mirror of
https://github.com/jezhiggins/arabica
synced 2024-11-15 19:48:00 +01:00
reworked absolutising, it's better, but I need to go back to the RFC at some point
This commit is contained in:
jez
parent
3d661b355c
commit
cec3c0c659
3 changed files with 77 additions and 18 deletions
|
|
@ -20,7 +20,8 @@ namespace Arabica
|
|||
scheme_(rhs.scheme_),
|
||||
host_(rhs.host_),
|
||||
path_(rhs.path_),
|
||||
port_(rhs.port_)
|
||||
port_(rhs.port_),
|
||||
is_absolute_(rhs.is_absolute_)
|
||||
{
|
||||
} // URI
|
||||
|
||||
|
|
@ -36,7 +37,8 @@ namespace Arabica
|
|||
return scheme_ == rhs.scheme_ &&
|
||||
host_ == rhs.host_ &&
|
||||
path_ == rhs.path_ &&
|
||||
port_ == rhs.port_;
|
||||
port_ == rhs.port_ &&
|
||||
is_absolute_ == rhs.is_absolute_;
|
||||
} // operator==
|
||||
|
||||
bool operator!=(const URI& rhs) const
|
||||
|
|
@ -52,17 +54,20 @@ namespace Arabica
|
|||
std::swap(host_, rhs.host_);
|
||||
std::swap(path_, rhs.path_);
|
||||
std::swap(port_, rhs.port_);
|
||||
std::swap(is_absolute_, rhs.is_absolute_);
|
||||
} // swap
|
||||
|
||||
const std::string& scheme() const { return scheme_; }
|
||||
const std::string& host() const { return host_; }
|
||||
const std::string& port() const;
|
||||
const std::string& path() const { return path_; }
|
||||
const bool& is_absolute() const { return is_absolute_; }
|
||||
|
||||
std::string as_string() const;
|
||||
|
||||
private:
|
||||
void parse(const std::string& URI);
|
||||
void parse_uri(const std::string& URI);
|
||||
void parseAuthority(std::string::const_iterator& u, std::string::const_iterator& ue);
|
||||
void absolutise(URI& relURI);
|
||||
void combinePath(const std::string& path);
|
||||
|
|
@ -71,6 +76,7 @@ namespace Arabica
|
|||
std::string host_;
|
||||
std::string path_;
|
||||
std::string port_;
|
||||
bool is_absolute_;
|
||||
}; // class URI
|
||||
} // namespace io
|
||||
} // namespace Arabica
|
||||
|
|
|
|||
|
|
@ -8,22 +8,30 @@ namespace {
|
|||
const std::string ZERO = "0";
|
||||
const std::string PORT_EIGHTY = "80";
|
||||
const std::string PORT_443 = "443";
|
||||
|
||||
const std::string SCHEME_HTTP = "http";
|
||||
const std::string SCHEME_HTTPS = "https";
|
||||
const std::string SCHEME_FILE = "file";
|
||||
|
||||
const std::string COLON = ":";
|
||||
const char FORWARD_SLASH = '/';
|
||||
|
||||
const std::string& wellKnownPort(const std::string& scheme)
|
||||
{
|
||||
if(scheme.empty())
|
||||
return ZERO;
|
||||
|
||||
if(scheme == "http")
|
||||
if(scheme == SCHEME_HTTP)
|
||||
return PORT_EIGHTY;
|
||||
if(scheme == "https")
|
||||
if(scheme == SCHEME_HTTPS)
|
||||
return PORT_443;
|
||||
|
||||
return ZERO;
|
||||
} // wellKnownPort
|
||||
} // namespace
|
||||
|
||||
URI::URI(const std::string& uri)
|
||||
URI::URI(const std::string& uri) :
|
||||
is_absolute_(false)
|
||||
{
|
||||
parse(uri);
|
||||
} // URI
|
||||
|
|
@ -32,7 +40,8 @@ URI::URI(const URI& base, const std::string& relativeUrl) :
|
|||
scheme_(base.scheme_),
|
||||
host_(base.host_),
|
||||
path_(base.path_),
|
||||
port_(base.port_)
|
||||
port_(base.port_),
|
||||
is_absolute_(base.is_absolute_)
|
||||
{
|
||||
if(!relativeUrl.empty())
|
||||
{
|
||||
|
|
