######################################################################
# diff experiments
#
# todo
# separate out html vs. email diff methods

from string import join, split, atoi
from DocumentTemplate.DT_Util import html_quote
from struct import pack, unpack
from OFS.History import historicalRevision
import TextFormatter
from OFS import ndiff
# Tim Peters' ndiff is now python 2.1's difflib module
# use the former for compatibility with older zopes
import re
def ISJUNK(line, pat=re.compile(r"\s*$").match):
    return pat(line) is not None

###########################################################################
# CLASS ZwikiDiffMixin
# RESPONSIBILITIES
# - encapsulate ZWikiPage diff functionality in a separate file
# - generate human-readable diffs between page versions
# COLLABORATORS ZWikiPage
###########################################################################

class ZWikiDiffMixin:
    """
    This mix-in class adds some diff methods to ZWikiPage.
    """
    
    ######################################################################
    # METHOD CATEGORY: diff
    ######################################################################

    def lasttext(self, versionsBack=1):
        """return text of the last or an earlier revision"""
        versionsBack = int(versionsBack)
        try:
            # problem here - manage_change_history only gets the last
            # 20 revisions
            lastrevision = self.manage_change_history()[versionsBack]
            key = lastrevision['key']
            serial=apply(pack, ('>HHHH',)+tuple(map(atoi, split(key,'.'))))
            lastself=historicalRevision(self, serial)
            return lastself.text()
        except:
            # return '' if we don't have a version that old
            return ''

    def textDiff(self,revA=1,revB=0,a=None,b=None):
        """
        generate a plain text diff, optimized for human readability,
        between two revisions of this page, numbering back from the latest.
        Alternately, a and/or b texts can be specified.
        """
        revA, revB = int(revA), int(revB)
        a = a or self.lasttext(versionsBack=revA)
        b = b or self.lasttext(versionsBack=revB)

        # wrap (but don't fill or pad) long lines
        formatter = TextFormatter.TextFormatter((
            {'width':70, 'fill':0, 'pad':0},
            ))
        a=formatter.compose((a,))
        b=formatter.compose((b,))
            
        a = split(a,'\n')
        b = split(b,'\n')
        cruncher=ndiff.SequenceMatcher(
            #isjunk=split,
            isjunk=ISJUNK,
            #isjunk=lambda x: x in " \\t", # requires newer difflib
            a=a,
            b=b)

        r = []
        for tag, alo, ahi, blo, bhi in cruncher.get_opcodes():
            if tag == 'replace':
                r.append('??changed:')
                r = r + self._abbreviateDiffLines(a[alo:ahi],'-',10)
                r = r + self._abbreviateDiffLines(b[blo:bhi],'',50)
                r.append('')
            elif tag == 'delete':
                r.append('--removed:')
                r = r + self._abbreviateDiffLines(a[alo:ahi],'-',10)
                r.append('')
            elif tag == 'insert':
                r.append('++added:')
                r = r + self._abbreviateDiffLines(b[blo:bhi],'',50)
                r.append('')
            elif tag == 'equal':
                pass
            else:
                raise ValueError, 'unknown tag ' + `tag`

        return join(r, '\n')+'\n'

    def _abbreviateDiffLines(self,lines,prefix,maxlines=5):
        output = []
        if maxlines and len(lines) > maxlines:
            extra = len(lines) - maxlines
            for i in xrange(maxlines - 1):
                output.append(prefix + lines[i])
            output.append(prefix + "[%d more line%s...]" %
                          (extra, ((extra == 1) and '') or 's')) # not working
        else:
            for line in lines:
                output.append(prefix + line)
        return output

    def diff(self,revA=1,revB=0,showSteps=0):
        """
        display a human-readable diff, formatted for web display,
        between two revisions of this page, numbering back from the latest.
        Also display some navigation links.
        """
        revA, revB = int(revA), int(revB)

        #if revA < revB:
        #    revA, revB = revB, revA
        #if showSteps and (revA-revB) != 1:
        #    for r in range(revA,revB):
        #        t = t + self.diff(r,r-1)
        #    return t
            
        t = self.textDiff(revA=revA,revB=revB)
        t = html_quote(t)
        
