46663 (588412), страница 6
Текст из файла (страница 6)
begin
frmClassifySettings.ShowModal;
end;
procedure TfrmMain.N9Click(Sender: TObject);
begin
Close;
end;
procedure TfrmMain.N11Click(Sender: TObject);
var tmpl, Template, NewTemplate, ItemIndex: olevariant;
ARange: Range;
pars: Paragraphs;
par: Paragraph;
st: string;
vcol: OleVariant;
iStat: integer;
begin
dmIAS.aqFiles.Close;
dmIAS.aqFiles.Open;
frmGetFileName.cbFiles.Items.Clear;
frmGetFileName.cbFiles.Text:= dmIAS.aqFiles.FieldByName('file').AsString;
while not dmIAS.aqFiles.Eof do
begin
frmGetFileName.cbFiles.Items.Add(dmIAS.aqFiles.FieldByName('file').AsString);
dmIAS.aqFiles.Next;
end;
frmGetFileName.iMD:= 0;
frmGetFileName.ShowModal;
if (frmGetFileName.iMD = 0) or (frmGetFileName.cbFiles.Text = '')
then exit;
dmIAS.aqExe.Close;
dmIAS.aqExe.SQL.Text:= 'Select * from qObrPubl where file = :file';
dmIAS.aqExe.Parameters.ParamByName('file').Value:= frmGetFileName.cbFiles.Text;
dmIAS.aqExe.Open;
try
wa.Connect;
wa.Visible := True;
except
MessageDlg('А у Вас Word не установлен :(', mtError, [mbOk], 0);
Abort;
end;
ItemIndex:= 1;
Template:= EmptyParam;
NewTemplate := False;
// Создание документа
wa.Documents.Add(Template, NewTemplate, EmptyParam, EmptyParam) ;
wd.ConnectTo(wa.Documents.Item(ItemIndex));
wd.PageSetup.Set_Orientation(wdOrientLandscape);
wa.Options.CheckSpellingAsYouType := False;
wa.Options.CheckGrammarAsYouType := False;
ARange:= wd.Range(EmptyParam, EmptyParam);
pars:= wd.Paragraphs;
tmpl:= ARange;
par:= pars.Add(tmpl);
wf.ConnectTo(wd.Sentences.Get_Last.Font);
wd.Range.Paragraphs.Set_Alignment(wdAlignParagraphLeft);
wa.Selection.Font.Bold:= 1;
wa.Selection.Font.Size:= 16;
st:= 'Отчет по обработанным статьям файла ' + frmGetFileName.cbFiles.Text;
wa.Selection.InsertAfter(st+#13);
wa.Selection.InsertAfter(' '+#13);
vcol := wdCollapseEnd;
wa.Selection.Collapse(vcol);
iStat:= -1;
while not dmIAS.aqExe.Eof do
begin
if iStat <> dmIAS.aqExe.FieldByName('id_publ').AsInteger
then
begin
vcol := wdCollapseEnd;
wa.Selection.Collapse(vcol);
wa.Selection.Font.Bold:= 0;
wa.Selection.Font.Size:= 14;
wa.Selection.InsertAfter(' '+#13);
wa.Selection.InsertAfter(' '+#13);
st:= 'Дата публикации: ';
wa.Selection.InsertAfter(st);
vcol := wdCollapseEnd;
wa.Selection.Collapse(vcol);
if (dmIAS.aqExe.FieldByName('data').AsString = '30.12.1899')
then st:= ' '
else st:= dmIAS.aqExe.FieldByName('data').AsString;
wa.Selection.InsertAfter(st+#13);
vcol := wdCollapseEnd;
wa.Selection.Collapse(vcol);
st:= 'Название публикации: ';
wa.Selection.InsertAfter(st);
vcol := wdCollapseEnd;
wa.Selection.Collapse(vcol);
st:= dmIAS.aqExe.FieldByName('name_publ').AsString;
wa.Selection.InsertAfter(st+#13);
vcol := wdCollapseEnd;
wa.Selection.Collapse(vcol);
st:= 'Ключевые слова: ';
wa.Selection.InsertAfter(st);
vcol := wdCollapseEnd;
wa.Selection.Collapse(vcol);
st:= dmIAS.aqExe.FieldByName('keywords').AsString;
wa.Selection.InsertAfter(st+#13);
vcol := wdCollapseEnd;
wa.Selection.Collapse(vcol);
st:= 'Автор публикации: ';
wa.Selection.InsertAfter(st);
vcol := wdCollapseEnd;
wa.Selection.Collapse(vcol);
st:= dmIAS.aqExe.FieldByName('author').AsString;
wa.Selection.InsertAfter(st+#13);
vcol := wdCollapseEnd;
wa.Selection.Collapse(vcol);
st:= 'Текст публикации: ';
wa.Selection.InsertAfter(st);
vcol := wdCollapseEnd;
wa.Selection.Collapse(vcol);
st:= dmIAS.aqExe.FieldByName('text_publ').AsString;
wa.Selection.InsertAfter(st+#13);
vcol := wdCollapseEnd;
wa.Selection.Collapse(vcol);
st:= 'Сопоставленные рубрики:';
wa.Selection.InsertAfter(st+#13);
vcol := wdCollapseEnd;
wa.Selection.Collapse(vcol);
st:= dmIAS.aqExe.FieldByName('indx').AsString
+ ' '
+ dmIAS.aqExe.FieldByName('name_r').AsString;
wa.Selection.InsertAfter(st);
vcol := wdCollapseEnd;
wa.Selection.Collapse(vcol);
iStat:= dmIAS.aqExe.FieldByName('id_publ').AsInteger;
end
else
begin
wa.Selection.InsertAfter(' '+#13);
st:= dmIAS.aqExe.FieldByName('indx').AsString
+ ' '
+ dmIAS.aqExe.FieldByName('name_r').AsString;
wa.Selection.InsertAfter(st);
vcol := wdCollapseEnd;
wa.Selection.Collapse(vcol);
end;
dmIAS.aqExe.Next;
end;
wa.Disconnect;
end;
procedure TfrmMain.N12Click(Sender: TObject);
begin
frmEditPubl.ShowModal;
end;
end.
