def GatherTexts(user_source_dir, corpus_source_dir, remove_old_plaintext = False):
# Make sure passed in paths are formatted correctly
user_source_dir = Utils_MalletInterpret.FormatPath(user_source_dir)
corpus_source_dir = Utils_MalletInterpret.FormatPath(corpus_source_dir)
# Copy files from user source directory to TWiC's corpus source directory
source_file_counter = 1
for user_filename in glob.glob(user_source_dir + "*.txt"):
user_text = TWiC_Text(user_filename)
#user_text.ConvertToPlainText("{0}{1}_{2}{3}".format(corpus_source_dir,\
# source_file_counter,\
# user_text.GetFilename(),\
# user_text.GetFileExtension()))
#source_file_counter += 1
source_file_counter += user_text.ConvertToPlainText_Chunks(corpus_source_dir, source_file_counter)
def Build_TextObjects(TextClass, mallet_script, tp_collection):
textobj_collection = []
for current_tp in tp_collection:
filename = Utils_MalletInterpret.GetFilename(current_tp.filename)
textobj_collection.append(TextClass('{0}{1}.tei'.format(mallet_script.tei_source, filename)))
return textobj_collection
def ConvertToPlainText_Chunks(self, p_output_dir, p_file_number, p_chunk=True, p_chunk_size=5000):
file_name = self.GetFilename()
file_ext = self.GetFileExtension()
output_lines = self.GetPreparedLines()
# Optional line chunking
chunks = []
if p_chunk:
chunks = Utils_MalletInterpret.GetChunkedLines(output_lines, p_chunk_size)
else:
chunks.append(output_lines)
# Write out files
for index in range(len(chunks)):
with open("{0}{1}_{2}_{3}{4}".format(p_output_dir, p_file_number, file_name, index, file_ext), 'w') as plaintext_output_file:
for line in chunks[index]:
plaintext_output_file.write(unidecode(line) + u"\n")
p_file_number += 1
return len(chunks)
def InterpretMalletOutput(mallet_script):
print "Interpreting MALLET output for TWiC visualization..."
myoutput_dir = os.path.join("..", "..", "..", "data", "input" + os.sep)
print "\tReading in MALLET output..."
####### 1. Reading corpus.topics.tsv
print "\t\tLoading topics for texts..."
# tp_collection = mallet_script.GetTopicsFileData("2.0.7")
tp_collection = mallet_script.GetTopicsFileData("2.0.9")
###### 2. Reading corpus.keys.tsv
print "\t\tLoading topic keys..."
topic_keys = mallet_script.GetKeysFileData()
###### 3. Reading corpus.topic-state.tsv
print "\t\tLoading topic words state file..."
fwt_collection = mallet_script.GetStateFileData()
###### 4. Reading corpus.wordweights.tsv
print "\t\tLoading topic word weights..."
ww_table = mallet_script.GetTopicWordWeights()
###### 5. Build a text object for each text
print "\tBuilding text objects..."
# start_time = int(round(time.time() * 1000))
# textobj_collection = TWiC_MalletInterpret.Build_TextObjects(TWiC_Poem, mallet_script, tp_collection)
# end_time = int(round(time.time() * 1000))
# print "Unoptimized: {0}ms".format(end_time - start_time)
# start_time = int(round(time.time() * 1000))
# text_obj_collection_opt = TWiC_MalletInterpret.Build_TextObjects_Opt(TWiC_Poem, mallet_script, tp_collection)
textobj_collection = TWiC_MalletInterpret.Build_TextObjects_Opt(TWiC_Text, mallet_script, tp_collection)
# end_time = int(round(time.time() * 1000))
# print "Optimized: {0}ms".format(end_time - start_time)
###### 6. Generate a list of unique colors for each topic
print "\tCreating color list..."
color_list = Utils_Color.Get_UniqueColorList(len(topic_keys.corpus_topic_proportions.keys()))
# color_list = Utils_Color.HCL_Fluo
###### 7. Build HTML and JSON files for each text for low and mid level TWiC representations
print "\tCreating JSON files for TWiC views of individual texts..."
for text in textobj_collection:
current_tp = None
for tp in tp_collection:
if text.GetFilename().split("_")[0] == Utils_MalletInterpret.GetFilenameWithUnderscore(tp.filename):
current_tp = tp
break
# TWiC_MalletInterpret.Build_HTMLandJSONForText(text, myoutput_dir, "{0}.css".format(mallet_script.corpus_name), \
# current_tp, fwt_collection, topic_keys, color_list, mallet_script, True)
TWiC_MalletInterpret.Build_JSONForTextwithForeignObject(text, myoutput_dir, "{0}.css".format(mallet_script.corpus_name), \
current_tp, fwt_collection, topic_keys, color_list, mallet_script, True)
###### 8. Build JSON files for visualization
print "\tBuilding corpus-level JSON map files..."
