Îêîí÷àòåëüíàÿ ïðîãðàììà

Íèæå ïðåäñòàâëåí ïîëíûé òåêñò ïðîãðàììû, ðàçðàáîòàííîé â ýòîé ãëàâå, ñ äâóìÿ ìîäèôèêàöèÿìè: ìû èíêàïñóëèðîâàëè âñå ñòðóêòóðû äàííûõ è ôóíêöèè â êëàññ TextQuery (â ïîñëåäóþùèõ ãëàâàõ ìû îáñóäèì ïîäîáíîå èñïîëüçîâàíèå êëàññîâ), êðîìå òîãî, òåêñò áûë èçìåíåí, òàê êàê íàø êîìïèëÿòîð ïîääåðæèâàë ñòàíäàðò Ñ++ íå ïîëíîñòüþ.

Íàïðèìåð, áèáëèîòåêà iostream íå ñîîòâåòñòâîâàëà òåêóùåìó ñòàíäàðòó. Øàáëîíû íå ïîääåðæèâàëè çíà÷åíèÿ àðãóìåíòîâ ïî óìîë÷àíèþ. Âîçìîæíî, âàì ïðèäåòñÿ èçìåíèòü êîå-÷òî â ýòîé ïðîãðàììå, ÷òîáû îíà êîìïèëèðîâàëàñü â âàøåé ñèñòåìå.

// ñòàíäàðòíûå çàãîëîâî÷íûå ôàéëû Ñ++

#include <algorithm>

#include <string>

#include <vector>

#include <utility>

#include <map>

#include <set>

// çàãîëîâî÷íûé ôàéë iostream, íå îòâå÷àþùèé ñòàíäàðòó

#include <fstream.h>

// çàãîëîâî÷íûå ôàéëû Ñ

#include <stddef.h>

#include <ctype.h>

// typedef äëÿ óäîáñòâà ÷òåíèÿ

typedef pair<short,short>           location;

typedef vector<location,allocator>  loc;

typedef vector<string,allocator>    text;

typedef pair<text*,loc*>            text_loc;

class TextQuery {

public:

    TextQuery() { memset( this, 0, sizeof( TextQuery )); }

    static void

      filter_elements( string felems ) { filt_elems = felems; }

    void query_text();

    void display_map_text();

    void display_text_locations();

    void doit() {

        retrieve_text();

        separate_words();

        filter_text();

        suffix_text();

        strip_caps();

        build_word_map();

    }

private:

    void retrieve_text();

    void separate_words():

    void filter_text();

    void strip_caps();

    void suffix_textQ;

    void suffix_s( string& );

    void build_word_map();

private:

    vector<string,allocator>      *lines_of_text;

    text_loc                      *text_locations;

    map< string,loc*,

        less<string>,allocator>   *word_map;

ààà static stringàààààààààààààààà filt_elems;

};

string TextQuery::filt_elems( "\", Õ;: !?)(\V" );

int main()

{

ààà TextQuery tq;

ààà tq.doit();

ààà tq.query_text();

ààà tq.display_map_text();

}

void

TextQuery::

retrieve_text()

{

ààà string file_name;

ààà cout << " please enter file name: ";

ààà cin >> file_name;

ààà ifstream infile( file_name.c_str(), ios::in );

ààà if ( !infile ) {

ààààààà cerr << "oops' unable to open file "

àààààààààààà << file_name <<ààà " -- bailing out!\n";

àààààà àexit( -1 );

ààà }

ààà else cout << "\n";

ààà lines_of_text = new vector<string,allocator>;

ààà string textline;

ààà while ( getline( infile, textline, '\n' ))

ààààààà lines_of_text->push_back( textline );

}

void

TextQuery::

separate_words()

{

ààà vector<string,allocator> *words =

àààààààààààààà new vector<string,allocator>;

ààà vector<location,allocator> *locations =

àààààààààààààà new vector<location,allocator>;

ààà for ( short line_pos = 0; line_pos < lines_of_text->size();

ààààààà line_pos++ )

ààà {

ààààààà short word_pos = 0;

ààààààà string textline = (*lines_of_text)[ line_pos ];

ààààààà string::size_type eol = textline.1ength();

ààààààà string::size_type pos = 0, prev_pos = 0;

ààààààà while (( pos = textline.find_first_of( ' ', pos ))

ààààààààà àààààààààà!= string::npos )

ààààààà {

ààààààààààà words->push_back(

ààààààààààààààà textline.substr( prev_pos, pos - prev_pos ));

