| [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 | |
|---|
| [393] | 34 | FORMFEED = chr(12) |
|---|
| [392] | 35 | ESCAPE = chr(27) |
|---|
| 36 | |
|---|
| 37 | class Parser(pdlparser.PDLParser) : |
|---|
| 38 | """A parser for PCL3, PCL4, PCL5 documents.""" |
|---|
| 39 | totiffcommand = 'pcl6 -sDEVICE=pdfwrite -dPARANOIDSAFER -dNOPAUSE -dBATCH -dQUIET -sOutputFile=- - | gs -sDEVICE=tiff24nc -dPARANOIDSAFER -dNOPAUSE -dBATCH -dQUIET -r%(dpi)i -sOutputFile="%(fname)s" -' |
|---|
| 40 | mediasizes = { # ESC&l####A |
|---|
| 41 | 0 : "Default", |
|---|
| 42 | 1 : "Executive", |
|---|
| 43 | 2 : "Letter", |
|---|
| 44 | 3 : "Legal", |
|---|
| 45 | 6 : "Ledger", |
|---|
| 46 | 25 : "A5", |
|---|
| 47 | 26 : "A4", |
|---|
| 48 | 27 : "A3", |
|---|
| 49 | 45 : "JB5", |
|---|
| 50 | 46 : "JB4", |
|---|
| 51 | 71 : "HagakiPostcard", |
|---|
| 52 | 72 : "OufukuHagakiPostcard", |
|---|
| 53 | 80 : "MonarchEnvelope", |
|---|
| 54 | 81 : "COM10Envelope", |
|---|
| 55 | 90 : "DLEnvelope", |
|---|
| 56 | 91 : "C5Envelope", |
|---|
| 57 | 100 : "B5Envelope", |
|---|
| 58 | 101 : "Custom", |
|---|
| 59 | } |
|---|
| 60 | |
|---|
| 61 | mediasources = { # ESC&l####H |
|---|
| 62 | 0 : "Default", |
|---|
| 63 | 1 : "Main", |
|---|
| 64 | 2 : "Manual", |
|---|
| 65 | 3 : "ManualEnvelope", |
|---|
| 66 | 4 : "Alternate", |
|---|
| 67 | 5 : "OptionalLarge", |
|---|
| 68 | 6 : "EnvelopeFeeder", |
|---|
| 69 | 7 : "Auto", |
|---|
| 70 | 8 : "Tray1", |
|---|
| 71 | } |
|---|
| 72 | |
|---|
| 73 | orientations = { # ESC&l####O |
|---|
| 74 | 0 : "Portrait", |
|---|
| 75 | 1 : "Landscape", |
|---|
| 76 | 2 : "ReversePortrait", |
|---|
| 77 | 3 : "ReverseLandscape", |
|---|
| 78 | } |
|---|
| 79 | |
|---|
| 80 | mediatypes = { # ESC&l####M |
|---|
| 81 | 0 : "Plain", |
|---|
| 82 | 1 : "Bond", |
|---|
| 83 | 2 : "Special", |
|---|
| 84 | 3 : "Glossy", |
|---|
| 85 | 4 : "Transparent", |
|---|
| 86 | } |
|---|
| 87 | |
|---|
| 88 | def isValid(self) : |
|---|
| 89 | """Returns True if data is PCL3/4/5, else False.""" |
|---|
| 90 | if self.firstblock.startswith("\033E\033") or \ |
|---|
| 91 | (self.firstblock.startswith("\033*rbC") and (not self.lastblock[-3:] == "\f\033@")) or \ |
|---|
| 92 | self.firstblock.startswith("\033%8\033") or \ |
|---|
| 93 | (self.firstblock.find("\033%-12345X") != -1) or \ |
|---|
| 94 | (self.firstblock.find("@PJL ENTER LANGUAGE=PCL\012\015\033") != -1) or \ |
|---|
| 95 | (self.firstblock.startswith(chr(0xcd)+chr(0xca)) and self.firstblock.find("\033E\033")) : |
|---|
| 96 | self.logdebug("DEBUG: Input file is in the PCL3/4/5 format.") |
|---|
| 97 | return True |
|---|
| 98 | else : |
|---|
| 99 | return False |
|---|
| 100 | |
|---|
| 101 | def setPageDict(self, pages, number, attribute, value) : |
|---|
| 102 | """Initializes a page dictionnary.""" |
|---|
| 103 | dic = pages.setdefault(number, { "copies" : 1, "mediasource" : "Main", "mediasize" : "Default", "mediatype" : "Plain", "orientation" : "Portrait", "escaped" : "", "duplex": 0}) |
|---|
| 104 | dic[attribute] = value |
|---|
| 105 | |
|---|
| [394] | 106 | def readByte(self) : |
|---|
| 107 | """Reads a byte from the input stream.""" |
|---|
| 108 | tag = ord(self.minfile[self.pos]) |
|---|
| 109 | self.pos += 1 |
|---|
| 110 | self.logdebug("BYTE %02x" % tag) |
|---|
| 111 | return tag |
|---|
| 112 | |
|---|
| [392] | 113 | def endPage(self) : |
|---|
| 114 | """Handle the FF marker.""" |
|---|
| [394] | 115 | self.logdebug("FORMFEED %i" % self.pagecount) |
|---|
| [392] | 116 | self.pagecount += 1 |
|---|
| 117 | |
|---|
| 118 | def escPercent(self) : |
|---|
| 119 | """Handles the ESC% sequence.""" |
|---|
| 120 | if self.minfile[self.pos : self.pos+7] == r"-12345X" : |
|---|
| 121 | self.logdebug("Generic ESCAPE sequence at %08x" % self.pos) |
|---|
| 122 | self.pos += 7 |
|---|
| 123 | |
|---|
| 124 | def handleTag(self, tagtable) : |
|---|
| 125 | """Handles tags.""" |
|---|
| [394] | 126 | tagtable[self.readByte()]() |
|---|
| [392] | 127 | |
|---|
| 128 | def escape(self) : |
