import re
import sys
import string
def starbox_catalog_disambiguation(s):
"""
Argument is article text.
Returns a list consisting of (1) new article text and (2) a
human-readable summary of the changes made.
"""
#
# List of acronyms and what they should disambiguate to
#
changes = [
['ADS', 'Aitken Double Star Catalogue'],
['AG', 'Astronomische Gesellschaft Katalog'],
['BD', 'Durchmusterung'],
['CCDM', 'Catalog of Components of Double and Multiple Stars'],
['CD', 'Durchmusterung'],
['CEL', 'Celescope Catalogue of Ultraviolet Magnitudes'], # redlink
# cf. http://adsabs.harvard.edu/abs/1973SAOSR.350....1D
['CPD', 'Durchmusterung'],
['CSI', 'Catalog of Stellar Identifications'], # redlink
['FK4', 'Fourth Fundamental Catalogue'],
['FK5', 'Fifth Fundamental Catalogue'],
['GC', 'Boss General Catalogue'],
['GCTP', 'General Catalogue of Trigonometric Parallaxes'],
['GSC', 'Guide Star Catalog'],
['HD', 'Henry Draper Catalogue'],
['HIC', 'Hipparcos Input Catalogue'], # redlink
['HIP', 'Hipparcos Catalogue'],
['HR', 'Bright Star Catalogue'],
['IDS', 'Index Catalogue of Visual Double Stars'],
['LFT', 'Luyten Five-Tenths catalogue'],
['LHS', 'Luyten Half-Second catalogue'],
['LTT', 'Luyten Two-Tenths catalogue'],
['N30', 'Catalog of 5,268 Standard Stars Based on the Normal System N30'], # redlink
# cf. http://cdsarc.u-strasbg.fr/viz-bin/Cat?I/80
['NLTT', 'New Luyten Two-Tenths catalogue'],
['NSV', 'New Catalogue of Suspected Variable Stars'], # redlink
['PHL', 'Palomar-Haro-Luyten catalogue'], # redlink
['PLX', 'General Catalogue of Trigonometric Parallaxes'],
['PMC', 'Tokyo Photoelectric Meridian Circle Catalog'], # redlink
['PPM', 'PPM Star Catalogue'],
['ROT', 'Catalogue of Rotational Velocities of the Stars'], # redlink
# cf. http://adsabs.harvard.edu/abs/1970crvs.book.....U
['SAO', 'Smithsonian Astrophysical Observatory Star Catalog'],
['TD1', 'TD1 Catalog of Stellar Ultraviolet Fluxes'], # redlink
['TYC', 'Tycho Catalogue'],
['UBV', 'UBV Photoelectric Catalogue'], # redlink
# cf. http://adsabs.harvard.edu/abs/1968PUSNO..21....0B
['WDS', 'Washington Double Star Catalog']
]
#
# Partial list of redlinked acronyms left as is:
#
# GCRV (General Catalogue of Stellar Radial Velocities)
# JP11
# MSX5C
# uvby98 (uvbyβ photoelectric photometric catalogue)
#
# Partial list of bluelinked acronyms that are left as is:
#
# AAVSO (article is on the organization that generated this database)
# 2MASS (destination article correct)
# EUVE (article is on the satellite that generated this catalog)
# IRAS (article is on the satellite that generated this catalog)
#
r = re.compile(r'({{Starbox catalog[^}]*}})')
ans = r.split(s)
summary = ''
for i in range((len(ans)-1)/2):
j=2*i+1
newansj = ans[j]
for k in changes:
newansj = newansj.replace('[['+k[0]+']]', '[['+k[1]+'|'+k[0]+']]')
if (newansj != ans[j]):
summary = summary + 'Disambiguated\n' + ans[j] + '\n\nto\n' + newansj + '\n.\n'
ans[j] = newansj
newstr = ''
for q in ans:
newstr += q
return [newstr, summary]
#
# Main program: tests starbox_catalog_disambiguation() function
#
# Article input name=arg 1, output name=arg 2
n1=sys.argv[1]
n2=sys.argv[2]
f1=open(n1,'r')
f2=open(n2,'w')
edits=starbox_catalog_disambiguation(f1.read())
f1.close()
print 'Edits:\n'
print edits[1]
f2.write(edits[0])
f2.close()
#
