[392] | 1 | #! /usr/bin/env python |
---|
| 2 | # -*- coding: ISO-8859-15 -*- |
---|
| 3 | # |
---|
| 4 | # pkpgcounter : a generic Page Description Language parser |
---|
| 5 | # |
---|
| 6 | # (c) 2003, 2004, 2005, 2006 Jerome Alet <alet@librelogiciel.com> |
---|
| 7 | # This program is free software; you can redistribute it and/or modify |
---|
| 8 | # it under the terms of the GNU General Public License as published by |
---|
| 9 | # the Free Software Foundation; either version 2 of the License, or |
---|
| 10 | # (at your option) any later version. |
---|
| 11 | # |
---|
| 12 | # This program is distributed in the hope that it will be useful, |
---|
| 13 | # but WITHOUT ANY WARRANTY; without even the implied warranty of |
---|
| 14 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
---|
| 15 | # GNU General Public License for more details. |
---|
| 16 | # |
---|
| 17 | # You should have received a copy of the GNU General Public License |
---|
| 18 | # along with this program; if not, write to the Free Software |
---|
| 19 | # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
---|
| 20 | # |
---|
| 21 | # $Id$ |
---|
| 22 | # |
---|
| 23 | |
---|
| 24 | """This modules implements a page counter for PCL3/4/5 documents.""" |
---|
| 25 | |
---|
| 26 | import sys |
---|
| 27 | import os |
---|
| 28 | import mmap |
---|
| 29 | from struct import unpack |
---|
| 30 | |
---|
| 31 | import pdlparser |
---|
| 32 | import pjl |
---|
| 33 | |
---|
| 34 | ESCAPE = chr(27) |
---|
| 35 | |
---|
| 36 | class Parser(pdlparser.PDLParser) : |
---|
| 37 | """A parser for PCL3, PCL4, PCL5 documents.""" |
---|
| 38 | totiffcommand = 'pcl6 -sDEVICE=pdfwrite -dPARANOIDSAFER -dNOPAUSE -dBATCH -dQUIET -sOutputFile=- - | gs -sDEVICE=tiff24nc -dPARANOIDSAFER -dNOPAUSE -dBATCH -dQUIET -r%(dpi)i -sOutputFile="%(fname)s" -' |
---|
| 39 | mediasizes = { # ESC&l####A |
---|
| 40 | 0 : "Default", |
---|
| 41 | 1 : "Executive", |
---|
| 42 | 2 : "Letter", |
---|
| 43 | 3 : "Legal", |
---|
| 44 | 6 : "Ledger", |
---|
| 45 | 25 : "A5", |
---|
| 46 | 26 : "A4", |
---|
| 47 | 27 : "A3", |
---|
| 48 | 45 : "JB5", |
---|
| 49 | 46 : "JB4", |
---|
| 50 | 71 : "HagakiPostcard", |
---|
| 51 | 72 : "OufukuHagakiPostcard", |
---|
| 52 | 80 : "MonarchEnvelope", |
---|
| 53 | 81 : "COM10Envelope", |
---|
| 54 | 90 : "DLEnvelope", |
---|
| 55 | 91 : "C5Envelope", |
---|
| 56 | 100 : "B5Envelope", |
---|
| 57 | 101 : "Custom", |
---|
| 58 | } |
---|
| 59 | |
---|
| 60 | mediasources = { # ESC&l####H |
---|
| 61 | 0 : "Default", |
---|
| 62 | 1 : "Main", |
---|
| 63 | 2 : "Manual", |
---|
| 64 | 3 : "ManualEnvelope", |
---|
| 65 | 4 : "Alternate", |
---|
| 66 | 5 : "OptionalLarge", |
---|
| 67 | 6 : "EnvelopeFeeder", |
---|
| 68 | 7 : "Auto", |
---|
| 69 | 8 : "Tray1", |
---|
| 70 | } |
---|
| 71 | |
---|
| 72 | orientations = { # ESC&l####O |
---|
| 73 | 0 : "Portrait", |
---|
| 74 | 1 : "Landscape", |
---|
| 75 | 2 : "ReversePortrait", |
---|
| 76 | 3 : "ReverseLandscape", |
---|
| 77 | } |
---|
| 78 | |
---|
| 79 | mediatypes = { # ESC&l####M |
---|
| 80 | 0 : "Plain", |
---|
| 81 | 1 : "Bond", |
---|
| 82 | 2 : "Special", |
---|
| 83 | 3 : "Glossy", |
---|
| 84 | 4 : "Transparent", |
---|
| 85 | } |
---|
| 86 | |
---|
| 87 | def isValid(self) : |
---|
| 88 | """Returns True if data is PCL3/4/5, else False.""" |
---|
| 89 | if self.firstblock.startswith("\033E\033") or \ |
---|
| 90 | (self.firstblock.startswith("\033*rbC") and (not self.lastblock[-3:] == "\f\033@")) or \ |
---|
| 91 | self.firstblock.startswith("\033%8\033") or \ |
---|
| 92 | (self.firstblock.find("\033%-12345X") != -1) or \ |
---|
| 93 | (self.firstblock.find("@PJL ENTER LANGUAGE=PCL\012\015\033") != -1) or \ |
---|
| 94 | (self.firstblock.startswith(chr(0xcd)+chr(0xca)) and self.firstblock.find("\033E\033")) : |
---|
| 95 | self.logdebug("DEBUG: Input file is in the PCL3/4/5 format.") |
---|
| 96 | return True |
---|
| 97 | else : |
---|
| 98 | return False |
---|
| 99 | |
---|
| 100 | def setPageDict(self, pages, number, attribute, value) : |
---|
