textstat.pyw

documents classification and words statiscs TF/IDF etc

        optFrame3.pack(fill=X, padx=5)
        self.maxFreq = StringVar()
        self.maxFreq.set(cfg['freqMax'])
        Spinbox(optFrame3, width=5, textvariable=self.maxFreq, values=('---',1,2,3,4,5,10,25,50,100,250,500,750,1000,2500,5000,10000)).pack(side=LEFT, anchor=W)
        Label(optFrame3, text=s['FTmaxFreq']).pack(side=LEFT, padx=5)

        self.grosskleinFreq = IntVar()
        Checkbutton(optFrame, text=s['FTignGross'], variable=self.grosskleinFreq).pack(pady=10, anchor=W)

        self.suchFreq = StringVar()
        Label(optFrame, text=s['FTsuchFreq'], justify=LEFT).pack(anchor=W, padx=5)
        self.suchFreqEntry = Entry(optFrame, textvariable=self.suchFreq)
        self.suchFreqEntry.pack(anchor=W, padx=5)
        self.suchFreqEntry.bind('<Return>', self.zeigWortfrequenz)

        Button(optFrame, text=s['FTfreqliste'], command=self.zeigWortfrequenz).pack(anchor=W, padx=5, pady=10)


    def FormenTabPopup(self, event):
        self.FormenPopupMenu = Menu(self.root, tearoff=0)
        self.FormenPopupMenu.add_command(label=s['FTkopieren'], command=self.copy2clipboard)
        self.FormenPopupMenu.add_command(label=s['FTsuchKonk'], command=self.zeigKonkordanz)
        self.FormenPopupMenu.add_command(label=s['FTzeigInfo'], command=self.zeigFormInfo)
        self.FormenPopupMenu.tk_popup(*self.root.winfo_pointerxy())


    def machKonkordanzTab(self):
        frame=self.tabPages.pages[s['ConcTab']]['page']
        suchFrame = Frame(frame)
        suchFrame.pack(fill=X, padx=10, pady=5)
        self.suchwort = Entry(suchFrame, width=40, takefocus=1)
        self.suchwort.pack(side=LEFT)
        self.suchwort.bind("<Return>", self.zeigKonkordanz)
        Button(suchFrame, text=s['CTsuchen'], takefocus=1, command=self.zeigKonkordanz).pack(side=LEFT, padx=5, pady=10)
        Button(suchFrame, text=s['CTmachQuery'], command=self.QueryDialog).pack(anchor=W, padx=20, pady=10)

        listframe = Frame(frame)
        listframe.pack(expand=YES, fill=BOTH, side=LEFT, padx=5, pady=5)
        self.konkliste = MultiListBox.MultiListbox(listframe, ((s['ConcTab'], 60),),
                            command=self.zeigZitat, borderwidth=1, height=20)
        self.konkliste.pack(side=LEFT, fill=BOTH, expand=YES)
        for l in self.konkliste.lists:
            l.bind('<Enter>', self.statusAktualisieren)
            l.bind('<Button-3>', self.KonkTabPopup)
            l.bind('<Double-Button-1>', self.zeigZitat)

        optFrame = LabelFrame(frame, text=s['CTOptionen'])
        optFrame.pack(fill=X, padx=5, pady=5)

        self.konkGanzewoerter = IntVar()
        Checkbutton(optFrame, text=s['CTganzWort'], variable=self.konkGanzewoerter).pack(anchor=W)
        self.konkGrossklein = IntVar()
        self.konkGrosskleinCheckbutton = Checkbutton(optFrame, text=s['CTignGross'], variable=self.konkGrossklein)
        self.konkGrosskleinCheckbutton.pack(anchor=W)
        self.markKonk = IntVar()
        Checkbutton(optFrame, text=s['CTmarkSuch'], variable=self.markKonk).pack(anchor=W)

