pdf::PDFSearcher Class Reference

#include <pdf_searcher.h>

Collaboration diagram for pdf::PDFSearcher:

[legend]

Public Member Functions
	PDFSearcher (PDFController *doc)
	~PDFSearcher ()
bool	begin_search_next (const PDFSearchCriteria &criteria, const string &from_anchor)
	construct a search context for the "search next" task
bool	begin_search_all (const PDFSearchCriteria &criteria)
	construct a search context for the "search all" task
SearchResult	search_next (PDFSearchDocument &results, PDFSearchTask *task)
	search the next word
SearchResult	seach_all (PDFSearchDocument &results, PDFSearchTask *task)
	search in the whole document
bool	dump_search_process (string &anchor)
	dump the current search process, for restarting search task
void	notify (SearchResult ret_code, PDFSearchDocument &results, unsigned int search_id)
	notify the listeners that the searching is done
PDFController *	get_doc_ctrl () const
	get the pointer of PDFController instance
Static Public Member Functions
static void	export_search_doc_to_coll (PDFSearchDocument &doc, PDFRangeCollection &collection)

---

Detailed Description

Definition at line 97 of file pdf_searcher.h.

---

Constructor & Destructor Documentation

pdf::PDFSearcher::PDFSearcher

(

PDFController *

doc

)

[inline, explicit]

Definition at line 100 of file pdf_searcher.h.

        : doc_controller(doc)
        , search_ctx()
    {}

pdf::PDFSearcher::~PDFSearcher

(

)

[inline]

Definition at line 105 of file pdf_searcher.h.

    {}

---

Member Function Documentation

bool pdf::PDFSearcher::begin_search_all

(

const PDFSearchCriteria &

criteria

)

construct a search context for the "search all" task

Definition at line 93 of file pdf_searcher.cpp.

References pdf::PDFSearchCriteria::case_sensitive, pdf::SearchContext::case_sensitive, pdf::SearchContext::char_cursor, pdf::SearchContext::dst_words, pdf::SearchContext::forward, pdf::PDFSearchCriteria::match_whole_word, pdf::SearchContext::match_whole_word, pdf::SearchContext::page_num, pdf::SearchContext::search_all, pdf::PDFSearchCriteria::text, and pdf::SearchContext::word_cursor.

Referenced by pdf::PDFSearchTask::execute().

{
    clear_search_ctx();

    // construct the search context
    search_ctx.case_sensitive = criteria.case_sensitive;
    search_ctx.match_whole_word = criteria.match_whole_word;
    
    // start from the first page
    search_ctx.page_num = 1;
    //search_ctx.forward = criteria.forward;
    search_ctx.forward = true;
    search_ctx.search_all = true;
    
    // start from the first word
    search_ctx.word_cursor = 0;
    search_ctx.char_cursor = 0;
    parse_dst_string(criteria.text, search_ctx.dst_words);

    return true;
}

Here is the caller graph for this function:

bool pdf::PDFSearcher::begin_search_next	(	const PDFSearchCriteria &	criteria,
		const string &	from_anchor
	)

construct a search context for the "search next" task

Definition at line 53 of file pdf_searcher.cpp.

References pdf::PDFSearchCriteria::case_sensitive, pdf::SearchContext::case_sensitive, pdf::SearchContext::char_cursor, pdf::PDFAnchor::char_idx, pdf::SearchContext::dst_words, pdf::PDFSearchCriteria::forward, pdf::SearchContext::forward, pdf::PDFSearchCriteria::match_whole_word, pdf::SearchContext::match_whole_word, pdf::PDFController::page_count(), pdf::SearchContext::page_num, pdf::PDFAnchor::page_num, pdf::SearchContext::search_all, pdf::PDFSearchCriteria::text, pdf::SearchContext::word_cursor, and pdf::PDFAnchor::word_num.

Referenced by pdf::PDFSearchTask::execute().

