rendertest/testpdfs.py

   1 import Image
   2 import ImageChops
   3 import ImageFilter
   4 import sys
   5 import os
   6 import traceback
   7 from athana import getTALstr
   8 import random
   9 import md5
  10
  11 filenames = []
  12 directories = ["pdfs"]
  13
  14 SWFRENDER="swfrender"
  15 PDFTOPPM="pdftoppm"
  16 CONVERT="convert"
  17 PDF2SWF="pdf2swf"
  18
  19 #COMPARE=["xpdf", PDF2SWF+" -s poly2bitmap"]
  20 #OUTPUTDIR = "results.poly2bitmap/"
  21
  22 COMPARE=["xpdf", PDF2SWF+" -s convertgradients"]
  23 OUTPUTDIR = "results.pdf2swf/"
  24
  25 #COMPARE=[PDF2SWF, PDF2SWF+" --flatten"]
  26 #OUTPUTDIR = "results.flatten/"
  27
  28 counter = 1
  29
  30 def randstr():
  31     return md5.md5(str(random.random())).hexdigest()[0:8]
  32
  33 def unlink(file):
  34     try:
  35         os.unlink(file)
  36     except:
  37         pass
  38
  39 def system(command):
  40     if ">" not in command:
  41         if os.system(command + " > /tmp/log.txt 2>&1") & 0xff00:
  42             error = open("/tmp/log.txt", "rb").read()
  43             print error
  44             return error
  45     else:
  46         if os.system(command) & 0xff00:
  47             return "Unknown error in "+command
  48
  49 class ConversionError:
  50     def __init__(self, msg):
  51         self.msg = msg
  52     def __str__(self):
  53         return self.msg
  54
  55 class TooComplexError:
  56     def __init__(self, msg):
  57         self.msg = msg
  58     def __str__(self):
  59         return self.msg
  60
  61 class BadMatch:
  62     def __init__(self, msg):
  63         self.msg = msg
  64     def __str__(self):
  65         return self.msg
  66
  67 def formatException():
  68     s = "Exception "+str(sys.exc_info()[0])
  69     info = sys.exc_info()[1]
  70     if info:
  71         s += " "+str(info)
  72     s += "\n"
  73     for l in traceback.extract_tb(sys.exc_info()[2]):
  74         s += "  File \"%s\", line %d, in %s\n" % (l[0],l[1],l[2])
  75         s += "    %s\n" % l[3]
  76     return s
  77
  78
  79 class PDFPage:
  80     def __init__(self, filename, page, width, height):
  81         self.filename = filename
  82         self.page = page
  83         self.rating = None
  84         self.message = None
  85         self.htmlmessage = None
  86         self.file1 = None
  87         self.file2 = None
  88         self.file12 = None
  89         self.html12 = None
  90         self.htmldiff = None
  91         self.width,self.height = width,height
  92
  93     def runtools(self, filename, page, file1, file2, file12):
  94
  95         badness = 0.0
  96
  97         if COMPARE[0] == "xpdf":
  98             unlink("/tmp/test-%06d.ppm" % page)
  99             error = system(PDFTOPPM + " -r 72 -f %d -l %d '%s' /tmp/test" % (page, page, filename))
 100             if error and "supports 65536" in error:
 101                 raise TooComplexError(error)
 102             if error:
 103                 raise ConversionError(error)
 104             unlink(file2)
 105             error = system(CONVERT + " /tmp/test-%06d.ppm  %s" % (page, file2))
 106             if error:
 107                 raise ConversionError(error)
 108             unlink("/tmp/test-%06d.ppm" % page)
 109         else:
 110             unlink("/tmp/test.swf")
 111             unlink("svp.ps")
 112             error = system(COMPARE[0]+ " -Q 300 -p%d '%s' -o /tmp/test.swf" % (page, filename))
 113             #system("mv svp.ps %s.ps" % randstr())
 114             if error and "supports 65536" in error:
 115                 raise TooComplexError(error)
 116             if error:
 117                 raise ConversionError(error)
 118             unlink(file2)
 119             error = system(SWFRENDER + " /tmp/test.swf -o %s" % file2)
 120             if error:
 121                 raise ConversionError(error)
 122             unlink("/tmp/test.swf")
 123
 124         unlink("/tmp/test.swf")
 125         error = system(COMPARE[1]+ " -Q 300 -p%d '%s' -o /tmp/test.swf" % (page, filename))
 126         if error:
 127             raise ConversionError(error)
 128         unlink(file1)
 129         error = system(SWFRENDER + " /tmp/test.swf -o %s" % file1)