        optFrame2 = Frame(optFrame)
        optFrame2.pack(fill=X, padx=5, pady=10)
        optFrame2a = Frame(optFrame2)
        optFrame2a.pack(fill=BOTH)
        self.konkContextL = StringVar()
        self.konkContextL.set(cfg['konkContextL'])
        Spinbox(optFrame2a, width=4, textvariable=self.konkContextL, from_=10, to=100, increment=10).pack(side=LEFT, anchor=W)
        Label(optFrame2a, text=s['CTcontextL']).pack(side=LEFT, padx=5)
        optFrame2b = Frame(optFrame2)
        optFrame2b.pack(fill=BOTH)
        self.konkContextR = StringVar()
        self.konkContextR.set(cfg['konkContextR'])
        Spinbox(optFrame2b, width=4, textvariable=self.konkContextR, from_=10, to=100, increment=10).pack(side=LEFT, anchor=W)
        Label(optFrame2b, text=s['CTcontextR']).pack(side=LEFT, padx=5)

        self.sortModusKonk = StringVar()
        Radiobutton(optFrame, text=s['CTsortAlpha'], value='alpha', variable=self.sortModusKonk).pack(anchor=W)
        Radiobutton(optFrame, text=s['CTsortR'], value='rechts', variable=self.sortModusKonk).pack(anchor=W)
        Radiobutton(optFrame, text=s['CTsortL'], value='links', variable=self.sortModusKonk).pack(anchor=W)
        self.sortModusKonk.set('alpha')

        Button(optFrame, text=s['CTaktualisier'], command=self.zeigKonkordanz).pack(anchor=W, padx=5, pady=10)


    def KonkTabPopup(self, event):
        self.KonkPopupMenu = Menu(self.root, tearoff=0)
        self.KonkPopupMenu.add_command(label=s['CTkopieren'], command=self.copy2clipboard)
        self.KonkPopupMenu.add_command(label=s['CTzeigZitat'], command=self.zeigZitat)
        self.KonkPopupMenu.tk_popup(*self.root.winfo_pointerxy())


    def machZitatTab(self):
        frame=self.tabPages.pages[s['ZitatTab']]['page']
        Label(frame, text=s['ZTinfo']).pack(padx=10, pady=10)

        txtframe = Frame(frame)
        txtframe.pack(expand=YES, fill=BOTH, side=LEFT, padx=5, pady=5)
        scrollbar = Scrollbar(txtframe, orient=VERTICAL)
        self.zitat = Text(txtframe, borderwidth=1, wrap=WORD, selectbackground='lightsteelblue', selectforeground='black', yscrollcommand=scrollbar.set)
        scrollbar.config(command=self.zitat.yview)
        scrollbar.pack(side=RIGHT, fill=Y, padx=0)
        self.zitat.pack(side=LEFT, expand=YES, fill=BOTH, padx=0)
        self.zitat.tag_configure('blau', foreground='blue', underline=1)
        self.zitat.tag_configure('rot', foreground='red')
        self.zitat.tag_bind('oeffnen', '<Double-Button-1>', self.zitatDatei_oeffnen)
        self.zitat.bind('<Button-3>', self.ZitatTabPopup)
        self.zitat.config(cursor='arrow')
        self.zitat.insert(END, s['ZTInit'])


    def ZitatTabPopup(self, event):
        self.ZitatPopupMenu = Menu(self.root, tearoff=0)
        self.ZitatPopupMenu.add_command(label=s['ZTkopieren'], command=self.copy2clipboard)
        self.ZitatPopupMenu.add_command(label=s['ZToeffnen'], command=self.zitatDatei_oeffnen)
        self.ZitatPopupMenu.tk_popup(*self.root.winfo_pointerxy())


    def WebDialog(self):
        try:
            self.wd.focus_set()
        except:
            self.wd = Toplevel(self.root)
            self.wd.title("Web2corpus - TextSTAT WebSpider")
            frame = Frame(self.wd)
            frame.pack(expand=YES, fill=BOTH, padx=10, pady=10)
            Label(frame, justify=LEFT, text=s['WDinfo']).pack(anchor=W, padx=10, pady=5)
            self.webUrl = StringVar()
            url = Entry(frame, width=50, textvariable=self.webUrl)
            url.pack(anchor=W, padx=10, pady=5)
            Label(frame, text=s['WDfrageSeiten']).pack(anchor=W, padx=10, pady=5)