unit uFuzzySearch;
interface
function IndistinctMatching(MaxMatching:Integer; strInputMatching:WideString; strInputStandart:WideString):Integer;
implementation
uses SysUtils;
type
TRetCount = packed record
lngSubRows : Word;
lngCountLike : Word;
end;
function Matching(StrInputA:WideString; StrInputB:WideString; lngLen:Integer):TRetCount;
Var
TempRet : TRetCount;
PosStrB : Integer;
PosStrA : Integer;
StrA : WideString;
StrB : WideString;
StrTempA : WideString;
StrTempB : WideString;
begin
StrA := String(StrInputA);
StrB := String(StrInputB);
For PosStrA:= 1 To Length(strA) - lngLen + 1 do
begin
StrTempA:= System.Copy(strA, PosStrA, lngLen);
For PosStrB:= 1 To Length(strB) - lngLen + 1 do
begin
StrTempB:= System.Copy(strB, PosStrB, lngLen);
If SysUtils.AnsiCompareText(StrTempA,StrTempB) = 0 Then
begin
Inc(TempRet.lngCountLike);
break;
end;
end;
Inc(TempRet.lngSubRows);
end; // PosStrA
Matching.lngCountLike:= TempRet.lngCountLike;
Matching.lngSubRows := TempRet.lngSubRows;
end; { function }
//------------------------------------------------------------------------------
function IndistinctMatching(MaxMatching:Integer; strInputMatching:WideString; strInputStandart:WideString):Integer;
Var
gret : TRetCount;
tret : TRetCount;
lngCurLen: Integer ; //текущая длина подстроки
begin
//если не передан какой-либо параметр, то выход
If (MaxMatching = 0) Or (Length(strInputMatching) = 0) Or
(Length(strInputStandart) = 0) Then
begin
IndistinctMatching:= 0;
exit;
end;
gret.lngCountLike:= 0;
gret.lngSubRows := 0;
// Цикл прохода по длине сравниваемой фразы
For lngCurLen:= 1 To MaxMatching do
begin
//Сравниваем строку A со строкой B
tret:= Matching(strInputMatching, strInputStandart, lngCurLen);
gret.lngCountLike := gret.lngCountLike + tret.lngCountLike;
gret.lngSubRows := gret.lngSubRows + tret.lngSubRows;
//Сравниваем строку B со строкой A
//tret:= Matching(strInputStandart, strInputMatching, lngCurLen);
//gret.lngCountLike := gret.lngCountLike + tret.lngCountLike;
//gret.lngSubRows := gret.lngSubRows + tret.lngSubRows;
end;
If gret.lngSubRows = 0 Then
begin
IndistinctMatching:= 0;
exit;
end;
IndistinctMatching:= Trunc((gret.lngCountLike / gret.lngSubRows) * 100);
end;
end.