|---|
| 129 | """Handles the ESC character.""" |
|---|
| 130 | self.logdebug("ESCAPE") |
|---|
| 131 | self.handleTag(self.esctags) |
|---|
| 132 | |
|---|
| 133 | def escAmp(self) : |
|---|
| 134 | """Handles the ESC& sequence.""" |
|---|
| 135 | self.logdebug("AMP") |
|---|
| 136 | self.handleTag(self.escamptags) |
|---|
| 137 | |
|---|
| 138 | def escStar(self) : |
|---|
| 139 | """Handles the ESC* sequence.""" |
|---|
| 140 | self.logdebug("STAR") |
|---|
| 141 | self.handleTag(self.escstartags) |
|---|
| 142 | |
|---|
| 143 | def escLeftPar(self) : |
|---|
| 144 | """Handles the ESC( sequence.""" |
|---|
| 145 | self.logdebug("LEFTPAR") |
|---|
| 146 | self.handleTag(self.escleftpartags) |
|---|
| 147 | |
|---|
| 148 | def escRightPar(self) : |
|---|
| 149 | """Handles the ESC( sequence.""" |
|---|
| 150 | self.logdebug("RIGHTPAR") |
|---|
| 151 | self.handleTag(self.escrightpartags) |
|---|
| 152 | |
|---|
| 153 | def escE(self) : |
|---|
| 154 | """Handles the ESCE sequence.""" |
|---|
| 155 | self.logdebug("RESET") |
|---|
| 156 | self.resets += 1 |
|---|
| 157 | |
|---|
| 158 | def escAmpl(self) : |
|---|
| 159 | """Handles the ESC&l sequence.""" |
|---|
| 160 | self.logdebug("l") |
|---|
| 161 | while 1 : |
|---|
| 162 | (value, end) = self.getInteger() |
|---|
| 163 | if value is None : |
|---|
| 164 | return |
|---|
| 165 | if end in ('h', 'H') : |
|---|
| 166 | mediasource = self.mediasources.get(value, str(value)) |
|---|
| 167 | self.mediasourcesvalues.append(mediasource) |
|---|
| 168 | self.logdebug("MEDIASOURCE %s" % mediasource) |
|---|
| 169 | elif end in ('a', 'A') : |
|---|
| 170 | mediasize = self.mediasizes.get(value, str(value)) |
|---|
| 171 | self.mediasizesvalues.append(mediasize) |
|---|
| 172 | self.logdebug("MEDIASIZE %s" % mediasize) |
|---|
| 173 | elif end in ('o', 'O') : |
|---|
| 174 | orientation = self.orientations.get(value, str(value)) |
|---|
| 175 | self.orientationsvalues.append(orientation) |
|---|
| 176 | self.logdebug("ORIENTATION %s" % orientation) |
|---|
| 177 | elif end in ('m', 'M') : |
|---|
| 178 | mediatype = self.mediatypes.get(value, str(value)) |
|---|
| 179 | self.mediatypesvalues.append(mediatype) |
|---|
| 180 | self.logdebug("MEDIATYPE %s" % mediatype) |
|---|
| 181 | elif end == 'X' : |
|---|
| 182 | self.copies.append(value) |
|---|
| 183 | self.logdebug("COPIES %i" % value) |
|---|
| 184 | elif end == 'L' : |
|---|
| 185 | self.logdebug("ESC&l%iL" % value) |
|---|
| 186 | |
|---|
| 187 | def escStarb(self) : |
|---|
| 188 | """Handles the ESC*b sequence.""" |
|---|
| 189 | self.logdebug("b") |
|---|
| 190 | while 1 : |
|---|
| 191 | (value, end) = self.getInteger() |
|---|
| 192 | self.logdebug("%s === %s" % (value, end)) |
|---|
| 193 | if (end is None) and (value is None) : |
|---|
| 194 | return |
|---|
| 195 | if end in ('V', 'W', 'v', 'w') : |
|---|
| 196 | self.pos += (value or 0) |
|---|
| 197 | self.logdebug("SKIPTO %08x" % self.pos) |
|---|
| 198 | |
|---|
| 199 | def escStaro(self) : |
|---|
| 200 | """Handles the ESC*o sequence.""" |
|---|
| 201 | self.logdebug("o") |
|---|
| 202 | while 1 : |
|---|
| 203 | (value, end) = self.getInteger() |
|---|
| 204 | if value is None : |
|---|
| 205 | return |
|---|
| 206 | if end == 'M' : |
|---|
| 207 | self.logdebug("ESC*o%iM" % value) |
|---|
| 208 | |
|---|
| 209 | def escStarp(self) : |
|---|
| 210 | """Handles the ESC*p sequence.""" |
|---|
| 211 | self.logdebug("p") |
|---|
| 212 | while 1 : |
|---|
| 213 | (value, end) = self.getInteger() |
|---|
| 214 | if value is None : |
|---|
| 215 | return |
|---|
| 216 | if end in ('X', 'Y') : |
|---|
| 217 | self.logdebug("ESC*p%i%s" % (value, end)) |
|---|
| 218 | |
|---|
| 219 | def escStarr(self) : |
|---|
| 220 | """Handles the ESC*r sequence.""" |
|---|
| 221 | self.logdebug("r") |
|---|
| 222 | while 1 : |
|---|
| 223 | (value, end) = self.getInteger() |
|---|
| 224 | if value is None : |
|---|
| 225 | if end is None : |
|---|
| 226 | return |
|---|
| 227 | elif end == 'b' : |
|---|