| 101 | """Initializes a page dictionnary.""" |
---|
| 102 | dic = pages.setdefault(number, { "copies" : 1, "mediasource" : "Main", "mediasize" : "Default", "mediatype" : "Plain", "orientation" : "Portrait", "escaped" : "", "duplex": 0}) |
---|
| 103 | dic[attribute] = value |
---|
| 104 | |
---|
| 105 | def endPage(self) : |
---|
| 106 | """Handle the FF marker.""" |
---|
| 107 | self.pagecount += 1 |
---|
| 108 | self.logdebug("ENDPAGE %i" % self.pagecount) |
---|
| 109 | |
---|
| 110 | def escPercent(self) : |
---|
| 111 | """Handles the ESC% sequence.""" |
---|
| 112 | if self.minfile[self.pos : self.pos+7] == r"-12345X" : |
---|
| 113 | self.logdebug("Generic ESCAPE sequence at %08x" % self.pos) |
---|
| 114 | self.pos += 7 |
---|
| 115 | |
---|
| 116 | def handleTag(self, tagtable) : |
---|
| 117 | """Handles tags.""" |
---|
| 118 | tag = ord(self.minfile[self.pos]) |
---|
| 119 | self.pos += 1 |
---|
| 120 | tagtable[tag]() |
---|
| 121 | |
---|
| 122 | def escape(self) : |
---|
| 123 | """Handles the ESC character.""" |
---|
| 124 | self.logdebug("ESCAPE") |
---|
| 125 | self.handleTag(self.esctags) |
---|
| 126 | |
---|
| 127 | def escAmp(self) : |
---|
| 128 | """Handles the ESC& sequence.""" |
---|
| 129 | self.logdebug("AMP") |
---|
| 130 | self.handleTag(self.escamptags) |
---|
| 131 | |
---|
| 132 | def escStar(self) : |
---|
| 133 | """Handles the ESC* sequence.""" |
---|
| 134 | self.logdebug("STAR") |
---|
| 135 | self.handleTag(self.escstartags) |
---|
| 136 | |
---|
| 137 | def escLeftPar(self) : |
---|
| 138 | """Handles the ESC( sequence.""" |
---|
| 139 | self.logdebug("LEFTPAR") |
---|
| 140 | self.handleTag(self.escleftpartags) |
---|
| 141 | |
---|
| 142 | def escRightPar(self) : |
---|
| 143 | """Handles the ESC( sequence.""" |
---|
| 144 | self.logdebug("RIGHTPAR") |
---|
| 145 | self.handleTag(self.escrightpartags) |
---|
| 146 | |
---|
| 147 | def escE(self) : |
---|
| 148 | """Handles the ESCE sequence.""" |
---|
| 149 | self.logdebug("RESET") |
---|
| 150 | self.resets += 1 |
---|
| 151 | |
---|
| 152 | def escAmpl(self) : |
---|
| 153 | """Handles the ESC&l sequence.""" |
---|
| 154 | self.logdebug("l") |
---|
| 155 | while 1 : |
---|
| 156 | (value, end) = self.getInteger() |
---|
| 157 | if value is None : |
---|
| 158 | return |
---|
| 159 | if end in ('h', 'H') : |
---|
| 160 | mediasource = self.mediasources.get(value, str(value)) |
---|
| 161 | self.mediasourcesvalues.append(mediasource) |
---|
| 162 | self.logdebug("MEDIASOURCE %s" % mediasource) |
---|
| 163 | elif end in ('a', 'A') : |
---|
| 164 | mediasize = self.mediasizes.get(value, str(value)) |
---|
| 165 | self.mediasizesvalues.append(mediasize) |
---|
| 166 | self.logdebug("MEDIASIZE %s" % mediasize) |
---|
| 167 | elif end in ('o', 'O') : |
---|
| 168 | orientation = self.orientations.get(value, str(value)) |
---|
| 169 | self.orientationsvalues.append(orientation) |
---|
| 170 | self.logdebug("ORIENTATION %s" % orientation) |
---|
| 171 | elif end in ('m', 'M') : |
---|
| 172 | mediatype = self.mediatypes.get(value, str(value)) |
---|
| 173 | self.mediatypesvalues.append(mediatype) |
---|
| 174 | self.logdebug("MEDIATYPE %s" % mediatype) |
---|
| 175 | elif end == 'X' : |
---|
| 176 | self.copies.append(value) |
---|
| 177 | self.logdebug("COPIES %i" % value) |
---|
| 178 | elif end == 'L' : |
---|
| 179 | self.logdebug("ESC&l%iL" % value) |
---|
| 180 | |
---|
| 181 | def escStarb(self) : |
---|
| 182 | """Handles the ESC*b sequence.""" |
---|
| 183 | self.logdebug("b") |
---|
| 184 | while 1 : |
---|
| 185 | (value, end) = self.getInteger() |
---|
| 186 | self.logdebug("%s === %s" % (value, end)) |
---|
| 187 | if (end is None) and (value is None) : |
---|
| 188 | return |
---|
| 189 | if end in ('V', 'W', 'v', 'w') : |
---|
| 190 | self.pos += (value or 0) |
---|
| 191 | self.logdebug("SKIPTO %08x" % self.pos) |
---|
| 192 | |
---|
| 193 | def escStaro(self) : |
---|
| 194 | """Handles the ESC*o sequence.""" |
---|
| 195 | self.logdebug("o") |
---|
| 196 | while 1 : |
---|
| 197 | (value, end) = self.getInteger() |
