summaryrefslogtreecommitdiffstatshomepage
diff options
context:
space:
mode:
authorAlex Robbins <alexdotrobbins@gmail.com>2017-08-04 17:30:34 -0500
committerPaul Sokolovsky <pfalcon@users.sourceforge.net>2017-08-17 09:25:51 +0300
commitc89254fd0f7f4517163e095f144bfe154f47758a (patch)
treeac2028090bd68d37d03d755ec8e097cbc0fd3ccf
parent025e5f2b339377ebc54ebc9cab2612946145a6fa (diff)
downloadmicropython-c89254fd0f7f4517163e095f144bfe154f47758a.tar.gz
micropython-c89254fd0f7f4517163e095f144bfe154f47758a.zip
extmod/modubinascii: Rewrite mod_binascii_a2b_base64.
This implementation ignores invalid characters in the input. This allows it to decode the output of b2a_base64, and also mimics the behavior of CPython.
-rw-r--r--extmod/modubinascii.c86
-rw-r--r--tests/extmod/ubinascii_a2b_base64.py7
2 files changed, 55 insertions, 38 deletions
diff --git a/extmod/modubinascii.c b/extmod/modubinascii.c
index d3092a4df1..b02d833185 100644
--- a/extmod/modubinascii.c
+++ b/extmod/modubinascii.c
@@ -105,54 +105,64 @@ mp_obj_t mod_binascii_unhexlify(mp_obj_t data) {
}
MP_DEFINE_CONST_FUN_OBJ_1(mod_binascii_unhexlify_obj, mod_binascii_unhexlify);
+// If ch is a character in the base64 alphabet, and is not a pad character, then
+// the corresponding integer between 0 and 63, inclusively, is returned.
+// Otherwise, -1 is returned.
+static int mod_binascii_sextet(byte ch) {
+ if (ch >= 'A' && ch <= 'Z') {
+ return ch - 'A';
+ } else if (ch >= 'a' && ch <= 'z') {
+ return ch - 'a' + 26;
+ } else if (ch >= '0' && ch <= '9') {
+ return ch - '0' + 52;
+ } else if (ch == '+') {
+ return 62;
+ } else if (ch == '/') {
+ return 63;
+ } else {
+ return -1;
+ }
+}
+
mp_obj_t mod_binascii_a2b_base64(mp_obj_t data) {
mp_buffer_info_t bufinfo;
mp_get_buffer_raise(data, &bufinfo, MP_BUFFER_READ);
- if (bufinfo.len % 4 != 0) {
- mp_raise_ValueError("incorrect padding");
- }
+ byte *in = bufinfo.buf;
vstr_t vstr;
- byte *in = bufinfo.buf;
- if (bufinfo.len == 0) {
- vstr_init_len(&vstr, 0);
- }
- else {
- vstr_init_len(&vstr, ((bufinfo.len / 4) * 3) - ((in[bufinfo.len-1] == '=') ? ((in[bufinfo.len-2] == '=') ? 2 : 1 ) : 0));
- }
- byte *out = (byte*)vstr.buf;
- for (mp_uint_t i = bufinfo.len; i; i -= 4) {
- char hold[4];
- for (int j = 4; j--;) {
- if (in[j] >= 'A' && in[j] <= 'Z') {
- hold[j] = in[j] - 'A';
- } else if (in[j] >= 'a' && in[j] <= 'z') {
- hold[j] = in[j] - 'a' + 26;
- } else if (in[j] >= '0' && in[j] <= '9') {
- hold[j] = in[j] - '0' + 52;
- } else if (in[j] == '+') {
- hold[j] = 62;
- } else if (in[j] == '/') {
- hold[j] = 63;
- } else if (in[j] == '=') {
- if (j < 2 || i > 4) {
- mp_raise_ValueError("incorrect padding");
- }
- hold[j] = 64;
- } else {
- mp_raise_ValueError("invalid character");
+ vstr_init(&vstr, (bufinfo.len / 4) * 3 + 1); // Potentially over-allocate
+ byte *out = (byte *)vstr.buf;
+
+ uint shift = 0;
+ int nbits = 0; // Number of meaningful bits in shift
+ bool hadpad = false; // Had a pad character since last valid character
+ for (size_t i = 0; i < bufinfo.len; i++) {
+ if (in[i] == '=') {
+ if ((nbits == 2) || ((nbits == 4) && hadpad)) {
+ nbits = 0;
+ break;
}
+ hadpad = true;
}
- in += 4;
- *out++ = (hold[0]) << 2 | (hold[1]) >> 4;
- if (hold[2] != 64) {
- *out++ = (hold[1] & 0x0F) << 4 | hold[2] >> 2;
- if (hold[3] != 64) {
- *out++ = (hold[2] & 0x03) << 6 | hold[3];
- }
+ int sextet = mod_binascii_sextet(in[i]);
+ if (sextet == -1) {
+ continue;
+ }
+ hadpad = false;
+ shift = (shift << 6) | sextet;
+ nbits += 6;
+
+ if (nbits >= 8) {
+ nbits -= 8;
+ out[vstr.len++] = (shift >> nbits) & 0xFF;
}
}
+
+ if (nbits) {
+ mp_raise_ValueError("incorrect padding");
+ }
+
return mp_obj_new_str_from_vstr(&mp_type_bytes, &vstr);
}
MP_DEFINE_CONST_FUN_OBJ_1(mod_binascii_a2b_base64_obj, mod_binascii_a2b_base64);
diff --git a/tests/extmod/ubinascii_a2b_base64.py b/tests/extmod/ubinascii_a2b_base64.py
index b35f265910..05a3169f3a 100644
--- a/tests/extmod/ubinascii_a2b_base64.py
+++ b/tests/extmod/ubinascii_a2b_base64.py
@@ -21,6 +21,13 @@ print(binascii.a2b_base64(b'f4D/'))
print(binascii.a2b_base64(b'f4D+')) # convert '+'
print(binascii.a2b_base64(b'MTIzNEFCQ0RhYmNk'))
+# Ignore invalid characters and pad sequences
+print(binascii.a2b_base64(b'Zm9v\n'))
+print(binascii.a2b_base64(b'Zm\x009v\n'))
+print(binascii.a2b_base64(b'Zm9v=='))
+print(binascii.a2b_base64(b'Zm9v==='))
+print(binascii.a2b_base64(b'Zm9v===YmFy'))
+
try:
print(binascii.a2b_base64(b'abc'))
except ValueError: