def parse(self, csvsrc):
text, encoding = self.detect_encoding(csvsrc, default_encodings=['utf-8', 'utf-16'])
#FIXME: raise parse error if encoding detection fails?
self.encoding = encoding or 'utf-8'
sniffer = csv.Sniffer()
sample = text[:1024]
try:
self.dialect = sniffer.sniff(sample)
if self.dialect.quoting == csv.QUOTE_MINIMAL:
#HACKISH: most probably a default, not real detection
self.dialect.quoting = csv.QUOTE_ALL
self.dialect.doublequote = True
except csv.Error:
self.dialect = 'default'
inputfile = csv.StringIO(text)
try:
fieldnames = detect_header(inputfile, self.dialect, self.fieldnames)
self.fieldnames = fieldnames
except csv.Error:
pass
inputfile.seek(0)
reader = try_dialects(inputfile, self.fieldnames, self.dialect)
first_row = True
for row in reader:
newce = self.UnitClass()
newce.fromdict(row)
if not first_row or not newce.match_header():
self.addunit(newce)
first_row = False
def getoutput(self):
output = csv.StringIO()
writer = csv.DictWriter(output, self.fieldnames,
extrasaction='ignore',
dialect=self.dialect)
writer.writeheader()
for ce in self.units:
writer.writerow(ce.todict())
return output.getvalue()
def serialize(self, out):
output = csv.StringIO()
writer = csv.DictWriter(output, FIELDNAMES, dialect="catkeys")
# No real headers, the first line contains metadata
writer.writerow(
dict(
zip(FIELDNAMES, [
self.header._header_dict[key] for key in FIELDNAMES_HEADER
])))
for unit in self.units:
writer.writerow(unit.dict)
out.write(output.getvalue().encode(self.encoding))
def serialize(self, out):
# Check first if there is at least one translated unit
translated_units = [u for u in self.units if u.istranslated()]
if not translated_units:
return
output = csv.StringIO()
writer = csv.DictWriter(output,
fieldnames=OMEGAT_FIELDNAMES,
dialect="omegat")
for unit in translated_units:
writer.writerow(unit.dict)
out.write(output.getvalue().encode(self.encoding))
def serialize(self, out):
# Check first if there is at least one translated unit
translated_units = [u for u in self.units if u.istranslated()]
if not translated_units:
return
output = csv.StringIO()
writer = csv.DictWriter(output, fieldnames=self._fieldnames, dialect="utx")
for unit in translated_units:
writer.writerow(unit.dict)
result = output.getvalue().encode(self.encoding)
out.write(self._write_header().encode(self.encoding))
out.write(result)
def serialize(self, out):
# Check first if there is at least one translated unit
translated_units = [u for u in self.units if u.istranslated()]
if not translated_units:
return
output = csv.StringIO()
writer = csv.DictWriter(output, fieldnames=WF_FIELDNAMES, dialect="wordfast")
# No real headers, the first line contains metadata
self.header.tucount = len(translated_units)
writer.writerow(dict(zip(WF_FIELDNAMES, [self.header.header[key] for key in WF_FIELDNAMES_HEADER])))
for unit in translated_units:
writer.writerow(unit.dict)
out.write(output.getvalue().encode(self.encoding))
def detect_header(sample, dialect, fieldnames):
"""Test if file has a header or not, also returns number of columns in first row"""
inputfile = csv.StringIO(sample)
try:
reader = csv.reader(inputfile, dialect)
except csv.Error:
try:
inputfile.seek(0)
reader = csv.reader(inputfile, 'default')
except csv.Error:
inputfile.seek(0)
reader = csv.reader(inputfile, 'excel')
header = next(reader)
columncount = max(len(header), 3)
if valid_fieldnames(header):
return header
return fieldnames[:columncount]
