Changeset 529 for pkpgcounter/trunk/pkpgpdls

Timestamp:

11/28/07 20:48:45 (16 years ago)

Author:

jerome

Message:

Finalized support for MS documents.

Location:

pkpgcounter/trunk/pkpgpdls

Files:

• analyzer.py (modified) (1 diff)
• mscrap.py (modified) (2 diffs)
• pdlparser.py (modified) (1 diff)
• version.py (modified) (1 diff)

pkpgcounter/trunk/pkpgpdls/analyzer.py

@@ -166 +166 @@
                        plain) :     # IMPORTANT : don't move this one up !
             try :               
-                return module.Parser(self, (firstblock, lastblock))
+                return module.Parser(self, self.filename, 
+                                           (firstblock, lastblock))
             except pdlparser.PDLParserError :
                 pass # try next parser

pkpgcounter/trunk/pkpgpdls/mscrap.py

@@ -24 +24 @@
 
 import os
-import urllib2
+import tempfile
 
 import pdlparser
@@ -56 +56 @@
             
     def getJobSize(self) :
-        """Counts pages in a Microsoft Word (r) (tm) (c) (etc...) document."""
-        return 0
+        """Counts pages in a Microsoft Word (r) (tm) (c) (etc...) document.
+
+           First we convert from .doc to .ps, then we use the PostScript parser.
+        """
+        doctops = 'xvfb-run -a abiword --import-extension=.doc --print="%(outfname)s" "%(infname)s"'
+        workfile = tempfile.NamedTemporaryFile(mode="w+b")
+        try :
+            outfname = workfile.name
+            infname = self.filename
+            status = os.system(doctops % locals())
+            if status or not os.stat(outfname).st_size :
+                raise pdlparser.PDLParserError, "Impossible to convert input document %(infname)s to PostScript" % locals()
+            psinputfile = open(outfname, "rb")
+            try :
+                (first, last) = self.parent.readFirstAndLastBlocks(psinputfile)
+                import postscript
+                return postscript.Parser(self.parent, 
+                                         outfname, 
+                                         (first, last)).getJobSize()
+            finally :
+                psinputfile.close()
+        finally :    
+            workfile.close()
+        raise pdlparser.PDLParserError, "Impossible to count pages in %(infname)s" % locals()

pkpgcounter/trunk/pkpgpdls/pdlparser.py

@@ -43 +43 @@
     required = []               # Default list of required commands
     openmode = "rb"             # Default file opening mode
-    def __init__(self, parent, (firstblock, lastblock)) :
+    def __init__(self, parent, filename, (firstblock, lastblock)) :
         """Initialize the generic parser."""
         self.parent = parent
         # We need some copies for later inclusion of parsers which
         # would modify the parent's values
-        self.filename = parent.filename[:]
+        self.filename = filename[:]
         self.firstblock = firstblock[:]
         self.lastblock = lastblock[:]

pkpgcounter/trunk/pkpgpdls/version.py

@@ -22 +22 @@
 
 
-__version__ = "3.40alpha"
+__version__ = "3.40"
 
 __doc__ = """pkpgcounter : a generic Page Description Languages parser."""