ààààààààààà locations->push_back(

ààààààààààààààà make_pair( line_pos, word_pos ));

ààààààààààà word_pos++; pos++; prev_pos = pos;

ààààààà }

ààààààà words->push_back(

ààààààààààà textline.substr( prev_pos, pos - prev_pos ));

ààààààà locations->push_back(make_pair(line_pos,word_pos));

ààà }

ààà text_locations = new text_loc( words, locations );



}

void

TextQuery::

filter_text()

{

ààà if ( filt_elems.empty() )

ààààààà return;

ààà vector<string,allocator> *words = text_locations->first;

ààà vector<string,allocator>:: iterator iter = words->begin();

ààà vector<string,allocator>::iterator iter_end = words->end();

ààà while ( iter != iter_end )

ààà {

ààààààà string::size_type pos = 0;

ààààààà while ((pos = (*iter).find_first_of(filt_elems, pos))

àààààààààààààààààà != string::npos )

ààààààààààà (*iter).erase(pos,l);

ààààààà ++iter;

ààà }

}

void

TextQuery::

suffix_text()

{

ààà vector<string,allocator> *words = text_locations->first;

ààà vector<string,allocator>::iterator iter = words->begin();

ààà vector<string,allocator>::iterator iter_end = words->end() ;

ààà while ( iter != iter_end ) {

ààààààà if ( (*iter).size() <= 3 )

àààààààà ààà{ iter++; continue; }

ààààààà if ( (*iter)[ (*iter).size()-l ] == 's' )

ààààààààààà suffix_s( *iter );

ààààààà // ôþÿþûýøªõû¹ýð  þñ¨ðñþªú𠸺¯¯øú¸þò...

ààààààà iter++;

ààà }

}

void

TextQuery::

suffix_s( string &word )

{

ààà string::size_type spos = 0;

ààà string::size_type pos3 = word.size()-3;

ààà // "ous", "ss", "is", "ius"

ààà string suffixes( "oussisius" );

ààà if ( ! word.compare( pos3, 3, suffixes, spos, 3 ) ||

àààààààà ! word.compare( pos3, 3, suffixes, spos+6, 3) ||

àààààààà ! word.compare( pos3+l, 2, suffixes, spos+2, 2 ) ||

àààààààà ! word.compare( pos3+l, 2, suffixes, spos+4, 2 ))

ààààààààààà return;

ààà string ies( "ies" );

ààà if ( ! word.compare( pos3, 3, ies ))

ààà {

ààààààà word.replace( pos3, 3, 1, 'º' );

ààààààà return;

ààà }

ààà string ses( "ses" );

ààà if ( ! word.compare( pos3, 3, ses ))

ààà {

ààààààà word.erase( pos3+l, 2 );

ààààààà return;

ààà }

ààà // ºôðûøü 's' ò úþý¡õ



ààà word.erase( pos3+2 );

ààà // ºôðûøü "'s"

ààà if ( word[ pos3+l ] == '\'' )

ààààààà word.erase( pos3+l );

}

void

TextQuery::

strip_caps()

{

ààà vector<string,allocator> *words = text_locations->first;

ààà vector<string,allocator>::iterator iter = words->begin();

ààà vector<string,allocator>::iterator iter_end = words->end();

ààà string caps( "ABCDEFGHI3KLMNOPQRSTUVWXYZ" );

ààà while ( iter != iter_end ) {

ààààààà string::size_type pos = 0;

ààààààà while (( pos = (*iter).find_first_of( caps, pos ))

ààààààààààààààààààà != string::npos )

ààààààààààà (*iter)[ pos ] = to1ower( (*iter)[pos] );

ààààààà ++iter;

ààà }

}

void

TextQuery::

build_word_map()

{

ààà word_map = new map<string,loc*,less<string>,allocator>;

ààà typedef map<string,loc*,less<string>,allocator>::value_type

ààààààà value_type;

ààà typedef set<string,less<string>,allocator>::difference_type

ààààààà diff_type;

ààà set<string,less<string>,allocator> exclusion_set;

ààà ifstream infile( "exclusion_set" );

ààà if ( !infile )

ààà {

ààààààà static string default_excluded_words[25] = {

ààààààààà "the","and","but","that","then","are","been",

ààààààààà "can","can't","cannot","could","did","for",

ààààààààà "had","have","him","his","her","its"."into",

ààààààààà "were","which","when","with","would"

ààààààà };

ààààààà cerr <<

àààààààààà "warning! unable to open word exclusion file! -- "