@ -52,12 +61,14 @@ std::string URI::as_string() const
|
|||
{
|
||||
std::string str;
|
||||
if(!scheme_.empty())
|
||||
str.append(scheme_).append("://");
|
||||
str.append(scheme_).append(COLON);
|
||||
if(is_absolute_)
|
||||
str.append("//");
|
||||
if(!host_.empty())
|
||||
{
|
||||
str.append(host_);
|
||||
if(!port_.empty())
|
||||
str.append(":").append(port_);
|
||||
str.append(COLON).append(port_);
|
||||
}
|
||||
str.append(path_);
|
||||
return str;
|
||||
|
|
@ -66,15 +77,24 @@ std::string URI::as_string() const
|
|||
void fixSlashes(std::string& path)
|
||||
{
|
||||
for(int i = path.find('\\'); i != std::string::npos; i = path.find('\\', i))
|
||||
path[i] = '/';
|
||||
path[i] = FORWARD_SLASH;
|
||||
} // fixSlashes
|
||||
|
||||
void URI::parse(const std::string& uri)
|
||||
{
|
||||
parse_uri(uri);
|
||||
|
||||
is_absolute_ = (!scheme_.empty() && !host_.empty()) ||
|
||||
(((scheme_ == SCHEME_FILE) && (!path_.empty())) &&
|
||||
((path_[0] == FORWARD_SLASH) || (path_[1] == ':')));
|
||||
} // parse
|
||||
|
||||
void URI::parse_uri(const std::string& uri)
|
||||
{
|
||||
// I'd like to use something a bit stronger - http://code.google.com/p/uri-grammar/
|
||||
// but that would put a Boost Spirit dependence right in the core, which I'm not prepared to do at the moment
|
||||
|
||||
int d = uri.find_first_of(":");
|
||||
int d = uri.find_first_of(COLON);
|
||||
if(d == std::string::npos)
|
||||
{
|
||||
path_ = uri;
|
||||
|
|
@ -87,7 +107,7 @@ void URI::parse(const std::string& uri)
|
|||
// looks like a windows file path
|
||||
path_ = uri;
|
||||
fixSlashes(path_);
|
||||
scheme_ = "file";
|
||||
scheme_ = SCHEME_FILE;
|
||||
return;
|
||||
} // if ...
|
||||
|
||||
|
|
@ -97,7 +117,7 @@ void URI::parse(const std::string& uri)
|
|||
std::string::const_iterator ue = uri.end();
|
||||
|
||||
++u;
|
||||
if(*u == '/' && *(u+1) == '/')
|
||||
if(*u == FORWARD_SLASH && *(u+1) == FORWARD_SLASH)
|
||||
{
|
||||
u += 2;
|
||||
parseAuthority(u, ue);
|
||||
|
|
@ -108,7 +128,7 @@ void URI::parse(const std::string& uri)
|
|||
|
||||
void URI::parseAuthority(std::string::const_iterator& u, std::string::const_iterator& ue)
|
||||
{
|
||||
std::string::const_iterator slash = std::find(u, ue, '/');
|
||||
std::string::const_iterator slash = std::find(u, ue, FORWARD_SLASH);
|
||||
if(slash == ue)
|
||||
return;
|
||||
|
||||
|
|
@ -123,13 +143,14 @@ void URI::parseAuthority(std::string::const_iterator& u, std::string::const_iter
|
|||
|
||||
void URI::absolutise(URI& relative)
|
||||
{
|
||||
if(!relative.scheme().empty())
|
||||
if((relative.is_absolute()) ||
|
||||
((!relative.scheme().empty()) && (relative.scheme() != scheme_)))
|
||||
{
|
||||
swap(relative);
|
||||
return;
|
||||
}
|
||||
|
||||
if(relative.path_[0] == '/')
|
||||
if(relative.path_[0] == FORWARD_SLASH)
|
||||
path_ = relative.path_;
|
||||
else
|
||||
combinePath(relative.path_);
|
||||
|
|
@ -137,8 +158,8 @@ void URI::absolutise(URI& relative)
|
|||
|
||||
void URI::combinePath(const std::string& relPath)
|
||||
{
|
||||
if(*(path_.rbegin()) != '/')
|
||||
path_.erase(path_.rfind('/')+1);
|
||||
if(*(path_.rbegin()) != FORWARD_SLASH)
|
||||