| 228 | if self.minfile[self.pos] == 'C' : |
|---|
| 229 | self.logdebug("Looks like it's PCL3.") |
|---|
| 230 | self.ispcl3 = True |
|---|
| 231 | self.pos += 1 |
|---|
| 232 | elif end == 'C' : |
|---|
| 233 | self.logdebug("EndGFX") |
|---|
| 234 | if not self.startgfx : |
|---|
| 235 | self.logdebug("EndGFX found before StartGFX, ignored.") |
|---|
| 236 | else : |
|---|
| 237 | self.endgfx.append(1) |
|---|
| 238 | if end == 'A' and (0 <= value <= 3) : |
|---|
| 239 | self.logdebug("StartGFX %i" % value) |
|---|
| 240 | self.startgfx.append(value) |
|---|
| 241 | elif end == 'U' : |
|---|
| 242 | self.logdebug("ESC*r%iU" % value) |
|---|
| 243 | elif end == 'S' : |
|---|
| 244 | self.logdebug("ESC*r%iS" % value) |
|---|
| 245 | |
|---|
| 246 | def escStart(self) : |
|---|
| 247 | """Handles the ESC*t sequence.""" |
|---|
| 248 | self.logdebug("t") |
|---|
| 249 | while 1 : |
|---|
| 250 | (value, end) = self.getInteger() |
|---|
| 251 | if value is None : |
|---|
| 252 | return |
|---|
| 253 | if end == 'R' : |
|---|
| 254 | self.logdebug("ESC*t%iR" % value) |
|---|
| 255 | |
|---|
| 256 | def escAmpu(self) : |
|---|
| 257 | """Handles the ESC&u sequence.""" |
|---|
| 258 | self.logdebug("u") |
|---|
| 259 | while 1 : |
|---|
| 260 | (value, end) = self.getInteger() |
|---|
| 261 | if value is None : |
|---|
| 262 | return |
|---|
| 263 | if end == 'D' : |
|---|
| 264 | self.logdebug("ESC&u%iD" % value) |
|---|
| 265 | |
|---|
| 266 | |
|---|
| 267 | def getInteger(self) : |
|---|
| 268 | """Returns an integer value and the end character.""" |
|---|
| 269 | sign = 1 |
|---|
| 270 | value = None |
|---|
| 271 | while 1 : |
|---|
| [394] | 272 | char = chr(self.readByte()) |
|---|
| [392] | 273 | if char == ESCAPE : |
|---|
| [394] | 274 | self.pos -= 1 # Adjust position |
|---|
| [392] | 275 | return (None, None) |
|---|
| 276 | if char == '-' : |
|---|
| 277 | sign = -1 |
|---|
| 278 | elif not char.isdigit() : |
|---|
| 279 | if value is not None : |
|---|
| 280 | return (sign*value, char) |
|---|
| 281 | else : |
|---|
| 282 | return (value, char) |
|---|
| 283 | else : |
|---|
| 284 | value = ((value or 0) * 10) + int(char) |
|---|
| 285 | |
|---|
| 286 | def getJobSize(self) : |
|---|
| 287 | """Count pages in a PCL5 document. |
|---|
| 288 | |
|---|
| 289 | Should also work for PCL3 and PCL4 documents. |
|---|
| 290 | |
|---|
| 291 | Algorithm from pclcount |
|---|
| 292 | (c) 2003, by Eduardo Gielamo Oliveira & Rodolfo Broco Manin |
|---|
| 293 | published under the terms of the GNU General Public Licence v2. |
|---|
| 294 | |
|---|
| 295 | Backported from C to Python by Jerome Alet, then enhanced |
|---|
| 296 | with more PCL tags detected. I think all the necessary PCL tags |
|---|
| 297 | are recognized to correctly handle PCL5 files wrt their number |
|---|
| 298 | of pages. The documentation used for this was : |
|---|
| 299 | |
|---|
| 300 | HP PCL/PJL Reference Set |
|---|
| 301 | PCL5 Printer Language Technical Quick Reference Guide |
|---|
| 302 | http://h20000.www2.hp.com/bc/docs/support/SupportManual/bpl13205/bpl13205.pdf |
|---|
| 303 | """ |
|---|
| 304 | infileno = self.infile.fileno() |
|---|
| 305 | self.minfile = minfile = mmap.mmap(infileno, os.fstat(infileno)[6], prot=mmap.PROT_READ, flags=mmap.MAP_SHARED) |
|---|
| 306 | self.ispcl3 = False |
|---|
| 307 | self.pagecount = 0 |
|---|
| 308 | self.resets = 0 |
|---|
| 309 | self.copies = [] |
|---|
| 310 | self.mediasourcesvalues = [] |
|---|
| 311 | self.mediasizesvalues = [] |
|---|
| 312 | self.orientationsvalues = [] |
|---|
| 313 | self.mediatypesvalues = [] |
|---|
| 314 | self.startgfx = [] |
|---|
| 315 | self.endgfx = [] |
|---|
| 316 | |
|---|
| 317 | tags = [ lambda : None] * 256 |
|---|
| [394] | 318 | tags[ord(FORMFEED)] = self.endPage |
|---|
| 319 | tags[ord(ESCAPE)] = self.escape |
|---|
| [392] | 320 | |
|---|
| 321 | self.esctags = [ lambda : None ] * 256 |
|---|
| 322 | self.esctags[ord('%')] = self.escPercent |
|---|
| 323 | self.esctags[ord('*')] = self.escStar |
|---|
