run.py

XML tutorial

#!/usr/bin/python

import os,re,sys


try:
    lang = str(sys.argv[1])
except IndexError:
    lang=""

print "lang: " + lang
if (not(lang == '')):
    langsuffix = "_" + lang
    langparam = "lang=" + lang
    langsuffixparam = "langsuffix=" + langsuffix
else:
    langsuffix = ""
    langparam = ""
    langsuffixparam = ""



if re.compile("Python\Z").search(os.getcwd()):
    os.chdir("..") 

os.chdir("ExampleSources" + langsuffix)

data=''

for f in os.listdir("."):
    if not(re.compile("\.xml\Z").search(f)):
           continue

    input=open(f)
    data = data + input.read()
    input.close()


os.chdir('..')
### data=re.compile('<description lang="cs">.*?</description>',re.S).sub('',data)

output=open("Data" + langsuffix + "/source.xml","w")

sys.stdout = output
print """<!-- generated by run.py  -->
<!DOCTYPE allExamples [
 <!ENTITY eacute '&#233;'>
 <!ENTITY ugrave '&#249;'>
 <!ENTITY agrave '&#224;'>
 <!ENTITY egrave '&#232;'>
 <!ENTITY ograve '&#242;'>
 <!ENTITY icirc  '&#238;'>
 <!ENTITY ecirc  '&#234;'>
 <!ENTITY ccedil '&#231;'>
]>
<allExamples>
"""
print data
print "</allExamples>"
output.close()

sys.stdout = sys.__stdout__
command = "saxonclient -o General" + langsuffix + "/contents.html Data" + langsuffix + "/source.xml XSL/stylesheet.xsl pwd=${PWD}/ " + langparam +" "+ langsuffixparam
print "command: " + command
os.system(command)


os.chdir("Output" + langsuffix)

def aaa(hit):
#    print hit.group(2)
    text = re.compile("\n",re.S).sub('<br></br>',hit.group(2))
    text = re.compile(" ",re.S).sub('&nbsp;',text)
    text = re.compile("\t",re.S).sub('&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;',text)
    text = re.compile("span&nbsp;class",re.S).sub('span class',text)
#    print text

    return hit.group(1) + text +hit.group(3)

for f in os.listdir("."):
#    print "file " + f
    if not(re.compile("\.html\Z").search(f)):
           continue

    input=open(f)
    data = input.read()
    input.close()

    data=re.compile('(<td class="doc">)(.*?)(</td>)',re.S).sub(aaa,data)
    
    output=open(f,'w')
    output.write(data)
    output.close()