Split quoted-printable parsing to q-p body parsing and "Q" header parsing.

Fixed also several other issues in quoted-printable parsing to get it fully
RFC compliant.

--HG--
branch : HEAD

diff --git a/src/lib-mail/Makefile.am b/src/lib-mail/Makefile.am
index 4faf083b91cc36990c44e6a8fda1058b0e7f3d93..486c62ff84231a94016486fbf9ede3f7fd346499 100644 (file)
--- a/src/lib-mail/Makefile.am
+++ b/src/lib-mail/Makefile.am
@@ -60,6 +60,7 @@ test_programs = \
        test-message-header-parser \
        test-message-id \
        test-message-parser \
+       test-quoted-printable \
        test-rfc2231-parser
 
 noinst_PROGRAMS = $(test_programs)
@@ -100,6 +101,10 @@ test_message_parser_SOURCES = test-message-parser.c
 test_message_parser_LDADD = message-parser.lo message-header-parser.lo message-size.lo rfc822-parser.lo rfc2231-parser.lo $(test_libs)
 test_message_parser_DEPENDENCIES = message-parser.lo message-header-parser.lo message-size.lo rfc822-parser.lo rfc2231-parser.lo $(test_libs)
 
+test_quoted_printable_SOURCES = test-quoted-printable.c
+test_quoted_printable_LDADD = quoted-printable.lo $(test_libs)
+test_quoted_printable_DEPENDENCIES = quoted-printable.lo $(test_libs)
+
 test_rfc2231_parser_SOURCES = test-rfc2231-parser.c
 test_rfc2231_parser_LDADD = rfc2231-parser.lo rfc822-parser.lo $(test_libs)
 test_rfc2231_parser_DEPENDENCIES = rfc2231-parser.lo rfc822-parser.lo $(test_libs)

diff --git a/src/lib-mail/message-header-decode.c b/src/lib-mail/message-header-decode.c
index f263deb2b8d92237977d4e431058def95f33656a..c0d0d2fd34c486463f2efd49a892a25d342e98f3 100644 (file)
--- a/src/lib-mail/message-header-decode.c
+++ b/src/lib-mail/message-header-decode.c
@@ -36,9 +36,9 @@ message_header_decode_encoded(const unsigned char *data, size_t size,
        switch (data[start_pos[0]+1]) {
        case 'q':
        case 'Q':
-               quoted_printable_decode(data + start_pos[1] + 1,
-                                       start_pos[2] - start_pos[1] - 1,
-                                       NULL, decodebuf);
+               quoted_printable_q_decode(data + start_pos[1] + 1,
+                                         start_pos[2] - start_pos[1] - 1,
+                                         decodebuf);
                break;
        case 'b':
        case 'B':

diff --git a/src/lib-mail/quoted-printable.c b/src/lib-mail/quoted-printable.c
index 61299f45769cae5839d2f1a2d1821d5709544ac0..a9b120226454dd800cbc81c2ad35fc2b6c72d909 100644 (file)
--- a/src/lib-mail/quoted-printable.c
+++ b/src/lib-mail/quoted-printable.c
@@ -5,33 +5,46 @@
 #include "hex-binary.h"
 #include "quoted-printable.h"
 
+#define QP_IS_TRAILING_SPACE(c) \
+       ((c) == ' ' || (c) == '\t')
+
 void quoted_printable_decode(const unsigned char *src, size_t src_size,
                             size_t *src_pos_r, buffer_t *dest)
 {
        char hexbuf[3];
-       size_t src_pos, next;
+       size_t src_pos, pos, next;
 
        hexbuf[2] = '\0';
 
        next = 0;
        for (src_pos = 0; src_pos < src_size; src_pos++) {
-               if (src[src_pos] != '_' && src[src_pos] != '=')
+               if (src[src_pos] != '=' && src[src_pos] != '\n')
                        continue;
 
-               buffer_append(dest, src + next, src_pos - next);
-               next = src_pos+1;
-
-               if (src[src_pos] == '_') {
-                       buffer_append_c(dest, ' ');
+               if (src[src_pos] == '\n') {
+                       /* drop trailing whitespace */
+                       pos = src_pos;
+                       if (pos > 0 && src[pos-1] == '\r')
+                               pos--;
+                       while (pos > 0 && QP_IS_TRAILING_SPACE(src[pos-1]))
+                               pos--;
+                       buffer_append(dest, src + next, pos - next);
+                       next = src_pos+1;
+                       buffer_append(dest, "\r\n", 2);
                        continue;
                }
 
+               /* '=' */
+               buffer_append(dest, src + next, src_pos - next);
+               next = src_pos;
+
                if (src_pos+1 >= src_size)
                        break;
 
                if (src[src_pos+1] == '\n') {
                        /* =\n -> skip both */
-                       src_pos++;
+                       src_pos += 2;
+                       next += 2;
                        continue;
                }
 
