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", "xpdf -C"]
  20 OUTPUTDIR = "results.clip/"
  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 runtool(self, cmd, filename, page, file):
  94         if cmd.startswith("xpdf"):
  95             unlink("/tmp/test-%06d.ppm" % page)
  96             args = cmd[4:]
  97             error = system(PDFTOPPM + "%s -r 72 -f %d -l %d '%s' /tmp/test" % (args, page, page, filename))
  98             if error:
  99                 raise ConversionError(error)
 100             unlink(file)
 101             error = system(CONVERT + " /tmp/test-%06d.ppm  %s" % (page, file))
 102             if error:
 103                 raise ConversionError(error)
 104             unlink("/tmp/test-%06d.ppm" % page)
 105         else:
 106             unlink("/tmp/test.swf")
 107             unlink("svp.ps")
 108             error = system(COMPARE[0]+ " -Q 300 -p%d '%s' -o /tmp/test.swf" % (page, filename))
 109             #system("mv svp.ps %s.ps" % randstr())
 110             if error and "supports 65536" in error:
 111                 raise TooComplexError(error)
 112             if error:
 113                 raise ConversionError(error)
 114             unlink(file)
 115             error = system(SWFRENDER + " /tmp/test.swf -o %s" % file)
 116             if error:
 117                 raise ConversionError(error)
 118             unlink("/tmp/test.swf")
 119
 120     def runtools(self, filename, page, file1, file2, file12):
 121
 122         badness = 0.0
 123
 124         self.runtool(COMPARE[0], filename, page, file2)
 125         self.runtool(COMPARE[1], filename, page, file1)
 126
 127         unlink(file12)
 128
 129         pic1 = Image.open(file1)
 130         pic1.load()
 131         self.width1 = pic1.size[0]
 132         self.height1 = pic1.size[1]
 133
 134         pic2 = Image.open(file2)
 135         pic2.load()
 136         self.width2 = pic2.size[0]
 137         self.height2 = pic2.size[1]
 138
 139         if abs(self.width1-self.width2)>5 or abs(self.height1!=self.height2)>5:
 140             badness += 65536*abs(self.width2-self.width1)*max(self.height1,self.height2)+65536*abs(self.height2-self.height1)*max(self.width1,self.width2)
 141
 142         minx = min(self.width1,self.width2)
 143         miny = min(self.height1,self.height2)
 144
 145         pic1 = pic1.crop((0,0,minx,miny))
 146         pic1 = pic1.convert("RGB")
 147         pic1 = pic1.filter(ImageFilter.BLUR)
 148         pic2 = pic2.crop((0,0,minx,miny))
 149         pic2 = pic2.convert("RGB")
 150         pic2 = pic2.filter(ImageFilter.BLUR)
 151
 152         diffimage = ImageChops.difference(pic1,pic2)
 153         diffimage.save(file12, "PNG")
 154
 155         # compute quadratical difference
 156         diff = diffimage.histogram()
 157         for i in range(1,128):
 158             badness += (diff[i] + diff[256-i])*float(i*i)
 159             badness += (diff[256+i] + diff[256+256-i])*float(i*i)
 160             badness += (diff[512+i] + diff[512+256-i])*float(i*i)
 161
 162         badness /= (minx*miny)*3
 163
 164         return badness
 165
 166     def compare(self):
 167         try:
 168             global counter
 169             self.file1 = str(counter) + ".png"
 170             counter = counter + 1
 171             self.file2 = str(counter) + ".png"
 172             counter = counter + 1
 173             self.file12 = str(counter) + ".png"
 174             counter = counter + 1
 175             self.rating = self.runtools(self.filename, self.page, OUTPUTDIR + self.file1, OUTPUTDIR + self.file2, OUTPUTDIR + self.file12)
 176         except BadMatch:
 177             self.rating = 65534.0
 178             self.message = formatException()
 179             print self.message
 180         except ConversionError:
 181             self.rating = 65535.0
 182             self.message = formatException()
 183             print self.message
 184         except TooComplexError:
 185             self.rating = 65536.0
 186             self.message = formatException()
 187             print self.message
 188         except:
 189             self.rating = 65537.0
 190             self.message = formatException()
 191             print self.message
 192
 193     def getsizes(self):
 194         if self.message:
 195             return ""
 196         if abs(self.width1 - self.width2) > 5 or \
 197            abs(self.height1 - self.height2) > 5:
 198                return '<font color="red">%dx%d <-> %dx%d</font>' % (self.width1, self.height1, self.width2, self.height2)
 199         else:
 200                return '%dx%d,%dx%d' % (self.width1, self.height1, self.width2, self.height2)
 201
 202     def generatehtml(self):
 203         global OUTPUTDIR
 204         global counter
 205         self.html12 = str(counter) + ".html"
 206         counter = counter + 1
 207         self.htmldiff = str(counter) + ".html"
 208         counter = counter + 1
