mirror of
https://git.tesses.org/tesses50/tessesframework.git
synced 2026-09-30 08:44:13 +00:00
Fixed a stoll bug that was there since day one, html parser, fix pkgconfig
This commit is contained in:
+99
-39
@@ -39,6 +39,8 @@
|
||||
#include "TessesFramework/Filesystem/VFS.hpp"
|
||||
#include "TessesFramework/Filesystem/VFSFix.hpp"
|
||||
// clang-format on
|
||||
|
||||
#include "TessesFramework/Serialization/BitConverter.hpp"
|
||||
using FileStream = Tesses::Framework::Streams::FileStream;
|
||||
using Stream = Tesses::Framework::Streams::Stream;
|
||||
using SeekOrigin = Tesses::Framework::Streams::SeekOrigin;
|
||||
@@ -607,12 +609,34 @@ uint16_t HttpServer::GetPort() {
|
||||
return server->GetPort();
|
||||
return 0;
|
||||
}
|
||||
|
||||
std::shared_ptr<Stream> ServerContext::OpenResponseStream() {
|
||||
if (sent)
|
||||
return nullptr;
|
||||
int64_t length = -1;
|
||||
if (!this->responseHeaders.TryGetFirstInt("Content-Length", length))
|
||||
length = -1;
|
||||
try {
|
||||
if (!this->responseHeaders.TryGetFirstInt("Content-Length", length))
|
||||
length = -1;
|
||||
} catch (std::invalid_argument &invalid) {
|
||||
std::string error =
|
||||
"<!DOCTYPE html><html lang=\"en\"><head><meta "
|
||||
"charset=\"UTF-8\"> <meta name=\"viewport\" "
|
||||
"content=\"width=device-width, "
|
||||
"initial-scale=1.0\"><title>Internal Server Error at " +
|
||||
HttpUtils::HtmlEncode(this->originalPath) +
|
||||
"</title><meta name=\"color-scheme\" content=\"dark "
|
||||
"light\"></head><body><h1>Internal Server Error at " +
|
||||
HttpUtils::HtmlEncode(this->originalPath) +
|
||||
"</h1><p>what(): " + HttpUtils::HtmlEncode(invalid.what()) +
|
||||
"</p></body></html>";
|
||||
this->responseHeaders.SetValue("Content-Length",
|
||||
static_cast<int64_t>(error.size()));
|
||||
this->statusCode = StatusCode::InternalServerError;
|
||||
this->WriteHeaders();
|
||||
this->strm->WriteBlock(reinterpret_cast<const uint8_t *>(error.data()),
|
||||
error.size());
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
if (this->version == "HTTP/1.1" && length == -1)
|
||||
this->responseHeaders.SetValue("Transfer-Encoding", "chunked");
|
||||
@@ -648,6 +672,10 @@ void HttpServer::StartAccepting() {
|
||||
uint16_t port;
|
||||
auto sock = svr->GetStream(ip, port);
|
||||
|
||||
if (!TF_IsRunning()) // we need to die, that continue is not what we
|
||||
// want if we must die from being done
|
||||
break;
|
||||
|
||||
TF_LOG("New Host IP: " + ip + ":" + std::to_string(port));
|
||||
|
||||
if (sock == nullptr) {
|
||||
@@ -708,9 +736,14 @@ ServerContext::ServerContext(std::shared_ptr<Stream> strm, bool debug)
|
||||
}
|
||||
std::shared_ptr<Stream> ServerContext::GetStream() { return this->strm; }
|
||||
void ServerContext::SendBytes(std::vector<uint8_t> buff) {
|
||||
std::shared_ptr<MemoryStream> strm = std::make_shared<MemoryStream>(false);
|
||||
strm->GetBuffer() = buff;
|
||||
SendStream(strm);
|
||||
|
||||
if (sent)
|
||||
return;
|
||||
this->responseHeaders.SetValue("Content-Length",
|
||||
static_cast<int64_t>(buff.size()));
|
||||
auto resp = OpenResponseStream();
|
||||
if (resp)
|
||||
resp->WriteBlock(buff.data(), buff.size());
|
||||
}
|
||||
ServerContext &ServerContext::WithLastModified(Date::DateTime dt) {
|
||||
this->responseHeaders.SetValue("Last-Modified", dt);
|
||||
@@ -718,11 +751,13 @@ ServerContext &ServerContext::WithLastModified(Date::DateTime dt) {
|
||||
}
|
||||
|
||||
void ServerContext::SendText(std::string text) {
|
||||
std::shared_ptr<MemoryStream> strm = std::make_shared<MemoryStream>(false);
|
||||
|
||||
auto &buff = strm->GetBuffer();
|
||||
buff.insert(buff.end(), text.begin(), text.end());
|
||||
SendStream(strm);
|
||||
if (sent)
|
||||
return;
|
||||
this->responseHeaders.SetValue("Content-Length",
|
||||
static_cast<int64_t>(text.size()));
|
||||
auto resp = OpenResponseStream();
|
||||
if (resp)
|
||||
resp->WriteBlock(reinterpret_cast<uint8_t *>(text.data()), text.size());
|
||||
}
|
||||
void ServerContext::SendErrorPage(bool showPath) {
|
||||
if (sent)
|
||||
@@ -792,22 +827,34 @@ void ServerContext::SendStream(std::shared_ptr<Stream> strm) {
|
||||
int64_t begin = 0;
|
||||
int64_t end = -1;
|
||||
|
||||
if (dash.size() == 1 &&
|
||||
res[0].find_first_of('-') != std::string::npos) {
|
||||
// NUMBER-
|
||||
begin = std::stoll(dash[0]);
|
||||
} else if (dash.size() == 2) {
|
||||
if (dash.size() == 2) {
|
||||
// NUMBER-NUMBER
|
||||
// or
|
||||
//-NUMBER
|
||||
|
||||
if (dash[0].size() > 0) {
|
||||
// NUMBER-NUMBER
|
||||
begin = std::stoll(dash[0]);
|
||||
end = std::stoll(dash[1]);
|
||||
} else {
|
||||
if (dash[0].empty()) {
|
||||
//-NUMBER
|
||||
end = std::stoll(dash[1]);
|
||||
|
||||
if (!Serialization::BitConverter::TryParseSigned(
|
||||
dash[1], end))
|
||||
throw std::runtime_error(
|
||||
"Failed to parse the end of the Range header");
|
||||
} else if (dash[1].empty()) {
|
||||
if (!Serialization::BitConverter::TryParseSigned(
|
||||
dash[0], begin))
|
||||
throw std::runtime_error(
|
||||
"Failed to parse the end of the Range header");
|
||||
} else {
|
||||
// NUMBER-NUMBER
|
||||
if (!Serialization::BitConverter::TryParseSigned(
|
||||
dash[0], begin))
|
||||
throw std::runtime_error(
|
||||
"Failed to parse the beginning of the Range "
|
||||
"header");
|
||||
|
||||
if (!Serialization::BitConverter::TryParseSigned(
|
||||
dash[1], end))
|
||||
throw std::runtime_error(
|
||||
"Failed to parse the end of the Range header");
|
||||
}
|
||||
} else {
|
||||
this->statusCode = BadRequest;
|
||||
@@ -886,6 +933,8 @@ void ServerContext::SendStream(std::shared_ptr<Stream> strm) {
|
||||
} else {
|
||||
|
||||
auto chunkedStream = this->OpenResponseStream();
|
||||
if (!chunkedStream)
|
||||
throw std::runtime_error("Can't open response stream");
|
||||
|
||||
if (method != "HEAD")
|
||||
strm->CopyTo(chunkedStream);
|
||||
@@ -945,24 +994,28 @@ ServerContext &ServerContext::WithStatusCode(StatusCode code) {
|
||||
this->statusCode = code;
|
||||
return *this;
|
||||
}
|
||||
void ServerContext::SendException(std::exception &ex) {
|
||||
if (this->debug) {
|
||||
void ServerContext::SendException(const std::exception &ex) {
|
||||
try {
|
||||
if (this->debug) {
|
||||
|
||||
this->WithMimeType("text/html")
|
||||
.WithStatusCode(StatusCode::InternalServerError)
|
||||
.SendText("<!DOCTYPE html><html lang=\"en\"><head><meta "
|
||||
"charset=\"UTF-8\"> <meta name=\"viewport\" "
|
||||
"content=\"width=device-width, "
|
||||
"initial-scale=1.0\"><title>Internal Server Error at " +
|
||||
HttpUtils::HtmlEncode(this->originalPath) +
|
||||
"</title><meta name=\"color-scheme\" content=\"dark "
|
||||
"light\"></head><body><h1>Internal Server Error at " +
|
||||
HttpUtils::HtmlEncode(this->originalPath) +
|
||||
"</h1><p>what(): " + HttpUtils::HtmlEncode(ex.what()) +
|
||||
"</p></body></html>");
|
||||
} else {
|
||||
this->WithStatusCode(StatusCode::InternalServerError)
|
||||
.SendErrorPage(true);
|
||||
this->WithMimeType("text/html")
|
||||
.WithStatusCode(StatusCode::InternalServerError)
|
||||
.SendText(
|
||||
"<!DOCTYPE html><html lang=\"en\"><head><meta "
|
||||
"charset=\"UTF-8\"> <meta name=\"viewport\" "
|
||||
"content=\"width=device-width, "
|
||||
"initial-scale=1.0\"><title>Internal Server Error at " +
|
||||
HttpUtils::HtmlEncode(this->originalPath) +
|
||||
"</title><meta name=\"color-scheme\" content=\"dark "
|
||||
"light\"></head><body><h1>Internal Server Error at " +
|
||||
HttpUtils::HtmlEncode(this->originalPath) +
|
||||
"</h1><p>what(): " + HttpUtils::HtmlEncode(ex.what()) +
|
||||
"</p></body></html>");
|
||||
} else {
|
||||
this->WithStatusCode(StatusCode::InternalServerError)
|
||||
.SendErrorPage(true);
|
||||
}
|
||||
} catch (...) {
|
||||
}
|
||||
}
|
||||
|
||||
@@ -1106,6 +1159,13 @@ void HttpServer::Process(std::shared_ptr<Stream> strm,
|
||||
ctx.requestHeaders.TryGetFirst("Content-Type", type) &&
|
||||
type == "application/x-www-form-urlencoded" &&
|
||||
ctx.requestHeaders.TryGetFirstInt("Content-Length", length)) {
|
||||
|
||||
if (length <= 0) {
|
||||
ctx.statusCode = StatusCode::BadRequest;
|
||||
ctx.SendErrorPage(true);
|
||||
return;
|
||||
}
|
||||
|
||||
size_t len = (size_t)length;
|
||||
std::vector<uint8_t> buffer(len);
|
||||
len = bStrm->ReadBlock(buffer.data(), len);
|
||||
|
||||
+152
-21
@@ -21,14 +21,19 @@
|
||||
|
||||
#include "TessesFramework/Http/HttpUtils.hpp"
|
||||
#include "TessesFramework/Filesystem/VFS.hpp"
|
||||
#include "TessesFramework/Serialization/BitConverter.hpp"
|
||||
#include "TessesFramework/Text/StringConverter.hpp"
|
||||
#include <algorithm>
|
||||
#include <iostream>
|
||||
#include <sstream>
|
||||
using VFSPath = Tesses::Framework::Filesystem::VFSPath;
|
||||
namespace Tesses::Framework::Http {
|
||||
|
||||
bool Uri::Relative(std::string url, Uri &uri) {
|
||||
auto index = url.find_first_of("//");
|
||||
bool Uri::Relative(std::string_view url, Uri &uri) {
|
||||
auto path = this->path;
|
||||
if (path.empty())
|
||||
path = "/";
|
||||
auto index = url.find("//");
|
||||
if (index != std::string::npos) {
|
||||
if (Uri::TryParse(url, uri)) {
|
||||
if (index == 0)
|
||||
@@ -63,7 +68,10 @@ bool Uri::Relative(std::string url, Uri &uri) {
|
||||
|
||||
auto fourthPart = HttpUtils::SplitString(thirdPart[1], "?", 2);
|
||||
|
||||
VFSPath p = VFSPath(this->path, fourthPart[0]);
|
||||
VFSPath p =
|
||||
VFSPath(path, (path.back() == '/' || fourthPart[0].empty())
|
||||
? fourthPart[0]
|
||||
: "../" + fourthPart[0]);
|
||||
uri.path =
|
||||
p.CollapseRelativeParents().ToString(); // this should be safe
|
||||
if (fourthPart.size() == 2) {
|
||||
@@ -98,21 +106,33 @@ uint16_t Uri::GetPort() {
|
||||
return 69;
|
||||
return 0;
|
||||
}
|
||||
bool Uri::TryParse(std::string url, Uri &uri) {
|
||||
bool Uri::TryParse(std::string_view url, Uri &uri) {
|
||||
uri.scheme = "";
|
||||
uri.port = 0;
|
||||
auto firstPart = HttpUtils::SplitString(url, "//", 2);
|
||||
if (firstPart.size() == 2)
|
||||
|
||||
uri.scheme = firstPart[0];
|
||||
else if (firstPart.empty())
|
||||
else
|
||||
return false;
|
||||
|
||||
auto secondPart = HttpUtils::SplitString(
|
||||
firstPart.size() == 2 ? firstPart[1] : firstPart[0], "/", 2);
|
||||
auto secondPart = HttpUtils::SplitString(firstPart[1], "/", 2);
|
||||
|
||||
if (secondPart.size() == 1) {
|
||||
uri.path = "/";
|
||||
auto hp = secondPart[0];
|
||||
auto fragPart = HttpUtils::SplitString(hp, "#", 2);
|
||||
if (fragPart.size() == 2) {
|
||||
uri.hash = fragPart[1];
|
||||
hp = fragPart[0];
|
||||
}
|
||||
auto queryPart = HttpUtils::SplitString(hp, "?", 2);
|
||||
if (queryPart.size() == 2) {
|
||||
HttpUtils::QueryParamsDecode(uri.query, queryPart[1]);
|
||||
hp = queryPart[0];
|
||||
}
|
||||
secondPart[0] = hp;
|
||||
|
||||
} else if (secondPart.size() == 2) {
|
||||
auto thirdPart = HttpUtils::SplitString(secondPart[1], "#", 2);
|
||||
if (thirdPart.empty())
|
||||
@@ -135,14 +155,25 @@ bool Uri::TryParse(std::string url, Uri &uri) {
|
||||
if (hostPortPart.empty())
|
||||
return false;
|
||||
if (hostPortPart.size() == 2) {
|
||||
uri.port = (uint16_t)std::stoul(hostPortPart[1]);
|
||||
uint64_t portNum;
|
||||
if (!Serialization::BitConverter::TryParseUnsigned(hostPortPart[1],
|
||||
portNum))
|
||||
return false;
|
||||
if (portNum > 65535)
|
||||
return false;
|
||||
uri.port = static_cast<uint16_t>(portNum);
|
||||
}
|
||||
uri.host = hostPortPart[0];
|
||||
|
||||
return true;
|
||||
}
|
||||
Uri::Uri() : query(true) {}
|
||||
std::string Uri::GetPathAndQuery() { return this->path + this->GetQuery(); }
|
||||
std::string Uri::GetPathAndQuery() {
|
||||
if (this->path.empty() || this->path.front() != '/')
|
||||
return '/' + this->path + this->GetQuery();
|
||||
|
||||
return this->path + this->GetQuery();
|
||||
}
|
||||
std::string Uri::GetQuery() {
|
||||
if (this->query.kvp.empty())
|
||||
return "";
|
||||
@@ -532,9 +563,16 @@ std::string HttpUtils::HtmlP(std::string_view text) {
|
||||
|
||||
auto flush = [&]() -> void {
|
||||
if (!builder.empty()) {
|
||||
// Collect trailing punctuation
|
||||
std::string trailing;
|
||||
static const char *punct = ".,;:!?'\"";
|
||||
while (!builder.empty() && std::strchr(punct, builder.back())) {
|
||||
trailing.insert(trailing.begin(), builder.back());
|
||||
builder.pop_back();
|
||||
}
|
||||
if (builder.find("http://") == 0 || builder.find("https://") == 0 ||
|
||||
builder.find("ftp://") == 0 || builder.find("ftps://") == 0 ||
|
||||
builder.find("magnet:") == 0 || builder.find("btmh:") == 0) {
|
||||
builder.find("magnet:") == 0) {
|
||||
newText += "<a href=\"" + HttpUtils::HtmlEncode(builder) +
|
||||
"\">" + HttpUtils::HtmlEncode(builder) + "</a>";
|
||||
} else if (builder.find("mailto:") == 0) {
|
||||
@@ -548,6 +586,7 @@ std::string HttpUtils::HtmlP(std::string_view text) {
|
||||
} else {
|
||||
newText += HttpUtils::HtmlEncode(builder);
|
||||
}
|
||||
newText += HttpUtils::HtmlEncode(trailing);
|
||||
builder = "";
|
||||
}
|
||||
};
|
||||
@@ -564,7 +603,7 @@ std::string HttpUtils::HtmlP(std::string_view text) {
|
||||
break;
|
||||
case '\t':
|
||||
flush();
|
||||
newText += "&tab;";
|
||||
newText += " ";
|
||||
break;
|
||||
case '\r':
|
||||
flush();
|
||||
@@ -578,6 +617,106 @@ std::string HttpUtils::HtmlP(std::string_view text) {
|
||||
|
||||
return newText;
|
||||
}
|
||||
static std::unordered_map<std::string, std::string> htmlencodings = {
|
||||
{""", "\""},
|
||||
{"'", "'"},
|
||||
{"&", "&"},
|
||||
{"<", "<"},
|
||||
{">", ">"},
|
||||
{" ", "\xC2\xA0"},
|
||||
{"©", "\xC2\xA9"},
|
||||
{"®", "\xC2\xAE"},
|
||||
{"™", "\xE2\x84\xA2"},
|
||||
{"—", "\xE2\x80\x94"},
|
||||
{"–", "\xE2\x80\x93"},
|
||||
{"…", "\xE2\x80\xA6"},
|
||||
{"’", "\xE2\x80\x99"},
|
||||
{"‘", "\xE2\x80\x98"},
|
||||
{"”", "\xE2\x80\x9D"},
|
||||
{"“", "\xE2\x80\x9C"},
|
||||
{"€", "\xE2\x82\xAC"},
|
||||
{"£", "\xC2\xA3"},
|
||||
{"¥", "\xC2\xA5"},
|
||||
{"°", "\xC2\xB0"},
|
||||
{"±", "\xC2\xB1"},
|
||||
{"×", "\xC3\x97"},
|
||||
{"÷", "\xC3\xB7"},
|
||||
{"«", "\xC2\xAB"},
|
||||
{"»", "\xC2\xBB"},
|
||||
{"•", "\xE2\x80\xA2"},
|
||||
{"·", "\xC2\xB7"},
|
||||
{"½", "\xC2\xBD"},
|
||||
{"¼", "\xC2\xBC"},
|
||||
{"¾", "\xC2\xBE"},
|
||||
};
|
||||
std::string HttpUtils::HtmlDecode(std::string_view view) {
|
||||
std::string myHtml = {};
|
||||
std::string tmp = {};
|
||||
bool inEscaped = false;
|
||||
for (size_t i = 0; i < view.size(); i++) {
|
||||
if (inEscaped) {
|
||||
if (view[i] == ';') {
|
||||
inEscaped = false;
|
||||
|
||||
if (tmp.size() > 2 && tmp[1] == '#') //&#N
|
||||
{
|
||||
if (tmp[2] == 'x' || tmp[2] == 'X') {
|
||||
auto res = tmp.substr(3);
|
||||
uint64_t code;
|
||||
if (Serialization::BitConverter::TryParseUnsigned(
|
||||
res, code, 16)) {
|
||||
if (code <= 0x10FFFF &&
|
||||
(code < 0xD800 || code > 0xDFFF)) {
|
||||
