def tag_cloud(dist, id_ = "", class_ = "", width=None, height=None,
max_size=None, min_size=None, max_words = None,
start_color=None, end_color=None, color_steps=None,
sort_order="random"):
''' returns a dict with style and body elements. style contains
defalt styling for the tag cloud, while body contains the html
markup. '''
# sort() returns a list of (word, count) tuples in order of decreasing frequency
dist = sort(dist)
# truncate the list of items if max_words was specified
if max_words:
max_words = int(max_words)
dist = dist[:max_words]
# get the number of words remaining in dist (ie the number of tags in the
# tagcloud) to pass back in the metadata
words_in_cloud = len(dist)
# get the equation of the line
#max_size = 70;
#min_size = 10;
if max_size and min_size:
font_size_fn = min_max_extrapolate(dist, max_size, min_size)
else:
if not (width and height):
width = 600
height = 800
font = 'times new roman'
font_size_fn = fit_to_area(width, height, dist, font)
# get the equation of the line between min_size and max_size. do this AFTER
# truncating to max_words and BEFORE shuffling the order around.
# determine the sort order. if the sort order is frequency, there's nothing
# to do since the distribution object is already sorted by frequency.
if sort_order not in ['random', 'frequency', 'alphabetical']:
print 'invalid sort order; using default = random'
sort_order = 'random'
if sort_order == 'random':
# shuffles in place
random.shuffle(dist)
if sort_order == 'alphabetical':
# not yet implemented
pass
# assemble the class and id tags for the tag cloud's wrapping div
divstyle = '''class="tagcloud"'''
if class_ != "":
divstyle = divstyle[:-1] + " " + class_ + ''' "'''
# the user can specify a unique id for the tag cloud if they want
# additional styling applied from their own style sheets.
if id_ != "": divstyle += ''' id="%s" ''' % id_
divstyle = divstyle.strip()
body = '''<div %s>''' % divstyle
for word, freq in dist:
# each word has a class of 'word' in addition to its frequency so that
# the user may specify additional styling. **note**: make sure the
# space after the span is maintained; otherwise the spans within the
# div won't wrap.
body += '''<span title="%d" class="word %s">%s</span> ''' % (freq, num_to_word(freq), word)
body += '''</div>'''
#print body
if start_color and end_color and color_steps:
num_colors = int(color_steps)
colors = color_scheme(start_color, end_color, color_steps)
else:
num_colors = 5
colors = color_scheme()
# get the distinct frequencies and specify a font-size and color for each,
# that corresponds to its size
freqs = []
for f in [x[1] for x in dist]:
if f not in freqs: freqs.append(f)
style = '''<style type="text/css">
.tagcloud {width: %s; height: %s; text-align: center; }
.word { text-align: center; vertical-align: middle; } ''' % (width, height)
for f in freqs:
freq_as_word = num_to_word(f)
color_index = f % num_colors
color = colors[color_index]
style += ('''
.%s {padding-left: 15px; padding-right: 15px; font-size: %s; color: %s }'''
% (freq_as_word, font_size_fn(f), color))
style += '''
</style>'''
# assemble the response
oid = pymongo.objectid.ObjectId()
uid = str(oid)
long_url = settings.HOME_PAGE + '/cloud/' + uid
short_url = bitly_shorten(long_url)
metadata = {
'utc_created': datetime.datetime.utcnow().strftime("%Y-%m-%dT%H:%M:%S"),
'total_tags' : words_in_cloud,
'short_url' : short_url,
}
record = {'_id': oid,
'body': body,
'style': style,
'metadata': metadata}
print style
# save to the database
con = pymongo.Connection()
collection = con.wordapi.tagclouds
collection.insert(record)
return record
def word_cloud(dist, css_id = "", css_class = "", layout = "svg", width=800,
height=600, max_words = None, start_color=None, end_color=None,
color_steps=None, sort_order="random", equn="linear", slope=0.15,
link_prefix=None):
''' returns a dict with style and body elements. style contains
defalt styling for the tag cloud, while body contains the html
markup. '''
# sort() returns a list of (word, count) tuples in order of decreasing frequency
dist = sort(dist)
# truncate the list of items if max_words was specified
if max_words:
max_words = int(max_words)
dist = dist[:max_words]
# get the number of words remaining in dist (ie the number of tags in the
# tagcloud) to pass back in the metadata
words_in_cloud = len(dist)