---|
| 198 | if value is None : |
---|
| 199 | return |
---|
| 200 | if end == 'M' : |
---|
| 201 | self.logdebug("ESC*o%iM" % value) |
---|
| 202 | |
---|
| 203 | def escStarp(self) : |
---|
| 204 | """Handles the ESC*p sequence.""" |
---|
| 205 | self.logdebug("p") |
---|
| 206 | while 1 : |
---|
| 207 | (value, end) = self.getInteger() |
---|
| 208 | if value is None : |
---|
| 209 | return |
---|
| 210 | if end in ('X', 'Y') : |
---|
| 211 | self.logdebug("ESC*p%i%s" % (value, end)) |
---|
| 212 | |
---|
| 213 | def escStarr(self) : |
---|
| 214 | """Handles the ESC*r sequence.""" |
---|
| 215 | self.logdebug("r") |
---|
| 216 | while 1 : |
---|
| 217 | (value, end) = self.getInteger() |
---|
| 218 | if value is None : |
---|
| 219 | if end is None : |
---|
| 220 | return |
---|
| 221 | elif end == 'b' : |
---|
| 222 | if self.minfile[self.pos] == 'C' : |
---|
| 223 | self.logdebug("Looks like it's PCL3.") |
---|
| 224 | self.ispcl3 = True |
---|
| 225 | self.pos += 1 |
---|
| 226 | elif end == 'C' : |
---|
| 227 | self.logdebug("EndGFX") |
---|
| 228 | if not self.startgfx : |
---|
| 229 | self.logdebug("EndGFX found before StartGFX, ignored.") |
---|
| 230 | else : |
---|
| 231 | self.endgfx.append(1) |
---|
| 232 | if end == 'A' and (0 <= value <= 3) : |
---|
| 233 | self.logdebug("StartGFX %i" % value) |
---|
| 234 | self.startgfx.append(value) |
---|
| 235 | elif end == 'U' : |
---|
| 236 | self.logdebug("ESC*r%iU" % value) |
---|
| 237 | elif end == 'S' : |
---|
| 238 | self.logdebug("ESC*r%iS" % value) |
---|
| 239 | |
---|
| 240 | def escStart(self) : |
---|
| 241 | """Handles the ESC*t sequence.""" |
---|
| 242 | self.logdebug("t") |
---|
| 243 | while 1 : |
---|
| 244 | (value, end) = self.getInteger() |
---|
| 245 | if value is None : |
---|
| 246 | return |
---|
| 247 | if end == 'R' : |
---|
| 248 | self.logdebug("ESC*t%iR" % value) |
---|
| 249 | |
---|
| 250 | def escAmpu(self) : |
---|
| 251 | """Handles the ESC&u sequence.""" |
---|
| 252 | self.logdebug("u") |
---|
| 253 | while 1 : |
---|
| 254 | (value, end) = self.getInteger() |
---|
| 255 | if value is None : |
---|
| 256 | return |
---|
| 257 | if end == 'D' : |
---|
| 258 | self.logdebug("ESC&u%iD" % value) |
---|
| 259 | |
---|
| 260 | |
---|
| 261 | def getInteger(self) : |
---|
| 262 | """Returns an integer value and the end character.""" |
---|
| 263 | sign = 1 |
---|
| 264 | value = None |
---|
| 265 | while 1 : |
---|
| 266 | char = self.minfile[self.pos] |
---|
| 267 | if char == ESCAPE : |
---|
| 268 | return (None, None) |
---|
| 269 | self.pos += 1 |
---|
| 270 | if char == '-' : |
---|
| 271 | sign = -1 |
---|
| 272 | elif not char.isdigit() : |
---|
| 273 | if value is not None : |
---|
| 274 | return (sign*value, char) |
---|
| 275 | else : |
---|
| 276 | return (value, char) |
---|
| 277 | else : |
---|
| 278 | value = ((value or 0) * 10) + int(char) |
---|
| 279 | |
---|
| 280 | def getJobSize(self) : |
---|
| 281 | """Count pages in a PCL5 document. |
---|
| 282 | |
---|
| 283 | Should also work for PCL3 and PCL4 documents. |
---|
| 284 | |
---|
| 285 | Algorithm from pclcount |
---|
| 286 | (c) 2003, by Eduardo Gielamo Oliveira & Rodolfo Broco Manin |
---|
| 287 | published under the terms of the GNU General Public Licence v2. |
---|
| 288 | |
---|
| 289 | Backported from C to Python by Jerome Alet, then enhanced |
---|
| 290 | with more PCL tags detected. I think all the necessary PCL tags |
---|
| 291 | are recognized to correctly handle PCL5 files wrt their number |
---|
| 292 | of pages. The documentation used for this was : |
---|
| 293 | |
---|
| 294 | HP PCL/PJL Reference Set |
---|
| 295 | PCL5 Printer Language Technical Quick Reference Guide |
---|
| 296 | http://h20000.www2.hp.com/bc/docs/support/SupportManual/bpl13205/bpl13205.pdf |
---|
| 297 | """ |
---|
| 298 | infileno = self.infile.fileno() |
---|
| 299 | self.minfile = minfile = mmap.mmap(infileno, os.fstat(infileno)[6], prot=mmap.PROT_READ, flags=mmap.MAP_SHARED) |
---|
| 300 | self.ispcl3 = False |
---|
| 301 | self.pagecount = 0 |
---|