path_.erase(path_.rfind(FORWARD_SLASH)+1);
|
||||
|
||||
std::string::size_type from = path_.length() - 1;
|
||||
path_.append(relPath);
|
||||
|
|
@ -146,7 +167,7 @@ void URI::combinePath(const std::string& relPath)
|
|||
int dots = path_.find("/../", from);
|
||||
while(dots != std::string::npos)
|
||||
{
|
||||
int preceding_slash = (dots > 0) ? path_.rfind('/', dots-1) : 0;
|
||||
int preceding_slash = (dots > 0) ? path_.rfind(FORWARD_SLASH, dots-1) : 0;
|
||||
path_.erase(preceding_slash, dots+3-preceding_slash);
|
||||
dots = path_.find("/../", preceding_slash);
|
||||
} // while
|
||||
|
|
|
|||
|
|
@ -21,6 +21,7 @@ class URITest : public TestCase
|
|||
assertEquals("", u.host());
|
||||
assertEquals("", u.scheme());
|
||||
assertEquals("0", u.port());
|
||||
assertEquals(false, u.is_absolute());
|
||||
assertEquals("woo", u.as_string());
|
||||
} // test1
|
||||
|
||||
|
|
@ -32,6 +33,7 @@ class URITest : public TestCase
|
|||
assertEquals("", u.host());
|
||||
assertEquals("", u.scheme());
|
||||
assertEquals("0", u.port());
|
||||
assertEquals(false, u.is_absolute());
|
||||
assertEquals("woo.xml", u.as_string());
|
||||
} // test2
|
||||
|
||||
|
|
@ -43,6 +45,7 @@ class URITest : public TestCase
|
|||
assertEquals("", u.host());
|
||||
assertEquals("", u.scheme());
|
||||
assertEquals("0", u.port());
|
||||
assertEquals(false, u.is_absolute());
|
||||
assertEquals("woo/woo.xml", u.as_string());
|
||||
} // test3
|
||||
|
||||
|
|
@ -54,6 +57,7 @@ class URITest : public TestCase
|
|||
assertEquals("", u.host());
|
||||
assertEquals("", u.scheme());
|
||||
assertEquals("0", u.port());
|
||||
assertEquals(false, u.is_absolute());
|
||||
assertEquals("/woo/woo.xml", u.as_string());
|
||||
} // test4
|
||||
|
||||
|
|
@ -65,6 +69,7 @@ class URITest : public TestCase
|
|||
assertEquals("localhost", u.host());
|
||||
assertEquals("http", u.scheme());
|
||||
assertEquals("80", u.port());
|
||||
assertEquals(true, u.is_absolute());
|
||||
assertEquals("http://localhost/woo/woo.xml", u.as_string());
|
||||
}
|
||||
|
||||
|
|
@ -76,6 +81,7 @@ class URITest : public TestCase
|
|||
assertEquals("localhost", u.host());
|
||||
assertEquals("http", u.scheme());
|
||||
assertEquals("8080", u.port());
|
||||
assertEquals(true, u.is_absolute());
|
||||
assertEquals("http://localhost:8080/woo/woo.xml", u.as_string());
|
||||
}
|
||||
|
||||
|
|
@ -87,6 +93,7 @@ class URITest : public TestCase
|
|||
assertEquals("www.jezuk.co.uk", u.host());
|
||||
assertEquals("http", u.scheme());
|
||||
assertEquals("80", u.port());
|
||||
assertEquals(true, u.is_absolute());
|
||||
assertEquals("http://www.jezuk.co.uk/arabica/news", u.as_string());
|
||||
}
|
||||
|
||||
|
|
@ -98,6 +105,7 @@ class URITest : public TestCase
|
|||
assertEquals("www.jezuk.co.uk", u.host());
|
||||
assertEquals("http", u.scheme());
|
||||
assertEquals("8000", u.port());
|
||||
assertEquals(true, u.is_absolute());
|
||||
assertEquals("http://www.jezuk.co.uk:8000/arabica/news", u.as_string());
|
||||
}
|
||||
|
||||
|
|
@ -110,6 +118,7 @@ class URITest : public TestCase
|
|||
assertEquals("localhost", r.host());
|
||||
assertEquals("http", r.scheme());
|
||||
assertEquals("80", r.port());
|