| 324 | self.esctags[ord('&')] = self.escAmp |
|---|
| 325 | self.esctags[ord('(')] = self.escLeftPar |
|---|
| 326 | self.esctags[ord(')')] = self.escRightPar |
|---|
| 327 | self.esctags[ord('E')] = self.escE |
|---|
| 328 | |
|---|
| 329 | self.escamptags = [lambda : None ] * 256 |
|---|
| 330 | self.escamptags[ord('l')] = self.escAmpl |
|---|
| 331 | self.escamptags[ord('u')] = self.escAmpu |
|---|
| 332 | |
|---|
| 333 | self.escstartags = [ lambda : None ] * 256 |
|---|
| 334 | self.escstartags[ord('b')] = self.escStarb |
|---|
| 335 | self.escstartags[ord('o')] = self.escStaro |
|---|
| 336 | self.escstartags[ord('p')] = self.escStarp |
|---|
| 337 | self.escstartags[ord('r')] = self.escStarr |
|---|
| 338 | self.escstartags[ord('t')] = self.escStart |
|---|
| 339 | |
|---|
| 340 | self.pos = 0 |
|---|
| 341 | try : |
|---|
| 342 | try : |
|---|
| 343 | while 1 : |
|---|
| [394] | 344 | tag = self.readByte() |
|---|
| 345 | self.logdebug("%08x ===> %02x" % (self.pos-1, tag)) |
|---|
| [392] | 346 | tags[tag]() |
|---|
| 347 | except IndexError : # EOF ? |
|---|
| 348 | pass |
|---|
| 349 | finally : |
|---|
| 350 | self.minfile.close() |
|---|
| 351 | |
|---|
| 352 | self.logdebug("Pagecount : \t\t%i" % self.pagecount) |
|---|
| 353 | self.logdebug("Resets : \t\t%i" % self.resets) |
|---|
| 354 | self.logdebug("Copies : \t\t%s" % self.copies) |
|---|
| 355 | self.logdebug("MediaTypes : \t\t%s" % self.mediatypesvalues) |
|---|
| 356 | self.logdebug("MediaSizes : \t\t%s" % self.mediasizesvalues) |
|---|
| 357 | self.logdebug("MediaSources : \t\t%s" % self.mediasourcesvalues) |
|---|
| 358 | self.logdebug("Orientations : \t\t%s" % self.orientationsvalues) |
|---|
| 359 | self.logdebug("StartGfx : \t\t%s" % len(self.startgfx)) |
|---|
| 360 | self.logdebug("EndGfx : \t\t%s" % len(self.endgfx)) |
|---|
| 361 | |
|---|
| 362 | return self.pagecount |
|---|
| 363 | |
|---|
| 364 | |
|---|
| 365 | |
|---|
| 366 | """ |
|---|
| 367 | tagsends = { "&n" : "W", |
|---|
| 368 | "&b" : "W", |
|---|
| 369 | "*i" : "W", |
|---|
| 370 | "*l" : "W", |
|---|
| 371 | "*m" : "W", |
|---|
| 372 | "*v" : "W", |
|---|
| 373 | "*c" : "W", |
|---|
| 374 | "(f" : "W", |
|---|
| 375 | "(s" : "W", |
|---|
| 376 | ")s" : "W", |
|---|
| 377 | "&p" : "X", |
|---|
| 378 | # "&l" : "XHAOM", # treated specially |
|---|
| 379 | "&a" : "G", # TODO : 0 means next side, 1 front side, 2 back side |
|---|
| 380 | "*g" : "W", |
|---|
| 381 | "*r" : "sbABC", |
|---|
| 382 | "*t" : "R", |
|---|
| 383 | # "*b" : "VW", # treated specially because it occurs very often |
|---|
| 384 | } |
|---|
| 385 | irmarker = chr(0xcd) + chr(0xca) # Marker for Canon ImageRunner printers |
|---|
| 386 | irmarker2 = chr(0x10) + chr(0x02) |
|---|
| 387 | wasirmarker = 0 |
|---|
| 388 | hasirmarker = (minfile[:2] == (irmarker)) |
|---|
| 389 | pagecount = resets = ejects = backsides = startgfx = endgfx = 0 |
|---|
| 390 | starb = ampl = ispcl3 = escstart = 0 |
|---|
| 391 | mediasourcecount = mediasizecount = orientationcount = mediatypecount = 0 |
|---|
| 392 | tag = None |
|---|
| 393 | endmark = chr(0x1b) + chr(0x0c) + chr(0x00) |
|---|
| 394 | asciilimit = chr(0x80) |
|---|
| 395 | pages = {} |
|---|
| 396 | pos = 0 |
|---|
| 397 | try : |
|---|
| 398 | try : |
|---|
| 399 | while 1 : |
|---|
| 400 | if hasirmarker and (minfile[pos:pos+2] == irmarker) : |
|---|
| 401 | codop = minfile[pos+2:pos+4] |
|---|
| 402 | # self.logdebug("Marker at 0x%08x (%s)" % (pos, wasirmarker)) |
|---|
| 403 | length = unpack(">H", minfile[pos+8:pos+10])[0] |
|---|
| 404 | pos += 20 |
|---|
| 405 | if codop != irmarker2 : |
|---|
| 406 | pos += length |
|---|
| 407 | wasirmarker = 1 |
|---|
| 408 | else : |
|---|
| 409 | wasirmarker = 0 |
|---|
| 410 | elif char == "\033" : |
|---|
| 411 | starb = ampl = 0 |
|---|
| 412 | if minfile[pos : pos+8] == r"%-12345X" : |
|---|
| 413 | endpos = pos + 9 |
|---|
| 414 | quotes = 0 |
|---|
| 415 | while (minfile[endpos] not in endmark) and \ |
|---|
| 416 | ((minfile[endpos] < asciilimit) or (quotes % 2)) : |