@@ -40,11 +53,61 @@ void quoted_printable_decode(const unsigned char *src, size_t src_size,
 
                if (src[src_pos+1] == '\r' && src[src_pos+2] == '\n') {
                        /* =\r\n -> skip both */
+                       src_pos += 3;
+                       next += 3;
+                       continue;
+               }
+
+               /* =<hex> */
+               hexbuf[0] = src[src_pos+1];
+               hexbuf[1] = src[src_pos+2];
+
+               if (hex_to_binary(hexbuf, dest) == 0) {
                        src_pos += 2;
+                       next = src_pos + 1;
+               } else {
+                       /* non-hex data, show as-is */
+                       next = src_pos;
+               }
+       }
+       if (src_pos == src_size) {
+               /* add everything but trailing spaces */
+               if (src_pos > 0 && src[src_pos-1] == '\r')
+                       src_pos--;
+               while (src_pos > 0 && QP_IS_TRAILING_SPACE(src[src_pos-1]))
+                       src_pos--;
+               buffer_append(dest, src + next, src_pos - next);
+               next = src_pos;
+       }
+
+       *src_pos_r = next;
+}
+
+void quoted_printable_q_decode(const unsigned char *src, size_t src_size,
+                              buffer_t *dest)
+{
+       char hexbuf[3];
+       size_t src_pos, next;
+
+       hexbuf[2] = '\0';
+
+       next = 0;
+       for (src_pos = 0; src_pos < src_size; src_pos++) {
+               if (src[src_pos] != '_' && src[src_pos] != '=')
+                       continue;
+
+               buffer_append(dest, src + next, src_pos - next);
+               next = src_pos;
+
+               if (src[src_pos] == '_') {
+                       buffer_append_c(dest, ' ');
                        next++;
                        continue;
                }
 
+               if (src_pos+2 >= src_size)
+                       break;
+
                /* =<hex> */
                hexbuf[0] = src[src_pos+1];
                hexbuf[1] = src[src_pos+2];
@@ -57,9 +120,5 @@ void quoted_printable_decode(const unsigned char *src, size_t src_size,
                        next = src_pos;
                }
        }
-
        buffer_append(dest, src + next, src_size - next);
-
-       if (src_pos_r != NULL)
-               *src_pos_r = src_pos;
 }

diff --git a/src/lib-mail/quoted-printable.h b/src/lib-mail/quoted-printable.h
index 7020bbc9fea40e00ace1f09eade4a35bea4910d4..38b50dc19af7897ca17cd7d2f7790ef67ea7e41b 100644 (file)
--- a/src/lib-mail/quoted-printable.h
+++ b/src/lib-mail/quoted-printable.h
@@ -5,9 +5,11 @@
    size of src, and may be same as src. Decoding errors are ignored.
 
    This function may be called multiple times for parsing the same stream.
-   If src_pos is non-NULL, it's updated to first non-translated character in
-   src. */
+   src_pos is updated to first non-translated character in src. */
 void quoted_printable_decode(const unsigned char *src, size_t src_size,
                             size_t *src_pos_r, buffer_t *dest);
+/* Decode MIME "Q" encoding. */
+void quoted_printable_q_decode(const unsigned char *src, size_t src_size,
+                              buffer_t *dest);
 
 #endif

diff --git a/src/lib-mail/test-quoted-printable.c b/src/lib-mail/test-quoted-printable.c
new file mode 100644 (file)
index 0000000..1e23d98
--- /dev/null
+++ b/src/lib-mail/test-quoted-printable.c
@@ -0,0 +1,79 @@
+/* Copyright (c) 2007-2009 Dovecot authors, see the included COPYING file */
+
+#include "lib.h"
+#include "buffer.h"
+#include "str.h"
+#include "quoted-printable.h"
+#include "test-common.h"
+
+static void test_quoted_printable_decode(void)
+{
+       const char *data[] = {
+               "foo  \r\nbar=", "foo\r\nbar",
+               "foo =\nbar", "foo bar",
+               "foo =\r\nbar", "foo bar",
+               "foo  \nbar=", "foo\r\nbar",
+               "=0A=0D  ", "\n\r",
+               "foo_bar", "foo_bar",
+               "foo=", "foo",
+               "foo=A", "foo",
+               "foo=Ax", "foo=Ax",
+               "foo=Ax=xy", "foo=Ax=xy"
+       };
+       buffer_t *buf;
+       unsigned int i, start, end, len;
+       size_t src_pos;
+
+       test_begin("quoted printable decode");
+       buf = buffer_create_dynamic(pool_datastack_create(), 128);
+       for (i = 0; i < N_ELEMENTS(data); i += 2) {
+               len = strlen(data[i]);
+               for (start = 0, end = 1; end <= len; ) {
+                       quoted_printable_decode(CONST_PTR_OFFSET(data[i], start),
+                                               end - start, &src_pos, buf);
+                       src_pos += start;
+                       start = src_pos;
+                       if (src_pos <= end)
+                               end++;
+                       else
+                               end = src_pos + 1;
+               }
+               test_assert(strcmp(data[i+1], str_c(buf)) == 0);
+               buffer_set_used_size(buf, 0);
+       }
+       test_end();
+}
+
+static void test_quoted_printable_q_decode(void)
+{
+       const char *data[] = {
+               "=0A=0D  ", "\n\r  ",
+               "__foo__bar__", "  foo  bar  ",
+               "foo=", "foo=",
+               "foo=A", "foo=A",
+               "foo=Ax", "foo=Ax",
+               "foo=Ax=xy", "foo=Ax=xy"
+       };
+       buffer_t *buf;
+       unsigned int i;
+
+       test_begin("quoted printable q decode");
+       buf = buffer_create_dynamic(pool_datastack_create(), 128);
+       for (i = 0; i < N_ELEMENTS(data); i += 2) {
+               quoted_printable_q_decode((const void *)data[i], strlen(data[i]),
+                                         buf);
+               test_assert(strcmp(data[i+1], str_c(buf)) == 0);
+               buffer_set_used_size(buf, 0);
+       }
+       test_end();
+}
+
+int main(void)
+{
+       static void (*test_functions[])(void) = {
+               test_quoted_printable_decode,
+               test_quoted_printable_q_decode,
+               NULL
+       };
+       return test_run(test_functions);
+}