pmuellr/progit-to-html.py

## progit-to-html.py
#!/usr/bin/env python

# convert the pro git book into HTML,
# suitable for reading in Dropbox on the iPhone

import os
import re
import sys
import base64
import subprocess

iDir = None

htmlHeader = """
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/>
<title>%(title)s</title>

<style>
h1,h2 {
    padding:               0.3em 0.3em 0.3em 1em;
    border:                1 solid #000;
    background:            #EEE;
   z-box-shadow:            0.3em 0.3em 0.3em 0em #DDD;
   z-moz-box-shadow:       0.3em 0.3em 0.3em 0em #DDD;
   z-webkit-box-shadow:    0.3em 0.3em 0.3em 0em #DDD;
    border-radius:         0.5em;
    -moz-border-radius:    0.5em;
    -webkit-border-radius: 0.5em;
}

.page-break {
    page-break-before:  always;
}

li.toc {
    font-size:          150%%;
    margin-top:         0.5em;
    margin-bottom:      0.5em;
}

pre, xmp {
    border-color:       #CBCBCB;
    background:         #FFF8ED;
    color:              #000000;
    overflow:           visible;
    border-width:       1px;
    border-style:       dotted;
    padding-top:        0.4em;
    padding-bottom:     0.4em;
    padding-left:       1.4em;
    padding-right:      1.4em;
    margin-left:        1.4em;
    margin-right:       1.4em;
  z-font-size:          80%%;
}

.caption {
    margin-left:        2em;
    font-style:         italic;
}

@media only screen and (max-device-width: 480px) {
    body {
        font-size: 200%%;
    }
}

@media only screen and (min-device-width: 481px) {
    body {
        font-size: 125%%;
    }
}

</style>
</head>
<body>

<h1>%(title)s</h1>
"""

htmlTrailer = """
</body>
</html>
"""

#----------------------------------------------------------------------------------------
class File:

    def __init__(self, fileName, chapName):
        self.fileName = fileName
        self.chapName = chapName.replace("-", " ")
        self.chapId   = chapName
        self.content  = ""

#----------------------------------------------------------------------------------------
def gather_input_files(iDir):
    result = []

    entries = os.listdir(iDir)

    for entry in entries:
        entryDir = os.path.join(iDir, entry)
        srcFiles = os.listdir(entryDir)
        if len(srcFiles) != 1:
            print "The directory %s does not appear to be a progit source directory" % entryDir

        fileName = os.path.join(entryDir, srcFiles[0])
        chapName = entry
        file = File(fileName, chapName)

        result.append(file)

    return result

# <p>Insert 18333fig0101.png

#----------------------------------------------------------------------------------------
def filter_images(content):
    lines = content.split("\n")

    pattern = re.compile(r'^<p>Insert (.*)\.png\s*$')

    for lineNo, line in enumerate(lines):
        match = pattern.match(line)
        if not match: continue


        imageName = "%s-tn.png" % match.group(1)
        imageName = os.path.join(iDir, "../figures", imageName)
        # print "   found an image: %s" % imageName

        imageFile    = open(imageName, "r")
        imageContent = imageFile.read()
        imageFile.close()

        imageBase64 = base64.b64encode(imageContent)

        lines[lineNo] = "<p><img src='data:image/png;base64,%s'></p><p class='caption'>" % imageBase64


    return "\n".join(lines)

#----------------------------------------------------------------------------------------
def generate_output_file(iFile):

    iFileName = iFile.fileName
    chapName  = iFile.chapName
    oFileName = chapName + ".html"

    print "reading %s, writing %s" % (iFileName, oFileName)

    content = subprocess.Popen(["Markdown.pl", iFileName], stdout=subprocess.PIPE).communicate()[0]

    content = filter_images(content)
    iFile.content = content

    oFile = open(oFileName, "w")
    title = "Pro Git - Chapter %s" % chapName
    oFile.write(htmlHeader % {"title": title})
    oFile.write(content)
    oFile.write(htmlTrailer)
    oFile.close()

    return content

#----------------------------------------------------------------------------------------

if len(sys.argv) != 2:
    print "Expecting directory name of progit files"
    sys.exit()

iDir = sys.argv[1]

iFiles = gather_input_files(iDir)


for iFile in iFiles:
    generate_output_file(iFile)

oFileName = "pro-git.html"
oFile = open(oFileName, "w")
oFile.write(htmlHeader % {"title": "Pro Git"})
oFile.write("<h1>Contents</h1>")
oFile.write("<ul>")

for iFile in iFiles:
    oFile.write("<li class='toc'><a href='#%s'>%s</a>" % (iFile.chapId, iFile.chapName))

oFile.write("</ul>")

for iFile in iFiles:
    oFile.write("<h1 class='page-break' id='%s'>%s</h1>\n%s" % (iFile.chapId, iFile.chapName, iFile.content))

oFile.write(htmlTrailer)
oFile.close()

print "wrote %s" % oFileName
	#!/usr/bin/env python

	# convert the pro git book into HTML,
	# suitable for reading in Dropbox on the iPhone

	import os
	import re
	import sys
	import base64
	import subprocess

	iDir = None

	htmlHeader = """
	<html>
	<head>
	<meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/>
	<title>%(title)s</title>