Text::StringConverter::UTF8::FromUTF32(
|
||||
myHtml, {static_cast<char32_t>(code)});
|
||||
} else {
|
||||
myHtml += "\xEF\xBF\xBD";
|
||||
}
|
||||
} else {
|
||||
myHtml += "\xEF\xBF\xBD";
|
||||
}
|
||||
} else {
|
||||
auto res = tmp.substr(2);
|
||||
uint64_t code;
|
||||
if (Serialization::BitConverter::TryParseUnsigned(
|
||||
res, code, 10)) {
|
||||
if (code <= 0x10FFFF &&
|
||||
(code < 0xD800 || code > 0xDFFF)) {
|
||||
Text::StringConverter::UTF8::FromUTF32(
|
||||
myHtml, {static_cast<char32_t>(code)});
|
||||
} else {
|
||||
myHtml += "\xEF\xBF\xBD";
|
||||
}
|
||||
} else {
|
||||
myHtml += "\xEF\xBF\xBD";
|
||||
}
|
||||
}
|
||||
} else {
|
||||
tmp += ';';
|
||||
auto result = htmlencodings.find(tmp);
|
||||
if (result != htmlencodings.end()) {
|
||||
myHtml += result->second;
|
||||
} else {
|
||||
myHtml += "\xEF\xBF\xBD";
|
||||
}
|
||||
}
|
||||
} else {
|
||||
tmp += view[i];
|
||||
}
|
||||
} else {
|
||||
if (view[i] == '&') {
|
||||
tmp = "&";
|
||||
inEscaped = true;
|
||||
} else {
|
||||
myHtml += view[i];
|
||||
}
|
||||
}
|
||||
}
|
||||
if (inEscaped)
|
||||
myHtml += tmp;
|
||||
return myHtml;
|
||||
}
|
||||
|
||||
std::string HttpUtils::HtmlEncode(std::string_view html) {
|
||||
std::string myHtml = {};
|
||||
for (auto item : html) {
|
||||
@@ -838,16 +977,8 @@ bool HttpDictionary::TryGetFirstInt(std::string key, int64_t &value) {
|
||||
std::string val;
|
||||
if (!TryGetFirst(key, val))
|
||||
return false;
|
||||
try {
|
||||
size_t off = 0;
|
||||
auto v = std::stoll(val, &off);
|
||||
if (off != val.size())
|
||||
return false;
|
||||
value = v;
|
||||
} catch (std::exception &ex) {
|
||||
return false;
|
||||
}
|
||||
return true;
|
||||
|
||||
return Serialization::BitConverter::TryParseSigned(val, value);
|
||||
}
|
||||
bool HttpDictionary::TryGetFirstDate(std::string key, Date::DateTime &dt) {
|
||||
std::string val;
|
||||
|
||||
@@ -284,4 +284,61 @@ void BitConverter::FromSint16LE(uint8_t &b, int16_t v) {
|
||||
memcpy(&dest, &v, sizeof(uint16_t));
|
||||
FromUint16LE(b, dest);
|
||||
}
|
||||
|
||||
bool BitConverter::TryParseSigned(const std::string &str, int64_t &val,
|
||||
int base) {
|
||||
try {
|
||||
if (str.find_first_not_of(" \t") != 0)
|
||||
return false;
|
||||
size_t pos = 0;
|
||||
int64_t v = std::stoll(str, &pos, base);
|
||||
if (pos < str.size())
|
||||
return false;
|
||||
val = v;
|
||||
return true;
|
||||
} catch (std::out_of_range &arg) {
|
||||
return false;
|
||||
} catch (std::invalid_argument &arg) {
|
||||
return false;
|
||||
}
|
||||
}
|
||||
bool BitConverter::TryParseUnsigned(const std::string &str, uint64_t &val,
|
||||
int base) {
|
||||
try {
|
||||
|
||||
if (str.find_first_not_of(" \t") != 0)
|
||||
return false;
|
||||
if (str[0] == '-' || str[0] == '+')
|
||||
return false;
|
||||
size_t pos = 0;
|
||||
uint64_t v = std::stoull(str, &pos, base);
|
||||
if (pos < str.size())
|
||||
return false;
|
||||
val = v;
|
||||
return true;
|
||||
} catch (std::out_of_range &arg) {
|
||||
return false;
|
||||
} catch (std::invalid_argument &arg) {
|
||||
return false;
|
||||
}
|
||||
}
|
||||
bool BitConverter::TryParseDouble(const std::string &str, double &val) {
|
||||
try {
|
||||
|
||||
if (str.find_first_not_of(" \t") != 0)
|
||||
return false;
|
||||
|
||||
size_t pos = 0;
|
||||
double v = std::stod(str, &pos);
|
||||
if (pos < str.size())
|
||||
return false;
|
||||
val = v;
|
||||
return true;
|
||||
} catch (std::out_of_range &arg) {
|
||||
return false;
|
||||
} catch (std::invalid_argument &arg) {
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
} // namespace Tesses::Framework::Serialization
|
||||
@@ -0,0 +1,429 @@
|
||||
#include "TessesFramework/Serialization/Html.hpp"