| 302 | self.resets = 0 |
---|
| 303 | self.copies = [] |
---|
| 304 | self.mediasourcesvalues = [] |
---|
| 305 | self.mediasizesvalues = [] |
---|
| 306 | self.orientationsvalues = [] |
---|
| 307 | self.mediatypesvalues = [] |
---|
| 308 | self.startgfx = [] |
---|
| 309 | self.endgfx = [] |
---|
| 310 | |
---|
| 311 | tags = [ lambda : None] * 256 |
---|
| 312 | tags[0x0c] = self.endPage |
---|
| 313 | tags[0x1b] = self.escape |
---|
| 314 | |
---|
| 315 | self.esctags = [ lambda : None ] * 256 |
---|
| 316 | self.esctags[ord('%')] = self.escPercent |
---|
| 317 | self.esctags[ord('*')] = self.escStar |
---|
| 318 | self.esctags[ord('&')] = self.escAmp |
---|
| 319 | self.esctags[ord('(')] = self.escLeftPar |
---|
| 320 | self.esctags[ord(')')] = self.escRightPar |
---|
| 321 | self.esctags[ord('E')] = self.escE |
---|
| 322 | |
---|
| 323 | self.escamptags = [lambda : None ] * 256 |
---|
| 324 | self.escamptags[ord('l')] = self.escAmpl |
---|
| 325 | self.escamptags[ord('u')] = self.escAmpu |
---|
| 326 | |
---|
| 327 | self.escstartags = [ lambda : None ] * 256 |
---|
| 328 | self.escstartags[ord('b')] = self.escStarb |
---|
| 329 | self.escstartags[ord('o')] = self.escStaro |
---|
| 330 | self.escstartags[ord('p')] = self.escStarp |
---|
| 331 | self.escstartags[ord('r')] = self.escStarr |
---|
| 332 | self.escstartags[ord('t')] = self.escStart |
---|
| 333 | |
---|
| 334 | self.pos = 0 |
---|
| 335 | try : |
---|
| 336 | try : |
---|
| 337 | while 1 : |
---|
| 338 | tag = ord(minfile[self.pos]) |
---|
| 339 | self.logdebug("%08x ===> %02x" % (self.pos, tag)) |
---|
| 340 | self.pos += 1 |
---|
| 341 | tags[tag]() |
---|
| 342 | except IndexError : # EOF ? |
---|
| 343 | pass |
---|
| 344 | finally : |
---|
| 345 | self.minfile.close() |
---|
| 346 | |
---|
| 347 | self.logdebug("Pagecount : \t\t%i" % self.pagecount) |
---|
| 348 | self.logdebug("Resets : \t\t%i" % self.resets) |
---|
| 349 | self.logdebug("Copies : \t\t%s" % self.copies) |
---|
| 350 | self.logdebug("MediaTypes : \t\t%s" % self.mediatypesvalues) |
---|
| 351 | self.logdebug("MediaSizes : \t\t%s" % self.mediasizesvalues) |
---|
| 352 | self.logdebug("MediaSources : \t\t%s" % self.mediasourcesvalues) |
---|
| 353 | self.logdebug("Orientations : \t\t%s" % self.orientationsvalues) |
---|
| 354 | self.logdebug("StartGfx : \t\t%s" % len(self.startgfx)) |
---|
| 355 | self.logdebug("EndGfx : \t\t%s" % len(self.endgfx)) |
---|
| 356 | |
---|
| 357 | return self.pagecount |
---|
| 358 | |
---|
| 359 | |
---|
| 360 | |
---|
| 361 | """ |
---|
| 362 | tagsends = { "&n" : "W", |
---|
| 363 | "&b" : "W", |
---|
| 364 | "*i" : "W", |
---|
| 365 | "*l" : "W", |
---|
| 366 | "*m" : "W", |
---|
| 367 | "*v" : "W", |
---|
| 368 | "*c" : "W", |
---|
| 369 | "(f" : "W", |
---|
| 370 | "(s" : "W", |
---|
| 371 | ")s" : "W", |
---|
| 372 | "&p" : "X", |
---|
| 373 | # "&l" : "XHAOM", # treated specially |
---|
| 374 | "&a" : "G", # TODO : 0 means next side, 1 front side, 2 back side |
---|
| 375 | "*g" : "W", |
---|
| 376 | "*r" : "sbABC", |
---|
| 377 | "*t" : "R", |
---|
| 378 | # "*b" : "VW", # treated specially because it occurs very often |
---|
| 379 | } |
---|
| 380 | irmarker = chr(0xcd) + chr(0xca) # Marker for Canon ImageRunner printers |
---|
| 381 | irmarker2 = chr(0x10) + chr(0x02) |
---|
| 382 | wasirmarker = 0 |
---|
| 383 | hasirmarker = (minfile[:2] == (irmarker)) |
---|
| 384 | pagecount = resets = ejects = backsides = startgfx = endgfx = 0 |
---|
| 385 | starb = ampl = ispcl3 = escstart = 0 |
---|
| 386 | mediasourcecount = mediasizecount = orientationcount = mediatypecount = 0 |
---|
| 387 | tag = None |
---|
| 388 | endmark = chr(0x1b) + chr(0x0c) + chr(0x00) |
---|
| 389 | asciilimit = chr(0x80) |
---|
| 390 | pages = {} |
---|
| 391 | pos = 0 |
---|
| 392 | try : |
---|
| 393 | try : |
---|
| 394 | while 1 : |
---|
| 395 | if hasirmarker and (minfile[pos:pos+2] == irmarker) : |
---|
| 396 | codop = minfile[pos+2:pos+4] |
---|
| 397 | # self.logdebug("Marker at 0x%08x (%s)" % (pos, wasirmarker)) |
---|
| 398 | length = unpack(">H", minfile[pos+8:pos+10])[0] |
