Using Python Regular Expressions with StyledTextCtrl

   1 ## Demo program to illustrate the interactive application of python regular
   2 #  expressions to utf-8 contents of a StyledTextCtrl in such a way that
   3 #  minimumum disturbance is caused to other data in the control and undo/redo
   4 #  can be used normally.
   5 
   6 # Author: Robert Ledger (ledgerbob at gmail.com) 2006.
   7 
   8 # Last Modification: 11 July 2006
   9 
  10 # 11 July 2006: Added test for zero length source text to avoid crashing.
  11 #               Added extra style and size information for GNU/Linux.
  12 
  13 # Program tested on:
  14 #    windows xp (python2.4 wx2.3.6)
  15 #    Debian GNU/Linux 3.1 (python2.3 wx2.3.6)    
  16 
  17 # Please be aware, I am not a professional programmer and nothing in
  18 # this code is guaranteed not to munch your vital data.
  19 
  20 import wx, re
  21 import wx.stc as stc
  22 
  23 MainEditor = None
  24 
  25 # The following functions are general purpose functions for regular
  26 # expression manipulation of the contents of StyledTextCtrls using python
  27 # regular expression syntax.
  28 #
  29 # They are independent of the SearchPanel gui and can be cut out and put into
  30 # their own module or included in a utility module if desired.
  31 #
  32 # All changes made by these functions can be undone by the STC's undo/redo
  33 # commands.
  34 #
  35 # In all the following, parameter sFind should be a regular expression object
  36 # or of type unicode.  The sReplace parameters are expected to be of type
  37 # unicode.  Where a unicode object is expected, on object of type string will do
  38 # as long as it only contains ascii characters.
  39 #
  40 
  41 def _norm(sFind, pos, endpos, ed):
  42 
  43     """
  44     Return (text, pos, endpos) where text is a string containing the
  45     text from pos to endpos in ed after real values have been substituted
  46     for default values in these variables.
  47     
  48     If sFind is not of type str or unicode it is assumed to be a
  49     regular expression object otherwise it is compiled into a regular
  50     expression object.
  51     
  52     
  53     """
  54     endtext = ed.GetTextLength()
  55 
  56     if pos is None:
  57         startSel, endSel = ed.GetSelection()
  58         pos = endSel
  59 
  60     if endpos is None:
  61         endpos = endtext
  62 
  63     if pos > endpos:
  64         pos, endpos = endpos, pos
  65 
  66     text = ed.GetTextRange(pos, endpos)
  67 
  68     if isinstance(sFind, (str, unicode)):
  69         sFind = re.compile(sFind)
  70 
  71     if pos == endpos:
  72         return '', sFind, 0, 0
  73 
  74     return text, sFind, pos, endpos
  75 
  76 
  77 def Search(sFind, pos, endpos, ed):
  78     """
  79     Search for re sFind from integer pos to integer endpos in STC ed.
  80     
  81     sFind can be a str, unicode or regular expression object. If type
  82     str then it must only contain ascii data.
  83 
  84     Returns a tuple (match, start, end), where match is an re match
  85     object, start and end are positions of the found text in the
  86     target STC.
  87         
  88     """
  89     text, rFind, pos, endpos = _norm(sFind, pos, endpos, ed)
  90     m = None
  91     if len(text):
  92     	m = rFind.search(text)
  93     if not m:
  94         return None, 0, 0
  95     uStart = pos + len(text[:m.start(0)].encode('utf8'))
  96     uLen = len(m.group(0).encode('utf8'))
  97     return m, uStart, uStart + uLen
  98 
  99 
 100 def SearchList(sFind, pos, endpos, ed):
 101     """
 102     Search the region of STC between pos and endpos for re sFind and
 103     return a list of tuples representing each find.
 104     
 105     sFind can be a str, unicode or regular expression object. If type
 106     str then it must only contain ascii data.
 107     
 108     The returned tuples are in the form (match, start, end) where match
 109     is an re match object and start and end are the start and end of the
 110     matched region in the target STC.
