#!/usr/bin/python
##****p* utils/fix_bib.py
##  NAME
##    fix_bib.py  --  fix formatting in a bib file
##  SYNOPSIS
##    python fix_bib.py in the directory where biblio.bib resides
##  DESCRIPTION
##    based on values of options,
##    open, read lines and close original file
##    
##    replace 'note' with 'availableas' keyword
##    
##    indentation:
##      loop over lines,
##        find lines with a '=' before the indent column,
##          insert spaces between the '=' and the word preceding it,
##          so that the '=' ends at the indent column
##        remove trailing newline character '\n' from all lines,
##      open new file and write all new_lines to it
##    
##    capitalization keywords: (title, publisher,booktitle,address,number,abstract:
##      loop over lines,
##        find line that start with any of the above keywords,
##          find first left and last right curly bracket, 
##            add a pair of curly brackets around them
##        remove trailing newline character '\n' from all lines,
##      open new file and write all new_lines to it
##  USES
##    string
##  MODIFICATION HISTORY
##    2008-07-10 -- WvH : - replace 'note' with 'availableas'
##                        - add curly braces around titles, publisher, number,...
##  TODO
##    add command-line options?
##  COPYRIGHT
##    Copyright (c) 2008 Wim Van Hoydonck
##  AUTHOR
##    Wim Van Hoydonck
##  CREATION DATE
##    2008-06-24
##  SOURCE
import string

orig_filename   = "test.bib"
new_filename    = "test_out.bib"
indent          = 17
spaces          = indent*" "
do_indent       = True
capitalize      = ["title",
                   "publisher",
                   "journal",
                   "booktitle",
                   "address",
                   "number",
                   "abstract"]
notekeyw        = "note"


# as python starts counting from 0, indent should be 1 less
indent          -= 1


# open, read lines and close original file
bib_file        = open(orig_filename,'r')
bib_lines       = bib_file.readlines()
bib_file.close()


# init new_lines as array
new_lines = []

new_lines2 = []
# replace 'note' by 'availableas'
# which is not a bibtex keyword
for line in bib_lines:
    sline = string.strip(line)
    idx   = sline.find(notekeyw)
    if ( idx == 0 ):
        lidx,ridx = line.find(notekeyw) , line.rfind(notekeyw)
        eidx = line.find(" =")
        # prepend line with a comment mark
        line = line[0:lidx]+"availableas"+line[ridx+4:]
    new_lines2.append(line)

new_lines = new_lines2

new_lines2 = []
# indent lines properly
for line in new_lines:
    # find lines with a '=' before the 17th column,
    idx = line.find("=")
    if (0 < idx < indent) and (line[0] != "%"):
        # insert spaces between the '=' and the word preceding it,
        # so that the '=' ends at the 17th column
        line = line[0:idx] + spaces[0:indent-idx] + line[idx:]
    # remove trailing newline character '\n' from all lines,
    new_lines2.append(line[0:-1])

new_lines = new_lines2

new_lines2 = []
# add a pair of curly braces around lines that start with certain keywords
for line in new_lines:
    for keyw in capitalize:
        # remove leading blanks
        sline   = string.strip(line)
        idx     = sline.find(keyw)
        # keyword is first thing in line
        if ( idx == 0 ):
            # find position of first left curly bracket and last right curly bracket
            lidx,ridx = line.find("{") , line.rfind("}")
            #print lidx,ridx , keyw
            # 
            if ( lidx < ridx ):
                line = line[0:lidx]+"{"+line[lidx:ridx]+"}"+line[ridx:]
    new_lines2.append(line)

new_lines = new_lines2


# open new file and write all new_lines to it
bib1_file = open(new_filename,'w')

for line in new_lines:
    print >> bib1_file , line
##******