def create_slp(code):
"""Create SLP file for a given dictionary code.
code is to be selected from dictcode.json.
"""
# ENSK -> ekaksharanamamala_sadhukalashagani
fullName = utils.code_to_dict(code)
# ekaksharanamamala, sadhukalashagani
bookName, author = fullName.split('_')
# Read the .txt file
filein = os.path.join('..', fullName, 'orig', bookName + '.txt')
fin = codecs.open(filein, 'r', 'utf-8')
data = fin.read()
fin.close()
# Convert the data to SLP1.
data = sanscript.transliterate(data, 'devanagari', 'slp1')
# Output directory
directory = os.path.join('..', fullName, 'slp')
# Create if the directory does not exist.
if not os.path.exists(directory):
os.mkdir(directory)
fileout = os.path.join(directory, bookName + '.txt')
# Create output file and save the SLP data in it.
fout = codecs.open(fileout, 'w', 'utf-8')
fout.write(data)
fout.close()
def apply_annotation(code):
"""Apply autoannotation to the dictionary corresponding to given code."""
# ENSK -> ekaksharanamamala_sadhukalashagani
fullName = utils.code_to_dict(code)
# ekaksharanamamala, sadhukalashagani
bookName, author = fullName.split('_')
# Read the .txt file
filein = os.path.join('..', fullName, 'orig', bookName + '.txt')
# Create output file.
directory = os.path.join('..', fullName, 'annotated')
if not os.path.exists(directory):
os.mkdir(directory)
fileout = os.path.join(directory, code + '0.txt')
# Annotate the filein and store in fileout.
annotate_kosha(filein, fileout)
def search_in_dict(query, code):
# ENSK -> ekaksharanamamala_sadhukalashagani
fullName = utils.code_to_dict(code)
# ekaksharanamamala, sadhukalashagani
bookName, author = fullName.split('_')
# Read the .txt file
filein = os.path.join('..', fullName, 'slp', bookName + '.txt')
result = []
verseDetails = utils.VerseInfo()
verse = ''
writeVerse = False
for lin in codecs.open(filein, 'r', 'utf-8'):
if lin.startswith(';'):
(tag, value) = utils.extract_tag(lin)
if tag == 'p':
verseDetails.update_pageNum(value)
if tag == 'k':
verseDetails.update_kanda(value)
if tag == 'v':
verseDetails.update_varga(value)
if tag == 'vv':
verseDetails.update_subvarga(value)
elif re.search('^[$#]', lin):
pass
else:
verse += lin
if query in lin:
writeVerse = True
if '..' in lin:
verseDetails.update_verseNum(verse)
if writeVerse:
page = verseDetails.give_page_details()
kanda = verseDetails.kanda
varga = verseDetails.varga
adhyaya = verseDetails.subvarga
versenum = verseDetails.verseNum
result.append({
'verse': verse,
'page': page,
'versenum': versenum,
'kanda': kanda,
'varga': varga,
'adhyaya': adhyaya
})
writeVerse = False
verse = ''
return result
def replace_colon(code):
# ENSK -> ekaksharanamamala_sadhukalashagani
fullName = utils.code_to_dict(code)
# ekaksharanamamala, sadhukalashagani
bookName, author = fullName.split('_')
# Read the .txt file
filein = os.path.join('..', fullName, 'orig', bookName + '.txt')
fin = codecs.open(filein, 'r', 'utf-8')
data = fin.read()
fin.close()
data = re.sub(r'([^ps]):', r'\g<1>рдГ', data)
directory = os.path.join('..', fullName, 'orig')
if not os.path.exists(directory):
os.mkdir(directory)
fileout = os.path.join(directory, bookName + '.txt')
fout = codecs.open(fileout, 'w', 'utf-8')
fout.write(data)
fout.close()
def create_slp(code):
# ENSK -> ekaksharanamamala_sadhukalashagani
fullName = utils.code_to_dict(code)
# ekaksharanamamala, sadhukalashagani
bookName, author = fullName.split('_')
# Read the .txt file
filein = os.path.join('..', fullName, 'orig', bookName + '.txt')
fin = codecs.open(filein, 'r', 'utf-8')
data = fin.read()
fin.close()
data = sanscript.transliterate(data, 'devanagari', 'slp1')
directory = os.path.join('..', fullName, 'slp')
if not os.path.exists(directory):
os.mkdir(directory)
fileout = os.path.join(directory, bookName + '.txt')
fout = codecs.open(fileout, 'w', 'utf-8')
fout.write(data)
fout.close()
def apply_changes(code):
"""Apply idempotent changes to file corresponding to the given code."""
# ENSK -> ekaksharanamamala_sadhukalashagani
fullName = utils.code_to_dict(code)
# ekaksharanamamala, sadhukalashagani
bookName, author = fullName.split('_')
# Read the .txt file
filein = os.path.join('..', fullName, 'orig', bookName + '.txt')
fin = codecs.open(filein, 'r', 'utf-8')
# Create path if does not exist.
directory = os.path.join('..', fullName, 'orig')
if not os.path.exists(directory):
os.mkdir(directory)
# Output file.
fileout = os.path.join(directory, bookName + '1.txt')
fout = codecs.open(fileout, 'w', 'utf-8')
# Apply idempotent corrections to each line.
for line in fin:
line = idempotent_corrections(line)
fout.write(line)
fin.close()
fout.close()
dom = ET.parse(xmlfile)
xslt = ET.parse(xsltfile)
transform = ET.XSLT(xslt)
newdom = transform(dom)
result = ET.tostring(newdom, pretty_print=True)
fout = codecs.open(htmlfile, 'wb')
fout.write(result)
fout.close()
print('HTML generated. Success!')
if __name__ == "__main__":
# Read the unique code of dictionary from arguments. ENSK
code = sys.argv[1]
# ENSK -> ekaksharanamamala_sadhukalashagani
fullName = utils.code_to_dict(code)
# ekaksharanamamala, sadhukalashagani
bookName, author = fullName.split('_')
# Read the .txt file
filein = os.path.join('..', fullName, 'orig', bookName + '.txt')
fin = codecs.open(filein, 'r', 'utf-8')
data = fin.read()
fin.close()
# Get filename of babylon to store the output
babylonfile = os.path.join('..', fullName, 'babylon',
bookName + '.babylon')
# Get directory to store MD.
mdDirectory = os.path.join('..', fullName, 'md')
# Get directory to store JSON.
jsonfile = os.path.join('..', fullName, 'json', bookName + '.json')
# Get filename of xml to store the output