 130         if error:
 131             raise ConversionError(error)
 132         unlink("/tmp/test.swf")
 133
 134         unlink(file12)
 135         pic1 = Image.open(file1)
 136         pic1.load()
 137         self.width1 = pic1.size[0]
 138         self.height1 = pic1.size[1]
 139
 140         pic2 = Image.open(file2)
 141         pic2.load()
 142         self.width2 = pic2.size[0]
 143         self.height2 = pic2.size[1]
 144
 145         if abs(self.width1-self.width2)>5 or abs(self.height1!=self.height2)>5:
 146             badness += 65536*abs(self.width2-self.width1)*max(self.height1,self.height2)+65536*abs(self.height2-self.height1)*max(self.width1,self.width2)
 147
 148         minx = min(self.width1,self.width2)
 149         miny = min(self.height1,self.height2)
 150
 151         pic1 = pic1.crop((0,0,minx,miny))
 152         pic1 = pic1.convert("RGB")
 153         pic1 = pic1.filter(ImageFilter.BLUR)
 154         pic2 = pic2.crop((0,0,minx,miny))
 155         pic2 = pic2.convert("RGB")
 156         pic2 = pic2.filter(ImageFilter.BLUR)
 157
 158         diffimage = ImageChops.difference(pic1,pic2)
 159         diffimage.save(file12, "PNG")
 160
 161         # compute quadratical difference
 162         diff = diffimage.histogram()
 163         for i in range(1,128):
 164             badness += (diff[i] + diff[256-i])*float(i*i)
 165             badness += (diff[256+i] + diff[256+256-i])*float(i*i)
 166             badness += (diff[512+i] + diff[512+256-i])*float(i*i)
 167
 168         badness /= (minx*miny)*3
 169
 170         return badness
 171
 172     def compare(self):
 173         try:
 174             global counter
 175             self.file1 = str(counter) + ".png"
 176             counter = counter + 1
 177             self.file2 = str(counter) + ".png"
 178             counter = counter + 1
 179             self.file12 = str(counter) + ".png"
 180             counter = counter + 1
 181             self.rating = self.runtools(self.filename, self.page, OUTPUTDIR + self.file1, OUTPUTDIR + self.file2, OUTPUTDIR + self.file12)
 182         except BadMatch:
 183             self.rating = 65534.0
 184             self.message = formatException()
 185             print self.message
 186         except ConversionError:
 187             self.rating = 65535.0
 188             self.message = formatException()
 189             print self.message
 190         except TooComplexError:
 191             self.rating = 65536.0
 192             self.message = formatException()
 193             print self.message
 194         except:
 195             self.rating = 65537.0
 196             self.message = formatException()
 197             print self.message
 198
 199     def getsizes(self):
 200         if self.message:
 201             return ""
 202         if abs(self.width1 - self.width2) > 5 or \
 203            abs(self.height1 - self.height2) > 5:
 204                return '<font color="red">%dx%d <-> %dx%d</font>' % (self.width1, self.height1, self.width2, self.height2)
 205         else:
 206                return '%dx%d,%dx%d' % (self.width1, self.height1, self.width2, self.height2)
 207
 208     def generatehtml(self):
 209         global OUTPUTDIR
 210         global counter
 211         self.html12 = str(counter) + ".html"
 212         counter = counter + 1
 213         self.htmldiff = str(counter) + ".html"
 214         counter = counter + 1
 215         fi = open(OUTPUTDIR + self.html12, "wb")
 216         fi.write(getTALstr("""
 217 <html><head></head>
 218 <body>
 219 <tal:block tal:replace="python:'File: '+self.filename"/><br>
 220 <tal:block tal:replace="python:'Page: '+str(self.page)"/><br>
 221 <tal:block tal:replace="python:'Rating: '+str(self.rating)"/><br>
 222 <pre tal:condition="python:self.message" tal:replace="python:'Message: '+str(self.message)"/><br>
 223 <hr>
 224 <table cellspacing="0" cellpadding="0">
 225 <tr><td><img tal:attributes="src python:self.file1"/></td><td><img tal:attributes="src python:self.file2"/></td></tr>
 226 <tr><td>pdf2swf Version</td><td>pdftoppm Version</td></tr>
 227 </table>
 228 <hr>
 229 </body>
 230 </html>""", {"self": self}))
 231         fi.close()
 232
 233         fi = open(OUTPUTDIR + self.htmldiff, "wb")