 111 
 112     """
 113     text, rFind, pos, endpos = _norm(sFind, pos, endpos, ed)
 114 
 115     ptr = 0
 116     oldend = 0
 117 
 118     lst=[]
 119     if len(text) == 0:
 120 	return lst
 121     for m in rFind.finditer(text):
 122         start, end = m.span(0)
 123         ofs=len(text[oldend:start].encode('utf8'))
 124         newstart=ptr+ofs
 125         uLen=len(m.group(0).encode('utf8'))
 126         ptr = newstart+uLen
 127         oldend = start + len(m.group(0))
 128         lst.append((m, pos + newstart, pos + ptr))
 129 
 130     return lst
 131 
 132 
 133 def ReplaceList(sReplace, lst, ed):
 134     """
 135     Takes a list of tuples as produced by SearchList and replaces the
 136     the region specified by the tuple with unicode string sReplace after
 137     re transformations.
 138 
 139     The list of regions must be ordered from the start of the document
 140     to the end with no overlapping regions. The regions are replaced
 141     in reverse order, (because once text is inserted all pointers following
 142     the insertion point will be invalid!)
 143 
 144     All the changes made to the STC by this method can be undone or
 145     redone with a single undo/redo operation on the STC.
 146 
 147     """
 148     lst = list(lst)
 149     lst.reverse()
 150     ed.BeginUndoAction()
 151     for m, start, end in lst:
 152         ed.SetTargetStart(start)
 153         ed.SetTargetEnd(end)
 154         ed.ReplaceTarget(m.expand(sReplace))
 155     ed.EndUndoAction()
 156 
 157 
 158 def ReverseSearch(sFind, pos, endpos, ed):
 159     """
 160     Find all matches of re sFind in the STC (ed) region specified by
 161     pos and endpos and return a tuple (match, pos, endpos) for the
 162     LAST match.
 163 
 164     """
 165     lst = SearchList(sFind, pos, endpos, ed)
 166     if len(lst)<1:
 167         return None, 0, 0
 168     return lst[-1]
 169 
 170 
 171 def ReplaceAll(sFind, sReplace, startpos, endpos, ed):
 172     """
 173     Replace all occurences of re sFind between startpos and endpos
 174     with re replacement expression sReplace after normal re transformations.
 175 
 176     """
 177     lst = self.SearchList(sFind, startpos, endpos, ed)
 178     ReplaceList( sReplace, lst, ed)
 179 
 180 
 181 def SearchReplace(sFind, sReplace, pos, endpos, ed):
 182     """
 183     Search the text in an STC between integer pos and integer endpos
 184     for re sFind and replace the matched text with sReplace after
 185     re transformation.
 186 
 187     Returns tuple (reMatchOject, start, end) where start and end represent
 188     positions of the replaced text in the STC ed.
 189 
 190     """
 191     m, uStart, uEnd = Search(sFind, pos, endpos, ed)
 192 
 193     ed.SetTargetStart(uStart)
 194     ed.SetTargetEnd(uEnd)
 195     ed.ReplaceTarget(m.expand(sReplace))
 196 
 197     return m, uStart, uEnd
 198 
 199 #
 200 #  End of independant functions for external use
 201 #
 202 
 203 class MessageButton(wx.Button):
 204     """
 205     A Button that invokes an OnMessageButton method in its
 206     parent with a user defined message as a parameter when it
 207     is clicked.
 208     """
 209 
 210     def __init__(self, parent, label, message=None):
 211         """
 212         Construct a MessageButton with an initial message
 213         set to label if no message is supplied
 214 
 215         """
 216         wx.Button.__init__(self, parent, -1, label)
 217         self.myparent = parent
 218         if message is None:
 219             message = label
 220         self.message = message
 221         self.Bind(wx.EVT_BUTTON, self.OnButton)
 222 
 223     def OnButton(self, event):
 224         self.myparent.OnMessageButton(self.message, self)
 225 
 226 def Alert( message ):
 227     dialog = wx.MessageDialog(None, message, style = wx.ICON_EXCLAMATION | wx.OK)
 228     dialog.ShowModal()
 229     dialog.Destroy()
 230 
 231 
 232 class MySTC(stc.StyledTextCtrl):
 233     """
 234     Derive a class from StyledTextCtrl and add some useful methods
 235     and initiializations.