|
||||
#include "TessesFramework/Console.hpp"
|
||||
#include "TessesFramework/Http/HttpUtils.hpp"
|
||||
#include <unordered_set>
|
||||
namespace Tesses::Framework::Serialization::Html {
|
||||
|
||||
static std::unordered_set<std::string> html_nodes_with_no_body = {
|
||||
"area", "base", "br", "col", "embed", "hr", "img",
|
||||
"input", "link", "meta", "source", "track", "wbr",
|
||||
};
|
||||
std::string HtmlNode::ToString() const {
|
||||
std::string text;
|
||||
HtmlNode::ToString(text);
|
||||
return text;
|
||||
}
|
||||
|
||||
void HtmlNode::InnerText(std::string &str) const {
|
||||
if (this->isText) {
|
||||
str += this->text_or_tag;
|
||||
} else {
|
||||
for (const auto &item : this->children)
|
||||
item->InnerText(str);
|
||||
}
|
||||
}
|
||||
|
||||
std::string HtmlNode::InnerText() const {
|
||||
std::string innerText;
|
||||
InnerText(innerText);
|
||||
return innerText;
|
||||
}
|
||||
|
||||
void HtmlNode::ToString(std::string &str) const {
|
||||
if (this->isText) {
|
||||
str += Http::HttpUtils::HtmlEncode(this->text_or_tag);
|
||||
} else {
|
||||
str += "<";
|
||||
str += this->text_or_tag;
|
||||
for (auto &attr : this->attributes) {
|
||||
str += ' ';
|
||||
str += attr.first;
|
||||
if (attr.second) {
|
||||
str += "=\"";
|
||||
str += Http::HttpUtils::HtmlEncode(attr.second.value());
|
||||
str += "\"";
|
||||
}
|
||||
}
|
||||
str += ">";
|
||||
|
||||
if (html_nodes_with_no_body.count(this->text_or_tag) == 0) {
|
||||
for (auto &child : this->children)
|
||||
child->ToString(str);
|
||||
|
||||
str += "</";
|
||||
str += this->text_or_tag;
|
||||
str += ">";
|
||||
}
|
||||
}
|
||||
}
|
||||
std::shared_ptr<HtmlNode> HtmlNode::FromText(std::string_view text) {
|
||||
auto node = std::make_shared<HtmlNode>();
|
||||
node->isText = true;
|
||||
node->text_or_tag = std::string{text};
|
||||
return node;
|
||||
}
|
||||
std::shared_ptr<HtmlNode> HtmlNode::FromNode(
|
||||
std::string_view name,
|
||||
std::initializer_list<std::pair<std::string, std::optional<std::string>>>
|
||||
attrs,
|
||||
std::initializer_list<std::shared_ptr<HtmlNode>> nodes) {
|
||||
return HtmlNode::FromNode(name, attrs.begin(), attrs.end(), nodes.begin(),
|
||||
nodes.end());
|
||||
}
|
||||
|
||||
std::shared_ptr<HtmlNode> HtmlNode::FromNode(
|
||||
std::string_view name,
|
||||
std::vector<std::pair<std::string, std::optional<std::string>>> attrs,
|
||||
std::vector<std::shared_ptr<HtmlNode>> nodes) {
|
||||
return HtmlNode::FromNode(name, attrs.begin(), attrs.end(), nodes.begin(),
|
||||
nodes.end());
|
||||
}
|
||||
|
||||
void HtmlNode::FindNodes(
|
||||
std::string_view tag,
|
||||
std::vector<std::shared_ptr<const HtmlNode>> &nodes) const {
|
||||
if (!this->isText) {
|
||||
if (this->text_or_tag == tag)
|
||||
nodes.push_back(this->shared_from_this());
|
||||
for (const auto &node : this->children)
|
||||
node->FindNodes(tag, nodes);
|
||||
}
|
||||
}
|
||||
|
||||
std::vector<std::shared_ptr<const HtmlNode>>
|
||||
HtmlNode::FindNodes(std::string_view tag) const {
|
||||
std::vector<std::shared_ptr<const HtmlNode>> nodes;
|
||||
FindNodes(tag, nodes);
|
||||
return nodes;
|
||||
}
|
||||
|
||||
enum class HtmlLexTokenType { SYMBOL, TEXT, STRING };
|
||||
struct HtmlLexToken {
|
||||
std::string text;
|
||||
HtmlLexTokenType type;
|
||||
HtmlLexToken(HtmlLexTokenType type, std::string_view text)
|
||||
: text(text), type(type) {}
|
||||
};
|
||||
|
||||
struct HtmlLexer {
|
||||
std::shared_ptr<TextStreams::TextReader> reader;
|
||||
std::vector<HtmlLexToken> &tokens;
|
||||
int32_t _peeked = -1;
|
||||
std::string builder;
|
||||
bool ignoreSpaceMode = false;
|
||||
std::string nodeName = {};
|
||||
|
||||
