fix the problem of fulldomain and remove host name

MohammadRaziei · Nov 13, 2023 · ae8e16c · ae8e16c
1 parent f029261
commit ae8e16c
Show file tree

Hide file tree

Showing 5 changed files with 34 additions and 13 deletions.
diff --git a/examples/main.cpp b/examples/main.cpp
@@ -56,7 +56,7 @@ int main() {
 
 
         tic;
-        const auto hostName = url.hostName();
+        const auto hostName = url.fulldomain();
         toc;
         tic;
         const auto host = url.host();
@@ -146,7 +146,14 @@ int main() {
     show(TLD::Url("http://mohammad:123@www.google.com?about", true).host());
     show(TLD::Url("https://www.p30download.ir", false).host());
 
-    show(TLD::Url("http://mohammad:123@www.google.com?about", true).hostName()); // TODO: fix hostName Issue for ignore_www
+    show(TLD::Url("http://mohammad:123@www.google.com?about", true).fulldomain()); // TODO: fix hostName Issue for ignore_www
+    show(TLD::Host("http://mohammad:123@www.google.com?about", true) == "google.com"); // TODO: fix hostName Issue for ignore_www
+    const TLD::Host host2 = TLD::Host::fromUrl("http://mohammad:123@www.google.com?about", true);
+    const TLD::Url url2("http://mohammad:123@www.google.com?about", true);
+    tic;
+    const auto a = url2.fulldomain();
+    toc;
+    show(a); // TODO: add clone version to copy without sharing
 
     printf("\ngood bye :)\n");
     return 0;

diff --git a/include/urlparser.h b/include/urlparser.h
@@ -42,7 +42,6 @@ class Url {
     const int port() const noexcept;
     QueryParams params() const noexcept;
     const Host& host() const;
-    const std::string& hostName() const;
 
    private:
     class Impl;
@@ -56,9 +55,11 @@ class Host {
     static void loadPslFromPath(const std::string& filepath);
     static void loadPslFromString(const std::string& filestr);
     static bool isPslLoaded() noexcept;
+    static std::string_view removeWWW(const std::string_view&) noexcept;
 
    public:
     Host(const std::string& host, const bool ignore_www = DEFAULT_IGNORE_WWW);
+    Host() = default;
 
     bool operator==(const Host&) const;
     bool operator==(const std::string&) const;

diff --git a/src/binding/main.cpp b/src/binding/main.cpp
@@ -102,11 +102,13 @@ NB_MODULE(_core, m) {
         .def_static("load_psl_from_string", &TLD::Host::loadPslFromString,
                     nb::arg("string"))
         .def_static("is_psl_loaded", &TLD::Host::isPslLoaded)
+        .def_static("removeWWW", &TLD::Host::removeWWW, nb::arg("hoststr"))
         .def_prop_ro("subdomain", &TLD::Host::subdomain)
         .def_prop_ro("domain", &TLD::Host::domain)
         .def_prop_ro("domain_name", &TLD::Host::domainName)
         .def_prop_ro("fulldomain", &TLD::Host::fulldomain)
         .def_prop_ro("suffix", &TLD::Host::suffix)
+//        .def("__eq__", &TLD::Host::operator==) // TODO: fix this operator for 2 types
         .def("to_dict", host_to_dict)
         .def("to_json", host_to_json)
         .def("__str__", &TLD::Host::str)
@@ -119,7 +121,6 @@ NB_MODULE(_core, m) {
         .def_prop_ro("protocol", &TLD::Url::protocol)
         .def_prop_ro("userinfo", &TLD::Url::userinfo)
         .def_prop_ro("host", &TLD::Url::host)
-        .def_prop_ro("host_name", &TLD::Url::hostName)
         .def_prop_ro("subdomain", &TLD::Url::subdomain)
         .def_prop_ro("domain", &TLD::Url::domain)
         .def_prop_ro("fulldomain", &TLD::Url::fulldomain)
@@ -129,6 +130,7 @@ NB_MODULE(_core, m) {
         .def_prop_ro("params", &TLD::Url::params)
         .def_prop_ro("query", &TLD::Url::query)
         .def_prop_ro("fragment", &TLD::Url::fragment)
+//        .def("__eq__", &TLD::Url::operator==) // TODO: fix this operator for 2 types
         .def("to_dict", url_to_dict)
         .def("to_json", url_to_json)
         .def("__str__", &TLD::Url::str)

diff --git a/src/urlparser_host.cpp b/src/urlparser_host.cpp
@@ -94,7 +94,7 @@ TLD::Host::Impl::Impl(const std::string& host_, const bool ignore_www)
                 if (www_pos != std::string::npos)
                     return;
             } else {
-                subdomain_pos = 4;
+                subdomain_pos = 4;  // length of "www."
                 fulldomain_ = fulldomain_.substr(4);
             }
         }
@@ -168,3 +168,10 @@ bool TLD::Host::operator==(const TLD::Host& other) const {
 bool TLD::Host::operator==(const std::string& host) const {
     return impl->fulldomain() == host;
 }
+std::string_view TLD::Host::removeWWW(const std::string_view& host) noexcept {
+    if (const size_t www_pos = host.find("www.");
+        www_pos == std::string_view::npos || www_pos != 0) {
+            return host;
+    }
+    return host.substr(4); // 4 is the length of the "www."
+}
diff --git a/src/urlparser_url.cpp b/src/urlparser_url.cpp
@@ -17,10 +17,11 @@ class TLD::Url::Impl : public URL::Url {
     Impl(const std::string& url, const bool ignore_www);
 
     const TLD::Host* getHost() noexcept;
+    inline const std::string& hostName();
 
    private:
     std::unique_ptr<TLD::Host> host_obj = nullptr;
-    const bool ignore_www;
+    const bool ignore_www = DEFAULT_IGNORE_WWW;
 };
 
 inline std::vector<std::string> split(const std::string& str,
@@ -43,25 +44,28 @@ bool TLD::Url::isPslLoaded() noexcept {
 ////////////////////////////////////////////////////////////////////
 
 TLD::Url::Impl::Impl(const std::string& url, const bool ignore_www)
-    : URL::Url(url), ignore_www(ignore_www) {}
+    : URL::Url(url) , ignore_www(ignore_www) {}
 
 TLD::Url::Url(const std::string& url, const bool ignore_www)
     : impl(std::make_unique<TLD::Url::Impl>(url, ignore_www)) {}
 
 const TLD::Host* TLD::Url::Impl::getHost() noexcept {
     if (!host_obj)
-        host_obj = std::make_unique<TLD::Host>(host_, ignore_www);
+        host_obj = std::make_unique<TLD::Host>(hostName(), false);
+    /// we set ignore_www to false because we remove www in hostName function
     return host_obj.get();
 }
+const std::string& TLD::Url::Impl::hostName() {
+    if (ignore_www){
+        host_ = TLD::Host::removeWWW(host_);
+    }
+    return host_;
+}
 
 const TLD::Host& TLD::Url::host() const {
     return *impl->getHost();
 }
 
-const std::string& TLD::Url::hostName() const {
-    return impl->host_;
-}
-
 /// suffix
 const std::string& TLD::Url::suffix() const noexcept {
     return impl->getHost()->suffix();
@@ -79,7 +83,7 @@ const std::string& TLD::Url::domain() const noexcept {
 
 /// fulldomain
 const std::string& TLD::Url::fulldomain() const noexcept {
-    return impl->getHost()->fulldomain();
+    return impl->hostName();
 }
 
 /// domainName