 236     
 237     """
 238     def __init__(self, parent, size=wx.DefaultSize, style=0):
 239         """
 240         Default Constructor
 241 
 242         """
 243         stc.StyledTextCtrl.__init__(self, parent, size=size, style=style)
 244 
 245         self.SetCaretLineBack('yellow')
 246         self.SetCaretLineVisible(True)
 247 
 248     def CenterPosInView(self, pos):
 249         """
 250         Given a position in the control, center the line that
 251         contins that position in the view.
 252         
 253         """
 254         line = self.LineFromPosition(pos)
 255         self.CenterLineInView(line)
 256         self.GotoLine(line)
 257         self.GotoPos(pos)
 258 
 259     def CenterLineInView(self, line):
 260         """
 261         Given a line in the control, center that line
 262         in the view.
 263         
 264         """
 265         nlines = self.LinesOnScreen()
 266         first=self.GetFirstVisibleLine()
 267         target = first + nlines/2
 268         self.LineScroll(0, line - target)
 269 
 270 
 271 class SearchEditBox(MySTC):
 272     """
 273     Subclass the edit control to provide  additions
 274     and initializations to make it useful as an input text box.
 275     
 276     """
 277     def __init__(self, parent):
 278         """ Default constructor """
 279         MySTC.__init__(self, parent, style=wx.SIMPLE_BORDER)
 280         self.SetUseHorizontalScrollBar(False)
 281         self.SetUseVerticalScrollBar(False)
 282 
 283 
 284 class SearchPanel(wx.Panel):
 285     """
 286     Class to privide a way to interactivly apply Python Regular Expressions
 287     to an STC in such a way that other data in the control is disturbed
 288     as little as possible and changes can be done and undone using
 289     normal STC commands.
 290     
 291     In addition to interactive controls, methods are supplied that can
 292     be used by external scripts.
 293     
 294     """
 295 
 296     FROMTOP = 0
 297     FORWARD = 1
 298     BACKWARD = 2
 299     WRAP = 3
 300     INSELECTION = 4
 301 
 302 
 303     def __init__(self, parent):
 304         """
 305         Default constructor. Sets up the gui componets of the panel.
 306         
 307         lastMatchObject:
 308             A place to store the match object returned from a python re
 309             search.
 310 
 311         """
 312         wx.Panel.__init__(self, parent, -1)
 313         
 314         self.lastMatchObject = None
 315    
 316         mainVbox = wx.BoxSizer(wx.VERTICAL)
 317     
 318         hbox=wx.BoxSizer(wx.HORIZONTAL)
 319     
 320         editBoxSizer =wx.BoxSizer(wx.VERTICAL)
 321         self.FindText = ed1 = SearchEditBox(self)
 322         self.ReplaceText = ed2 = SearchEditBox(self)
 323         editBoxSizer.Add(ed1,1,wx.EXPAND)
 324         editBoxSizer.Add(ed2,1,wx.EXPAND)
 325     
 326         hbox.Add(editBoxSizer,1,wx.EXPAND | wx.ALL, 5)
 327      
 328         rboxsizer=wx.BoxSizer(wx.VERTICAL)
 329         self.directionRadioBox = wx.RadioBox(self,-1,
 330             label='Direction',choices=['From Top','Forward','Backward','Wrap','In Selection'],
 331             style= wx.RA_SPECIFY_ROWS
 332         )
 333         rboxsizer.Add(self.directionRadioBox,1,wx.EXPAND | wx.ALL, 5)
 334         hbox.Add(rboxsizer,0,wx.EXPAND )
 335 
 336         buttonBoxSizer=wx.BoxSizer(wx.VERTICAL)
 337         for label, message in [
 338             ('Search', 'Search'),
 339             ('Replace', 'Replace'),
 340             ('R && S', 'ReplaceAndSearch',),
 341             ('Replace All','ReplaceAll'),
 342             ('Count', 'Count'),
 343             ('List', 'List')
 344         ]:
 345             buttonBoxSizer.Add(MessageButton(self, label, message),1,wx.EXPAND)
 346             
 347         hbox.Add(buttonBoxSizer,0,wx.EXPAND | wx.ALL, 5)
 348         mainVbox.Add(hbox,1,wx.EXPAND)
 349     
 350         self.SetSizer(mainVbox)
 351         mainVbox.Fit(self)
 352 
 353 
 354     def OnMessageButton(self, message, object):
 355         """
 356         Callback function for MessageButtons.
 357         
 358         Initialize some variables from the gui then call a method
 359         to perform the operation requested by the user.