            frame1 = Frame(frame)
            frame1.pack(expand=YES, fill=BOTH)
            self.webAnzahl = StringVar()
            self.webAnzahl.set('1')
            Spinbox(frame1, width=5, textvariable=self.webAnzahl, values=(1,10,25,50,100,250,500,750,1000)).pack(side=LEFT, anchor=W, padx=10, pady=5)
            Label(frame1, text=s['WDanzahl']).pack(side=LEFT)
            self.webBasis = StringVar()
            self.webBasis.set('server')
            Radiobutton(frame, text=s['WDinSubdir'], value='subdir', variable=self.webBasis).pack(anchor=W, padx=10 )
            Radiobutton(frame, text=s['WDaufServer'], value='server', variable=self.webBasis).pack(anchor=W, padx=10 )
            Button(frame, text=s['WDok'], width=10, command=self.htmlFile_hinzu).pack(side=LEFT, padx=10, pady=10)
            Button(frame, text=s['WDcancel'], width=10, command=self.wd.destroy).pack(side=RIGHT, padx=10, pady=10)
            url.focus_set()


    def NewsDialog(self):
        try:
            self.nd.focus_set()
        except:
            self.nd = Toplevel(self.root)
            self.nd.title("News2corpus - TextSTAT NewsGrabber")
            frame = Frame(self.nd)
            frame.pack(expand=YES, fill=BOTH, padx=10, pady=10)
            Label(frame, justify=LEFT, text=s['NDinfo']).pack(anchor=W, padx=10, pady=5)

            frame1 = Frame(frame)
            frame1.pack(expand=YES, fill=BOTH)
            self.newsServer = StringVar()
            Entry(frame1, width=40, textvariable=self.newsServer).pack(side=LEFT, anchor=W, padx=10, pady=5)
            Label(frame1, text=s['NDserver']).pack(side=LEFT, padx=10, pady=5)

            frame2 = Frame(frame)
            frame2.pack(expand=YES, fill=BOTH)
            self.newsGruppe = StringVar()
            Entry(frame2, width=40, textvariable=self.newsGruppe).pack(side=LEFT, anchor=W, padx=10, pady=5)
            Label(frame2, text=s['NDgruppe']).pack(side=LEFT, padx=10, pady=5)

            frame3 = Frame(frame)
            frame3.pack(expand=YES, fill=BOTH)
            self.newsAnzahl = StringVar()
            self.newsAnzahl.set('50')
            Spinbox(frame3, width=5, textvariable=self.newsAnzahl, values=(1,10,25,50,100,250,500,750,1000)).pack(side=LEFT, anchor=W, padx=10, pady=5)
            Label(frame3, text=s['NDanzahl']).pack(side=LEFT)
            self.newsZitate = IntVar()
            self.newsZitate.set(1)
            Checkbutton(frame3, text=s['NDzitWeg'], variable=self.newsZitate).pack(side=LEFT, padx=20)

            Button(frame, text=s['NDok'], width=10, command=self.newsgroup_hinzu).pack(side=LEFT, padx=10, pady=10)
            Button(frame, text=s['NDcancel'], width=10, command=self.nd.destroy).pack(side=RIGHT, padx=10, pady=10)
            self.nd.focus_set()


    def QueryDialog(self):
        try:
            self.qd.focus_set()
        except:
            self.qd = Toplevel(self.root)
            self.qd.title("TextSTAT QueryEditor")
            frame = Frame(self.qd)
            frame.pack(expand=YES, fill=BOTH, padx=10, pady=10)
            Label(frame, justify=LEFT, text=s['QDinfo'], fg='blue').pack(anchor=W, padx=5, pady=15)

            frame1 = LabelFrame(frame, text=s['QDsuchFelder'])
            frame1.pack(fill=BOTH, expand=YES)
            frame1a = Frame(frame1)
            frame1a.pack(expand=YES, fill=BOTH)
            self.qAusdruck1 = Entry(frame1a)
            self.qAusdruck1.pack(side=LEFT, anchor=W, padx=5, pady=5)
            Label(frame1a, text=s['QAusdruck1']).pack(side=LEFT, padx=5, pady=5)
            frame1b = Frame(frame1)
            frame1b.pack(expand=YES, fill=BOTH)
            self.qAusdruck2 = Entry(frame1b)
            self.qAusdruck2.pack(side=LEFT, anchor=W, padx=5, pady=5)
            Label(frame1b, text=s['QAusdruck2']).pack(side=LEFT, padx=5, pady=5)

            frame2 = LabelFrame(frame, text=s['QAbstand'])
            frame2.pack(fill=BOTH, expand=YES, pady=30)
            self.qAbstandMin = Spinbox(frame2, width=5, from_=0, to=10)
            self.qAbstandMin.pack(side=LEFT, anchor=W, padx=5, pady=5)
            Label(frame2, text=s['QAbstandMin']).pack(side=LEFT, padx=5, pady=5)
            self.qAbstandMax = Spinbox(frame2, width=5, from_=0, to=10)
            self.qAbstandMax.pack(side=LEFT, anchor=W, padx=5, pady=5)
            Label(frame2, text=s['QAbstandMax']).pack(side=LEFT, padx=5, pady=5)