||||
assertEquals(true, u.is_absolute());
|
||||
assertEquals("http://localhost/nobby", r.as_string());
|
||||
} // test9
|
||||
|
||||
|
|
@ -122,6 +131,7 @@ class URITest : public TestCase
|
|||
assertEquals("localhost", r.host());
|
||||
assertEquals("http", r.scheme());
|
||||
assertEquals("80", r.port());
|
||||
assertEquals(true, r.is_absolute());
|
||||
assertEquals("http://localhost/", r.as_string());
|
||||
} // test10
|
||||
|
||||
|
|
@ -134,6 +144,7 @@ class URITest : public TestCase
|
|||
assertEquals("www.jezuk.co.uk", r.host());
|
||||
assertEquals("http", r.scheme());
|
||||
assertEquals("8000", r.port());
|
||||
assertEquals(true, r.is_absolute());
|
||||
assertEquals("http://www.jezuk.co.uk:8000/trouser/press", r.as_string());
|
||||
} // test11
|
||||
|
||||
|
|
@ -146,6 +157,7 @@ class URITest : public TestCase
|
|||
assertEquals("www.jezuk.co.uk", r.host());
|
||||
assertEquals("http", r.scheme());
|
||||
assertEquals("8000", r.port());
|
||||
assertEquals(true, r.is_absolute());
|
||||
assertEquals("http://www.jezuk.co.uk:8000/arabica/trouser/press", r.as_string());
|
||||
} // test12
|
||||
|
||||
|
|
@ -158,6 +170,7 @@ class URITest : public TestCase
|
|||
assertEquals("www.jezuk.co.uk", r.host());
|
||||
assertEquals("http", r.scheme());
|
||||
assertEquals("8000", r.port());
|
||||
assertEquals(true, r.is_absolute());
|
||||
assertEquals("http://www.jezuk.co.uk:8000/arabica/trouser/press", r.as_string());
|
||||
} // test13
|
||||
|
||||
|
|
@ -170,6 +183,7 @@ class URITest : public TestCase
|
|||
assertEquals("www.jezuk.co.uk", r.host());
|
||||
assertEquals("http", r.scheme());
|
||||
assertEquals("8000", r.port());
|
||||
assertEquals(true, r.is_absolute());
|
||||
assertEquals("http://www.jezuk.co.uk:8000/trouser/press", r.as_string());
|
||||
} // test14
|
||||
|
||||
|
|
@ -182,6 +196,7 @@ class URITest : public TestCase
|
|||
assertEquals("www.jezuk.co.uk", r.host());
|
||||
assertEquals("http", r.scheme());
|
||||
assertEquals("8000", r.port());
|
||||
assertEquals(true, r.is_absolute());
|
||||
assertEquals("http://www.jezuk.co.uk:8000/trouser/press", r.as_string());
|
||||
} // test15
|
||||
|
||||
|
|
@ -194,6 +209,7 @@ class URITest : public TestCase
|
|||
assertEquals("www.jezuk.co.uk", r.host());
|
||||
assertEquals("http", r.scheme());
|
||||
assertEquals("8000", r.port());
|
||||
assertEquals(true, r.is_absolute());
|
||||
assertEquals("http://www.jezuk.co.uk:8000/arabica/news/trouser/press", r.as_string());
|
||||
} // test16
|
||||
|
||||
|
|
@ -206,6 +222,7 @@ class URITest : public TestCase
|
|||
assertEquals("www.jezuk.co.uk", r.host());
|
||||
assertEquals("http", r.scheme());
|
||||
assertEquals("8000", r.port());
|
||||
assertEquals(true, r.is_absolute());
|
||||
assertEquals("http://www.jezuk.co.uk:8000/arabica/news/trouser/press", r.as_string());
|
||||
} // test17
|
||||
|
||||
|
|
@ -218,6 +235,7 @@ class URITest : public TestCase
|
|||
assertEquals("www.jezuk.co.uk", r.host());
|
||||
assertEquals("http", r.scheme());
|
||||
assertEquals("8000", r.port());
|
||||
assertEquals(true, r.is_absolute());
|
||||
assertEquals("http://www.jezuk.co.uk:8000/arabica/trouser/press", r.as_string());