 360         
 361         """
 362         self.direction = self.GetDirection()
 363         self.sFind = self.GetFindStr()
 364         self.sReplace = self.GetReplaceStr()
 365         f = getattr(self, 'On'+message, self.OnUnknownSignal)
 366         self.ed = MainEditor
 367         self.ed.SetFocus()
 368         return f()
 369 
 370 
 371     def GetDirection(self):
 372         """
 373         Get the 'search direction' value from the gui component.
 374 
 375         """
 376         return self.directionRadioBox.GetSelection()
 377 
 378 
 379     def SetDirection(self, i):
 380         """
 381         Set value for 'search direction' gui component.
 382 
 383         """
 384         self.directionRadioBox.SetSelection(i)
 385 
 386 
 387     def GetFindStr(self):
 388         """
 389         Get the search string from the 'find' text box and escape
 390         all unicode chars, then uncescape escaped unicode chars.
 391         
 392         This allows a mixture of unicode charachters and escaped
 393         unicode charachters of the form \u00A3 to be entered in the
 394         imput box.
 395 
 396         """
 397         s = self.FindText.GetText().encode('raw_unicode_escape')
 398         s = s.decode('raw_unicode_escape')
 399         return s
 400 
 401 
 402     def GetReplaceStr(self):
 403         """
 404         Get the replace string from the 'replace' text box and convert
 405         it to python internal representation.
 406         
 407         Unicode escapes of the form \u00A3 can be used.
 408         
 409         """
 410         s = self.ReplaceText.GetText().encode('raw_unicode_escape')
 411         s = s.decode('raw_unicode_escape')
 412         return s
 413 
 414 
 415     def GetSearchRegion(self):
 416         """
 417         Returns the (start, end) of the region, in the target editor,
 418         to be searched.  Result depends on the state of the gui controls.
 419         
 420         A value of None is returned for end to  indicate that search
 421         should go to the end of the text in the control.
 422 
 423         If the search direction is FORWARD or WRAP, the search region will
 424         begin after any current selection or at the current cursor position
 425         if there is no selection.
 426         
 427         """
 428         startSel, endSel = self.ed.GetSelection()
 429         
 430         if self.direction == self.INSELECTION:
 431             return startSel, endSel
 432             
 433         if self.direction == self.FORWARD or self.direction == self.WRAP:
 434         
 435             if startSel != endSel:
 436                 startSel = endSel
 437                 
 438             return startSel, None
 439             
 440         if self.direction == self.BACKWARD:
 441             return 0, startSel
 442             
 443         if self.direction == self.FROMTOP:
 444                 return 0, None
 445                 
 446 
 447     def OnSearch(self):
 448         """
 449         Method invoked by 'Search' button in gui. Searches forward from
 450         the current position or, if text is selected, from just after the
 451         end of the selected text.
 452         
 453         """
 454         try:
 455             rFind = re.compile(self.sFind)
 456         except:
 457             Alert('Error in regular expression.')
 458             return False
 459             
 460         self.lastMatchObject=None
 461 
 462         startSel, endSel = self.GetSearchRegion()
 463 
 464         if self.direction == self.BACKWARD:
 465             m, uStart, uEnd = ReverseSearch(rFind, startSel, endSel, ed=self.ed)
 466             if not m:
 467                 Alert('No match found')
 468                 return
 469 
 470         else:
 471     
 472             if self.direction == self.FROMTOP:
 473                 self.SetDirection(self.FORWARD)
 474     
 475             m, uStart, uEnd = Search(rFind, startSel, endSel, ed=self.ed)
 476             if not m:
 477                 Alert('No match found')
 478                 if not self.direction == self.WRAP:
 479                     return False
 480                 text = self.ed.GetTextRange(0,startSel)
 481                 pos, endpos = 0, startSel
 482                 m, uStart, uEnd = Search(rFind, pos, endpos, ed=self.ed)
 483                 if not m:
 484                     Alert('No match found after wrapping')
 485                     return
 486                 
 487         self.lastMatchObject=m
 488         self.ed.CenterPosInView(uStart)
 489         self.ed.SetSelection(uStart, uEnd)
 490 
 491 
 492     def OnReplace(self):
 493         """
 494         Method invoked by 'Replace' button in gui.
 495         
 496         Replaces the text found by a previous press of the search buton
 497         with the contents of the sReplace text box after re substitutions.