 209         fi = open(OUTPUTDIR + self.html12, "wb")
 210         fi.write(getTALstr("""
 211 <html><head></head>
 212 <body>
 213 <tal:block tal:replace="python:'File: '+self.filename"/><br>
 214 <tal:block tal:replace="python:'Page: '+str(self.page)"/><br>
 215 <tal:block tal:replace="python:'Rating: '+str(self.rating)"/><br>
 216 <pre tal:condition="python:self.message" tal:replace="python:'Message: '+str(self.message)"/><br>
 217 <hr>
 218 <table cellspacing="0" cellpadding="0">
 219 <tr><td><img tal:attributes="src python:self.file1"/></td><td><img tal:attributes="src python:self.file2"/></td></tr>
 220 <tr><td>pdf2swf Version</td><td>pdftoppm Version</td></tr>
 221 </table>
 222 <hr>
 223 </body>
 224 </html>""", {"self": self}))
 225         fi.close()
 226
 227         fi = open(OUTPUTDIR + self.htmldiff, "wb")
 228         fi.write(getTALstr("""
 229 <html><head></head>
 230 <body>
 231 <tal:block tal:replace="python:'File: '+self.filename"/><br>
 232 <tal:block tal:replace="python:'Page: '+str(self.page)"/><br>
 233 <tal:block tal:replace="python:'Rating: '+str(self.rating)"/><br>
 234 <pre tal:condition="python:self.message" tal:replace="python:'Message: '+str(self.message)"/><br>
 235 <hr>
 236 <img tal:attributes="src python:self.file12"/>
 237 <hr>
 238 </body>
 239 </html>""", {"self": self}))
 240         fi.close()
 241
 242         if self.message:
 243             self.htmlmessage = str(counter) + ".html"
 244             counter = counter + 1
 245             fi = open(OUTPUTDIR + self.htmlmessage, "wb")
 246             fi.write(getTALstr("""
 247 <html><head></head>
 248 <body>
 249 <pre tal:content="raw python:self.message">
 250 </pre>
 251 </body>
 252 </html>""", {"self": self}))
 253             fi.close()
 254
 255
 256
 257 def compare_pages(page1,page2):
 258     if page1.rating < page2.rating:
 259         return 1
 260     elif page1.rating > page2.rating:
 261         return -1
 262     else:
 263         return 0
 264
 265
 266 def add_directory(directory):
 267     if not os.path.isdir(directory):
 268         print "bad directory:",directory
 269         return
 270     for file in os.listdir(directory):
 271         global filenames
 272         filename = os.path.join(directory, file)
 273         if file.lower().endswith(".pdf"):
 274             filenames += [filename]
 275             print "+",filename
 276         elif os.path.isdir(filename):
 277             add_directory(filename)
 278
 279 pages = []
 280 try:
 281     os.mkdir(OUTPUTDIR)
 282 except: pass
 283
 284 for file in filenames:
 285     print "+",file
 286
 287 for dir in directories:
 288     add_directory(dir)
 289
 290 for filename in filenames:
 291     try:
 292         unlink("/tmp/test.txt")
 293         error = system(PDF2SWF + " -I %s -o /tmp/test.txt" % filename)
 294         if error:
 295             raise ConversionError(error)
 296         fi = open("/tmp/test.txt", "rb")
 297         for line in fi.readlines():
 298             p = {}
 299             for param in line.split(" "):
 300                 key,value = param.split("=")
 301                 p[key] = value
 302             page = int(p["page"])
 303             width = int(float(p["width"]))
 304             height = int(float(p["height"]))
 305             print filename, page, "%dx%d" % (width, height)
 306             pdfpage = PDFPage(filename, page, width, height)
 307             pdfpage.compare()
 308
 309             if width < 2000 and height < 2000:
 310                 pages += [pdfpage]
 311
 312             # only consider the first 3 pages
 313             if page > 3:
 314                 break
 315         fi.close()
 316     except KeyboardInterrupt:
 317         break
 318     except:
 319         pdfpage = PDFPage(filename, -1, -1, -1)
 320         pdfpage.rating = 65536.0
 321         pdfpage.message = formatException()
 322         pages += [pdfpage]
 323
 324 pages.sort(compare_pages)
 325
 326 position = 1
 327 for page in pages:
 328     page.generatehtml()
 329     page.position = position
 330     position = position + 1
 331
 332 fi = open(OUTPUTDIR + "index.html", "wb")
 333 fi.write(getTALstr("""<html>
 334 <head></head>
 335 <body>
 336 <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>
 337 <tal:block tal:repeat="page pages">
 338 <tr>
 339 <td tal:content="python:page.position"/>
 340 <td tal:content="python:page.rating"/>
 341 <td tal:content="python:page.filename"/>
 342 <td tal:content="raw python:page.getsizes()"/>
 343 <td tal:content="python:page.page"/>
 344 <td><a tal:attributes="href python:page.html12">Side by Side</a></td>
 345 <td><a tal:attributes="href python:page.htmldiff">Difference</a></td>
 346 <td><a tal:condition="python:page.message" tal:attributes="href python:page.htmlmessage">Error message</a></td>
 347 </tr>
 348 </tal:block>
 349 </table>
 350 </body>
 351 </html>""", {"pages": pages}))
 352 fi.close()
 353