            Button(frame, text=s['QmachAusdruck'], command=self.query_machen).pack(side=LEFT, padx=10, pady=10)
            Button(frame, text=s['QDcancel'], width=10, command=self.qd.destroy).pack(side=RIGHT, padx=10, pady=10)
            self.qd.bind("<Return>", self.query_machen)
            self.qAusdruck1.focus_set()


    def query_machen(self, event=''):
        a1 = self.qAusdruck1.get()
        a1 = a1.replace('*', '\w*')
        a1 = a1.replace('?', '\w')
        a2 = self.qAusdruck2.get()
        a2 = a2.replace('*', '\w*')
        a2 = a2.replace('?', '\w')
        ab1 = self.qAbstandMin.get()
        ab2 = self.qAbstandMax.get()
        if a1 == '' and a2 == '':
            self.warnung(titel=s['Error'], text=s['ErrQuery'])
            self.qAusdruck1.focus_set()
            return

        if ab1 == '0' and ab2 == '0':
            such = "%s\W+%s" % (a1, a2)
        else:
            such = "%s(\W+\w+){%s,%s}\W+%s" % (a1, ab1, ab2, a2)
        self.busyCursorOn()
        self.suchwort.delete(0, END)
        self.suchwort.insert(END, such)
        self.konkGrossklein.set(1)
        self.konkGanzewoerter.set(1)
        self.zeigKonkordanz()
        self.busyCursorOff()


    def TextDialog(self, text='', titel='TextSTAT'):
        try:
            self.td.focus_set()
        except:
            self.td = Toplevel(self.root)
            self.td.transient(self.root)
            self.td.title(titel)
            self.td.geometry('600x400+50+50')
            frame = Frame(self.td)
            frame.pack(expand=YES, fill=BOTH)

            scrollbar = Scrollbar(frame, orient=VERTICAL)
            txt = Text(frame, borderwidth=1, height=15, wrap=WORD, font=(('Courier', 'New'), '9'), yscrollcommand=scrollbar.set)
            scrollbar.config(command=txt.yview)
            scrollbar.pack(side=RIGHT, fill=Y, padx=0)
            txt.pack(side=LEFT, expand=YES, fill=BOTH)
            Button(self.td, text=' OK ', command=self.td.destroy).pack(pady=10)
            txt.insert(END, text)
            self.td.focus_set()


    def zeigHistory(self):
        f = os.path.join(SkriptDir, 'History.txt')
        text = open(f).read()
        t = s['Titel'] + ' - ' + s['History']
        self.TextDialog(text=text, titel=t)


    def zeigLicense(self):
        f = os.path.join(SkriptDir, 'License.txt')
        text = open(f).read()
        t = s['Titel'] + ' - ' + s['License']
        self.TextDialog(text=text, titel=t)


    # Das Korpus
    # Dictionary von Dictionaries, der Dateipfad dient als Key
    # Format: db[dateipfad] = {'path':f, 'file':ff}
    # ansprechen als z.B.: self.korpus[dateipfad]['file']
    # ACHTUNG: in fruehen Versionen war das eine Liste von Listen [dateiname, datei]

    def korpus_new(self):
        # Wenn anderes Korpus ge鰂fnet ist
        if self.korpus != None and self.korpus.dbGeaendert == 1: 
            jn = tkMessageBox.askquestion(s['KorpusSave'], s['AbfrageSave'])
            if jn == 'yes':
                self.korpus_save()
        try:
            f = tkFileDialog.asksaveasfilename(initialdir=cfg['KorpDir'], defaultextension='.crp', title=s['KorpusNew2'])
            if f:
                # Default-Dir neu setzen
                cfg['KorpDir'] = os.path.dirname(f)
                self.korpus = Korpus.Korpus(self)
                self.korpus.newDb(f)