moin_to_markdown.py

#!/usr/bin/python
# This is a simple script to convert a very limited set of MoinMoin wiki syntax
#to markdown. I am using it in my migration from MoinMoin to a gitit+markdown
#wiki.
DO_GIT = False
#DISABLED#DO_GIT = True
import os
import re
REPLACEMENTS = (
    ("(./)", "[x]"),
)
WIKI_WORDS = """SomePage
AnotherPage""".split()

class HeaderConverter(object):
    expression = re.compile("^(=+)(.*?)=+")

    def process(self, line):
        match = self.expression.match(line)
        if match:
            level = len(match.group(1))
            return ("#" * level) +  match.group(2).rstrip()
        return line

class UnorderedListConverter(object):
    expression = re.compile("^( +)\*(.*)")

    def process(self, line):
        match = self.expression.match(line)
        if match:
            level = len(match.group(1))
            return ("    " * (level - 1)) + "*" +  match.group(2).rstrip()
        return line


class OrderedListConverter(object):
    expression = re.compile("^( +)\\d+\.(.*)")

    def process(self, line):
        match = self.expression.match(line)
        if match:
            level = len(match.group(1))
            return ("    " * (level - 1)) + "1." +  match.group(2).rstrip()
        return line

class WikiWordConverter(object):

    def process(self, line):
        for word in WIKI_WORDS:
            if word in line:
                return line.replace(word, "[%s]()" % word)
        return line

def convert(moin):
    lineConverters = [HeaderConverter(), UnorderedListConverter(),
        OrderedListConverter(), WikiWordConverter()]
    markdownLines = []
    for line in moin.split("\n"):
        for converter in lineConverters:
            line = converter.process(line)
        markdownLines.append(line)
    markdown = "\n".join(markdownLines)
    for pair in REPLACEMENTS:
        markdown = markdown.replace(pair[0], pair[1])
    return markdown

def visit(wikiWords, dirName, fileNames):
    fileNames.sort()
    if not dirName.endswith("/revisions"):
        return
    parentDir = os.path.basename(os.path.dirname(dirName))
    if parentDir not in wikiWords and not parentDir.startswith("WhiteBoard"):
        return
    print "Processing %s revisions of page %s" %(len(fileNames), parentDir)
    for revision in fileNames:
        inFile = open(os.path.join(dirName, revision))
        outFile = open(parentDir + ".page", "w")
        outFile.write(convert(inFile.read()))
        inFile.close()
        outFile.close()
        if DO_GIT:
            os.system("git add '%s.page'" % parentDir)
            os.system("git commit -m 'Automatic conversion from MoinMoin to" \
        " gitit/Markdown via the moin_to_markdown.py script." \
        " Page: %s Revision: %s'" % (parentDir, revision))

if __name__ == "__main__":
    os.path.walk("/var/local/plwiki/data/pages", visit, WIKI_WORDS)