TextSearch.h

Go to the documentation of this file.

//---------------------------------------------------------------------------------------
// Copyright (c) 2001-2023 by Apryse Software Inc. All Rights Reserved.
// Consult legal.txt regarding legal and license information.
//---------------------------------------------------------------------------------------
#ifndef PDFTRON_H_CPPPDFTextSearch
#define PDFTRON_H_CPPPDFTextSearch

#include <PDF/PDFDoc.h>
#include <Common/UString.h>
#include <PDF/Highlights.h>
#include <C/PDF/TRN_TextSearch.h>

namespace pdftron { 
    namespace PDF {

class SearchResult 
{
public:

    operator bool () { return IsFound(); }

    bool IsFound() const;

    bool IsDocEnd() const;

    bool IsPageEnd() const;

    int GetPageNumber() const;

    UString GetMatch() const;

    UString GetAmbientString() const;

    Highlights GetHighlights() const;



// @cond PRIVATE_DOC

#ifndef SWIGHIDDEN
    SearchResult();
    SearchResult(const SearchResult& b);
    SearchResult(int, TRN_TextSearchResultCode rc, UString r, UString a, Highlights h);
#endif
// @endcond

private:
    int page_num;
    TRN_TextSearchResultCode resCode;
    UString result;
    UString ambient;
    Highlights hlts;
};


class TextSearch 
{
public:

    TextSearch();
    ~TextSearch();

    typedef TRN_UInt32 Mode;

    enum TextSearchModes
    {
        //use regular expressions
        e_reg_expression = 0x0001,
        
        //match case-sensitively
        e_case_sensitive = e_reg_expression << 1,

        //match the entire word
        e_whole_word     = e_case_sensitive << 1,

        //search upward (from the end of the file and from the bottom of a page)
        e_search_up      = e_whole_word << 1,

        //tells the search process to return when each page is finished; this is
        //useful when a user needs Run() to return periodically so that certain
        //things (e.g., UI) can be updated from time to time.
        e_page_stop      = e_search_up << 1,        

        //tells the search process to compute Highlight information.
        e_highlight      = e_page_stop << 1,

        //tells the search process to compute the ambient string of the found pattern.
        //This is useful if a user wants to examine or display what surrounds the
        //found pattern.
        e_ambient_string = e_highlight << 1,

        //refrain from replacing newlines with spaces
        e_raw_text_search = e_ambient_string << 1,

        //use Z-order as reading order for text
        e_search_using_zorder = e_raw_text_search << 1
    };

    bool Begin(PDFDoc& doc, const UString& pattern, Mode mode, int start_page = -1, int end_page = -1);

    SearchResult Run();

    bool SetPattern(const UString& pattern);

    Mode GetMode() const;

    void SetMode( Mode mode );

    void SetRightToLeftLanguage(bool flag);

    int GetCurrentPage() const;

    void SetOCGContext(OCG::Context* context);

    void Destroy();


/* {{codegen: TextSearch}} */
        
    void SetAmbientLettersBefore(int ambient_letters_before);
    
    void SetAmbientLettersAfter(int ambient_letters_after);
    
    void SetAmbientWordsBefore(int ambient_words_before);
    
    void SetAmbientWordsAfter(int ambient_words_after);
/* {{codegen: TextSearch}} */

// @cond PRIVATE_DOC

    #ifndef SWIGHIDDEN
    TRN_TextSearch mp_textsearch;
    #endif /* SWIGHIDDEN */
private:
    TextSearch(const TextSearch&);
    TextSearch& operator= (const TextSearch&);
// @endcond

};


#include <Impl/TextSearch.inl>

    };  // namespace PDF
};  // namespace pdftron

#endif // PDFTRON_H_CPPPDFTextSearch