# Build a json that shows the hierarchy of Corpus -> Text clusters -> Texts based on Jensen-Shannon Distance
TWiC_MalletInterpret.Build_CorpusMapJSON_Avg(mallet_script.corpus_title, topic_keys.corpus_topic_proportions, tp_collection, myoutput_dir + "json" + os.sep)
# Output a JSON of the topic-color list
# TWiC_MalletInterpret.Build_TopicColorMapJSON(color_list, myoutput_dir + "json/")
# Generate topic list JSON based on the used_topics_list
# TWiC_MalletInterpret.Build_TopicWordsJSON(topic_keys, myoutput_dir + "json/")
# New JSON format for client side
TWiC_MalletInterpret.Build_CorpusInfoJSON(mallet_script.corpus_title, textobj_collection, tp_collection, topic_keys, color_list, myoutput_dir + "json" + os.sep)
# Build a json that lists the distribution weights of words likely to appear in each topic
TWiC_MalletInterpret.Build_WordWeightJSON(ww_table, myoutput_dir + "json" + os.sep)
print "Finished processing {0} for TWiC.".format(mallet_script.corpus_title)
def Build_CorpusInfoJSON(corpus_title, text_collection, tp_collection, topic_keys, color_list, output_dir):
# Output JSON format
# {
# "topic_info" : [ # indexed by int topic ID number
# [
# ["habit", "dash", "torn",...], # topic words
# 022440" # hex color
# ],...
# ],
# "corpus_info" : [
# "Corpus Title",
# [0.5, 0.2, ...] # topic proportions [topic0,...topicN]
# ],
# "file_info" : {
# "0" : [ # indexed by str file ID number
# "Filename",
# "Text Title",
# [0.5, 0.2, ...], # topic proportions [topic0,...topicN]
# 3, # stanza count
# 65, # line count
# 400 # word count
# ],...
# }
# }
# Indexers Info
# "topic_info" (indexed by int topic ID)
topic_info = "topic_info"
TI_TopicWords = 0
TI_Color = 1
# "corpus_info"
corpus_info = "corpus_info"
CI_CorpusTitle = 0
CI_TopicProportions = 1
# "file_info" (indexed by str numeric file ID)
file_info = "file_info"
FI_Filename = 0
FI_TextTitle = 1
FI_TopicProportions = 2
FI_StanzaCount = 3
FI_LineCount = 4
FI_WordCount = 5
FI_FieldCount = 6
json_output = { topic_info : [ ], corpus_info : ["", []], file_info : { } }
topic_count = len(topic_keys.corpus_topic_proportions.keys())
# Fill out topic_info and corpus_info
json_output[corpus_info][CI_CorpusTitle] = corpus_title
json_output[corpus_info][CI_TopicProportions] = [topic_keys.corpus_topic_proportions[str(topic_index)]
for topic_index in range(topic_count)]
json_output[topic_info] = [[topic_keys.corpus_topic_words[str(topic_index)] for topic_index in range(topic_count)],
[color_list[topic_index] for topic_index in range(topic_count)]]
# print "COLOR LIST COMPLETION"
# print [color_list[topic_index] for topic_index in range(topic_count)]
# Fill out file_info
for tp in tp_collection:
json_output[file_info][tp.fileid] = []
json_output[file_info][tp.fileid] = [0 for index in range(FI_FieldCount)]
# json_output[file_info][tp.fileid][FI_Filename] = tp.filename
json_output[file_info][tp.fileid][FI_Filename] = Utils_MalletInterpret.GetFilename(tp.filename)
json_output[file_info][tp.fileid][FI_TopicProportions] = []
for topic_index in range(0, topic_count):
json_output[file_info][tp.fileid][FI_TopicProportions].append(tp.topic_guide[str(topic_index)])
for text in text_collection:
text_filename = text.GetFilename()
for fileid in json_output[file_info].keys():
if text_filename == json_output[file_info][fileid][FI_Filename]:
json_output[file_info][fileid][FI_TextTitle] = text.GetTitle()
# Still have to fill out
# FI_StanzaCount = 3
# FI_LineCount = 4
# FI_WordCount = 5
# Output JSON
with open(output_dir + "twic_corpusinfo.json", "w") as output_file:
output_file.write(json.dumps(json_output))
def Build_HTMLandJSONForText(text, output_dir, css_filename, current_tp, fwt_collection, topic_keys, color_list, mallet_script, split_filename=False):
file_id = text.GetFilename()
if split_filename:
file_id = text.GetFilename().split("_")[0]
output_html = open(output_dir + "html" + os.sep + file_id + '.html', 'w')
output_html.write('<html>\n')
output_html.write('\t<head>\n')
output_html.write('\t\t<link rel="stylesheet" type="text/css" href="{0}">\n'.format(css_filename))
