GpgFrontend/vmime
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge pull request #210 from jengelh/twoaddr

Browse Source 
Handle parsing of further non-conformant From lines
This commit is contained in:
Vincent Richard 2019-02-04 21:10:47 +01:00 committed by GitHub
commit 414661858d
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 191 additions and 189 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

341
src/vmime/mailbox.cpp
View File

@ -80,189 +80,73 @@ void mailbox::parseImpl(
const char* const pstart = buffer.data() + position; const char* const pstart = buffer.data() + position;
const char* p = pstart; const char* p = pstart;
// Ignore blank spaces at the beginning
while (p < pend && parserHelpers::isSpace(*p)) ++p;
// Current state for parsing machine // Current state for parsing machine
enum States { enum States {
State_None, State_None,
State_String,
State_Name, State_Name,
State_Address State_Bracket,
State_Address_Bracketed,
State_Address_Unbracketed,
State_NameAt,
}; };
States state = State_Name; // let's start with name, we will see later (*) States state = State_None;
// Temporary buffers for extracted name and address // Temporary buffers for extracted name and address
string name; string name, address;
string address; unsigned int hold = 0;
bool hadBrackets = false; bool hadBrackets = false;
while (p < pend) { while (p < pend) {
if (state == State_None) {
if (state == State_Name) { while (p < pend && parserHelpers::isSpace(*p))
++p;
if (*p == ',' || *p == ';')
/* Completely new mailbox -- stop parsing this one. */
break;
if (*p == '<') { if (*p == '<') {
state = State_Bracket;
state = State_Address;
continue; continue;
} }
if (*p == '"') { // Quoted string if (*p == '"') { // Quoted string
state = State_String;
++p; ++p;
continue;
}
state = State_Name;
} else if (state == State_String) {
bool escaped = false; bool escaped = false;
while (p < pend) { for (; p < pend; ++p) {
if (escaped) { if (escaped) {
name += *p; name += *p;
escaped = false; escaped = false;
continue;
} else if (*p == '\\') { } else if (*p == '\\') {
escaped = true; escaped = true;
continue;
} else { }
if (*p == '"') { if (*p == '"') {
++p; ++p;
break;
} else {
name += *p;
}
}
++p;
}
} else {
bool escaped = false;
int comment = 0;
while (p < pend) {
if (escaped) {
if (!comment) name += *p;
escaped = false;
} else if (comment) {
if (*p == '\\') {
escaped = true;
} else if (*p == '(') {
++comment;
} else if (*p == ')') {
--comment;
}
} else if (*p == '\\') {
escaped = true;
} else if (*p == '(') {
++comment;
} else if (*p == '<') {
// Erase any space between display name and <address>
string::iterator q = name.end();
while (q != name.begin() && parserHelpers::isSpace(*(q - 1))) {
--q;
}
name.erase(q, name.end());
break;
} else if (/* parserHelpers::isSpace(*p) || */ *p == '@') {
break;
} else {
name += *p;
}
++p;
}
}
if (p < pend && *p == '@') {
// (*) Actually, we were parsing the local-part of an address
// and not a display name...
address = name;
name.clear();
bool escaped = false;
int comment = 0;
while (p < pend) {
if (escaped) {
if (!comment) address += *p;
escaped = false;
} else if (comment) {
if (*p == '\\') {
escaped = true;
} else if (*p == '(') {
++comment;
} else if (*p == ')') {
--comment;
}
} else if (*p == '\\') {
escaped = true;
} else if (*p == '(') {
++comment;
} else if (parserHelpers::isSpace(*p)) {
break;
} else {
address += *p;
}
++p;
}
} else {
while (p < pend && parserHelpers::isSpace(*p)) ++p;
state = State_None; state = State_None;
break;
} }
name += *p;
} else if (state == State_Address) { }
} else if (state == State_Address_Bracketed) {
bool escaped = false; bool escaped = false;
int comment = 0; int comment = 0;
while (p < pend) { for (; p < pend; ++p) {
if (escaped) { if (escaped) {
if (!comment) {
if (!comment) address += *p; address += *p;
name += *p;
++hold;
}
escaped = false; escaped = false;
} else if (comment) { } else if (comment) {
if (*p == '\\') { if (*p == '\\') {
escaped = true; escaped = true;
} else if (*p == '(') { } else if (*p == '(') {
@ -270,48 +154,151 @@ void mailbox::parseImpl(
} else if (*p == ')') { } else if (*p == ')') {
--comment; --comment;
} }
} else if (*p == '(') { } else if (*p == '(') {
++comment; ++comment;
} else if (*p == '\\') { } else if (*p == '\\') {
escaped = true; escaped = true;
} else if (*p == '<') { } else if (*p == '<') {
state = State_Bracket;
break;
} else if (*p == '>') {
hadBrackets = true;
name += *p++;
++hold;
state = State_Name;
break;
} else if (parserHelpers::isSpace(*p)) {
name += *p;
++hold;
} else {
address += *p;
name += *p;
++hold;
}
}
} else if (state == State_Address_Unbracketed) {
bool escaped = false;
int comment = 0;
for (; p < pend; ++p) {
if (escaped) {
if (!comment) {
address += *p;
name += *p;
++hold;
}
escaped = false;
} else if (comment) {
if (*p == '\\')
escaped = true;
else if (*p == '(')
++comment;
else if (*p == ')')
--comment;
} else if (*p == '(') {
++comment;
} else if (*p == '\\') {
escaped = true;
} else if (*p == '<') {
state = State_Bracket;
break;
} else if (*p == '>') {
name += *p++;
++hold;
state = State_None;
break;
} else if (*p == ',' || *p == ';') {
state = State_None;
break;
} else if (parserHelpers::isSpace(*p)) {
state = State_Name;
break;
} else {
address += *p;
name += *p;
++hold;
}
}
} else if (state == State_Name) {
bool escaped = false;
unsigned int comment = 0, at_hold = 0;
for (; p < pend; ++p) {
if (escaped) {
if (!comment) {
name += *p;
++at_hold;
}
escaped = false;
} else if (comment) {
if (*p == '\\')
escaped = true;
else if (*p == '(')
++comment;
else if (*p == ')')
--comment;
} else if (*p == '\\') {
escaped = true;
} else if (*p == '(') {
++comment;
} else if (*p == '<') {
state = State_Bracket;
break;
} else if (*p == '@') {
hold = at_hold;
state = State_NameAt;
break;
} else if (parserHelpers::isSpace(*p)) {
name += *p;
++hold;
at_hold = 1;
} else {
name += *p;
hold = 0;
++at_hold;
}
}
} else if (state == State_Bracket) {
string::const_iterator q = name.cend();
unsigned int nh = 0;
while (nh < hold && q != name.cbegin() && parserHelpers::isSpace(*(q - 1))) {
--q;
++nh;
}
hold = nh;
name += *p++;
++hold;
if (!address.empty())
// If we found a '<' here, it means that the address // If we found a '<' here, it means that the address
// starts _only_ here...and the stuff we have parsed // starts _only_ here...and the stuff we have parsed
// before belongs actually to the display name! // before belongs actually to the display name!
name += address;
address.clear(); address.clear();
state = State_Address_Bracketed;
} else if (*p == '>') { } else if (state == State_NameAt) {
name += *p++;
hadBrackets = true; ++hold;
break; // (*) Actually, we were parsing the local-part of an address
// and not a display name...
} else if (!parserHelpers::isSpace(*p)) { // Back up to the last whitespace and treat as address
string::const_iterator q = name.cend();
address += *p; unsigned int nh = 0;
while (nh < hold && q != name.cbegin() && !parserHelpers::isSpace(*(q - 1))) {
--q;
++nh;
} }
address.assign(q, name.cend());
++p; while (nh < hold && q != name.cbegin() && parserHelpers::isSpace(*(q - 1))) {
} --q;
++nh;
break;
} else {
while (p < pend && parserHelpers::isSpace(*p)) ++p;
if (p < pend) {
state = State_Address;
} }
hold = nh;
state = State_Address_Unbracketed;
} }
} }
if (hold <= name.size())
name.erase(name.size() - hold);
// Swap name and address when no address was found // Swap name and address when no address was found
// (email address is mandatory, whereas name is optional). // (email address is mandatory, whereas name is optional).
if (address.empty() && !name.empty() && !hadBrackets) { if (address.empty() && !name.empty() && !hadBrackets) {

