打印HTML到PDF

#! /usr/bin/env python

# Convert HTML pages to PDF.
# Copyright 2012 by Akkana Peck -- share and enjoy under the GPL v2 or later.
#
# Thx to http://notes.alexdong.com/xhtml-to-pdf-using-pyqt4-webkit-and-headless
#
# To convert all the output files into a single multi-page PDF, use this:
# gs -dNOPAUSE -sDEVICE=pdfwrite -sOUTPUTFILE=multipage.pdf -dBATCH file*.pdf

import sys
from PyQt4.QtCore import *
from PyQt4.QtGui import *
from PyQt4.QtWebKit import *
from pyPdf import PdfFileWriter, PdfFileReader

pagelist = sys.argv[1:]
pagelistindex = 0
pdflist = []

app = QApplication(sys.argv)

web = QWebView()

# QPrinter always prints too small, and I can't figure out why.
# Empirically, I can get dot-for-dot with this zoom factor:
# Printing to 1366 (with speaker notes), use zoom factor 1.24.
# Printing to 1024 to remove the speaker notes, use zoom factor 2.0.
web.setZoomFactor(2.0)

# You can show the window, but it's not necessary:
#web.show()

printer = QPrinter(mode = QPrinter.ScreenResolution)
printer.setPaperSize(QSizeF(1024, 768), QPrinter.DevicePixel)
printer.setOutputFormat(QPrinter.PdfFormat)
printer.setPageMargins(0 , 0 , 0 , 0 , QPrinter.DevicePixel)
printer.setFullPage(True)

def makeMultipage(outfile, inlist) :
    # Make a multi-page document from all the pages we've created:
    output = PdfFileWriter()
    for infile in inlist :
        inpdf = PdfFileReader(file(infile, "rb"))
        output.addPage(inpdf.getPage(0))

    # finally, write "output" to document-output.pdf
    outputStream = file(outfile, "wb")
    output.write(outputStream)
    outputStream.close()
    print 'Wrote all pages to', outfile

def print_next() :
    '''Print the page that has just loaded, then start the load
       of the next page in the list. Called from loadFinished signal.
    '''
    global pagelist, pagelistindex, web, printer

    outputfile = "../pdf/file%03d.pdf" % pagelistindex
    printer.setOutputFileName(outputfile)
    pdflist.append(outputfile)
    web.print_(printer)
    print "Generated", outputfile, "from", pagelist[pagelistindex]

    pagelistindex += 1
    if pagelistindex >= len(pagelist) :
        makeMultipage('../pdf/all.pdf', pdflist)

        print "Exiting"
        QApplication.exit()
        # That doesn't always work, so hedge our bets:
        sys.exit(0)

    # Load the next URL in the list
    #print "Loading", pagelist[pagelistindex]
    web.load(QUrl(pagelist[pagelistindex]))

QObject.connect(web, SIGNAL("loadFinished(bool)"), print_next)
web.load(QUrl(pagelist[pagelistindex]))

sys.exit(app.exec_())

http://shallowsky.com/software/presentation/qhtmlprint

https://github.com/antialize/wkhtmltopdf

https://github.com/gmanny/Pechkin