|---|
| 417 | if minfile[endpos] == '"' : |
|---|
| 418 | quotes += 1 |
|---|
| 419 | endpos += 1 |
|---|
| 420 | self.setPageDict(pages, pagecount, "escaped", minfile[pos : endpos]) |
|---|
| 421 | pos += (endpos - pos) |
|---|
| 422 | else : |
|---|
| 423 | # |
|---|
| 424 | # <ESC>*b###y#m###v###w... -> PCL3 raster graphics |
|---|
| 425 | # <ESC>*b###W -> Start of a raster data row/block |
|---|
| 426 | # <ESC>*b###V -> Start of a raster data plane |
|---|
| 427 | # <ESC>*c###W -> Start of a user defined pattern |
|---|
| 428 | # <ESC>*i###W -> Start of a viewing illuminant block |
|---|
| 429 | # <ESC>*l###W -> Start of a color lookup table |
|---|
| 430 | # <ESC>*m###W -> Start of a download dither matrix block |
|---|
| 431 | # <ESC>*v###W -> Start of a configure image data block |
|---|
| 432 | # <ESC>*r1A -> Start Gfx |
|---|
| 433 | # <ESC>(s###W -> Start of a characters description block |
|---|
| 434 | # <ESC>)s###W -> Start of a fonts description block |
|---|
| 435 | # <ESC>(f###W -> Start of a symbol set block |
|---|
| 436 | # <ESC>&b###W -> Start of configuration data block |
|---|
| 437 | # <ESC>&l###X -> Number of copies for current page |
|---|
| 438 | # <ESC>&n###W -> Starts an alphanumeric string ID block |
|---|
| 439 | # <ESC>&p###X -> Start of a non printable characters block |
|---|
| 440 | # <ESC>&a2G -> Back side when duplex mode as generated by rastertohp |
|---|
| 441 | # <ESC>*g###W -> Needed for planes in PCL3 output |
|---|
| 442 | # <ESC>&l###H (or only 0 ?) -> Eject if NumPlanes > 1, as generated by rastertohp. Also defines mediasource |
|---|
| 443 | # <ESC>&l###A -> mediasize |
|---|
| 444 | # <ESC>&l###O -> orientation |
|---|
| 445 | # <ESC>&l###M -> mediatype |
|---|
| 446 | # <ESC>*t###R -> gfx resolution |
|---|
| 447 | # |
|---|
| 448 | tagstart = minfile[pos] ; pos += 1 |
|---|
| 449 | if tagstart in "E9=YZ" : # one byte PCL tag |
|---|
| 450 | if tagstart == "E" : |
|---|
| 451 | resets += 1 |
|---|
| 452 | continue # skip to next tag |
|---|
| 453 | tag = tagstart + minfile[pos] ; pos += 1 |
|---|
| 454 | if tag == "*b" : |
|---|
| 455 | starb = 1 |
|---|
| 456 | tagend = "VW" |
|---|
| 457 | elif tag == "&l" : |
|---|
| 458 | ampl = 1 |
|---|
| 459 | tagend = "XHAOM" |
|---|
| 460 | else : |
|---|
| 461 | try : |
|---|
| 462 | tagend = tagsends[tag] |
|---|
| 463 | except KeyError : |
|---|
| 464 | continue # Unsupported PCL tag |
|---|
| 465 | # Now read the numeric argument |
|---|
| 466 | size = 0 |
|---|
| 467 | while 1 : |
|---|
| 468 | char = minfile[pos] ; pos += 1 |
|---|
| 469 | if not char.isdigit() : |
|---|
| 470 | break |
|---|
| 471 | size = (size * 10) + int(char) |
|---|
| 472 | if char in tagend : |
|---|
| 473 | if tag == "&l" : |
|---|
| 474 | if char == "X" : |
|---|
| 475 | self.setPageDict(pages, pagecount, "copies", size) |
|---|
| 476 | elif char == "H" : |
|---|
| 477 | self.setPageDict(pages, pagecount, "mediasource", self.mediasources.get(size, str(size))) |
|---|
| 478 | mediasourcecount += 1 |
|---|
| 479 | ejects += 1 |
|---|
| 480 | elif char == "A" : |
|---|
| 481 | self.setPageDict(pages, pagecount, "mediasize", self.mediasizes.get(size, str(size))) |
|---|
| 482 | mediasizecount += 1 |
|---|
| 483 | elif char == "O" : |
|---|
| 484 | self.setPageDict(pages, pagecount, "orientation", self.orientations.get(size, str(size))) |
|---|
| 485 | orientationcount += 1 |
|---|
| 486 | elif char == "M" : |
|---|
| 487 | self.setPageDict(pages, pagecount, "mediatype", self.mediatypes.get(size, str(size))) |
|---|
| 488 | mediatypecount += 1 |
|---|
| 489 | elif tag == "*r" : |
|---|
| 490 | # Special tests for PCL3 |
|---|
| 491 | if (char == "s") and size : |
|---|
| 492 | while 1 : |
|---|
| 493 | char = minfile[pos] ; pos += 1 |
|---|
| 494 | if char == "A" : |
|---|
| 495 | break |
|---|
| 496 | elif (char == "b") and (minfile[pos] == "C") and not size : |
|---|
| 497 | ispcl3 = 1 # Certainely a PCL3 file |
|---|
| 498 | startgfx += (char == "A") and (minfile[pos - 2] in ("0", "1", "2", "3")) # Start Gfx |
