Python parser and editor for wiki tables

wikitableparser.py

#/usr/bin/python
import re

class BorderedTextList:
	def __init__(self,lst):
		self.lst = lst
	def text(self):
		return "".join([l.text() for l in self.lst])
	def split(self,re):
		return BorderedTextList([l.split(re) for l in self.lst])
	def __repr__(self):
		return "["+(", ".join([str(l) for l in self.lst]))+"]";
	def __getitem__(self, key):
		if isinstance(key,tuple):
			cur = self.lst
			for t in key:
				cur = cur[t]
			return cur
		else:
			return self.lst[key]
	def __iter__(self):
		return iter(self.lst)
	def __len__(self):
		return len(self.lst)

class BorderedText(object):
	def __init__(self,txt,left="",right=""):
		self.txt = txt
		self.left = left
		self.right = right
	def split(self,regex):
		splt = re.split("("+regex+")",self.txt)
		if len(splt)==1:
			return BorderedTextList([BorderedText(self.txt,self.left,self.right)])
		first = BorderedText(splt[0],self.left)
		rest = zip(*[splt[i::2] for i in range(1,3)]) 
		rest = [BorderedText(txt,left) for left,txt in rest]
		rest[-1].right = self.right
		return BorderedTextList([first] + rest)
	def text(self):
		return self.left + self.txt + self.right
	def __repr__(self):
		def nlrep(txt):
			return txt.replace("\n","\\n").replace("\r","\\r")
		return '[left="%s" txt="%s" right="%s"]' % (nlrep(self.left),nlrep(self.txt),nlrep(self.right))

class BorderedTextSpaned(BorderedText):
	def __init__(self,txt,left="",right="",spanParent=None):
		super(BorderedTextSpaned, self).__init__(txt,left,right)
		if spanParent == None:
			self.spanParent = self
		else:
			self.spanParent = spanParent

class WikiTableEditor:
	def __init__(self,txt):
		self.parseTable(txt)
	def parseTable(self,txt):
		self.rows = BorderedText(txt).split("(?:^|\r\n|\n)\|- *").split("(?:(?:^|\r\n|\n)[|!]| *\|\|) *")
		def moveDefinitionToLeft(t):
			l = re.split("(\| *)",t.txt,1)
			if len(l) == 1 or "[" in l[0]:
				return t
			else:
				return BorderedText(l[2],t.left+l[0]+l[1],t.right)
		self.rows.lst = [[moveDefinitionToLeft(f) for f in r] for r in self.rows]
		def fillColspans(inRow):
			outRow = []
			for f in inRow:
				outRow.append(BorderedTextSpaned(f.txt,f.left,f.right))
				m = re.search('colspan="?([0-9]+)',f.left)
				if m:
					for i in range(0,int(m.group(1))-1):
						outRow.append(BorderedTextSpaned("","","",f))
			return BorderedTextList(outRow)
		self.rows.lst = [fillColspans(r) for r in self.rows]
		for row in range(0,len(self.rows)):
			for col in range(0,len(self.rows[row])):
				m = re.search('rowspan="?([0-9]+)',self.rows[row,col].left)
				if m:
					span = int(m.group(1))
					if span > 1:
						for k in range(row+1,min(row+span,len(self.rows))):
							self.rows[k].lst = self.rows[k].lst[0:col] + [BorderedTextSpaned("","","",self.rows[row,col])] + self.rows[k].lst[col:]

	def __repr__(self):
		cw = []
		for r in self.rows:
			if len(cw) < len(r):
				cw += (len(r)-len(cw))*[2]
			for k,v in enumerate(r):
				cl = min(len(v.txt),40)
				if cw[k] < cl:
					cw[k] = cl
		return "\n".join(["I ".join([re.sub("[\n\r]","",f.txt)[:cw[k]].ljust(cw[k]) for k,f in enumerate(row)]) for row in self.rows])
	def __getitem__(self,key):
		e = self.rows[key]
		if isinstance(e,BorderedText):
			return self.rows[key].spanParent.txt
		else:
			return e
	def __setitem__(self,key,txt):
		self.rows[key].spanParent.txt = txt
	def text(self):
		return self.rows.text()
	def __len__(self):
		return len(self.rows)
	def insertRow(self,before,numCols=-1):
		if numCols == -1:
			numCols = max([len(i) for i in self.rows])
		newRow = [BorderedTextSpaned("","\n|-")]
		for i in range(1,numCols):
			newRow.append(BorderedTextSpaned("","\n| "))
		newRows = (self.rows.lst[0:before] +
			[BorderedTextList(newRow)] +
			self.rows.lst[before:])

		self.rows = BorderedTextList(newRows)


if __name__ == "__main__":
	import sys
	txt = open(sys.argv[1],"rb").read().decode("iso-8859-1")
	def printable(txt):
		return txt.replace("\n","\\n").replace("\r","\\r")
	for m in re.findall("\{\|(.*?)\|\}",txt,re.DOTALL):
		e = WikiTableEditor(m)
		print(e)
		for i in sys.argv[2:]:
			row,col = i.split("/")
			row = int(row)
			col = int(col)
			print("row=%d col=%d val='%s' left='%s' right='%s'" % (row,col,e[row,col],printable(e.rows[row,col].left),printable(e.rows[row,col].right)))