	<style>
	h1,h2 {
	padding: 0.3em 0.3em 0.3em 1em;
	border: 1 solid #000;
	background: #EEE;
	z-box-shadow: 0.3em 0.3em 0.3em 0em #DDD;
	z-moz-box-shadow: 0.3em 0.3em 0.3em 0em #DDD;
	z-webkit-box-shadow: 0.3em 0.3em 0.3em 0em #DDD;
	border-radius: 0.5em;
	-moz-border-radius: 0.5em;
	-webkit-border-radius: 0.5em;
	}

	.page-break {
	page-break-before: always;
	}

	li.toc {
	font-size: 150%%;
	margin-top: 0.5em;
	margin-bottom: 0.5em;
	}

	pre, xmp {
	border-color: #CBCBCB;
	background: #FFF8ED;
	color: #000000;
	overflow: visible;
	border-width: 1px;
	border-style: dotted;
	padding-top: 0.4em;
	padding-bottom: 0.4em;
	padding-left: 1.4em;
	padding-right: 1.4em;
	margin-left: 1.4em;
	margin-right: 1.4em;
	z-font-size: 80%%;
	}

	.caption {
	margin-left: 2em;
	font-style: italic;
	}

	@media only screen and (max-device-width: 480px) {
	body {
	font-size: 200%%;
	}
	}

	@media only screen and (min-device-width: 481px) {
	body {
	font-size: 125%%;
	}
	}

	</style>
	</head>
	<body>

	<h1>%(title)s</h1>
	"""

	htmlTrailer = """
	</body>
	</html>
	"""

	#----------------------------------------------------------------------------------------
	class File:

	def __init__(self, fileName, chapName):
	self.fileName = fileName
	self.chapName = chapName.replace("-", " ")
	self.chapId = chapName
	self.content = ""

	#----------------------------------------------------------------------------------------
	def gather_input_files(iDir):
	result = []

	entries = os.listdir(iDir)

	for entry in entries:
	entryDir = os.path.join(iDir, entry)
	srcFiles = os.listdir(entryDir)
	if len(srcFiles) != 1:
	print "The directory %s does not appear to be a progit source directory" % entryDir

	fileName = os.path.join(entryDir, srcFiles[0])
	chapName = entry
	file = File(fileName, chapName)

	result.append(file)

	return result

	# <p>Insert 18333fig0101.png

	#----------------------------------------------------------------------------------------
	def filter_images(content):
	lines = content.split("\n")

	pattern = re.compile(r'^<p>Insert (.)\.png\s$')

	for lineNo, line in enumerate(lines):
	match = pattern.match(line)
	if not match: continue


	imageName = "%s-tn.png" % match.group(1)
	imageName = os.path.join(iDir, "../figures", imageName)
	# print " found an image: %s" % imageName

	imageFile = open(imageName, "r")
	imageContent = imageFile.read()
	imageFile.close()

	imageBase64 = base64.b64encode(imageContent)

	lines[lineNo] = "<p><img src='data:image/png;base64,%s'></p><p class='caption'>" % imageBase64


	return "\n".join(lines)

	#----------------------------------------------------------------------------------------
	def generate_output_file(iFile):

	iFileName = iFile.fileName
	chapName = iFile.chapName
	oFileName = chapName + ".html"

	print "reading %s, writing %s" % (iFileName, oFileName)

	content = subprocess.Popen(["Markdown.pl", iFileName], stdout=subprocess.PIPE).communicate()[0]

	content = filter_images(content)
	iFile.content = content

	oFile = open(oFileName, "w")
	title = "Pro Git - Chapter %s" % chapName
	oFile.write(htmlHeader % {"title": title})
	oFile.write(content)
	oFile.write(htmlTrailer)
	oFile.close()

	return content

	#----------------------------------------------------------------------------------------

	if len(sys.argv) != 2:
	print "Expecting directory name of progit files"
	sys.exit()

	iDir = sys.argv[1]

	iFiles = gather_input_files(iDir)


	for iFile in iFiles:
	generate_output_file(iFile)

	oFileName = "pro-git.html"
	oFile = open(oFileName, "w")
	oFile.write(htmlHeader % {"title": "Pro Git"})
	oFile.write("<h1>Contents</h1>")
	oFile.write("<ul>")

	for iFile in iFiles:
	oFile.write("<li class='toc'><a href='#%s'>%s</a>" % (iFile.chapId, iFile.chapName))

	oFile.write("</ul>")

	for iFile in iFiles:
	oFile.write("<h1 class='page-break' id='%s'>%s</h1>\n%s" % (iFile.chapId, iFile.chapName, iFile.content))

	oFile.write(htmlTrailer)
	oFile.close()

	print "wrote %s" % oFileName