#!/usr/bin/env python
'''
Read two shell files with specific formatting to parse out 
common records and produce and HTML table.
 
It is used to extract the common records from a bash shell script file
and tcsh shell script file.
 
The idea is to create records in the bash and tcsh files with the same
id so that this script can scoop them up and put them in a table.
 
Records are surrounded by:
 
  # BEGIN: <id>
  <statements>
  # END
 
The id can be any string but it must be the same in both files.
Anything in between is printed.
 
THe #@@ token is used to specify non-executing records.
 
Here is a simple example:
 
   $ # ================================================================
   $ # bash commands
   $ # ================================================================
   $ cat compare.sh <<EOF
   #!/bin/bash
 
   # BEGIN: alias
   alias foo=bar
   # END
 
   # BEGIN: array: iterate by item, single line
   A=("a" "b" "c")
   for x in ${A[@]} ; do echo $x; done
   # END
 
   # BEGIN: array: iterate by index
   for i in ${!ARRAY[@]}; do
       printf "%d %s\n" $i ${ARRAY[i]}
   done
   # END
   EOF
 
   $ # ================================================================
   $ # tcsh commands
   $ # ================================================================
   $ cat compare.csh <<EOF
   # BEGIN: alias
   alias foo bar
   # END
 
   # BEGIN: array: iterate by item, single line
   #@@ N/A
   # END
 
   # BEGIN: array: iterate by index
   foreach i ( `seq $#ARRAY` )
       echo $i $ARRAY[$i] |\
           awk '{printf("%d %s\n",$1,$2);}'
   end
   # END
   EOF
 
   $ # ================================================================
   $ # Generate the HTML.
   $ # ================================================================
   $ ./genhtml.py compare.sh compare.csh > compare.html
'''
import sys
import os
import re
 
 
# ================================================================
# err
# ================================================================
def err(msg):
    '''
    Print an error message and exit.
    '''
    print('ERROR: %s' % (msg))
    sys.exit(1)
 
 
# ================================================================
# err
# ================================================================
def parse(fn):
    '''
    Parse a shell script file looking for specific format
    instructions in the comments.
    '''
    fp = open(fn,'r')
    rows = {}
    row = None
    n=0
    for line in fp.readlines():
        n += 1
        line = line.rstrip()
 
        # begin
        m = re.search('^# BEGIN: (.+)$',line)
        if m:
            if row != None:
                err('missing END statement around line: %d in %s' % (n,fn))
 
            title = m.group(1)
            if title in rows:
                err('duplicate entry "%s" at line %d in %s' % (title,n,fn))
 
            row = {'title':title,
                   'file':fn,
                   'lineno':n,
                   'lines':[],
                   }
            continue
 
        # end
        m = re.search('^# END',line)
        if m:
            rows[title]=row
            row = None
            continue
 
        # between begin and end, everything else is ignored
        if row:
            row['lines'].append(line)
 
    fp.close()
    return rows
 
 
# ================================================================
# htmlize
# ================================================================
def htmlize(line):
    '''
    HTMLize a string.
    '''
    line = line.replace('&','&amp;')
    line = line.replace('"','&quot;')
    line = line.replace("'",'&apos;')
    line = line.replace('<','&lt;')
    line = line.replace('>','&gt;')
    line = line.replace(' ','&nbsp;')
    line = line.replace('`','&#96;')
    line = line.replace('-','&ndash;')
    return line
 
# ================================================================
# myprint
# ================================================================
def myprint(m):
    '''
    Write a string to stdout.
    '''
    sys.stdout.write(str(m))
 
 
# ================================================================
# cell
# ================================================================
def cell(indent, rec):
    '''
    Write out a column cell.
    '''
    myprint('%s<td class="mycell-code">' % (indent))
    first=True
    for line in rec['lines']:
        if not first:
            myprint('<br />')
        first = False
        line = line.replace('#@@ ','')
        myprint(htmlize(line))
 
    myprint('</td>\n')
 
 
# ================================================================
# gencss
# ================================================================
def gencss():
    '''
    Generate the CSS styling information specifically for
    wordpress.
    '''
 
    css ='''
    <style type="text/css">
      table.mytable {
        font-family: Arial, Helvetica; 
        font-size: 1.2em;
        border-collapse: separate;
        border-spacing: 1px;
        *border-collapse: expression('separate', cellSpacing = '1px');
        border: 2px solid lightgray;
      }
      tr.myrow-even {
        vertical-align: top;
        background-color: #ffffea;
        border: 2px solid lightgray;
      }
      tr.myrow-odd {
        vertical-align: top;
        background-color: none;
        border: 2px solid lightgray;
      }
      th.mycell-hdr {
        vertical-align: top;
        text-align: center;
        padding: 4px 4px;
        border: 2px solid lightgray;
        /*font-size: 1.3em;*/
      }
      td.mycell-id {
        vertical-align: top;
        text-align: right;
        padding: 2px 2px;
        border: 1px solid lightgray;
        font-size: 0.9em;
        line-height: 100%;
      }
      td.mycell-title {
        vertical-align: top;
        text-align: left;
        padding: 2px 2px;
        border: 1px solid lightgray;
        font-size: 0.9em;
        line-height: 100%;
      }
      td.mycell-code {
        vertical-align: top;
        font-family: "Courier New", monospace; 
        text-align: left;
        padding: 2px 2px;
        border: 1px solid lightgray;
        font-size: 0.9em;
        line-height: 100%;
      }
      div#mydiv {
        margin-left: -30px;
        width: 110%;
      }
    </style>
 '''
    return css
 
# ================================================================
# main
# ================================================================
def main():
    '''
    Main entry point.
    '''
    # This is a simple script, no need for fancy argument handling.
    assert len(sys.argv) == 3
 
    bash_shell = sys.argv[1]
    tcsh_shell = sys.argv[2]
 
    bash_recs = parse(bash_shell)
    tcsh_recs = parse(tcsh_shell)
    
    if len(bash_recs) != len(tcsh_recs):
        err('Mismatched number of records: %d != %d' % (len(bash_recs),
                                                        len(tcsh_recs)))
 
    for title in sorted(bash_recs):
        if title not in tcsh_recs:
            err('missing entry "%s"' % (title))
 
    myprint(gencss())
    print '''
    <div id="mydiv">
      <table>
        <thead>
          <tr bgcolor="#DDDDDD">
            <th class="mycell-hdr">ID</th> 
            <th class="mycell-hdr">Operation</th> 
            <th class="mycell-hdr">bash</th>
            <th class="mycell-hdr">tcsh</th>
          </tr>
        </thead>
        <tbody>
'''
    tr_indent = '          '
    td_indent = tr_indent + '  '
    i=0
    for title in sorted(bash_recs, key=lambda x: x.lower()):
        bash_rec=bash_recs[title]
        tcsh_rec=tcsh_recs[title]
        
        if (i%2) == 0:
            myprint('%s<tr class="myrow-even">\n' % (tr_indent))
        else:
            myprint('%s<tr class="myrow-odd">\n' % (tr_indent))
        myprint('%s<td class="mycell-id">%d</td>\n' % (td_indent, i+1))
        myprint('%s<td class="mycell-title">%s</td>\n' % (td_indent, title))
        cell(td_indent, bash_rec)
        cell(td_indent, tcsh_rec)
        myprint('%s</tr>\n' % (tr_indent))
        i += 1
 
    print '''
        </tbody>
      </table>
    </div>
'''
 
if __name__ == '__main__':
    main()