 234         fi.write(getTALstr("""
 235 <html><head></head>
 236 <body>
 237 <tal:block tal:replace="python:'File: '+self.filename"/><br>
 238 <tal:block tal:replace="python:'Page: '+str(self.page)"/><br>
 239 <tal:block tal:replace="python:'Rating: '+str(self.rating)"/><br>
 240 <pre tal:condition="python:self.message" tal:replace="python:'Message: '+str(self.message)"/><br>
 241 <hr>
 242 <img tal:attributes="src python:self.file12"/>
 243 <hr>
 244 </body>
 245 </html>""", {"self": self}))
 246         fi.close()
 247
 248         if self.message:
 249             self.htmlmessage = str(counter) + ".html"
 250             counter = counter + 1
 251             fi = open(OUTPUTDIR + self.htmlmessage, "wb")
 252             fi.write(getTALstr("""
 253 <html><head></head>
 254 <body>
 255 <pre tal:content="raw python:self.message">
 256 </pre>
 257 </body>
 258 </html>""", {"self": self}))
 259             fi.close()
 260
 261
 262
 263 def compare_pages(page1,page2):
 264     if page1.rating < page2.rating:
 265         return 1
 266     elif page1.rating > page2.rating:
 267         return -1
 268     else:
 269         return 0
 270
 271
 272 def add_directory(directory):
 273     if not os.path.isdir(directory):
 274         print "bad directory:",directory
 275         return
 276     for file in os.listdir(directory):
 277         global filenames
 278         filename = os.path.join(directory, file)
 279         if file.lower().endswith(".pdf"):
 280             filenames += [filename]
 281             print "+",filename
 282         elif os.path.isdir(filename):
 283             add_directory(filename)
 284
 285 pages = []
 286 try:
 287     os.mkdir(OUTPUTDIR)
 288 except: pass
 289
 290 for file in filenames:
 291     print "+",file
 292
 293 for dir in directories:
 294     add_directory(dir)
 295
 296 for filename in filenames:
 297     try:
 298         unlink("/tmp/test.txt")
 299         error = system(PDF2SWF + " -I %s -o /tmp/test.txt" % filename)
 300         if error:
 301             raise ConversionError(error)
 302         fi = open("/tmp/test.txt", "rb")
 303         for line in fi.readlines():
 304             p = {}
 305             for param in line.split(" "):
 306                 key,value = param.split("=")
 307                 p[key] = value
 308             page = int(p["page"])
 309             width = int(float(p["width"]))
 310             height = int(float(p["height"]))
 311             print filename, page, "%dx%d" % (width, height)
 312             pdfpage = PDFPage(filename, page, width, height)
 313             pdfpage.compare()
 314
 315             if width < 2000 and height < 2000:
 316                 pages += [pdfpage]
 317
 318             # only consider the first 3 pages
 319             if page > 3:
 320                 break
 321         fi.close()
 322     except KeyboardInterrupt:
 323         break
 324     except:
 325         pdfpage = PDFPage(filename, -1, -1, -1)
 326         pdfpage.rating = 65536.0
 327         pdfpage.message = formatException()
 328         pages += [pdfpage]
 329
 330 pages.sort(compare_pages)
 331
 332 position = 1
 333 for page in pages:
 334     page.generatehtml()
 335     page.position = position
 336     position = position + 1
 337
 338 fi = open(OUTPUTDIR + "index.html", "wb")
 339 fi.write(getTALstr("""<html>
 340 <head></head>
 341 <body>
 342 <table border="1"><tr><th>Position</th><th>Rating</th><th>File</th><th>Size</th><th>Page</th><th>Images</th><th>Diff</th><th>Further Info</th></tr>
 343 <tal:block tal:repeat="page pages">
 344 <tr>
 345 <td tal:content="python:page.position"/>
 346 <td tal:content="python:page.rating"/>
 347 <td tal:content="python:page.filename"/>
 348 <td tal:content="raw python:page.getsizes()"/>
 349 <td tal:content="python:page.page"/>
 350 <td><a tal:attributes="href python:page.html12">Side by Side</a></td>
 351 <td><a tal:attributes="href python:page.htmldiff">Difference</a></td>
 352 <td><a tal:condition="python:page.message" tal:attributes="href python:page.htmlmessage">Error message</a></td>
 353 </tr>
 354 </tal:block>
 355 </table>
 356 </body>
 357 </html>""", {"pages": pages}))
 358 fi.close()
 359