lib/lyx2lyx/LyX.py

   1 # This file is part of lyx2lyx
   2 # -*- coding: iso-8859-1 -*-
   3 # Copyright (C) 2002-2004 Dekel Tsur <dekel@lyx.org>, José Matos <jamatos@lyx.org>
   4 #
   5 # This program is free software; you can redistribute it and/or
   6 # modify it under the terms of the GNU General Public License
   7 # as published by the Free Software Foundation; either version 2
   8 # of the License, or (at your option) any later version.
   9 #
  10 # This program is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License
  16 # along with this program; if not, write to the Free Software
  17 # Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
  18
  19 from parser_tools import get_value, check_token, find_token, find_tokens, find_end_of, find_end_of_inset
  20 import os.path
  21 import gzip
  22 import sys
  23 import re
  24 import string
  25
  26 ##
  27 # file format version
  28 #
  29 version = "1.4.0cvs"
  30 default_debug_level = 2
  31 format_re = re.compile(r"(\d)[\.,]?(\d\d)")
  32 fileformat = re.compile(r"\\lyxformat\s*(\S*)")
  33 original_version = re.compile(r"\#LyX (\S*)")
  34
  35 format_relation = [("0_10",  [210], ["0.10.7","0.10"]),
  36                    ("0_12",  [215], ["0.12","0.12.1","0.12"]),
  37                    ("1_0_0", [215], ["1.0.0","1.0"]),
  38                    ("1_0_1", [215], ["1.0.1","1.0.2","1.0.3","1.0.4", "1.1.2","1.1"]),
  39                    ("1_1_4", [215], ["1.1.4","1.1"]),
  40                    ("1_1_5", [216], ["1.1.5","1.1.5fix1","1.1.5fix2","1.1"]),
  41                    ("1_1_6", [217], ["1.1.6","1.1.6fix1","1.1.6fix2","1.1"]),
  42                    ("1_1_6fix3", [218], ["1.1.6fix3","1.1.6fix4","1.1"]),
  43                    ("1_2", [220], ["1.2.0","1.2.1","1.2.3","1.2.4","1.2"]),
  44                    ("1_3", [221], ["1.3.0","1.3.1","1.3.2","1.3.3","1.3.4","1.3.5","1.3"]),
  45                    ("1_4", range(223,238), ["1.4.0cvs","1.4"])]
  46
  47
  48 def formats_list():
  49     formats = []
  50     for version in format_relation:
  51         for format in version[1]:
  52             if format not in formats:
  53                 formats.append(format)
  54     return formats
  55
  56
  57 def get_end_format():
  58     return format_relation[-1][1][-1]
  59
  60
  61 def get_backend(textclass):
  62     if textclass == "linuxdoc" or textclass == "manpage":
  63         return "linuxdoc"
  64     if textclass[:7] == "docbook":
  65         return "docbook"
  66     return "latex"
  67
  68
  69 ##
  70 # Class
  71 #
  72 class LyX_Base:
  73     """This class carries all the information of the LyX file."""
  74     def __init__(self, end_format = 0, input = "", output = "", error = "", debug = default_debug_level):
  75         if input and input != '-':
  76             self.input = self.open(input)
  77         else:
  78             self.input = sys.stdin
  79         if output:
  80             self.output = open(output, "w")
  81         else:
  82             self.output = sys.stdout
  83
  84         if error:
  85             self.err = open(error, "w")
  86         else:
  87             self.err = sys.stderr
  88
  89         self.debug = debug
  90
  91         if end_format:
  92             self.end_format = self.lyxformat(end_format)
  93         else:
  94             self.end_format = get_end_format()
  95
  96         self.backend = "latex"
  97         self.textclass = "article"
  98         self.header = []
  99         self.body = []
 100
 101
 102     def warning(self, message, debug_level= default_debug_level):
 103         if debug_level <= self.debug:
 104             self.err.write(message + "\n")
 105
 106
 107     def error(self, message):
 108         self.warning(message)
 109         self.warning("Quiting.")
 110         sys.exit(1)
 111
 112
 113     def read(self):
 114         """Reads a file into the self.header and self.body parts"""
 115         preamble = 0
 116
 117         while 1:
 118             line = self.input.readline()
 119             if not line:
 120                 self.error("Invalid LyX file.")