|---|
| 499 | endgfx += (not size) and (char in ("C", "B")) # End Gfx |
|---|
| 500 | elif tag == "*t" : |
|---|
| 501 | escstart += 1 |
|---|
| 502 | elif (tag == "&a") and (size == 2) : |
|---|
| 503 | # We are on the backside, so mark current page as duplex |
|---|
| 504 | self.setPageDict(pages, pagecount, "duplex", 1) |
|---|
| 505 | backsides += 1 # Back side in duplex mode |
|---|
| 506 | else : |
|---|
| 507 | # we just ignore the block. |
|---|
| 508 | if tag == "&n" : |
|---|
| 509 | # we have to take care of the operation id byte |
|---|
| 510 | # which is before the string itself |
|---|
| 511 | size += 1 |
|---|
| 512 | pos += size |
|---|
| 513 | else : |
|---|
| 514 | if starb : |
|---|
| 515 | # special handling of PCL3 in which |
|---|
| 516 | # *b introduces combined ESCape sequences |
|---|
| 517 | size = 0 |
|---|
| 518 | while 1 : |
|---|
| 519 | char = minfile[pos] ; pos += 1 |
|---|
| 520 | if not char.isdigit() : |
|---|
| 521 | break |
|---|
| 522 | size = (size * 10) + int(char) |
|---|
| 523 | if char in ("w", "v") : |
|---|
| 524 | ispcl3 = 1 # certainely a PCL3 document |
|---|
| 525 | pos += size - 1 |
|---|
| 526 | elif char in ("y", "m") : |
|---|
| 527 | ispcl3 = 1 # certainely a PCL3 document |
|---|
| 528 | pos -= 1 # fix position : we were ahead |
|---|
| 529 | elif ampl : |
|---|
| 530 | # special handling of PCL3 in which |
|---|
| 531 | # &l introduces combined ESCape sequences |
|---|
| 532 | size = 0 |
|---|
| 533 | while 1 : |
|---|
| 534 | char = minfile[pos] ; pos += 1 |
|---|
| 535 | if not char.isdigit() : |
|---|
| 536 | break |
|---|
| 537 | size = (size * 10) + int(char) |
|---|
| 538 | if char in ("a", "o", "h", "m") : |
|---|
| 539 | ispcl3 = 1 # certainely a PCL3 document |
|---|
| 540 | pos -= 1 # fix position : we were ahead |
|---|
| 541 | if char == "h" : |
|---|
| 542 | self.setPageDict(pages, pagecount, "mediasource", self.mediasources.get(size, str(size))) |
|---|
| 543 | mediasourcecount += 1 |
|---|
| 544 | elif char == "a" : |
|---|
| 545 | self.setPageDict(pages, pagecount, "mediasize", self.mediasizes.get(size, str(size))) |
|---|
| 546 | mediasizecount += 1 |
|---|
| 547 | elif char == "o" : |
|---|
| 548 | self.setPageDict(pages, pagecount, "orientation", self.orientations.get(size, str(size))) |
|---|
| 549 | orientationcount += 1 |
|---|
| 550 | elif char == "m" : |
|---|
| 551 | self.setPageDict(pages, pagecount, "mediatype", self.mediatypes.get(size, str(size))) |
|---|
| 552 | mediatypecount += 1 |
|---|
| 553 | except IndexError : # EOF ? |
|---|
| 554 | pass |
|---|
| 555 | finally : |
|---|
| 556 | minfile.close() |
|---|
| 557 | |
|---|
| 558 | # if pagecount is still 0, we will use the number |
|---|
| 559 | # of resets instead of the number of form feed characters. |
|---|
| 560 | # but the number of resets is always at least 2 with a valid |
|---|
| 561 | # pcl file : one at the very start and one at the very end |
|---|
| 562 | # of the job's data. So we substract 2 from the number of |
|---|
| 563 | # resets. And since on our test data we needed to substract |
|---|
| 564 | # 1 more, we finally substract 3, and will test several |
|---|
| 565 | # PCL files with this. If resets < 2, then the file is |
|---|
| 566 | # probably not a valid PCL file, so we use 0 |
|---|
| 567 | |
|---|
| 568 | if self.debug : |
|---|
| 569 | sys.stderr.write("pagecount : %s\n" % pagecount) |
|---|
| 570 | sys.stderr.write("resets : %s\n" % resets) |
|---|
| 571 | sys.stderr.write("ejects : %s\n" % ejects) |
|---|
| 572 | sys.stderr.write("backsides : %s\n" % backsides) |
|---|
| 573 | sys.stderr.write("startgfx : %s\n" % startgfx) |
|---|
| 574 | sys.stderr.write("endgfx : %s\n" % endgfx) |
|---|
| 575 | sys.stderr.write("mediasourcecount : %s\n" % mediasourcecount) |
|---|
| 576 | sys.stderr.write("mediasizecount : %s\n" % mediasizecount) |
|---|
| 577 | sys.stderr.write("orientationcount : %s\n" % orientationcount) |
|---|
| 578 | sys.stderr.write("mediatypecount : %s\n" % mediatypecount) |