àààààààààààà << "using default set\n";

ààààààà copy( default_excluded_words,

ààààààààààààà default_excluded_words+25,

ààààààààààààà inserter(exclusion_set, exclusion_set.begin()));



ààà }

ààà else {

ààààààà istream_iterator< string, diff_type >

ààààààààààà input_set( infile ), eos;

ààààààà copy( input_set, eos,

ààààààààààà inserter( exclusion_set, exclusion_set.begin() ));

ààà }

ààà // ÿ¨þñõöøü¸  ÿþ ò¸õü ¸ûþòðü, ò¸ªðòû   ÿð¨v

ààà vector<string,allocator> *text_words =

ààààààà text_locations->first;

ààà vector<location,allocator> *text.locs =

ààààààà text_locations->second;

ààà register int elem_cnt = text_words->size();

ààà for ( int ix = 0; ix < elem_cnt; ++-ix )

ààà {

ààààààà string textword = ( *text_words )[ ix ];

ààààààà if ( textword.size() < 3 ||

ààààààààààà exclusion_set.count( textword ))

ààààààààààààààà continue;

ààààààà if ( ! word_map->count((*text_words)[ix] ))

ààààààà { // ¸ûþòþ þª¸ºª¸ªòºõª, ôþñðòøü:

ààààààààààà loc *ploc = new vector<location,allocator>;

ààààààààààà ploc->push_back( (*text_locs)[ix] );

ààààààààààà word_map->

ààààààààààààààà insert( value_type( (*text_words)[ix],ploc ));

ààààààà }

ààààààà else (*word_map) [(*text_words) [ix]]->

ààààààààààààààà push_back( (*text_locs) [ix] );

ààà }

}

void

TextQuery::

query_text()

{

ààà string query_text;

ààà do {

ààààààà cout

ààààààà << "enter a word against which to search the text.\n"

ààààààà << "to quit, enter a single character ==> ";

ààààààà cin >> query_text;

ààààààà if ( query_text.size() < 2 ) break;

ààààààà string caps( "ABCDEFGHIJKLMNOPQRSTUVWXYZ" );

ààààààà string::size_type pos = 0;

ààààààà while (( pos = query_text.find_first_of( caps, pos ))

ààààààààààààààààà àà!= string::npos )

ààààààààààà query_text[ pos ] = to1ower( query_text[pos] );

ààààààà // query_text ôþûöýþ ñvª¹ òòõôõýþ

ààààààà if ( !word_map->count( query_text )) {

ààààààààààà cout << "\nSorry. There are no entries for "

àààààààààààààààà << query_text << ".\n\n";



ààààààààààà continue;

ààààààà }

ààààààà loc *ploc = (*word_map) [ query_text ];

ààààààà set<short,less<short>,allocator> occurrence_1i nes;

ààààààà loc::iterator liter = ploc->begin(),

ààààààààààààààààààààà liter_end = ploc->end();

ààààà ààwhile ( liter != liter_end ) {

àààààààààààà occurrence_lines.1nsert(

àààààààààààààààààà occurrence_lines.end(), (*liter).first);

àààààààààààà ++liter;

ààààààà }

ààààààà register int size = occurrence_lines.size();

ààààààà cout << "\n" << query_text

ààà ààààààààà<< " occurs " << size

àààààààààààà << (size == 1 ? " time:" : " times:")

àààààààààààà << "\n\n";

ààààààà set<short,less<short>,allocator>::iterator

ààààààààààààà it=occurrence_lines.begin();

ààààààà for ( ; it != occurrence_"lines.end(); ++it ) {

ààààààààààà int line = *it;

ààààààààààà cout << "\t( line "

àààààààààààààààà // ñºôõü ýºüõ¨þò𪹠¸ª¨þúø ¸ 1,

àààààààààààààààà // úðú ¤ªþ ÿ¨øý ªþ òõ÷ôõ

àààààààààààààààà << line + 1 << " ) "

àààààààààààààààà << (*lines_of_text)[line] << endl;

ààààààà }

ààààààà cout << endl;

ààà }

ààà while ( ! query_text.empty() );

ààà cout << "Ok, bye!\n";

}

void

TextQuery::

display_map_text()

{

ààà typedef map<string,loc*, less<string>, allocator> map_text;

ààà map_text::iterator iter = word_map->begin(),

àààààààààààààààààààààà iter_end = word_map->end();

ààà while ( iter != iter_end ) {

ààààààà cout << "word: " << (*iter).first << " (";