 498         
 499         """
 500         startSel, endSel = self.ed.GetSelection()
 501         if startSel == endSel:
 502             Alert('Can\'t replace text -non selected')
 503             return
 504         m = self.lastMatchObject
 505         if m is None:
 506             Alert('Can\'t replace text!\nDo a search first')
 507             return
 508         selectedText=self.ed.GetSelectedText()
 509         matchedText = m.group(0)
 510         if matchedText != selectedText:
 511             self.lastMatchObject = None
 512             Alert('Can\'t replace text.\nSelected text does not match last found text ')
 513             return
 514         newText = m.expand(self.sReplace)
 515         self.ed.ReplaceSelection(newText)
 516         self.lastMatchObject = None
 517 
 518 
 519     def OnReplaceAndSearch(self):
 520         """
 521         Method invoked by 'Replace & Search' button in gui.
 522         
 523         Equivelent of pressing the Search button followed by
 524         pressing the Replace button.
 525         
 526         """
 527         self.OnReplace()
 528         self.OnSearch()
 529 
 530 
 531     def OnReplaceAll(self):
 532         """
 533         Method invoked by 'Replace All' button in gui.
 534         
 535         Replaces all text matched by sFind with sReplace after re
 536         substitutions.  Operates on the entire text or on the selected
 537         text if the 'in selection' option is set.
 538         
 539         """
 540         lst = self.OnList(log=False)
 541         n=len(lst)
 542         ReplaceList(self.sReplace, lst, ed=self.ed)
 543         Alert('Made %s replacements'%len(lst))
 544 
 545 
 546     def OnCount(self):
 547         """
 548         Method invoked by 'Count' button in gui.
 549         
 550         Counts the number of matches for sFind in the entire text, or in
 551         the selected text if the 'in selection' option is set.
 552         
 553         """
 554         startSel=self.GetSearchRegion()
 555         count = len(self.OnList(log=False))
 556         Alert( 'Found search string %s times.'%count)
 557 
 558 
 559     def OnList(self, log=True):
 560         """
 561         Method invoked by 'List' button in gui.
 562         
 563         Creates a list of matches and prints a list of line numbers
 564         and found strings. The search is done either on the entire text
 565         or on the selected text if the 'in selection' option is active.
 566         
 567         """
 568         try:
 569             rFind = re.compile(self.sFind)
 570         except:
 571             Alert('Error in regular expression.')
 572             return []
 573     
 574         dir = self.GetDirection()
 575         if dir == self.INSELECTION:
 576             startSel, endSel = self.GetSearchRegion()
 577         else:
 578             startSel, endSel = 0, None
 579         
 580         lst = SearchList(rFind, startSel, endSel, ed=self.ed)
 581         
 582         if log:
 583             for m, uStart, uEnd in lst:
 584                 print '%s: %r'%(
 585                         self.ed.LineFromPosition(uStart),
 586                         self.ed.GetTextRange(uStart,uEnd)
 587                       )
 588                       
 589             Alert('List Search found %s matches.\n\nResults were printed on the console.'%len(lst))
 590         
 591         return lst
 592 
 593     def OnUnknownSignal(self):
 594         pass
 595 
 596 
 597 
 598 class TestFrame(wx.Frame):
 599 
 600     def __init__(self):
 601         global MainEditor
 602         wx.Frame.__init__(self, None, -1, 'Ledgerbob\'s Python Search Demo')
 603 
 604         MainEditor = MySTC(self)
 605         sizer = wx.BoxSizer(wx.VERTICAL)
 606         sizer.Add(MainEditor, 1, wx.EXPAND)
 607         sizer.Add(SearchPanel(self), 0, wx.EXPAND)
 608         self.SetSizer(sizer)
 609 	self.SetSize((400, 400))
 610         
 611 if __name__=="__main__":
 612     app = wx.PySimpleApp()
 613     win = TestFrame()
 614     win.Show(True)
 615     app.MainLoop()