---|
| 399 | pos += 20 |
---|
| 400 | if codop != irmarker2 : |
---|
| 401 | pos += length |
---|
| 402 | wasirmarker = 1 |
---|
| 403 | else : |
---|
| 404 | wasirmarker = 0 |
---|
| 405 | elif char == "\033" : |
---|
| 406 | starb = ampl = 0 |
---|
| 407 | if minfile[pos : pos+8] == r"%-12345X" : |
---|
| 408 | endpos = pos + 9 |
---|
| 409 | quotes = 0 |
---|
| 410 | while (minfile[endpos] not in endmark) and \ |
---|
| 411 | ((minfile[endpos] < asciilimit) or (quotes % 2)) : |
---|
| 412 | if minfile[endpos] == '"' : |
---|
| 413 | quotes += 1 |
---|
| 414 | endpos += 1 |
---|
| 415 | self.setPageDict(pages, pagecount, "escaped", minfile[pos : endpos]) |
---|
| 416 | pos += (endpos - pos) |
---|
| 417 | else : |
---|
| 418 | # |
---|
| 419 | # <ESC>*b###y#m###v###w... -> PCL3 raster graphics |
---|
| 420 | # <ESC>*b###W -> Start of a raster data row/block |
---|
| 421 | # <ESC>*b###V -> Start of a raster data plane |
---|
| 422 | # <ESC>*c###W -> Start of a user defined pattern |
---|
| 423 | # <ESC>*i###W -> Start of a viewing illuminant block |
---|
| 424 | # <ESC>*l###W -> Start of a color lookup table |
---|
| 425 | # <ESC>*m###W -> Start of a download dither matrix block |
---|
| 426 | # <ESC>*v###W -> Start of a configure image data block |
---|
| 427 | # <ESC>*r1A -> Start Gfx |
---|
| 428 | # <ESC>(s###W -> Start of a characters description block |
---|
| 429 | # <ESC>)s###W -> Start of a fonts description block |
---|
| 430 | # <ESC>(f###W -> Start of a symbol set block |
---|
| 431 | # <ESC>&b###W -> Start of configuration data block |
---|
| 432 | # <ESC>&l###X -> Number of copies for current page |
---|
| 433 | # <ESC>&n###W -> Starts an alphanumeric string ID block |
---|
| 434 | # <ESC>&p###X -> Start of a non printable characters block |
---|
| 435 | # <ESC>&a2G -> Back side when duplex mode as generated by rastertohp |
---|
| 436 | # <ESC>*g###W -> Needed for planes in PCL3 output |
---|
| 437 | # <ESC>&l###H (or only 0 ?) -> Eject if NumPlanes > 1, as generated by rastertohp. Also defines mediasource |
---|
| 438 | # <ESC>&l###A -> mediasize |
---|
| 439 | # <ESC>&l###O -> orientation |
---|
| 440 | # <ESC>&l###M -> mediatype |
---|
| 441 | # <ESC>*t###R -> gfx resolution |
---|
| 442 | # |
---|
| 443 | tagstart = minfile[pos] ; pos += 1 |
---|
| 444 | if tagstart in "E9=YZ" : # one byte PCL tag |
---|
| 445 | if tagstart == "E" : |
---|
| 446 | resets += 1 |
---|
| 447 | continue # skip to next tag |
---|
| 448 | tag = tagstart + minfile[pos] ; pos += 1 |
---|
| 449 | if tag == "*b" : |
---|
| 450 | starb = 1 |
---|
| 451 | tagend = "VW" |
---|
| 452 | elif tag == "&l" : |
---|
| 453 | ampl = 1 |
---|
| 454 | tagend = "XHAOM" |
---|
| 455 | else : |
---|
| 456 | try : |
---|
| 457 | tagend = tagsends[tag] |
---|
| 458 | except KeyError : |
---|
| 459 | continue # Unsupported PCL tag |
---|
| 460 | # Now read the numeric argument |
---|
| 461 | size = 0 |
---|
| 462 | while 1 : |
---|
| 463 | char = minfile[pos] ; pos += 1 |
---|
| 464 | if not char.isdigit() : |
---|
| 465 | break |
---|
| 466 | size = (size * 10) + int(char) |
---|
| 467 | if char in tagend : |
---|
| 468 | if tag == "&l" : |
---|
| 469 | if char == "X" : |
---|
| 470 | self.setPageDict(pages, pagecount, "copies", size) |
---|
| 471 | elif char == "H" : |
---|
| 472 | self.setPageDict(pages, pagecount, "mediasource", self.mediasources.get(size, str(size))) |
---|
| 473 | mediasourcecount += 1 |
---|
| 474 | ejects += 1 |
---|
| 475 | elif char == "A" : |
---|
| 476 | self.setPageDict(pages, pagecount, "mediasize", self.mediasizes.get(size, str(size))) |
---|
| 477 | mediasizecount += 1 |
---|
| 478 | elif char == "O" : |
---|
| 479 | self.setPageDict(pages, pagecount, "orientation", self.orientations.get(size, str(size))) |
---|
| 480 | orientationcount += 1 |
---|
| 481 | elif char == "M" : |
---|
| 482 | self.setPageDict(pages, pagecount, "mediatype", self.mediatypes.get(size, str(size))) |
---|
| 483 | mediatypecount += 1 |
---|
| 484 | elif tag == "*r" : |
---|
| 485 | # Special tests for PCL3 |
