# This file is part of lyx2lyx # -*- coding: iso-8859-1 -*- # Copyright (C) 2002-2004 Dekel Tsur , José Matos # # This program is free software; you can redistribute it and/or # modify it under the terms of the GNU General Public License # as published by the Free Software Foundation; either version 2 # of the License, or (at your option) any later version. # # This program is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with this program; if not, write to the Free Software # Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. import string def check_token(line, token): if line[:len(token)] == token: return 1 return 0 # We need to check that the char after the token is space, but I think # we can ignore this def find_token(lines, token, start, end = 0): if end == 0: end = len(lines) m = len(token) for i in xrange(start, end): if lines[i][:m] == token: return i return -1 def find_token2(lines, token, start, end = 0): if end == 0: end = len(lines) for i in xrange(start, end): x = string.split(lines[i]) if len(x) > 0 and x[0] == token: return i return -1 def find_tokens(lines, tokens, start, end = 0): if end == 0: end = len(lines) for i in xrange(start, end): line = lines[i] for token in tokens: if line[:len(token)] == token: return i return -1 def find_re(lines, rexp, start, end = 0): if end == 0: end = len(lines) for i in xrange(start, end): if rexp.match(lines[i]): return i return -1 def find_token_backwards(lines, token, start): m = len(token) for i in xrange(start, -1, -1): line = lines[i] if line[:m] == token: return i return -1 def find_tokens_backwards(lines, tokens, start): for i in xrange(start, -1, -1): line = lines[i] for token in tokens: if line[:len(token)] == token: return i return -1 def get_value(lines, token, start, end = 0): i = find_token2(lines, token, start, end) if i == -1: return "" if len(string.split(lines[i])) > 1: return string.split(lines[i])[1] else: return "" def del_token(lines, token, i, j): k = find_token2(lines, token, i, j) if k == -1: return j else: del lines[k] return j-1 # Finds the paragraph that contains line i. def get_paragraph(lines, i): while i != -1: i = find_tokens_backwards(lines, ["\\end_inset", "\\layout"], i) if i == -1: return -1 if check_token(lines[i], "\\layout"): return i i = find_beginning_of_inset(lines, i) # Finds the paragraph after the paragraph that contains line i. def get_next_paragraph(lines, i): while i != -1: i = find_tokens(lines, ["\\begin_inset", "\\layout", "\\end_float", "\\the_end"], i) if not check_token(lines[i], "\\begin_inset"): return i i = find_end_of_inset(lines, i) def find_end_of(lines, i, start_token, end_token): count = 1 n = len(lines) while i < n: i = find_tokens(lines, [end_token, start_token], i+1) if check_token(lines[i], start_token): count = count+1 else: count = count-1 if count == 0: return i return -1 # Finds the matching \end_inset def find_beginning_of(lines, i, start_token, end_token): count = 1 while i > 0: i = find_tokens_backwards(lines, [start_token, end_token], i-1) if check_token(lines[i], end_token): count = count+1 else: count = count-1 if count == 0: return i return -1 # Finds the matching \end_inset def find_end_of_inset(lines, i): return find_end_of(lines, i, "\\begin_inset", "\\end_inset") # Finds the matching \end_inset def find_beginning_of_inset(lines, i): return find_beginning_of(lines, i, "\\begin_inset", "\\end_inset") def find_end_of_tabular(lines, i): return find_end_of(lines, i, "