|---|
| 579 | sys.stderr.write("escstart : %s\n" % escstart) |
|---|
| 580 | sys.stderr.write("hasirmarker : %s\n" % hasirmarker) |
|---|
| 581 | |
|---|
| 582 | if hasirmarker : |
|---|
| 583 | self.logdebug("Rule #20 (probably a Canon ImageRunner)") |
|---|
| 584 | pagecount += 1 |
|---|
| 585 | elif (orientationcount == (pagecount - 1)) and (resets == 1) : |
|---|
| 586 | if resets == ejects == startgfx == mediasourcecount == escstart == 1 : |
|---|
| 587 | self.logdebug("Rule #19") |
|---|
| 588 | else : |
|---|
| 589 | self.logdebug("Rule #1") |
|---|
| 590 | pagecount -= 1 |
|---|
| 591 | elif pagecount and (pagecount == orientationcount) : |
|---|
| 592 | self.logdebug("Rule #2") |
|---|
| 593 | elif resets == ejects == mediasourcecount == mediasizecount == escstart == 1 : |
|---|
| 594 | #if ((startgfx and endgfx) and (startgfx != endgfx)) or (startgfx == endgfx == 0) : |
|---|
| 595 | if (startgfx and endgfx) or (startgfx == endgfx == 0) : |
|---|
| 596 | self.logdebug("Rule #3") |
|---|
| 597 | pagecount = orientationcount |
|---|
| 598 | elif (endgfx and not startgfx) and (pagecount > orientationcount) : |
|---|
| 599 | self.logdebug("Rule #4") |
|---|
| 600 | pagecount = orientationcount |
|---|
| 601 | else : |
|---|
| 602 | self.logdebug("Rule #5") |
|---|
| 603 | pagecount += 1 |
|---|
| 604 | elif (ejects == mediasourcecount == orientationcount) and (startgfx == endgfx) : |
|---|
| 605 | if (resets == 2) and (orientationcount == (pagecount - 1)) and (orientationcount > 1) : |
|---|
| 606 | self.logdebug("Rule #6") |
|---|
| 607 | pagecount = orientationcount |
|---|
| 608 | elif pagecount == mediasourcecount == escstart : |
|---|
| 609 | self.logdebug("Rule #7") |
|---|
| 610 | elif resets == startgfx == endgfx == mediasizecount == orientationcount == escstart == 1 : |
|---|
| 611 | self.logdebug("Rule #8") |
|---|
| 612 | elif resets == startgfx == endgfx == (pagecount - 1) : |
|---|
| 613 | self.logdebug("Rule #9") |
|---|
| 614 | elif (not startgfx) and (not endgfx) : |
|---|
| 615 | self.logdebug("Rule #10") |
|---|
| 616 | elif (resets == 2) and (startgfx == endgfx) and (mediasourcecount == 1) : |
|---|
| 617 | if orientationcount == (pagecount - 1) : |
|---|
| 618 | self.logdebug("Rule #11") |
|---|
| 619 | pagecount = orientationcount |
|---|
| 620 | elif not pagecount : |
|---|
| 621 | self.logdebug("Rule #17") |
|---|
| 622 | pagecount = ejects |
|---|
| 623 | elif (resets == 1) and (startgfx == endgfx) and (mediasourcecount == 0) : |
|---|
| 624 | if (startgfx > 1) and (startgfx != (pagecount - 1)) : |
|---|
| 625 | self.logdebug("Rule #12") |
|---|
| 626 | pagecount -= 1 |
|---|
| 627 | else : |
|---|
| 628 | self.logdebug("Rule #18") |
|---|
| 629 | elif startgfx == endgfx : |
|---|
| 630 | self.logdebug("Rule #13") |
|---|
| 631 | pagecount = startgfx |
|---|
| 632 | elif startgfx == (endgfx - 1) : |
|---|
| 633 | self.logdebug("Rule #14") |
|---|
| 634 | pagecount = startgfx |
|---|
| 635 | elif (startgfx == 1) and not endgfx : |
|---|
| 636 | self.logdebug("Rule #15") |
|---|
| 637 | pass |
|---|
| 638 | else : |
|---|
| 639 | self.logdebug("Rule #16") |
|---|
| 640 | pagecount = abs(startgfx - endgfx) |
|---|
| 641 | |
|---|
| 642 | defaultpjlcopies = 1 |
|---|
| 643 | defaultduplexmode = "Simplex" |
|---|
| 644 | defaultpapersize = "" |
|---|
| 645 | oldpjlcopies = -1 |
|---|
| 646 | oldduplexmode = "" |
|---|
| 647 | oldpapersize = "" |
|---|
| 648 | for pnum in range(pagecount) : |
|---|
| 649 | # if no number of copies defined, take the preceding one else the one set before any page else 1. |
|---|
| 650 | page = pages.get(pnum, pages.get(pnum - 1, pages.get(0, { "copies" : 1, "mediasource" : "Main", "mediasize" : "Default", "mediatype" : "Plain", "orientation" : "Portrait", "escaped" : "", "duplex": 0}))) |
|---|
| 651 | pjlstuff = page["escaped"] |
|---|
| 652 | if pjlstuff : |
|---|
| 653 | pjlparser = pjl.PJLParser(pjlstuff) |
|---|
| 654 | nbdefaultcopies = int(pjlparser.default_variables.get("COPIES", -1)) |