---|
| 486 | if (char == "s") and size : |
---|
| 487 | while 1 : |
---|
| 488 | char = minfile[pos] ; pos += 1 |
---|
| 489 | if char == "A" : |
---|
| 490 | break |
---|
| 491 | elif (char == "b") and (minfile[pos] == "C") and not size : |
---|
| 492 | ispcl3 = 1 # Certainely a PCL3 file |
---|
| 493 | startgfx += (char == "A") and (minfile[pos - 2] in ("0", "1", "2", "3")) # Start Gfx |
---|
| 494 | endgfx += (not size) and (char in ("C", "B")) # End Gfx |
---|
| 495 | elif tag == "*t" : |
---|
| 496 | escstart += 1 |
---|
| 497 | elif (tag == "&a") and (size == 2) : |
---|
| 498 | # We are on the backside, so mark current page as duplex |
---|
| 499 | self.setPageDict(pages, pagecount, "duplex", 1) |
---|
| 500 | backsides += 1 # Back side in duplex mode |
---|
| 501 | else : |
---|
| 502 | # we just ignore the block. |
---|
| 503 | if tag == "&n" : |
---|
| 504 | # we have to take care of the operation id byte |
---|
| 505 | # which is before the string itself |
---|
| 506 | size += 1 |
---|
| 507 | pos += size |
---|
| 508 | else : |
---|
| 509 | if starb : |
---|
| 510 | # special handling of PCL3 in which |
---|
| 511 | # *b introduces combined ESCape sequences |
---|
| 512 | size = 0 |
---|
| 513 | while 1 : |
---|
| 514 | char = minfile[pos] ; pos += 1 |
---|
| 515 | if not char.isdigit() : |
---|
| 516 | break |
---|
| 517 | size = (size * 10) + int(char) |
---|
| 518 | if char in ("w", "v") : |
---|
| 519 | ispcl3 = 1 # certainely a PCL3 document |
---|
| 520 | pos += size - 1 |
---|
| 521 | elif char in ("y", "m") : |
---|
| 522 | ispcl3 = 1 # certainely a PCL3 document |
---|
| 523 | pos -= 1 # fix position : we were ahead |
---|
| 524 | elif ampl : |
---|
| 525 | # special handling of PCL3 in which |
---|
| 526 | # &l introduces combined ESCape sequences |
---|
| 527 | size = 0 |
---|
| 528 | while 1 : |
---|
| 529 | char = minfile[pos] ; pos += 1 |
---|
| 530 | if not char.isdigit() : |
---|
| 531 | break |
---|
| 532 | size = (size * 10) + int(char) |
---|
| 533 | if char in ("a", "o", "h", "m") : |
---|
| 534 | ispcl3 = 1 # certainely a PCL3 document |
---|
| 535 | pos -= 1 # fix position : we were ahead |
---|
| 536 | if char == "h" : |
---|
| 537 | self.setPageDict(pages, pagecount, "mediasource", self.mediasources.get(size, str(size))) |
---|
| 538 | mediasourcecount += 1 |
---|
| 539 | elif char == "a" : |
---|
| 540 | self.setPageDict(pages, pagecount, "mediasize", self.mediasizes.get(size, str(size))) |
---|
| 541 | mediasizecount += 1 |
---|
| 542 | elif char == "o" : |
---|
| 543 | self.setPageDict(pages, pagecount, "orientation", self.orientations.get(size, str(size))) |
---|
| 544 | orientationcount += 1 |
---|
| 545 | elif char == "m" : |
---|
| 546 | self.setPageDict(pages, pagecount, "mediatype", self.mediatypes.get(size, str(size))) |
---|
| 547 | mediatypecount += 1 |
---|
| 548 | except IndexError : # EOF ? |
---|
| 549 | pass |
---|
| 550 | finally : |
---|
| 551 | minfile.close() |
---|
| 552 | |
---|
| 553 | # if pagecount is still 0, we will use the number |
---|
| 554 | # of resets instead of the number of form feed characters. |
---|
| 555 | # but the number of resets is always at least 2 with a valid |
---|
| 556 | # pcl file : one at the very start and one at the very end |
---|
| 557 | # of the job's data. So we substract 2 from the number of |
---|
| 558 | # resets. And since on our test data we needed to substract |
---|
| 559 | # 1 more, we finally substract 3, and will test several |
---|
| 560 | # PCL files with this. If resets < 2, then the file is |
---|
| 561 | # probably not a valid PCL file, so we use 0 |
---|
| 562 | |
---|
| 563 | if self.debug : |
---|
| 564 | sys.stderr.write("pagecount : %s\n" % pagecount) |
---|
| 565 | sys.stderr.write("resets : %s\n" % resets) |
---|
| 566 | sys.stderr.write("ejects : %s\n" % ejects) |
---|
| 567 | sys.stderr.write("backsides : %s\n" % backsides) |
---|
| 568 | sys.stderr.write("startgfx : %s\n" % startgfx) |
---|
| 569 | sys.stderr.write("endgfx : %s\n" % endgfx) |
---|
