Convert Word-style annotations to Agile Author style annotations for Manning books

convert_agile_author_annotations.py

'''convert_agile_author_annotations.py

Written February 25, 2012 by Josiah Carlson
Released into the public domain.

Did you write your annotions for Agile Author using the Word style?

<example id="...">
<title>title</title>
<programlisting>
def foo():
    line1 #A
    line2 #B
    line3 #B

#A annotation 1
#B annotation 2
</programlisting></example>

Do you now need to convert them using the Agile Author co, callout, and
calloutlist elements? I feel your pain. This module will pull the code and
annotations from inside the programlisting element, and using the id given
for the example element, produce the following output:

def foo():
    line1 <co id="...-A" linkends="...-Ar"/>
    line2 <co id="...-B" linkends="...-Br"/>
    line3 <co id="...-B2" linkends="...-Br"/>
</programlisting>
<calloutlist>
    <callout id="...-Ar" arearefs="...-A"><para>annotation 1</para></callout>
    <callout id="...-Br" arearefs="...-B ...-B2"><para>annotation 2</para></callout>
</calloutlist>
</example>

You can take this output and replace everything after the <programlisting>
element until the end of the </example> element.

Note: this code will only print the first code listing it comes across, then it
will quit. This is on purpose, primarily to allow you to take one example at
a time to convert.
'''

from collections import defaultdict
import re
import sys

from BeautifulSoup import BeautifulStoneSoup as Soup

def main(files):
    for f in files:
        parsed = Soup(open(f, 'rb').read())
        for tag in parsed.findAll('example'):
            text = tag.find('programlisting').text
            if '#' not in text:
                continue
            pieces = re.split('#([A-Z])', text)
            id = [v for k,v in tag.attrs if k == 'id'][0]
            code_done = False
            counts = defaultdict(int)
            out1 = []
            for i, piece in enumerate(pieces):
                odd = i & 1
                if not code_done and not odd and piece[-1:] == '\n':
                    code_done = True
                    out1.append(piece.rstrip() + '\n')
                    out1.append('</programlisting>\n')
                    out1.append('<calloutlist>\n')
                    continue
                if code_done:
                    if odd:
                        out1.append('    <callout id="%s-%sr" arearefs="%s"><para>'%(
                            id, piece, ' '.join(
                                id+'-'+piece+(str(i+1) if i else '') for i in xrange(counts[piece]))))
                    else:
                        out1.append(' '.join(piece.split()) + '</para></callout>\n')
                else:
                    if odd:
                        counts[piece] += 1
                        out1.append('<co id="%s-%s%s" linkends="%s-%sr"/>'%(
                            id, piece, counts[piece] if counts[piece] > 1 else '', id, piece))
                    else:
                        out1.append(piece)
            out1.append('</calloutlist>\n</example>')
            print "".join(out1)
            
            
            raise SystemExit

if __name__ == '__main__':
    if len(sys.argv) > 1:
        main(sys.argv[1:])
    else:
        print "usage: python convert_agile_author_annotations.py [agile author xml files]"