HtmlLexer(std::shared_ptr<TextStreams::TextReader> reader,
|
||||
std::vector<HtmlLexToken> &tokens)
|
||||
: reader(std::move(reader)), tokens(tokens) {}
|
||||
|
||||
int32_t Peek() {
|
||||
if (_peeked != -1)
|
||||
return _peeked;
|
||||
_peeked = Read();
|
||||
return _peeked;
|
||||
}
|
||||
|
||||
int64_t Read() {
|
||||
if (_peeked != -1) {
|
||||
auto tmp = _peeked;
|
||||
_peeked = -1;
|
||||
return tmp;
|
||||
}
|
||||
|
||||
return reader->ReadChar();
|
||||
}
|
||||
|
||||
void Flush() {
|
||||
if (!builder.empty()) {
|
||||
this->tokens.emplace_back(HtmlLexTokenType::TEXT, builder);
|
||||
|
||||
if (ignoreSpaceMode && nodeName.empty())
|
||||
nodeName = builder;
|
||||
|
||||
builder = "";
|
||||
}
|
||||
}
|
||||
|
||||
void SkipTo(std::string_view fmt) {
|
||||
if (fmt.empty())
|
||||
return;
|
||||
std::string window;
|
||||
window.resize(fmt.size());
|
||||
|
||||
while (fmt != window) {
|
||||
int32_t read = Read();
|
||||
if (read == -1)
|
||||
break;
|
||||
for (size_t i = 0; i < window.size() - 1; i++) {
|
||||
window[i] = window[i + 1];
|
||||
}
|
||||
|
||||
window.back() = static_cast<char>(read);
|
||||
}
|
||||
}
|
||||
|
||||
void Lex() {
|
||||
int read;
|
||||
while ((read = Read()) != -1) {
|
||||
int peek = Peek();
|
||||
switch (read) {
|
||||
case '"':
|
||||
case '\'': {
|
||||
if (ignoreSpaceMode) {
|
||||
Flush();
|
||||
char quote = static_cast<char>(read);
|
||||
std::string val;
|
||||
int c;
|
||||
while ((c = Read()) != -1 && c != quote)
|
||||
val += static_cast<char>(c);
|
||||
tokens.emplace_back(HtmlLexTokenType::STRING, val);
|
||||
} else {
|
||||
builder += static_cast<char>(read);
|
||||
}
|
||||
break;
|
||||
}
|
||||
case '/': {
|
||||
if (ignoreSpaceMode && peek == '>') {
|
||||
Flush();
|
||||
Read();
|
||||
ignoreSpaceMode = false;
|
||||
tokens.emplace_back(HtmlLexTokenType::SYMBOL, "/>");
|
||||
} else if (ignoreSpaceMode) {
|
||||
builder += '/';
|
||||
} else {
|
||||
builder += '/';
|
||||
}
|
||||
break;
|
||||
}
|
||||
case '>': {
|
||||
if (ignoreSpaceMode) {
|
||||
Flush();
|
||||
ignoreSpaceMode = false;
|
||||
tokens.emplace_back(HtmlLexTokenType::SYMBOL, ">");
|
||||
|
||||
if (nodeName == "script" || nodeName == "style") {
|
||||
nodeName = "_";
|
||||
std::string builder2 = "";
|
||||
bool eof = false;
|
||||
while ((read = Read()) != -1) {
|
||||
if (read == '<') {
|
||||
peek = Peek();
|
||||
if (peek == '/') {
|
||||
Read();
|
||||
break;
|
||||
} else if (peek == -1) {
|
||||
eof = true;
|
||||
break;
|
||||
} else {
|
||||
builder2 += '<';
|
||||
}
|
||||
} else {
|
||||
builder2 += static_cast<char>(read);
|
||||
}
|
||||
}
|
||||
|
||||
if (read != -1 && !eof) {
|
||||
tokens.emplace_back(HtmlLexTokenType::TEXT,
|
||||
builder2);
|
||||
|
||||
ignoreSpaceMode = true;
|
||||
tokens.emplace_back(HtmlLexTokenType::SYMBOL, "</");
|
||||
|
||||
} else {
|
||||
return;
|
||||
}
|
||||
}
|
||||
|
||||
} else {
|
||||
builder += '>';
|
||||
}
|
||||
} break;
|
||||
case '<': {
|
||||
Flush();
|
||||
if (peek == '!') {
|
||||
//<!-- -->
|
||||
//<!DOCTYPE html>
|
||||
Read();
|
||||
peek = Peek();
|
||||
if (peek == '-') {
|
||||
Read();
|
||||
peek = Peek();
|
||||
if (peek == '-') {
|
||||
Read();
|
||||
SkipTo("-->");
|
||||
|
||||
} else {
|
||||
SkipTo(">");
|
||||
}
|
||||
} else {
|
||||
SkipTo(">");
|
||||
}
|
||||
|
||||
} else if (peek == '/') {
|
||||
Read();
|
||||
|
||||
ignoreSpaceMode = true;
|
||||
tokens.emplace_back(HtmlLexTokenType::SYMBOL, "</");
|
||||
} else {
|
||||
|
||||
ignoreSpaceMode = true;
|
||||
tokens.emplace_back(HtmlLexTokenType::SYMBOL, "<");
|
||||
nodeName = {};
|
||||
}
|
||||
} break;
|
||||
case '=': {
|
||||
if (!ignoreSpaceMode) {
|
||||
builder += "=";
|
||||
} else {
|