| 570 | sys.stderr.write("mediasourcecount : %s\n" % mediasourcecount) |
---|
| 571 | sys.stderr.write("mediasizecount : %s\n" % mediasizecount) |
---|
| 572 | sys.stderr.write("orientationcount : %s\n" % orientationcount) |
---|
| 573 | sys.stderr.write("mediatypecount : %s\n" % mediatypecount) |
---|
| 574 | sys.stderr.write("escstart : %s\n" % escstart) |
---|
| 575 | sys.stderr.write("hasirmarker : %s\n" % hasirmarker) |
---|
| 576 | |
---|
| 577 | if hasirmarker : |
---|
| 578 | self.logdebug("Rule #20 (probably a Canon ImageRunner)") |
---|
| 579 | pagecount += 1 |
---|
| 580 | elif (orientationcount == (pagecount - 1)) and (resets == 1) : |
---|
| 581 | if resets == ejects == startgfx == mediasourcecount == escstart == 1 : |
---|
| 582 | self.logdebug("Rule #19") |
---|
| 583 | else : |
---|
| 584 | self.logdebug("Rule #1") |
---|
| 585 | pagecount -= 1 |
---|
| 586 | elif pagecount and (pagecount == orientationcount) : |
---|
| 587 | self.logdebug("Rule #2") |
---|
| 588 | elif resets == ejects == mediasourcecount == mediasizecount == escstart == 1 : |
---|
| 589 | #if ((startgfx and endgfx) and (startgfx != endgfx)) or (startgfx == endgfx == 0) : |
---|
| 590 | if (startgfx and endgfx) or (startgfx == endgfx == 0) : |
---|
| 591 | self.logdebug("Rule #3") |
---|
| 592 | pagecount = orientationcount |
---|
| 593 | elif (endgfx and not startgfx) and (pagecount > orientationcount) : |
---|
| 594 | self.logdebug("Rule #4") |
---|
| 595 | pagecount = orientationcount |
---|
| 596 | else : |
---|
| 597 | self.logdebug("Rule #5") |
---|
| 598 | pagecount += 1 |
---|
| 599 | elif (ejects == mediasourcecount == orientationcount) and (startgfx == endgfx) : |
---|
| 600 | if (resets == 2) and (orientationcount == (pagecount - 1)) and (orientationcount > 1) : |
---|
| 601 | self.logdebug("Rule #6") |
---|
| 602 | pagecount = orientationcount |
---|
| 603 | elif pagecount == mediasourcecount == escstart : |
---|
| 604 | self.logdebug("Rule #7") |
---|
| 605 | elif resets == startgfx == endgfx == mediasizecount == orientationcount == escstart == 1 : |
---|
| 606 | self.logdebug("Rule #8") |
---|
| 607 | elif resets == startgfx == endgfx == (pagecount - 1) : |
---|
| 608 | self.logdebug("Rule #9") |
---|
| 609 | elif (not startgfx) and (not endgfx) : |
---|
| 610 | self.logdebug("Rule #10") |
---|
| 611 | elif (resets == 2) and (startgfx == endgfx) and (mediasourcecount == 1) : |
---|
| 612 | if orientationcount == (pagecount - 1) : |
---|
| 613 | self.logdebug("Rule #11") |
---|
| 614 | pagecount = orientationcount |
---|
| 615 | elif not pagecount : |
---|
| 616 | self.logdebug("Rule #17") |
---|
| 617 | pagecount = ejects |
---|
| 618 | elif (resets == 1) and (startgfx == endgfx) and (mediasourcecount == 0) : |
---|
| 619 | if (startgfx > 1) and (startgfx != (pagecount - 1)) : |
---|
| 620 | self.logdebug("Rule #12") |
---|
| 621 | pagecount -= 1 |
---|
| 622 | else : |
---|
| 623 | self.logdebug("Rule #18") |
---|
| 624 | elif startgfx == endgfx : |
---|
| 625 | self.logdebug("Rule #13") |
---|
| 626 | pagecount = startgfx |
---|
| 627 | elif startgfx == (endgfx - 1) : |
---|
| 628 | self.logdebug("Rule #14") |
---|
| 629 | pagecount = startgfx |
---|
| 630 | elif (startgfx == 1) and not endgfx : |
---|
| 631 | self.logdebug("Rule #15") |
---|
| 632 | pass |
---|
| 633 | else : |
---|
| 634 | self.logdebug("Rule #16") |
---|
| 635 | pagecount = abs(startgfx - endgfx) |
---|
| 636 | |
---|
| 637 | defaultpjlcopies = 1 |
---|
| 638 | defaultduplexmode = "Simplex" |
---|
| 639 | defaultpapersize = "" |
---|
| 640 | oldpjlcopies = -1 |
---|
| 641 | oldduplexmode = "" |
---|
| 642 | oldpapersize = "" |
---|
| 643 | for pnum in range(pagecount) : |
---|
| 644 | # if no number of copies defined, take the preceding one else the one set before any page else 1. |
---|
| 645 | page = pages.get(pnum, pages.get(pnum - 1, pages.get(0, { "copies" : 1, "mediasource" : "Main", "mediasize" : "Default", "mediatype" : "Plain", "orientation" : "Portrait", "escaped" : "", "duplex": 0}))) |
---|
| 646 | pjlstuff = page["escaped"] |
---|
| 647 | if pjlstuff : |
