pixwhile/blankie/murl.cpp

#include <regex>
#include <vector>
#include <ctype.h>
#include <climits>

#include "murl.h"
#include "../numberhelper.h"

#define UNRESERVED "[\\w\\d\\-._~]"
#define PCT_ENCODED "%[\\da-f]{2}"
#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
// Hostname does not allow for stuff like "//The", but it is not important here
#define HOSTCHAR "[\\w\\d\\-.]"
#define HOST "(\\[[\\da-f:.]+\\]|localhost|" HOSTCHAR "+(?:\\." HOSTCHAR "+)+)"
#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_REGEX \
    "(?:(https?)?:)?(?://" AUTHORITY ")?" \
    PATH_ABEMPTY \
    "(?:\\?" QUERY ")?" \
    "(?:#" FRAGMENT ")?"

static inline void hexencode(char c, char out[2]);
static inline char hexdecode(char nibble1, char nibble2);
static void handle_segment(std::vector<std::string>& segments, const std::string& str, size_t offset, size_t length);
static std::string tolower(std::string str);

namespace blankie {
namespace murl {

const std::regex url_regex(HTTP_HTTPS_REGEX, std::regex::icase);

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;
}

bool Url::is_host_equal(std::string other) const {
    return tolower(this->hostname) == tolower(std::move(other));
}

std::string escape(const std::string& in) {
    std::string out;
    char encoded[2];
    size_t pos = 0;
    size_t last_pos = 0;

    out.reserve(in.size());
    while ((pos = in.find_first_not_of("abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789", pos)) != std::string::npos) {
        out.append(in, last_pos, pos - last_pos);
        hexencode(in[pos], encoded);
        out += '%';
        out.append(encoded, 2);
        pos++;
        last_pos = pos;
    }

    if (in.size() > last_pos) {
        out.append(in, last_pos);
    }

    return out;
}

std::string unescape(const std::string& in) {
    std::string out;
    size_t pos = 0;
    size_t last_pos = 0;

    out.reserve(in.size());
    while ((pos = in.find('%', pos)) != std::string::npos) {
        if (pos + 2 >= in.size()) {
            throw std::invalid_argument("String abruptly terminated while finding percent-encoded nibbles");
        }
        out.append(in, last_pos, pos - last_pos);
        out += hexdecode(in[pos + 1], in[pos + 2]);
        pos += 3;
        last_pos = pos;
    }

    if (in.size() > last_pos) {
        out.append(in, last_pos);
    }

    return out;
}

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 void hexencode(char c, char out[2]) {
    char nibble1 = (c >> 4) & 0xF;
    char nibble2 = c & 0xF;

    auto hexencode = [](char nibble) {
        return static_cast<char>(nibble < 10
            ? '0' + nibble
            : 'A' + nibble - 10);
    };
    out[0] = hexencode(nibble1);
    out[1] = hexencode(nibble2);
}

static inline char hexdecode(char nibble1, char nibble2) {
    auto hexdecode = [](char nibble) {
        if (nibble >= '0' && nibble <= '9') return static_cast<char>(nibble - '0');
        if (nibble >= 'A' && nibble <= 'F') return static_cast<char>(nibble - 'A' + 10);
        if (nibble >= 'a' && nibble <= 'f') return static_cast<char>(nibble - 'a' + 10);
        throw std::invalid_argument("Invalid percent-encoded nibble received");
    };

    return static_cast<char>((hexdecode(nibble1) << 4) | hexdecode(nibble2));
}

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));
    }
}

static std::string tolower(std::string str) {
    for (size_t i = 0; i < str.size(); i++) {
        str[i] = static_cast<char>(tolower(str[i]));
    }
    return str;
}