Line data    Source code

       1              : #ifndef WEBPAGEGRABBER_H
       2              : #define WEBPAGEGRABBER_H
       3              : 
       4              : #include <QObject>
       5              : #include <QString>
       6              : #include <QUrl>
       7              : #include <QTimer>
       8              : #include "../FangObject.h"
       9              : #include "SimpleHTTPDownloader.h"
      10              : 
      11              : #define DEFAULT_HANDLE_META_REFRESH true
      12              : #define DEFAULT_TIMEOUT_MS 5000
      13              : 
      14              : /*!
      15              :  * @brief Loads a web page at a given URL and signals with the XHTML document when done.
      16              :  *
      17              :  * Note that this class is not rentrant.
      18              :  */
      19              : class WebPageGrabber : public FangObject
      20              : {
      21              :     Q_OBJECT
      22              : public:
      23              :     /**
      24              :      * @brief WebPageGrabber creates an XHTML document from either a string or a URL.
      25              :      * @param handleMetaRefresh If true, handles refreshes from within HTML documents rather than
      26              :      *                          just HTTP communication.
      27              :      * @param timeoutMS         Timeout after last download activity in milliseconds
      28              :      * @param parent
      29              :      * @param networkManager    Optional network manager for dependency injection (for testing)
      30              :      */
      31              :     explicit WebPageGrabber(bool handleMetaRefresh = DEFAULT_HANDLE_META_REFRESH,
      32              :                             int timeoutMS = DEFAULT_TIMEOUT_MS,
      33              :                             QObject *parent = nullptr,
      34              :                             QNetworkAccessManager* networkManager = nullptr);
      35              : 
      36              :     /*!
      37              :      * \brief Just like the above but with all the defaults specified *except* for parent.
      38              :      */
      39              :     explicit WebPageGrabber(QObject *parent);
      40              : 
      41              : 
      42              : signals:
      43              :     // If you requested a URL, ready() will be emitted when it's ready!
      44              :     // If document is null, an error happened. :(
      45              :     void ready(WebPageGrabber* grabber, QString* document);
      46              : 
      47              : public slots:
      48              :     // Fetches the webpage and emits ready() with the XHTML document.
      49              :     // Signals with null on an error.
      50              :     virtual void load(const QUrl &url);
      51              : 
      52              :     // Load the HTML string into a Tidy'd XHTML document and returns it (no signal is emmitted.)
      53              :     // Returns null on an error.
      54              :     QString* load(const QString& htmlString);
      55              : 
      56              :     // Returns the previously loaded document, or the empty string if there was an error.
      57            3 :     inline QString* getDocument() { return error ? nullptr : &document; }
      58              : 
      59            0 :     inline QUrl getOriginalURL() const { return originalUrl; }
      60              : 
      61            0 :     inline bool isDone() const { return done; }
      62              :     
      63              : private slots:
      64              :     // Internal load methods.
      65              :     void loadInternal(const QUrl &url);
      66              :     QString* loadInternal(const QString& htmlString, bool handleRefresh);
      67              : 
      68              :     // Uh oh, an error!
      69              :     void onDownloadError(QString err);
      70              : 
      71              :     // We got some HTTP content!
      72              :     void onDownloadFinished(QByteArray array);
      73              : 
      74              :     // Searches the XHTML'd document for a redirect URL.
      75              :     // Returns the redirect URL, or the empty string.
      76              :     QString searchForRedirect(const QString& document);
      77              : 
      78              :     // Handles the boilerplate involved with emitting the ready() signal.
      79              :     void emitReadySignal(QString* document);
      80              :     
      81              : private:
      82              :     void init();
      83              : 
      84              :     SimpleHTTPDownloader downloader;
      85              :     QString document;
      86              :     bool handleMetaRefresh;
      87              :     int redirectAttempts;
      88              :     bool error;
      89              :     QUrl originalUrl;
      90              :     bool done;
      91              : };
      92              : 
      93              : #endif // WEBPAGEGRABBER_H