ obj = pdf_new_name(ctx, buf->scratch);
pdf_array_push(ary, obj);
pdf_drop_obj(obj);
obj = NULL;
@@ -372,7 +372,7 @@ pdf_parse_dict(pdf_document *xref, fz_stream *file, pdf_lexbuf *buf)
if (tok != PDF_TOK_NAME)
fz_throw(ctx, "invalid key in dict");
- key = fz_new_name(ctx, buf->scratch);
+ key = pdf_new_name(ctx, buf->scratch);
tok = pdf_lex(file, buf);
@@ -386,7 +386,7 @@ pdf_parse_dict(pdf_document *xref, fz_stream *file, pdf_lexbuf *buf)
val = pdf_parse_dict(xref, file, buf);
break;
- case PDF_TOK_NAME: val = fz_new_name(ctx, buf->scratch); break;
+ case PDF_TOK_NAME: val = pdf_new_name(ctx, buf->scratch); break;
case PDF_TOK_REAL: val = pdf_new_real(ctx, buf->f); break;
case PDF_TOK_STRING: val = pdf_new_string(ctx, buf->scratch, buf->len); break;
case PDF_TOK_TRUE: val = pdf_new_bool(ctx, 1); break;
@@ -455,7 +455,7 @@ pdf_parse_stm_obj(pdf_document *xref, fz_stream *file, pdf_lexbuf *buf)
return pdf_parse_array(xref, file, buf);
case PDF_TOK_OPEN_DICT:
return pdf_parse_dict(xref, file, buf);
- case PDF_TOK_NAME: return fz_new_name(ctx, buf->scratch); break;
+ case PDF_TOK_NAME: return pdf_new_name(ctx, buf->scratch); break;
case PDF_TOK_REAL: return pdf_new_real(ctx, buf->f); break;
case PDF_TOK_STRING: return pdf_new_string(ctx, buf->scratch, buf->len); break;
case PDF_TOK_TRUE: return pdf_new_bool(ctx, 1); break;
@@ -506,7 +506,7 @@ pdf_parse_ind_obj(pdf_document *xref,
obj = pdf_parse_dict(xref, file, buf);
break;
- case PDF_TOK_NAME: obj = fz_new_name(ctx, buf->scratch); break;
+ case PDF_TOK_NAME: obj = pdf_new_name(ctx, buf->scratch); break;
case PDF_TOK_REAL: obj = pdf_new_real(ctx, buf->f); break;
case PDF_TOK_STRING: obj = pdf_new_string(ctx, buf->scratch, buf->len); break;
case PDF_TOK_TRUE: obj = pdf_new_bool(ctx, 1); break;
diff --git a/pdf/pdf_write.c b/pdf/pdf_write.c
index 5d03ffe3..24d328d6 100644
--- a/pdf/pdf_write.c
+++ b/pdf/pdf_write.c
@@ -1054,7 +1054,7 @@ add_linearization_objs(pdf_document *xref, pdf_write_options *opts)
/* FIXME: Do we have document information? Do an I entry */
/* FIXME: Do we have logical structure heirarchy? Do a C entry */
/* FIXME: Do L, Page Label hint table */
- o = fz_new_name(ctx, "FlateDecode");
+ o = pdf_new_name(ctx, "FlateDecode");
pdf_dict_puts(hint_obj, "Filter", o);
pdf_drop_obj(o);
o = NULL;
@@ -1426,7 +1426,7 @@ static void addhexfilter(pdf_document *xref, pdf_obj *dict)
pdf_obj *ahx, *nullobj;
fz_context *ctx = xref->ctx;
- ahx = fz_new_name(ctx, "ASCIIHexDecode");
+ ahx = pdf_new_name(ctx, "ASCIIHexDecode");
nullobj = pdf_new_null(ctx);
newf = newdp = NULL;
diff --git a/pdf/pdf_xobject.c b/pdf/pdf_xobject.c
index afa86527..19de0363 100644
--- a/pdf/pdf_xobject.c
+++ b/pdf/pdf_xobject.c
@@ -153,11 +153,11 @@ pdf_new_xobject(pdf_document *xref, fz_rect *bbox, fz_matrix *mat)
res = pdf_new_dict(ctx, 0);
procset = pdf_new_array(ctx, 2);
- obj = fz_new_name(ctx, "PDF");
+ obj = pdf_new_name(ctx, "PDF");
pdf_array_push(procset, obj);
pdf_drop_obj(obj);
obj = NULL;
- obj = fz_new_name(ctx, "Text");
+ obj = pdf_new_name(ctx, "Text");
pdf_array_push(procset, obj);
pdf_drop_obj(obj);
obj = NULL;
@@ -166,12 +166,12 @@ pdf_new_xobject(pdf_document *xref, fz_rect *bbox, fz_matrix *mat)
procset = NULL;
pdf_dict_puts(dict, "Resources", res);
- obj = fz_new_name(ctx, "Form");
+ obj = pdf_new_name(ctx, "Form");
pdf_dict_puts(dict, "Subtype", obj);
pdf_drop_obj(obj);
obj = NULL;
- obj = fz_new_name(ctx, "XObject");
+ obj = pdf_new_name(ctx, "XObject");
pdf_dict_puts(dict, "Type", obj);
pdf_drop_obj(obj);
obj = NULL;
generated by cgit v1.2.3 (git 2.25.1) at 2025-02-25 07:16:18 +0000