# -*- coding: utf-8 -*-

"""Generates, from the tables of the file "units.mdwn", a xml file for
use with udunits2.

Author: Pierre de Buyl
License: BSD 3-clause
"""

import codecs

import re

# r is regular expression to match unit symbols and the associated exponents.
r = re.compile(r"(?P<symbol>\w+)(?P<exp>-?\d)?")

def convert_def(s):
    """
    Convert a unit definition from the H5MD table format to a udunits2
    acceptable format.
    """
    res = u''
    for i in s.split():
        symbol, exp = r.findall(i)[0]
        if len(exp)>0:
            if exp[:1] == u'-':
                res += u' 1/'+symbol
                if exp!=u'-1': res += u'^'+exp[1:]
            else:
                if exp!=u'1': res += symbol+u'^'+exp
        else:
            res += u' '+symbol
    return res.strip()

def convert_superscript(s):
    tmp = u''.join(dict(zip(u"⁻⁰¹²³⁴⁵⁶⁷⁸⁹", u"-0123456789")).get(c, c) for c in s)
    if tmp.lstrip()[:1]==u'1':
        return tmp.lstrip()[1:]
    else:
        return tmp


# Initialize the variables for the table parser below.
IN_TABLE=1
state = 0
tables=[]

# Parse the file 'units.mdwn' to extract the tables.
with codecs.open('units.mdwn', 'rb', encoding='utf8') as f:
    ll = f.readlines()
    for l in ll:
        if state==0 and l.lstrip()[:2]=='+-':
            state = IN_TABLE
            t = []
            tables.append(t)
            continue
        if state==IN_TABLE:
            if l.lstrip()[:1]=='|':
                #print "table line = ", l
                t.append( ";".join([i.strip() for i in l.strip().split('|') if len(i)>0]))
            elif (len(l.strip())==0):
                state=0
                continue

print """<?xml version="1.0" encoding="US-ASCII"?>
<!--
Copyright 2014 The H5MD Developers

This file is part of H5MD.  See the file LICENSE in the top-level
source-directory of the package for copying and redistribution conditions.

Units for the module 'units' system 'SI'.

-->
<unit-system>
<!-- SI base units -->"""

# The following units require a special handling because the carry no
# dimension.
dimensionless = [u'rad', u'sr']
# The following units require a special handling because they are
# aliases to existing units.
aliases = [u'Bq', u'Sv']

def format_alias(flag):
    if flag:
        return """
       </aliases>"""
    else:
        return ""

def write_xml(table, base=False):
    for i in table:
        j = i.split(';')
        if j[1]==u'symbol': continue
        print """    <unit>"""
        if j[1] in dimensionless:
            print """    <dimensionless/>"""
        else:
            if base:
                print """    <base/>"""
            else:
                print u"""        <def>%s</def>""" % convert_def(convert_superscript(j[3].split('=')[1]))
                if j[1] in aliases:
                    print """        <aliases>"""
        print """        <name><singular>%s</singular></name>""" % j[2]
        if j[1]!=j[2]: print """        <symbol>%s</symbol>%s""" % (j[1], format_alias(j[1] in aliases))
        print """    </unit>"""

def convert_prefix(s):
    tmp = u''.join(dict(zip(u"⁻⁰¹²³⁴⁵⁶⁷⁸⁹", u"-0123456789")).get(c, c) for c in s)
    if tmp[:2]==u'10':
        tmp = u'1E' + tmp[2:]
    else:
        raise ValueError("Bad prefix.")
    return tmp
def write_xml_prefixes(table):
    for i in table[1:]:
        j = i.split(';')
        print u"""    <prefix>
        <value>%s</value> <name>%s</name>  <symbol>%s</symbol>
    </prefix>""" % (convert_prefix(j[2]), j[0][:-1], j[1])
write_xml(tables[1], base=True)
print "<!-- SI derived units -->"
write_xml(tables[2])
print "<!-- SI prefixes -->"
write_xml_prefixes(tables[3])

print """</unit-system>"""