||||
Flush();
|
||||
tokens.emplace_back(HtmlLexTokenType::SYMBOL, "=");
|
||||
}
|
||||
} break;
|
||||
case ' ':
|
||||
case '\n':
|
||||
case '\t':
|
||||
case '\r':
|
||||
if (!ignoreSpaceMode) {
|
||||
builder += static_cast<char>(read);
|
||||
} else {
|
||||
Flush();
|
||||
}
|
||||
break;
|
||||
default:
|
||||
builder += static_cast<char>(read);
|
||||
break;
|
||||
}
|
||||
}
|
||||
Flush();
|
||||
}
|
||||
};
|
||||
|
||||
struct HtmlParser {
|
||||
|
||||
std::vector<HtmlLexToken> &tokens;
|
||||
size_t index = 0;
|
||||
HtmlParser(std::vector<HtmlLexToken> &tokens) : tokens(tokens) {}
|
||||
|
||||
bool IsSymbol(std::string_view type) {
|
||||
if (Done())
|
||||
return false;
|
||||
auto &obj = Peak();
|
||||
if (obj.type != HtmlLexTokenType::SYMBOL)
|
||||
return false;
|
||||
|
||||
if (obj.text != type)
|
||||
return false;
|
||||
index++;
|
||||
return true;
|
||||
}
|
||||
|
||||
bool Done() { return index >= tokens.size(); }
|
||||
|
||||
HtmlLexToken &Read() { return tokens[index++]; }
|
||||
HtmlLexToken &Peak() { return tokens[index]; }
|
||||
|
||||
void Parse(std::string_view name, std::shared_ptr<HtmlNode> &node,
|
||||
bool root = false) {
|
||||
while (!Done()) {
|
||||
auto &curToken = Peak();
|
||||
switch (curToken.type) {
|
||||
case HtmlLexTokenType::TEXT: {
|
||||
Read();
|
||||
node->children.push_back(HtmlNode::FromText(
|
||||
Http::HttpUtils::HtmlDecode(curToken.text)));
|
||||
} break;
|
||||
case HtmlLexTokenType::SYMBOL: {
|
||||
if (curToken.text == "<") {
|
||||
Read();
|
||||
if (Done()) {
|
||||
return;
|
||||
}
|
||||
auto &val = Peak();
|
||||
if (val.type == HtmlLexTokenType::TEXT) {
|
||||
Read();
|
||||
|
||||
std::shared_ptr<HtmlNode> newNode =
|
||||
std::make_shared<HtmlNode>();
|
||||
newNode->isText = false;
|
||||
newNode->text_or_tag = val.text;
|
||||
|
||||
while (!Done()) {
|
||||
auto &k = Peak();
|
||||
|
||||
if (k.type == HtmlLexTokenType::TEXT) {
|
||||
Read();
|
||||
if (IsSymbol("=")) {
|
||||
if (!Done()) {
|
||||
auto &v = Peak();
|
||||
|
||||
if (v.type ==
|
||||
HtmlLexTokenType::STRING) {
|
||||
Read();
|
||||
newNode->attributes.emplace_back(
|
||||
k.text,
|
||||
Http::HttpUtils::HtmlDecode(
|
||||
v.text));
|
||||
continue;
|
||||
}
|
||||
}
|
||||
}
|
||||
newNode->attributes.emplace_back(k.text,
|
||||
std::nullopt);
|
||||
|
||||
continue;
|
||||
}
|
||||
|
||||
if (k.type == HtmlLexTokenType::SYMBOL &&
|
||||
(k.text == ">" || k.text == "/>"))
|
||||
break;
|
||||
}
|
||||
if (!IsSymbol("/>")) {
|
||||
IsSymbol(">");
|
||||
if (html_nodes_with_no_body.count(val.text) == 0)
|
||||
Parse(val.text, newNode);
|
||||
}
|
||||
|
||||
node->children.push_back(newNode);
|
||||
}
|
||||
} else if (curToken.text == "</") {
|
||||
Read();
|
||||
if (Done()) {
|
||||
node->children.push_back(HtmlNode::FromText("</"));
|
||||
return;
|
||||
}
|
||||
auto &val = Peak();
|
||||
Read();
|
||||
IsSymbol(">");
|
||||
|
||||
if (val.text == name && !root)
|
||||
return;
|
||||
}
|
||||
} break;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
std::shared_ptr<HtmlNode> Parse() {
|
||||
std::shared_ptr<HtmlNode> node = std::make_shared<HtmlNode>();
|
||||
node->isText = false;
|
||||
node->text_or_tag = ":root";
|
||||
Parse(":root", node, true);
|
||||
return node;
|
||||
}
|
||||
};
|
||||
|
||||
std::shared_ptr<HtmlNode>
|
||||
Parse(std::shared_ptr<TextStreams::TextReader> reader) {
|
||||
std::vector<HtmlLexToken> tokens;
|
||||
|
||||
HtmlLexer lexer(reader, tokens);
|
||||
|
||||
lexer.Lex();
|
||||
|
||||
HtmlParser parser(tokens);
|
||||
|
||||
return parser.Parse();
|
||||
}
|
||||
|
||||
} // namespace Tesses::Framework::Serialization::Html
|
||||
Reference in New Issue
Block a user