{
    PDFAnchor from_param(from_anchor);
    if (from_param.page_num <= 0 
        || from_param.page_num > 
        static_cast<int>(doc_controller->page_count()))
    {
        from_param.page_num = 1;
    }

    clear_search_ctx();
    // construct the search context
    
    search_ctx.case_sensitive = criteria.case_sensitive;
    search_ctx.match_whole_word = criteria.match_whole_word;
    search_ctx.page_num = from_param.page_num;
    search_ctx.forward = criteria.forward;
    search_ctx.search_all = false;
    search_ctx.word_cursor = from_param.word_num;
    search_ctx.char_cursor = from_param.char_idx;

    if (search_ctx.forward)
    {
        // move forward the start char index
        search_ctx.char_cursor++;
    }
    else
    {
        // move backward the start char index
        // NOTE: We should care about the boundary problem
        // the start anchor might be the last word and the last char
        search_ctx.char_cursor--;
    }

    parse_dst_string(criteria.text, search_ctx.dst_words);

    return true;
}

Here is the call graph for this function:

Here is the caller graph for this function:

bool pdf::PDFSearcher::dump_search_process

(

string &

anchor

)

dump the current search process, for restarting search task

Definition at line 225 of file pdf_searcher.cpp.

References pdf::SearchContext::char_cursor, pdf::PDFAnchor::char_idx, pdf::PDFAnchor::get_string(), pdf::SearchContext::page_num, pdf::PDFAnchor::page_num, pdf::SearchContext::word_cursor, and pdf::PDFAnchor::word_num.

Referenced by pdf::PDFSearchTask::execute().

{
    PDFAnchor process;
    process.page_num = search_ctx.page_num;
    process.word_num = search_ctx.word_cursor;
    process.char_idx = search_ctx.char_cursor;
    //process.file_name = get_doc_ctrl()->name();

    anchor = process.get_string();
    return true;
}

Here is the call graph for this function:

Here is the caller graph for this function:

void pdf::PDFSearcher::export_search_doc_to_coll	(	PDFSearchDocument &	doc,
		PDFRangeCollection &	collection
	)			[static]

export the PDFSearchDocument to PDFSearchCollection for UDS. content of the doc would be cleared after this function. TODO. This function and PDFSearchDocument should be removed due to the low efficiency.

Definition at line 332 of file pdf_searcher.cpp.

References pdf::PDFCollection< T >::add(), pdf::PDFCollection< T >::clear(), pdf::PluginRangeImpl::end_anchor, pdf::PDFCollection< T >::get(), _UDSString::get_buffer, pdf::PDFCollection< T >::size(), and pdf::PluginRangeImpl::start_anchor.

Referenced by notify().

{
    for(int i = 0; i < doc.size(); ++i)
    {
        PDFSearchPage *page = doc.get(i);
        for(int k = 0; k < page->size(); ++k)
        {
            PluginRangeImpl *range = page->get(k);
            if (range == 0)
            {
                break;
            }

            PluginRangeImpl *save_range = new PluginRangeImpl;
            save_range->start_anchor = new StringImpl(
                range->start_anchor->get_buffer(range->start_anchor));
            save_range->end_anchor = new StringImpl(
                range->end_anchor->get_buffer(range->end_anchor));

            collection.add(save_range);
        }
        page->clear();
    }
    doc.clear();
}

Here is the call graph for this function:

Here is the caller graph for this function:

PDFController* pdf::PDFSearcher::get_doc_ctrl

(

)

const [inline]

get the pointer of PDFController instance

Definition at line 129 of file pdf_searcher.h.

Referenced by pdf::PDFSearchTask::get_user_data().

{ return doc_controller; }

Here is the caller graph for this function:

void pdf::PDFSearcher::notify	(	SearchResult	ret_code,
		PDFSearchDocument &	results,
		unsigned int	search_id
	)

notify the listeners that the searching is done

Definition at line 278 of file pdf_searcher.cpp.

References pdf::Signal< R, A1, A2, A3, A4, A5 >::broadcast(), export_search_doc_to_coll(), and pdf::PDFController::sig_search_results_ready.

Referenced by pdf::PDFSearchTask::execute().