|
||||
} // test18
|
||||
|
||||
|
|
@ -230,6 +248,7 @@ class URITest : public TestCase
|
|||
assertEquals("www.jezuk.co.uk", r.host());
|
||||
assertEquals("http", r.scheme());
|
||||
assertEquals("8000", r.port());
|
||||
assertEquals(true, r.is_absolute());
|
||||
} // test19
|
||||
|
||||
void test20()
|
||||
|
|
@ -241,6 +260,7 @@ class URITest : public TestCase
|
|||
assertEquals("www.jezuk.co.uk", r.host());
|
||||
assertEquals("http", r.scheme());
|
||||
assertEquals("8000", r.port());
|
||||
assertEquals(true, u.is_absolute());
|
||||
assertEquals("http://www.jezuk.co.uk:8000/trouser/press", r.as_string());
|
||||
} // test20
|
||||
|
||||
|
|
@ -255,6 +275,7 @@ class URITest : public TestCase
|
|||
assertEquals("www.jezuk.co.uk", u.host());
|
||||
assertEquals("http", u.scheme());
|
||||
assertEquals("80", u.port());
|
||||
assertEquals(true, u.is_absolute());
|
||||
} // test21
|
||||
|
||||
void test22()
|
||||
|
|
@ -348,6 +369,7 @@ class URITest : public TestCase
|
|||
assertEquals("file", u.scheme());
|
||||
assertEquals("0", u.port());
|
||||
assertEquals("file://d:/work/jezuk/arabica/file.xml", u.as_string());
|
||||
assertEquals(true, u.is_absolute());
|
||||
} // test32
|
||||
|
||||
void test33()
|
||||
|
|
@ -357,6 +379,8 @@ class URITest : public TestCase
|
|||
assertEquals("", u.host());
|
||||
assertEquals("file", u.scheme());
|
||||
assertEquals("/woot.txt", u.path());
|
||||
assertEquals(true, u.is_absolute());
|
||||
assertEquals("file:///woot.txt", u.as_string());
|
||||
} // test33
|
||||
|
||||
void test34()
|
||||
|
|
@ -366,6 +390,8 @@ class URITest : public TestCase
|
|||
assertEquals("", u.host());
|
||||
assertEquals("file", u.scheme());
|
||||
assertEquals("/woot.txt", u.path());
|
||||
assertEquals(true, u.is_absolute());
|
||||
assertEquals("file:///woot.txt", u.as_string());
|
||||
} // test34
|
||||
|
||||
void test35()
|
||||
|
|
@ -375,6 +401,8 @@ class URITest : public TestCase
|
|||
assertEquals("", u.host());
|
||||
assertEquals("file", u.scheme());
|
||||
assertEquals("woot.txt", u.path());
|
||||
assertEquals(false, u.is_absolute());
|
||||
assertEquals("file:woot.txt", u.as_string());
|
||||
} // URI
|
||||
|
||||
void test36()
|
||||
|
|
@ -384,6 +412,8 @@ class URITest : public TestCase
|
|||
assertEquals("localhost", u.host());
|
||||
assertEquals("file", u.scheme());
|
||||
assertEquals("/woot.txt", u.path());
|
||||
assertEquals(true, u.is_absolute());
|
||||
assertEquals("file://localhost/woot.txt", u.as_string());
|
||||
} // test36
|
||||
|
||||
void test37()
|
||||
|
|
@ -394,6 +424,7 @@ class URITest : public TestCase
|
|||
assertEquals("", r.host());
|
||||
assertEquals("file", r.scheme());
|
||||
assertEquals("/woot/woot/woo.txt", r.path());
|
||||
assertEquals(true, r.is_absolute());
|
||||
} // test37
|
||||
|
||||
void test38()
|
||||
|
|
@ -404,6 +435,7 @@ class URITest : public TestCase
|
|||
assertEquals("", r.host());
|
||||
assertEquals("file", r.scheme());
|
||||
assertEquals("/woot/woot/woo.txt", r.path());
|
||||
assertEquals(true, r.is_absolute());
|
||||
} // test38
|
||||
|
||||
}; // class URITest
|
||||
|
|
|
|||
Loading…
Reference in a new issue