 121
 122             line = line[:-1]
 123             # remove '\r' from line's end, if present
 124             if line[-1:] == '\r':
 125                 line = line[:-1]
 126
 127             if check_token(line, '\\begin_preamble'):
 128                 preamble = 1
 129             if check_token(line, '\\end_preamble'):
 130                 preamble = 0
 131
 132             if not preamble:
 133                 line = string.strip(line)
 134
 135             if not line and not preamble:
 136                 break
 137
 138             self.header.append(line)
 139
 140         while 1:
 141             line = self.input.readline()
 142             if not line:
 143                 break
 144             # remove '\r' from line's end, if present
 145             if line[-2:-1] == '\r':
 146                 self.body.append(line[:-2])
 147             else:
 148                 self.body.append(line[:-1])
 149
 150         self.textclass = get_value(self.header, "\\textclass", 0)
 151         self.backend = get_backend(self.textclass)
 152         self.format  = self.read_format()
 153         self.language = get_value(self.header, "\\language", 0)
 154         if self.language == "":
 155             self.language = "english"
 156         self.initial_version = self.read_version()
 157
 158
 159     def write(self):
 160         self.set_version()
 161         self.set_format()
 162
 163         for line in self.header:
 164             self.output.write(line+"\n")
 165         self.output.write("\n")
 166         for line in self.body:
 167             self.output.write(line+"\n")
 168
 169
 170     def open(self, file):
 171         """Transparently deals with compressed files."""
 172
 173         self.dir = os.path.dirname(os.path.abspath(file))
 174         try:
 175             gzip.open(file).readline()
 176             self.output = gzip.GzipFile("","wb",6,self.output)
 177             return gzip.open(file)
 178         except:
 179             return open(file)
 180
 181
 182     def lyxformat(self, format):
 183         result = format_re.match(format)
 184         if result:
 185             format = int(result.group(1) + result.group(2))
 186         else:
 187             self.error(str(format) + ": " + "Invalid LyX file.")
 188
 189         if format in formats_list():
 190             return format
 191
 192         self.error(str(format) + ": " + "Format not supported.")
 193         return None
 194
 195
 196     def read_version(self):
 197         for line in self.header:
 198             if line[0] != "#":
 199                 return None
 200
 201             result = original_version.match(line)
 202             if result:
 203                 return result.group(1)
 204         return None
 205
 206
 207     def set_version(self):
 208         self.header[0] = "#LyX %s created this file. For more info see http://www.lyx.org/" % version
 209         if self.header[1][0] == '#':
 210             del self.header[1]
 211
 212
 213     def read_format(self):
 214         for line in self.header:
 215             result = fileformat.match(line)
 216             if result:
 217                 return self.lyxformat(result.group(1))
 218         else:
 219             self.error("Invalid LyX File.")
 220         return None
 221
 222
 223     def set_format(self):
 224         if self.format <= 217:
 225             format = str(float(format)/100)
 226         else:
 227             format = str(self.format)
 228         i = find_token(self.header, "\\lyxformat", 0)
 229         self.header[i] = "\\lyxformat %s" % format
 230
 231
 232     def set_parameter(self, param, value):
 233         i = find_token(self.header, '\\' + param, 0)
 234         if i == -1:
 235             self.warning(3, 'Parameter not found in the header: %s' % param)
 236             return
 237         self.header[i] = '\\%s %s' % (param, str(value))
 238
 239
 240     def convert(self):
 241         "Convert from old to new format."
 242         mode, convertion_chain = self.chain()
 243         self.warning("convertion chain: " + str(convertion_chain), 3)
 244
 245         for step in convertion_chain:
 246             convert_step = getattr(__import__("lyx_" + step), mode)
 247             convert_step(self)
 248
 249
 250     def chain(self):
 251         """ This is where all the decisions related with the convertion are taken"""
 252
 253         self.start =  self.format
 254         format = self.format
 255         correct_version = 0
 256
 257         for rel in format_relation:
 258             if self.initial_version in rel[2]:
 259                 if format in rel[1]:
 260                     initial_step = rel[0]
 261                     correct_version = 1
 262                     break
 263
 264         if not correct_version:
 265             if format <= 215:
 266                 self.warning("Version does not match file format, discarding it.")
 267             for rel in format_relation:
 268                 if format in rel[1]:
 269                     initial_step = rel[0]
 270                     break
 271             else:
 272                 # This should not happen, really.
 273                 self.error("Format not supported.")
 274
 275         # Find the final step
 276         for rel in format_relation:
 277             if self.end_format in rel[1]:
 278                 final_step = rel[0]
 279                 break
 280         else:
 281             self.error("Format not supported.")