23
tests/parser/mailboxTest.cpp
View File

@ -30,7 +30,7 @@ VMIME_TEST_SUITE_BEGIN(mailboxTest)
VMIME_TEST(testParse) VMIME_TEST(testParse)
VMIME_TEST(testEmptyEmailAddress) VMIME_TEST(testEmptyEmailAddress)
VMIME_TEST(testSeparatorInComment) VMIME_TEST(testSeparatorInComment)
VMIME_TEST(testAddressInName) VMIME_TEST(testMalformations)
VMIME_TEST_LIST_END VMIME_TEST_LIST_END
@ -146,13 +146,28 @@ VMIME_TEST_SUITE_BEGIN(mailboxTest)
VASSERT_EQ("email2", "bbb@vmime.org", mbox2->getEmail()); VASSERT_EQ("email2", "bbb@vmime.org", mbox2->getEmail());
} }
void testAddressInName() { void testMalformations() {
vmime::mailbox mbox; vmime::mailbox mbox;
mbox.parse("a@b.c <e@f.g>");
mbox.parse("a@b.c <e@f.g>");
VASSERT_EQ("name", vmime::text("a@b.c"), mbox.getName()); VASSERT_EQ("name", vmime::text("a@b.c"), mbox.getName());
VASSERT_EQ("email", "e@f.g", mbox.getEmail()); VASSERT_EQ("email", "e@f.g", mbox.getEmail());
mbox.parse("a@b.c e@f.g <h@i.j>");
VASSERT_EQ("name", vmime::text("a@b.c e@f.g"), mbox.getName());
VASSERT_EQ("email", "h@i.j", mbox.getEmail());
mbox.parse("Foo <bar<baz@quux.com>");
VASSERT_EQ("name", vmime::text("Foo <bar"), mbox.getName());
VASSERT_EQ("email", "baz@quux.com", mbox.getEmail());
mbox.parse("Foo <foo@x.com> <bar@x.com>");
VASSERT_EQ("name", vmime::text("Foo <foo@x.com>"), mbox.getName());
VASSERT_EQ("email", "bar@x.com", mbox.getEmail());
mbox.parse("Foo <foo@x.com> Bar <bar@y.com>");
VASSERT_EQ("name", vmime::text("Foo <foo@x.com> Bar"), mbox.getName());
VASSERT_EQ("email", "bar@y.com", mbox.getEmail());
} }
VMIME_TEST_SUITE_END VMIME_TEST_SUITE_END