{
    // export the results from PDFSearchDocument to PDFRangeCollection
    PDFRangeCollection *coll = new PDFRangeCollection;
    export_search_doc_to_coll(results, *coll);

    doc_controller->sig_search_results_ready.broadcast(ret_code, coll
        , search_id);
}

Here is the call graph for this function:

Here is the caller graph for this function:

SearchResult pdf::PDFSearcher::seach_all	(	PDFSearchDocument &	results,
		PDFSearchTask *	task
	)

search in the whole document

Definition at line 168 of file pdf_searcher.cpp.

References pdf::PDFCollection< T >::add(), pdf::Task::is_aborted(), pdf::Task::is_paused(), LOGPRINTF, pdf::PDFController::page_count(), pdf::SearchContext::page_num, pdf::RES_ABORTED, pdf::RES_NOT_FOUND, pdf::RES_OK, pdf::RES_PAUSED, pdf::PDFElemCollection< T, E >::set_element(), pdf::PDFCollection< T >::size(), and pdf::SearchContext::word_cursor.

Referenced by pdf::PDFSearchTask::execute().

{
    // return code of this function
    SearchResult res = RES_NOT_FOUND;

    // return code of searching every page
    SearchResult res_once = res;

    PDFSearchPage *search_page = new PDFSearchPage;

    while ( search_ctx.page_num > 0
            && search_ctx.page_num <=
               static_cast<int>(doc_controller->page_count()))
    {
        
        // search the whole page if it is not the current page
        res_once = search_current_page(search_ctx, *search_page);

        if (res_once == RES_OK)
        {
            search_page->set_element(search_ctx.page_num);
            results.add(search_page);
            search_page = new PDFSearchPage;
        }
        
        // forward : increase page number; otherwise decrease page number
        search_ctx.page_num++;
 
        // reset the index of start word to be 0
        search_ctx.word_cursor = 0;

         // abort current task
        if (task->is_aborted())
        {
            LOGPRINTF("Task Search All canceled!\n");
            res = RES_ABORTED;
            break;
        }
        else if (task->is_paused())
        {
            LOGPRINTF("Task Search All paused!\n");
            res = RES_PAUSED;
            break;
        }
    }

    delete search_page;

    if (results.size() > 0)
    {
        res = RES_OK;
    }
    
    return res;
}

Here is the call graph for this function:

Here is the caller graph for this function:

SearchResult pdf::PDFSearcher::search_next	(	PDFSearchDocument &	results,
		PDFSearchTask *	task
	)

search the next word

Definition at line 115 of file pdf_searcher.cpp.

References pdf::PDFCollection< T >::add(), pdf::SearchContext::forward, pdf::Task::is_aborted(), pdf::Task::is_paused(), LOGPRINTF, pdf::PDFController::page_count(), pdf::SearchContext::page_num, pdf::RES_ABORTED, pdf::RES_NOT_FOUND, pdf::RES_OK, pdf::RES_PAUSED, pdf::PDFElemCollection< T, E >::set_element(), and pdf::SearchContext::word_cursor.

Referenced by pdf::PDFSearchTask::execute().

{

    PDFSearchPage *search_page = new PDFSearchPage;
    SearchResult res = RES_NOT_FOUND;
    while(res != RES_OK
          && search_ctx.page_num > 0
          && search_ctx.page_num <=
             static_cast<int>(doc_controller->page_count()))
    {
        res = search_current_page(search_ctx, *search_page);

        if (res != RES_OK)
        {
            // forward : increase page number; otherwise decrease page number
            search_ctx.forward ? search_ctx.page_num++
                : search_ctx.page_num--;

            // if it is not the first page, start from the first word if forward
            // else start from the last word
            search_ctx.word_cursor = -1;
        }

        // abort current task
        if (task->is_aborted())
        {
            LOGPRINTF("Task Search Next canceled!\n");
            res = RES_ABORTED;
            break;
        }
        else if (task->is_paused())
        {
            LOGPRINTF("Task Search Next paused!\n");
            res = RES_PAUSED;
            break;
        }
    }

    if (res != RES_OK)
    {
        delete search_page;
    }
    else
    {
        search_page->set_element(search_ctx.page_num);
        results.add(search_page);
    }

    return res;

}

Here is the call graph for this function:

Here is the caller graph for this function:

---

The documentation for this class was generated from the following files:

• pdf_searcher.h
• pdf_searcher.cpp