def random_filename(fnamelength=8, ext_prob=1):
"""Create a filename as a random sequance of letters and numbers, possibly with an extension."""
fname = random_string(fnamelength)
ext = ''
if random.random() < ext_prob:
ext = '.' + random_string(3)
return fname + ext
def create_fake_email_contents(n_headers=5):
"""Create email contents similar to what is requested by email RFC."""
contents = []
# Create required headers
for header, prob, generator in GENERATED_EMAIL_HEADERS:
if random.random() < prob:
contents.append(header + generator())
# Create additional fake headers
for i in range(n_headers):
contents.append(random_string(20, chars=FNAMECHARS) + ': ' + random_string(30, FNAMECHARS))
# Add empty line separating headers and body
contents.append('')
# Add body
contents.append(random_string(200, chars=FCONTENTSCHARS))
return '\n'.join(contents)
def create_fake_email_contents(n_headers=5):
"""Create email contents similar to what is requested by email RFC."""
contents = []
# Create required headers
for header, prob, generator in GENERATED_EMAIL_HEADERS:
if random.random() < prob:
contents.append(header + generator())
# Create additional fake headers
for i in range(n_headers):
contents.append(
random_string(20, chars=FNAMECHARS) + ': ' +
random_string(30, FNAMECHARS))
# Add empty line separating headers and body
contents.append('')
# Add body
contents.append(random_string(200, chars=FCONTENTSCHARS))
return '\n'.join(contents)
def create_corpus_dictionary(nitems=NEMAILS):
"""Create a random dictionary of email file names and their contents."""
d = {}
for i in range(nitems):
filename = random_filename()
contents = random_string(200, chars=FCONTENTSCHARS)
d[filename] = contents
return d
def create_corpus_dictionary(nitems=N_EMAILS):
"""Create a random dictionary of email file names and their contents."""
d = {}
for i in range(nitems):
filename = random_filename()
contents = random_string(200, chars=FCONTENTS_CHARS)
d[filename] = contents
return d
def random_email_address(namelength=5, domain2length=7, domain1length=3):
return random_string(namelength, LCCHARS) + \
'@' + \
random_string(domain2length, LCCHARS) + \
'.' + \
random_string(domain1length, LCCHARS)
"""Create a random dictionary of email file names and their contents."""
d = {}
for i in range(nitems):
filename = random_filename()
contents = create_fake_email_contents()
d[filename] = contents
return d
# The following structure contains a triple for each header:
# (header string, probability of generating this header, function used to generate the contents)
GENERATED_EMAIL_HEADERS = [('Date: ', 1.0, random_date),
('From: ', 1.0, random_email_address),
('To: ', 0.5, random_email_address),
('Subject: ', 0.5,
lambda: random_string(30, chars=FNAMECHARS + ' '))]
def create_fake_email_contents(n_headers=5):
"""Create email contents similar to what is requested by email RFC."""
contents = []
# Create required headers
for header, prob, generator in GENERATED_EMAIL_HEADERS:
if random.random() < prob:
contents.append(header + generator())
# Create additional fake headers
for i in range(n_headers):
contents.append(
random_string(20, chars=FNAMECHARS) + ': ' +
random_string(30, FNAMECHARS))
# Add empty line separating headers and body
def random_email_address(namelength=5, domain2length=7, domain1length=3):
return random_string(namelength, LCCHARS) + \
'@' + \
random_string(domain2length, LCCHARS) + \
'.' + \
random_string(domain1length, LCCHARS)
def create_corpus_dictionary(nitems=N_EMAILS):
"""Create a random dictionary of email file names and their contents."""
d = {}
for i in range(nitems):
filename = random_filename()
contents = create_fake_email_contents()
d[filename] = contents
return d
# The following structure contains a triple for each header:
# (header string, probability of generating this header, function used to generate the contents)
GENERATED_EMAIL_HEADERS = [
('Date: ', 1.0, random_date),
('From: ', 1.0, random_email_address),
('To: ', 0.5, random_email_address),
('Subject: ', 0.5, lambda: random_string(30, chars=FNAMECHARS+' '))
]
def create_fake_email_contents(n_headers=5):
"""Create email contents similar to what is requested by email RFC."""
contents = []
# Create required headers
for header, prob, generator in GENERATED_EMAIL_HEADERS:
if random.random() < prob:
contents.append(header + generator())
# Create additional fake headers
for i in range(n_headers):
contents.append(random_string(20, chars=FNAMECHARS) + ': ' + random_string(30, FNAMECHARS))
# Add empty line separating headers and body
contents.append('')
# Add body