|---|
| 655 | nbcopies = int(pjlparser.environment_variables.get("COPIES", -1)) |
|---|
| 656 | nbdefaultqty = int(pjlparser.default_variables.get("QTY", -1)) |
|---|
| 657 | nbqty = int(pjlparser.environment_variables.get("QTY", -1)) |
|---|
| 658 | if nbdefaultcopies > -1 : |
|---|
| 659 | defaultpjlcopies = nbdefaultcopies |
|---|
| 660 | if nbdefaultqty > -1 : |
|---|
| 661 | defaultpjlcopies = nbdefaultqty |
|---|
| 662 | if nbcopies > -1 : |
|---|
| 663 | pjlcopies = nbcopies |
|---|
| 664 | elif nbqty > -1 : |
|---|
| 665 | pjlcopies = nbqty |
|---|
| 666 | else : |
|---|
| 667 | if oldpjlcopies == -1 : |
|---|
| 668 | pjlcopies = defaultpjlcopies |
|---|
| 669 | else : |
|---|
| 670 | pjlcopies = oldpjlcopies |
|---|
| 671 | if page["duplex"] : |
|---|
| 672 | duplexmode = "Duplex" |
|---|
| 673 | else : |
|---|
| 674 | defaultdm = pjlparser.default_variables.get("DUPLEX", "") |
|---|
| 675 | if defaultdm : |
|---|
| 676 | if defaultdm.upper() == "ON" : |
|---|
| 677 | defaultduplexmode = "Duplex" |
|---|
| 678 | else : |
|---|
| 679 | defaultduplexmode = "Simplex" |
|---|
| 680 | envdm = pjlparser.environment_variables.get("DUPLEX", "") |
|---|
| 681 | if envdm : |
|---|
| 682 | if envdm.upper() == "ON" : |
|---|
| 683 | duplexmode = "Duplex" |
|---|
| 684 | else : |
|---|
| 685 | duplexmode = "Simplex" |
|---|
| 686 | else : |
|---|
| 687 | duplexmode = oldduplexmode or defaultduplexmode |
|---|
| 688 | defaultps = pjlparser.default_variables.get("PAPER", "") |
|---|
| 689 | if defaultps : |
|---|
| 690 | defaultpapersize = defaultps |
|---|
| 691 | envps = pjlparser.environment_variables.get("PAPER", "") |
|---|
| 692 | if envps : |
|---|
| 693 | papersize = envps |
|---|
| 694 | else : |
|---|
| 695 | if not oldpapersize : |
|---|
| 696 | papersize = defaultpapersize |
|---|
| 697 | else : |
|---|
| 698 | papersize = oldpapersize |
|---|
| 699 | else : |
|---|
| 700 | if oldpjlcopies == -1 : |
|---|
| 701 | pjlcopies = defaultpjlcopies |
|---|
| 702 | else : |
|---|
| 703 | pjlcopies = oldpjlcopies |
|---|
| 704 | |
|---|
| 705 | duplexmode = (page["duplex"] and "Duplex") or oldduplexmode or defaultduplexmode |
|---|
| 706 | if not oldpapersize : |
|---|
| 707 | papersize = defaultpapersize |
|---|
| 708 | else : |
|---|
| 709 | papersize = oldpapersize |
|---|
| 710 | papersize = oldpapersize or page["mediasize"] |
|---|
| 711 | if page["mediasize"] != "Default" : |
|---|
| 712 | papersize = page["mediasize"] |
|---|
| 713 | if not duplexmode : |
|---|
| 714 | duplexmode = oldduplexmode or defaultduplexmode |
|---|
| 715 | oldpjlcopies = pjlcopies |
|---|
| 716 | oldduplexmode = duplexmode |
|---|
| 717 | oldpapersize = papersize |
|---|
| 718 | copies = pjlcopies * page["copies"] |
|---|
| 719 | pagecount += (copies - 1) |
|---|
| 720 | self.logdebug("%s*%s*%s*%s*%s*%s*BW" % (copies, \ |
|---|
| 721 | page["mediatype"], \ |
|---|
| 722 | papersize, \ |
|---|
| 723 | page["orientation"], \ |
|---|
| 724 | page["mediasource"], \ |
|---|
| 725 | duplexmode)) |
|---|
| 726 | |
|---|
| 727 | return pagecount |
|---|
| 728 | """ |
|---|
| 729 | |
|---|
| 730 | def test() : |
|---|
| 731 | """Test function.""" |
|---|
| 732 | if (len(sys.argv) < 2) or ((not sys.stdin.isatty()) and ("-" not in sys.argv[1:])) : |
|---|
| 733 | sys.argv.append("-") |
|---|
| 734 | totalsize = 0 |
|---|
| 735 | for arg in sys.argv[1:] : |
|---|
| 736 | if arg == "-" : |
|---|
| 737 | infile = sys.stdin |
|---|
| 738 | mustclose = 0 |
|---|
| 739 | else : |
|---|
| 740 | infile = open(arg, "rb") |
|---|
| 741 | mustclose = 1 |
|---|
| 742 | try : |
|---|
| 743 | parser = Parser(infile, debug=1) |
|---|
| 744 | totalsize += parser.getJobSize() |
|---|
| 745 | except pdlparser.PDLParserError, msg : |
|---|
| 746 | sys.stderr.write("ERROR: %s\n" % msg) |
|---|
| 747 | sys.stderr.flush() |
|---|
| 748 | if mustclose : |
|---|
| 749 | infile.close() |
|---|
| 750 | print "%s" % totalsize |
|---|
| 751 | |
|---|
| 752 | if __name__ == "__main__" : |
|---|
| 753 | test() |
|---|