# the equation of the line used here determines the *relative* sizes of
# different frequency words. actual font sizes are computed with javascript
# to dynamically fit the words into the specified area.
# get the equation of the line AFTER truncating to max_words and BEFORE
# shuffling the order around.
if equn == "log":
# shifting the logarithm up by 1 gives a word of frequency one size one.
font_size_fn = lambda freq: 10*math.log(freq,2) + 1
elif equn == "exp":
font_size_fn = lambda freq: math.pow(freq, 1.2)
else: # linear
min_size = 1
font_size_fn = extrapolate_linear(dist, min_size, slope)
# determine the sort order. if the sort order is frequency, there's nothing
# to do since the distribution object is already sorted by frequency.
if sort_order == 'random':
# shuffles in place
random.shuffle(dist)
if sort_order == 'alphabetical':
dist.sort()
# assemble the class and id tags for the tag cloud's wrapping div
divstyle = '''id="tagcloud"'''
if css_class != "":
divstyle = divstyle[:-1] + " " + class_ + ''' "'''
# the user can specify a unique id for the tag cloud if they want
# additional styling applied from their own style sheets.
if css_id != "": divstyle += ''' id="%s" ''' % id_
divstyle = divstyle.strip()
if layout == "svg":
body = svg_based_layout(divstyle, dist)
else:
body = div_based_layout(divstyle, dist)
if start_color and end_color and color_steps:
num_colors = int(color_steps)
colors = color_scheme(start_color, end_color, color_steps)
else:
num_colors = 5
colors = color_scheme()
# get the distinct frequencies and specify a font-size and color for each,
# that corresponds to its size
freqs = []
for f in [x[1] for x in dist]:
if f not in freqs: freqs.append(f)
style = '''<style type="text/css">
// important clearfix for divs which wrap floating elements
// see http://nicolasgallagher.com/micro-clearfix-hack/
#hidden-resizer { zoom: 1; }
#hidden-resizer:before, #hidden-resizer:after { content: ""; display: table; }
#hidden-resizer:after { clear: both; }
#tagcloud { font-size: 10px; text-align: center; zoom: 1; }
#tagcloud:before, #tagcloud:after { content: ""; display: table; }
#tagcloud:after { clear: both; }
.word { text-align: center; vertical-align: middle; line-height:1; padding-right:5px; float:left; } '''
for f in freqs:
freq_as_word = num_to_word(f)
color_index = f % num_colors
color = colors[color_index]
style += ('''
.%s {font-size: %sem; color: %s }'''
% (freq_as_word, font_size_fn(f), color))
style += '''
</style>'''
# assemble the response (if you get an error on this line, probably mongo is
# not running)
oid = pymongo.objectid.ObjectId()
uid = str(oid)
long_url = settings.HOME_PAGE + '/cloud/' + uid
if settings.DEBUG:
short_url = long_url
else:
short_url = bitly_shorten(long_url)
metadata = {
'utc_created': datetime.datetime.utcnow().strftime("%Y-%m-%dT%H:%M:%S"),
'total_tags' : words_in_cloud,
'short_url' : short_url,
}
record = {
'_id': oid,
'body': body,
'style': style,
# hidden div to pass target width and height information onto javascript resizer
# TODO not all tagclouds will necessarily have width and height?
'wordcloud_size': '''
<div id="tagcloud_size" style="width:%dpx; height:%dpx; position: absolute; left:-999em;top:-999em;"></div>''' % (width, height),
'script': '''<script type="text/javascript" src="http://code.jquery.com/jquery-1.7.1.min.js"></script>
<script type="text/javascript" src="/media/js/wordcloud.js"></script>
''',
'metadata': metadata
}
print style
# save to the database
con = pymongo.Connection()
collection = con.wordapi.tagclouds
collection.insert(record)
return record