152 lines
4.1 KiB
C++
152 lines
4.1 KiB
C++
#include <regex>
|
|
#include <vector>
|
|
#include <climits>
|
|
|
|
#include "murl.h"
|
|
|
|
#define UNRESERVED "[\\w\\d\\-._~]"
|
|
#define PCT_ENCODED "%[\\da-f]{2}"
|
|
// A space is added to sub-delims to make it work with URLs that have a space
|
|
#define SUB_DELIMS "[!$&'()*+,;= ]"
|
|
|
|
#define USERINFO "((?:" UNRESERVED "|" PCT_ENCODED "|" SUB_DELIMS "|:)*)"
|
|
// A lot looser than RFC 3986, but things will go very wrong very quickly if I comply
|
|
#define HOST "([^/?#]+?)"
|
|
#define PORT "(\\d*)"
|
|
#define AUTHORITY "(?:" USERINFO "@)?" HOST "(?::" PORT ")?"
|
|
|
|
#define PCHAR "(?:" UNRESERVED "|" PCT_ENCODED "|" SUB_DELIMS "|[:@])"
|
|
#define SEGMENT PCHAR "*"
|
|
#define PATH_ABEMPTY "((?:/" SEGMENT ")*)"
|
|
|
|
#define QUERY "((?:" PCHAR "|[/?])*)"
|
|
// Looser than RFC 3986, but fragments might as well own everything
|
|
#define FRAGMENT "(.*)"
|
|
|
|
#define HTTP_HTTPS_URL \
|
|
"(?:(https?)?:)?(?://" AUTHORITY ")?" \
|
|
PATH_ABEMPTY \
|
|
"(?:\\?" QUERY ")?" \
|
|
"(?:#" FRAGMENT ")?"
|
|
|
|
static std::regex url_regex(HTTP_HTTPS_URL, std::regex::icase);
|
|
static inline int to_int(const std::string& str);
|
|
static void handle_segment(std::vector<std::string>& segments, const std::string& str, size_t offset, size_t length);
|
|
|
|
namespace blankie {
|
|
namespace murl {
|
|
|
|
Url::Url(const std::string& str) {
|
|
std::smatch sm;
|
|
if (!std::regex_match(str, sm, url_regex)) {
|
|
throw std::invalid_argument(str + " is not a URL");
|
|
}
|
|
this->scheme = sm.str(1);
|
|
this->userinfo = sm.str(2);
|
|
this->hostname = sm.str(3);
|
|
this->port = sm[4].length() > 0 ? to_int(sm.str(4)) : -1;
|
|
this->path = sm.str(5);
|
|
this->query = sm.str(6);
|
|
this->fragment = sm.str(7);
|
|
}
|
|
|
|
std::string Url::to_string() const {
|
|
std::string str;
|
|
if (!this->hostname.empty()) {
|
|
if (!this->scheme.empty()) {
|
|
str += this->scheme + ':';
|
|
}
|
|
str += "//";
|
|
if (!this->userinfo.empty()) {
|
|
str += this->userinfo + '@';
|
|
}
|
|
str += this->hostname;
|
|
if (this->port != -1) {
|
|
str += ':';
|
|
str += std::to_string(this->port);
|
|
}
|
|
}
|
|
str += this->path;
|
|
if (!this->query.empty()) {
|
|
str += '?';
|
|
str += this->query;
|
|
}
|
|
if (!this->fragment.empty()) {
|
|
str += '#';
|
|
str += this->fragment;
|
|
}
|
|
return str;
|
|
}
|
|
|
|
std::string normalize_path(const std::string& str) {
|
|
std::vector<std::string> segments;
|
|
std::string res;
|
|
size_t pos = 0;
|
|
size_t last_pos = 0;
|
|
bool starts_with_slash = false;
|
|
bool ends_with_slash = false;
|
|
|
|
if (str.size() > 0 && str[0] == '/') {
|
|
starts_with_slash = true;
|
|
last_pos = pos = 1;
|
|
}
|
|
|
|
while ((pos = str.find('/', pos)) != std::string::npos) {
|
|
handle_segment(segments, str, last_pos, pos - last_pos);
|
|
last_pos = ++pos;
|
|
}
|
|
if (str.size() > last_pos) {
|
|
handle_segment(segments, str, last_pos, str.size() - last_pos);
|
|
}
|
|
|
|
if (str.size() > 1 && str.back() == '/' && !segments.empty()) {
|
|
ends_with_slash = true;
|
|
}
|
|
|
|
if (starts_with_slash) {
|
|
res += '/';
|
|
}
|
|
for (size_t i = 0; i < segments.size(); i++) {
|
|
if (i != 0) {
|
|
res += '/';
|
|
}
|
|
res += std::move(segments[i]);
|
|
}
|
|
if (ends_with_slash) {
|
|
res += '/';
|
|
}
|
|
|
|
return res;
|
|
}
|
|
|
|
}; // namespace murl
|
|
}; // namespace blankie
|
|
|
|
static inline int to_int(const std::string& str) {
|
|
char* endptr;
|
|
|
|
long res = strtol(str.c_str(), &endptr, 10);
|
|
if (res > INT_MAX) {
|
|
throw std::overflow_error(str + " is too big");
|
|
} else if (res < INT_MIN) {
|
|
throw std::underflow_error(str + " is too small");
|
|
} else if (endptr[0] != '\0') {
|
|
throw std::invalid_argument(str + " has trailing text");
|
|
}
|
|
|
|
return static_cast<int>(res);
|
|
}
|
|
|
|
static void handle_segment(std::vector<std::string>& segments, const std::string& str, size_t offset, size_t length) {
|
|
if (length == 2 && str[offset] == '.' && str[offset + 1] == '.') {
|
|
if (segments.empty()) {
|
|
return;
|
|
}
|
|
segments.pop_back();
|
|
} else if (length == 1 && str[offset] == '.') {
|
|
// do nothing
|
|
} else if (length != 0) {
|
|
segments.push_back(str.substr(offset, length));
|
|
}
|
|
}
|