chenx221/Textractor
1
0
Fork 0
forked from Public-Mirror/Textractor
Code Pull Requests Activity

add user-agent switch method

Browse Source 
Add method that changes user-agent when in the headless mode
This commit is contained in:
zeheyler 2020-10-18 16:16:06 +03:00
parent 73d1f21bc1
commit 2ab780a491
3 changed files with 52 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

55
extensions/devtools.cpp
View File

@ -13,20 +13,31 @@ void DevTools::startDevTools(QString path, bool headless, int port)
{
if (startChrome(path, headless, port))
{
QJsonDocument doc;
QString webSocketDebuggerUrl;
if (GetwebSocketDebuggerUrl(webSocketDebuggerUrl, port))
if (GetJsonfromHTTP(doc, "/json/list", port))
{
connect(&webSocket, &QWebSocket::stateChanged, this, &DevTools::stateChanged);
connect(&webSocket, &QWebSocket::textMessageReceived, this, &DevTools::onTextMessageReceived);
webSocket.open(webSocketDebuggerUrl);
session += 1;
for (const auto obj : doc.array())
if (obj.toObject().value("type") == "page")
{
webSocketDebuggerUrl.append(obj.toObject().value("webSocketDebuggerUrl").toString());
break;
}
if (!webSocketDebuggerUrl.isEmpty())
{
if (GetJsonfromHTTP(doc, "/json/version", port))
{
useragent = doc.object().value("User-Agent").toString();
}
connect(&webSocket, &QWebSocket::stateChanged, this, &DevTools::stateChanged);
connect(&webSocket, &QWebSocket::textMessageReceived, this, &DevTools::onTextMessageReceived);
webSocket.open(webSocketDebuggerUrl);
session += 1;
return;
}
}
else
{
status = "Failed to find chrome debug port!";
emit statusChanged(status);
}
status = "Failed to find chrome debug port!";
emit statusChanged(status);
}
else
{
@ -45,6 +56,11 @@ QString DevTools::getStatus()
return status;
}
QString DevTools::getUserAgent()
{
return useragent;
}
DevTools::~DevTools()
{
closeDevTools();
@ -71,14 +87,13 @@ bool DevTools::startChrome(QString path, bool headless, int port)
return true;
}
bool DevTools::GetwebSocketDebuggerUrl(QString& url, int port)
bool DevTools::GetJsonfromHTTP(QJsonDocument& doc, QString object, int port)
{
url.clear();
if (HttpRequest httpRequest{
L"Mozilla/5.0 Textractor",
L"127.0.0.1",
L"POST",
FormatString(L"/json/list").c_str(),
object.toStdWString().c_str(),
"",
NULL,
NULL,
@ -89,16 +104,8 @@ bool DevTools::GetwebSocketDebuggerUrl(QString& url, int port)
})
{
QString qtString = QString::fromStdWString(httpRequest.response);
QJsonDocument doc = QJsonDocument::fromJson(qtString.toUtf8());
QJsonArray rootObject = doc.array();
for (const auto obj : rootObject)
if (obj.toObject().value("type") == "page")
{
url.append(obj.toObject().value("webSocketDebuggerUrl").toString());
break;
}
if (!url.isEmpty())
doc = QJsonDocument::fromJson(qtString.toUtf8());
if (!doc.isEmpty())
return true;
else
return false;

5
extensions/devtools.h
View File

@ -1,4 +1,3 @@
#include <fstream>
#include <QtCore>
#include <QtWebSockets/QWebSocket>
#include <ppltasks.h>
@ -30,11 +29,12 @@ public:
bool SendRequest(QString method, QJsonObject params, QJsonObject& root);
long methodToReceive(QString method, QJsonObject params = {});
QString getStatus();
QString getUserAgent();
private:
bool isConnected();
bool startChrome(QString path, bool headless = false, int port = 9222);
bool GetwebSocketDebuggerUrl(QString& url, int port = 9222);
bool GetJsonfromHTTP(QJsonDocument& doc, QString object, int port = 9222);
long idIncrement();
long idmIncrement();
bool compareJson(QJsonValue storedparams, QJsonValue params);
@ -47,4 +47,5 @@ private:
long idmethod;
PROCESS_INFORMATION processInfo;
QString status;
QString useragent;
};

20
extensions/devtoolsdeepltranslate.cpp
View File

@ -1,6 +1,5 @@
#include "qtcommon.h"
#include "extension.h"
#include "network.h"
#include "devtools.h"
extern const wchar_t* TRANSLATION_ERROR;
@ -33,7 +32,7 @@ QStringList languages
"Spanish: es",
};
int docfound = -1, targetNodeId = -1, session = -1, pageenabled = -1;
int docfound = -1, targetNodeId = -1, session = -1, pageenabled = -1, useragentflag = -1;
std::pair<bool, std::wstring> Translate(const std::wstring& text, DevTools* devtools)
{
@ -65,6 +64,7 @@ std::pair<bool, std::wstring> Translate(const std::wstring& text, DevTools* devt
docfound = -1;
targetNodeId = -1;
pageenabled = -1;
useragentflag = -1;
}
// Add spaces near ellipsis for better translation and check for quotes
@ -89,6 +89,22 @@ std::pair<bool, std::wstring> Translate(const std::wstring& text, DevTools* devt
}
pageenabled = 1;
}
// Change user-agent if in headless mode
if (useragentflag == -1)
{
QString useragent = devtools->getUserAgent();
useragent.replace(QRegularExpression("HeadlessChrome"), "Chrome");
if (!useragent.isEmpty())
{
if (!devtools->SendRequest("Network.setUserAgentOverride", { {"userAgent", useragent} }, root))
{
return { false, FormatString(L"%s", ERROR_COMMAND_FAIL) };
}
}
useragentflag = 1;
}
long navigate = devtools->methodToReceive("Page.navigatedWithinDocument");
long target = devtools->methodToReceive("DOM.attributeModified", { { "value" , "lmt__mobile_share_container" } });