        # careful, don't feed the regexps..
        t = re.sub(r'(?s)(\?\?changed.*?)\n((-.*?\n)+?)((-\[[0-9]+ more.*?\]\n)?)([^-].*?)((\[[0-9]+ more.*?\]\n)?)(?=(<b>)?\+\+added|(<b>)?--removed|(<b>)?\?\?changed|$)',
                   r'<b>\1</b><span style="color:red; text-decoration:line-through"><overstrike>\n\2</overstrike></span>\4<span style="color:green">\6</span>\7',
                   t)

        t = re.sub(r'(?s)(\+\+added.*?\n)(.*?)((\[[0-9]+ more.*?\]\n)?)(?=(<b>)?\+\+added|(<b>)?--removed|(<b>)?\?\?changed|$)',
                   r'<b>\1</b><span style="color:green">\2</span>\3',
                   t)
        # -[43 more lines...] should not get matched in \2


        t = re.sub(r'(?s)(\-\-removed.*?\n)((-.*?\n)+?)((-\[[0-9]+ more.*?\]\n)?)(?=(<b>)?\+\+added|(<b>)?--removed|(<b>)?\?\?changed|$)',
                   r'<b>\1</b><span style="color:red; text-decoration:line-through"><overstrike>\2</overstrike></span>\4',
                   t)

        # move this out to dtml/* or something
        return """\
<html>
<body>
<div align="right">
<a href="%s/manage_change_history_page">full history</a> | 
%s | 
%s | 
<a href="%s">return to page</a>
</div>
<pre>
%s
</pre>
</body>
</html>
""" % (self.page_url(),
       (revA>=19 and '<< previous edit') # we only see 20 revs right now
         or '<a href="%s/diff?revA=%d&revB=%d"><< previous edit</a>' %(
           self.page_url(), revA+1, revA),
       (revB==0 and '<a href="%s">next edit >></a>' % (self.page_url()))
         or '<a href="%s/diff?revA=%d&revB=%d">next edit >></a>' % (
           self.page_url(), revB, max(revB-1,0)),
       self.page_url(),
       t)

    def oldDiff(self,revA=1,revB=0):
        """
        display a zope-page-history-style html-formatted diff 
        between two revisions of this page, numbering back from the latest.
        """
        revA, revB = int(revA), int(revB)
        a=split(self.lasttext(versionsBack=revA),'\n')
        b=split(self.lasttext(versionsBack=revB),'\n')
        cruncher=ndiff.SequenceMatcher(isjunk=split, a=a, b=b)

        r = ['<table border=1 width="100%">']
        for tag, alo, ahi, blo, bhi in cruncher.get_opcodes():
            if tag == 'replace':
                replace(a, alo, ahi, b, blo, bhi, r)
            elif tag == 'delete':
                dump('-', a, alo, ahi, r)
            elif tag == 'insert':
                dump('+', b, blo, bhi, r)
            elif tag == 'equal':
                pass     #dump(' ', a, alo, ahi, r)
            else:
                raise ValueError, 'unknown tag ' + `tag`
        r.append('</table>')
        diff = join(r, '\n')
        return '<html>\n<body>\n'+diff+'</body>\n</html>\n'



######################################################################
# FUNCTION CATEGORY: diff helper functions
######################################################################

def dump(tag, x, lo, hi, r):
    r1=[]
    r2=[]
    for i in xrange(lo, hi):
        r1.append(tag)
        r2.append(x[i])
    r.append("<tr>\n"
            "<td valign=top width=1%%><pre>\n%s\n</pre></td>\n"
            "<td valign=top width=99%%><pre>\n%s\n</pre></td>\n"
            "</tr>\n"
            % (join(r1,'\n'), html_quote(join(r2, '\n'))))

def replace(x, xlo, xhi, y, ylo, yhi, r):

    rx1=[]
    rx2=[]
    for i in xrange(xlo, xhi):
        rx1.append('-')
        rx2.append(x[i])

    ry1=[]
    ry2=[]
    for i in xrange(ylo, yhi):
        ry1.append('+')
        ry2.append(y[i])


    r.append("<tr>\n"
            "<td valign=top width=1%%><pre>\n%s\n%s\n</pre></td>\n"
            "<td valign=top width=99%%><pre>\n%s\n%s\n</pre></td>\n"
            "</tr>\n"
            % (join(rx1, '\n'), join(ry1, '\n'),
               html_quote(join(rx2, '\n')), html_quote(join(ry2, '\n'))))