 282
 283         # Convertion mode, back or forth
 284         steps = []
 285         if (initial_step, self.start) < (final_step, self.end_format):
 286             mode = "convert"
 287             first_step = 1
 288             for step in format_relation:
 289                 if  initial_step <= step[0] <= final_step:
 290                     if first_step and len(step[1]) == 1:
 291                         first_step = 0
 292                         continue
 293                     steps.append(step[0])
 294         else:
 295             mode = "revert"
 296             relation_format = format_relation
 297             relation_format.reverse()
 298             last_step = None
 299
 300             for step in relation_format:
 301                 if  final_step <= step[0] <= initial_step:
 302                     steps.append(step[0])
 303                     last_step = step
 304
 305             if last_step[1][-1] == self.end_format:
 306                 steps.pop()
 307
 308         return mode, steps
 309
 310
 311     def get_toc(self, depth = 4):
 312         " Returns the TOC of a lyx document."
 313         paragraphs_filter = {'Title' : 0,'Chapter' : 1, 'Section' : 2, 'Subsection' : 3, 'Subsubsection': 4}
 314         allowed_insets = ['Quotes']
 315
 316         sections = []
 317         for section in paragraphs_filter.keys():
 318             sections.append('\\begin_layout %s' % section)
 319
 320         toc_par = []
 321         i = 0
 322         while 1:
 323             i = find_tokens(self.body, sections, i)
 324             if i == -1:
 325                 break
 326
 327             j = find_end_of(self.body,  i + 1, '\\begin_layout', '\\end_layout')
 328             if j == -1:
 329                 self.warning('Incomplete file.', 0)
 330                 break
 331
 332             section = string.split(self.body[i])[1]
 333             if section[-1] == '*':
 334                 section = section[:-1]
 335
 336             par = []
 337
 338             k = i + 1
 339             # skip paragraph parameters
 340             while not self.body[k] or self.body[k][0] == '\\':
 341                 k = k +1
 342
 343             while k < j:
 344                 if check_token(self.body[k], '\\begin_inset'):
 345                     inset = string.split(self.body[k])[1]
 346                     end = find_end_of_inset(self.body, k)
 347                     if end == -1 or end > j:
 348                         self.warning('Malformed file.', 0)
 349
 350                     if inset in allowed_insets:
 351                         par.extend(self.body[k: end+1])
 352                     k = end + 1
 353                 else:
 354                     par.append(self.body[k])
 355                     k = k + 1
 356
 357             # trim empty lines in the end.
 358             while string.strip(par[-1]) == '' and par:
 359                 par.pop()
 360
 361             toc_par.append(Paragraph(section, par))
 362
 363             i = j + 1
 364
 365         return toc_par
 366
 367
 368 class File(LyX_Base):
 369     def __init__(self, end_format = 0, input = "", output = "", error = "", debug = default_debug_level):
 370         LyX_Base.__init__(self, end_format, input, output, error, debug)
 371         self.read()
 372
 373
 374 class NewFile(LyX_Base):
 375     def set_header(self, **params):
 376         # set default values
 377         self.header.extend([
 378             "#LyX xxxx created this file. For more info see http://www.lyx.org/",
 379             "\\lyxformat xxx",
 380             "\\begin_document",
 381             "\\begin_header",
 382             "\\textclass article",
 383             "\\language english",
 384             "\\inputencoding auto",
 385             "\\fontscheme default",
 386             "\\graphics default",
 387             "\\paperfontsize default",
 388             "\\papersize default",
 389             "\\paperpackage none",
 390             "\\use_geometry false",
 391             "\\use_amsmath 1",
 392             "\\cite_engine basic",
 393             "\\use_bibtopic false",
 394             "\\paperorientation portrait",
 395             "\\secnumdepth 3",
 396             "\\tocdepth 3",
 397             "\\paragraph_separation indent",
 398             "\\defskip medskip",
 399             "\\quotes_language english",
 400             "\\quotes_times 2",
 401             "\\papercolumns 1",
 402             "\\papersides 1",
 403             "\\paperpagestyle default",
 404             "\\tracking_changes false",
 405             "\\end_header"])
 406
 407         self.format = get_end_format()
 408         for param in params:
 409             self.set_parameter(param, params[param])
 410
 411
 412     def set_body(self, paragraphs):
 413         self.body.extend(['\\begin_body',''])
 414
 415         for par in paragraphs:
 416             self.body.extend(par.asLines())
 417
 418         self.body.extend(['','\\end_body', '\\end_document'])
 419
 420
 421 class Paragraph:
 422     def __init__(self, name, body=[], settings = [], child = []):
 423         self.name = name
 424         self.body = body
 425         self.settings = settings
 426         self.child = child
 427
 428     def asLines(self):
 429         result = ['','\\begin_layout %s' % self.name]
 430         result.extend(self.settings)
 431         result.append('')
 432         result.extend(self.body)
 433         result.append('\\end_layout')
 434
 435         if not self.child:
 436             return result
 437
 438         result.append('\\begin_deeper')
 439         for node in self.child:
 440             result.extend(node.asLines())
 441         result.append('\\end_deeper')
 442
 443         return result