ààààààà intàààààààààà loc_cnt = 0;

ààààààà locààààààààà *text_locs = (*iter).second;

ààààààà loc::iterator literàààà = text_locs->begin(),

ààààààààààààààààààààà liter_end = text_locs->end();

ààààààà while ( liter != liter_end )

ààààààà {



ààààààààààà if ( loc_cnt )

ààààààààààààààà cout << ",";

ààààààààààà else ++loc_cnt;

ààààààààààà cout << "(" << (*liter).first

àààààààààààààààà << "," << (*liter).second << ")";

ààààààààààà ++"liter;

ààààààà }

ààààààà cout << ")\n";

ààààààà ++iter;

ààà }

ààà cout << endl;

}

void

TextQuery::

disp1ay_text_locations()

{

ààà vector<string,allocator> *text_words =

ààààààà text_locations->first;

ààà vector<location,allocator> *text_locs =

ààààààà text_locations->second;

ààà register int elem_cnt = text_words->size();

ààà if ( elem_cnt != text_locs->size() )

ààà {

ààààààà cerr

àààààààà << "oops! internal error: word and position vectors "

àààààààà << "are of unequal size\n"

àààààààà << "words: " << elem_cnt << " "

àààààààà << "locs: " << text_locs->size()

àààààààà << " -- bailing out!\n";

ààààààà exit( -2 );

ààà }

ààà for ( int ix=0; ix < elem_cnt; ix++ )

ààà {

ààààààà cout << "word: " << (*text_words)[ ix ] << "\t"

àààààààààààà << "location: ("

àààààààààààà << (*text_locs)[ix].first << ","

àààààààààààà << (*text.locs)[ix].second << ")"

àààààààààààà << "\n";

ààà }

ààà cout << endl;

}

Lÿ¨ðöýõýøõ 6.25

+ñ· ¸ýøªõ, ÿþ¢õüº ýðü ÿþª¨õñþòðû¸  ¸ÿõ¡øðû¹ývù úû𸸠inserter ôû  ÷ðÿþûýõýø  ýðñþ¨ð ¸ªþÿ-¸ûþò (¤ªþ ºÿþüøýðõª¸  ò ¨ð÷ôõûõ 6.13.1, ð ôõªðû¹ýþ ¨ð¸¸üðª¨øòðõª¸  ò 12.4.1).

set<string> exclusion_set;

ifstreamààà infile( "exclusion_set" );

copy( default_excluded_words, default_excluded_words+25,

ààààà inserter(exclusion_set, exclusion_set.begin() ));

Lÿ¨ðöýõýøõ 6.26

¦õ¨òþýð¢ðû¹ýð  ¨õðûø÷ð¡ø  ÿþø¸úþòþù ¸ø¸ªõüv þª¨ðöðõª ÿ¨þ¡õôº¨ývù ÿþô¿þô: ýðñþ¨ óûþñðû¹ýv¿ ¯ºýú¡øù þÿõ¨ø¨ºõª ýðñþ¨þü ýõ÷ðòø¸øüv¿ ¸ª¨ºúªº¨ ôðýýv¿. +úþý¢ðªõû¹ývù òð¨øðýª ÿ¨õô¸ªðòû õª ¸þñþù ðû¹ªõ¨ýðªøòývù ÿþô¿þô, úþóôð üv øýúðÿ¸ºûø¨ºõü ¯ºýú¡øø ø ôðýývõ ò úû𸸠TextQuery. T¨ðòýøªõ þñð ¸ÿþ¸þñð. ¦ðúþòv ýõôþ¸ªðªúø ø ÿ¨õøüº•õ¸ªòð úðöôþóþ?

Lÿ¨ðöýõýøõ 6.27

T ôðýýþù òõ¨¸øø ÿ¨þó¨ðüüv øü  ¯ðùûð ¸ ªõú¸ªþü òòþôøª¸  ÿþ ÷ðÿ¨þ¸º. +þûõõ ºôþñýþ ñvûþ ñv ÷ðôðò𪹠õóþ úðú ÿð¨ðüõª¨ úþüðýôýþù ¸ª¨þúø; ò óûðòõ 7 üv ÿþúðöõü, úðú ¤ªþ ôõûðõª¸ . ¦ðúøõ õ•õ ÿð¨ðüõª¨v úþüðýôýþù ¸ª¨þúø öõûðªõû¹ýþ ¨õðûø÷þòðª¹?


Ñîäåðæàíèå ðàçäåëà

---