---|
| 648 | pjlparser = pjl.PJLParser(pjlstuff) |
---|
| 649 | nbdefaultcopies = int(pjlparser.default_variables.get("COPIES", -1)) |
---|
| 650 | nbcopies = int(pjlparser.environment_variables.get("COPIES", -1)) |
---|
| 651 | nbdefaultqty = int(pjlparser.default_variables.get("QTY", -1)) |
---|
| 652 | nbqty = int(pjlparser.environment_variables.get("QTY", -1)) |
---|
| 653 | if nbdefaultcopies > -1 : |
---|
| 654 | defaultpjlcopies = nbdefaultcopies |
---|
| 655 | if nbdefaultqty > -1 : |
---|
| 656 | defaultpjlcopies = nbdefaultqty |
---|
| 657 | if nbcopies > -1 : |
---|
| 658 | pjlcopies = nbcopies |
---|
| 659 | elif nbqty > -1 : |
---|
| 660 | pjlcopies = nbqty |
---|
| 661 | else : |
---|
| 662 | if oldpjlcopies == -1 : |
---|
| 663 | pjlcopies = defaultpjlcopies |
---|
| 664 | else : |
---|
| 665 | pjlcopies = oldpjlcopies |
---|
| 666 | if page["duplex"] : |
---|
| 667 | duplexmode = "Duplex" |
---|
| 668 | else : |
---|
| 669 | defaultdm = pjlparser.default_variables.get("DUPLEX", "") |
---|
| 670 | if defaultdm : |
---|
| 671 | if defaultdm.upper() == "ON" : |
---|
| 672 | defaultduplexmode = "Duplex" |
---|
| 673 | else : |
---|
| 674 | defaultduplexmode = "Simplex" |
---|
| 675 | envdm = pjlparser.environment_variables.get("DUPLEX", "") |
---|
| 676 | if envdm : |
---|
| 677 | if envdm.upper() == "ON" : |
---|
| 678 | duplexmode = "Duplex" |
---|
| 679 | else : |
---|
| 680 | duplexmode = "Simplex" |
---|
| 681 | else : |
---|
| 682 | duplexmode = oldduplexmode or defaultduplexmode |
---|
| 683 | defaultps = pjlparser.default_variables.get("PAPER", "") |
---|
| 684 | if defaultps : |
---|
| 685 | defaultpapersize = defaultps |
---|
| 686 | envps = pjlparser.environment_variables.get("PAPER", "") |
---|
| 687 | if envps : |
---|
| 688 | papersize = envps |
---|
| 689 | else : |
---|
| 690 | if not oldpapersize : |
---|
| 691 | papersize = defaultpapersize |
---|
| 692 | else : |
---|
| 693 | papersize = oldpapersize |
---|
| 694 | else : |
---|
| 695 | if oldpjlcopies == -1 : |
---|
| 696 | pjlcopies = defaultpjlcopies |
---|
| 697 | else : |
---|
| 698 | pjlcopies = oldpjlcopies |
---|
| 699 | |
---|
| 700 | duplexmode = (page["duplex"] and "Duplex") or oldduplexmode or defaultduplexmode |
---|
| 701 | if not oldpapersize : |
---|
| 702 | papersize = defaultpapersize |
---|
| 703 | else : |
---|
| 704 | papersize = oldpapersize |
---|
| 705 | papersize = oldpapersize or page["mediasize"] |
---|
| 706 | if page["mediasize"] != "Default" : |
---|
| 707 | papersize = page["mediasize"] |
---|
| 708 | if not duplexmode : |
---|
| 709 | duplexmode = oldduplexmode or defaultduplexmode |
---|
| 710 | oldpjlcopies = pjlcopies |
---|
| 711 | oldduplexmode = duplexmode |
---|
| 712 | oldpapersize = papersize |
---|
| 713 | copies = pjlcopies * page["copies"] |
---|
| 714 | pagecount += (copies - 1) |
---|
| 715 | self.logdebug("%s*%s*%s*%s*%s*%s*BW" % (copies, \ |
---|
| 716 | page["mediatype"], \ |
---|
| 717 | papersize, \ |
---|
| 718 | page["orientation"], \ |
---|
| 719 | page["mediasource"], \ |
---|
| 720 | duplexmode)) |
---|
| 721 | |
---|
| 722 | return pagecount |
---|
| 723 | """ |
---|
| 724 | |
---|
| 725 | def test() : |
---|
| 726 | """Test function.""" |
---|
| 727 | if (len(sys.argv) < 2) or ((not sys.stdin.isatty()) and ("-" not in sys.argv[1:])) : |
---|
| 728 | sys.argv.append("-") |
---|
| 729 | totalsize = 0 |
---|
| 730 | for arg in sys.argv[1:] : |
---|
| 731 | if arg == "-" : |
---|
| 732 | infile = sys.stdin |
---|
| 733 | mustclose = 0 |
---|
| 734 | else : |
---|
| 735 | infile = open(arg, "rb") |
---|
| 736 | mustclose = 1 |
---|
| 737 | try : |
---|
| 738 | parser = Parser(infile, debug=1) |
---|
| 739 | totalsize += parser.getJobSize() |
---|
| 740 | except pdlparser.PDLParserError, msg : |
---|
| 741 | sys.stderr.write("ERROR: %s\n" % msg) |
---|
| 742 | sys.stderr.flush() |
---|
| 743 | if mustclose : |
---|
| 744 | infile.close() |
---|
| 745 | print "%s" % totalsize |
---|
| 746 | |
---|
| 747 | if __name__ == "__main__" : |
---|
| 748 | test() |
---|