unit uClassify;
interface
uses
Windows, Messages, SysUtils, Variants, Classes, Graphics, Controls, Forms,
Dialogs, StdCtrls, ExtCtrls, Mask, DBCtrls, Grids, DBGridEh, ComCtrls,
WordXP, OleServer;
type
TfrmClassify = class(TForm)
Panel1: TPanel;
Panel2: TPanel;
Label1: TLabel;
DBNavigator1: TDBNavigator;
dbmText: TDBMemo;
Panel3: TPanel;
Label7: TLabel;
Button2: TButton;
Panel4: TPanel;
Label6: TLabel;
Label3: TLabel;
dbAuthor: TDBEdit;
Label2: TLabel;
dbDate: TDBEdit;
Label4: TLabel;
dbName: TDBEdit;
Label5: TLabel;
dbKeywords: TDBEdit;
pbClassify: TProgressBar;
dbgClassify: TDBGridEh;
Button1: TButton;
Label8: TLabel;
dbFile: TDBEdit;
lblCountArticles: TLabel;
Button3: TButton;
wd: TWordDocument;
wa: TWordApplication;
wf: TWordFont;
procedure SetDBElemColor(flColor: boolean);
procedure Button1Click(Sender: TObject);
procedure Button2Click(Sender: TObject);
procedure FormCreate(Sender: TObject);
procedure FormShow(Sender: TObject);
procedure Button3Click(Sender: TObject);
private
{ Private declarations }
public
{ Public declarations }
end;
var
frmClassify: TfrmClassify;
implementation
uses uDataModule, uFuzzySearch, uClassifySettings, uStructs;
{$R *.dfm}
procedure TfrmClassify.SetDBElemColor(flColor: boolean);
begin
if frmClassify = nil
then exit;
if flColor
then
begin
dbName.Color:= clMoneyGreen;
dbKeywords.Color:= clMoneyGreen;
dbmText.Color:= clMoneyGreen;
end
else
begin
dbName.Color:= clWindow;
dbKeywords.Color:= clWindow;
dbmText.Color:= clWindow;
end;
end;
procedure TfrmClassify.Button1Click(Sender: TObject);
var id_publ, id_rubr, level,
Name_cnt, Text_cnt, Key_cnt, im,
iNamePorog12, iTextPorog12, iKeyPorog12,
iNamePorog23, iTextPorog23, iKeyPorog23,
iFuzzyPorog: integer;
stWord, res: string;
function GetRes(NameCnt, TextCnt, KeyCnt: integer): String;
begin
if (NameCnt < iNamePorog12)
and (TextCnt < iTextPorog12)
and (KeyCnt < iKeyPorog12)
then result:= '-'
else
if (NameCnt >= iNamePorog23)
or (TextCnt >= iTextPorog23)
or (KeyCnt >= iKeyPorog23)
then result:= '+'
else result:= '?';
end;
begin
// Устанавливаем пороги срабатывания
iNamePorog12:= frmClassifySettings.GetNamePorog(12);
iTextPorog12:= frmClassifySettings.GetTextPorog(12);
iKeyPorog12:= frmClassifySettings.GetKeyPorog(12);
iNamePorog23:= frmClassifySettings.GetNamePorog(23);
iTextPorog23:= frmClassifySettings.GetTextPorog(23);
iKeyPorog23:= frmClassifySettings.GetKeyPorog(23);
iFuzzyPorog:= frmClassifySettings.GetFuzzyPorog;
// Начинаем классификацию
dmIAS.aqDict.Close;
dmIAS.aqDict.Open;
dmIAS.TruncClassify;
pbClassify.Position:= 0;
pbClassify.Max:= dmIAS.atPublikatsii.RecordCount * (dmIAS.aqDict.RecordCount+1);
with dmIAS.atPublikatsii do
begin
First;
while not Eof do
begin
id_publ:= FieldByName('id_publ').AsInteger;
Name_cnt:= 0;
Text_cnt:= 0;
Key_cnt:= 0;
dmIAS.aqDict.First;
id_rubr:= dmIAS.aqDict.FieldByName('id').AsInteger;
level:= dmIAS.aqDict.FieldByName('level').AsInteger;
while not dmIAS.aqDict.Eof do
begin
// Классифицируем по словарю для каждой рубрики
if id_rubr <> dmIAS.aqDict.FieldByName('id').AsInteger
then
begin
res:= GetRes(Name_cnt, Text_cnt, Key_cnt);
dmIAS.InsertClassify(id_publ, id_rubr, level, Name_cnt,
Text_cnt, Key_cnt, res);
id_rubr:= dmIAS.aqDict.FieldByName('id').AsInteger;
level:= dmIAS.aqDict.FieldByName('level').AsInteger;
Name_cnt:= 0;
Text_cnt:= 0;
Key_cnt:= 0;
end;
stWord:= AnsiUpperCase(dmIAS.aqDict.FieldByName('Word').AsString);
// Классификация по наименованию
im:= IndistinctMatching(length(stWord),
stWord,
AnsiUpperCase(FieldByName('name_publ').AsString));
if im > iFuzzyPorog
then Inc(Name_cnt);
// Классификация по тексту
im:= IndistinctMatching(length(stWord),
stWord,
AnsiUpperCase(FieldByName('text_publ').AsString));
if im > iFuzzyPorog
then Inc(Text_cnt);
// Классификация по ключевым словам
im:= IndistinctMatching(length(stWord),
stWord,
AnsiUpperCase(FieldByName('keywords').AsString));
if im > iFuzzyPorog
then Inc(Key_cnt);