output_html.write('\t</head>\n')
output_html.write('\t<body>\n')
output_html.write('\t\t<div class="left">\n')
output_html.write('\t\t\t<div class="title">\n')
output_html.write('\t\t\t\t{0}<br>\n'.format(text.GetTitle()))
output_html.write('\t\t\t</div>\n')
output_html.write('\t\t\t{0}<br>\n'.format(text.GetPublication()))
output_html.write('\t\t</div>\n')
output_html.write('\t\t<div class="center">\n')
# Create a JSON for each text (for mid-level twic, text tile)
# Figure out the possible topics for each word based on the topic state file
current_fwt = None
for fwt in fwt_collection:
fwt_file_id = Utils_MalletInterpret.GetFilename(fwt.GetFilename())
if split_filename:
fwt_file_id = Utils_MalletInterpret.GetFilenameWithUnderscore(fwt.GetFilename())
if fwt_file_id == file_id:
current_fwt = fwt
break
# Convert text to JSON readable by the high-level TWiC visualization
TWiC_MalletInterpret.ConvertTextToJSON(text, output_dir + "json" + os.sep + "texts" + os.sep, mallet_script, current_fwt)
# Read in the plain text file
input_file = open(current_tp.filename, 'r')
data = input_file.readlines()
input_file.close()
# If there was no state file entry, output HTML lines without topics
used_topics_list = []
if None == current_fwt:
for line in data:
output_line = ''
words = line.split(' ')
for actual_word_index in range(0, len(words)):
output_line += words[actual_word_index] + ' '
output_line = output_line.strip()
output_html.write('\t\t\t' + output_line + '<br>\n')
else:
statefile_word_index = 0
for line in data:
output_line = ''
words = line.split(' ')
if statefile_word_index < len(current_fwt.word_info):
lowercase_state_word = clean_word(current_fwt.word_info[statefile_word_index].word.lower())
# Go through each word in the line
for actual_word_index in range(0, len(words)):
# Lowercase only for comparison
lowercase_word = clean_word(words[actual_word_index].lower())
if statefile_word_index < len(current_fwt.word_info) and \
lowercase_word == lowercase_state_word:
output_line += '<span title="Topic {0}"><font color="{1}"><b>{2}</b></font></span>'.format(current_fwt.word_info[statefile_word_index].topic,
color_list[int(current_fwt.word_info[statefile_word_index].topic)], words[actual_word_index])
if current_fwt.word_info[statefile_word_index].topic not in used_topics_list:
used_topics_list.append(current_fwt.word_info[statefile_word_index].topic)
statefile_word_index += 1
if statefile_word_index < len(current_fwt.word_info):
lowercase_state_word = clean_word(current_fwt.word_info[statefile_word_index].word.lower())
else:
output_line += words[actual_word_index]
output_line += ' '
output_line = output_line.strip()
output_html.write('\t\t\t' + output_line + '<br>\n')
output_html.write('\t\t</div><br><br>\n')
output_html.write('\t\t<div class="topics">\n')
for used_topic in used_topics_list:
output_html.write('\t\t\t<font color="{0}">Topic {1}: {2}</font><br>\n'.format(color_list[int(used_topic)], used_topic, topic_keys.corpus_topic_words[used_topic]))
output_html.write('\t\t</div>\n')
output_html.write('\t</body>\n')
output_html.write('</html>')
output_html.close()
def Build_JSONForTextwithForeignObject(text, output_dir, css_filename, current_tp, fwt_collection, topic_keys, color_list, mallet_script, split_filename=False):
file_id = text.GetFilename()
if split_filename:
file_id = text.GetFilename().split("_")[0]
# Figure out the possible topics for each word based on the topic state file
current_fwt = None
for fwt in fwt_collection:
fwt_file_id = Utils_MalletInterpret.GetFilename(fwt.GetFilename())
if split_filename:
fwt_file_id = Utils_MalletInterpret.GetFilenameWithUnderscore(fwt.GetFilename())
if fwt_file_id == file_id:
current_fwt = fwt
break
# Retrieve json data of a line-word-topic map from ConvertTextToJSON
json_data = TWiC_MalletInterpret.ConvertTextToJSON(text, output_dir + "json" + os.sep + "texts" + os.sep, mallet_script, current_fwt, False)
if "No state file data" == json_data:
print "Warning: No state file data for {0}".format(text.GetFilename())
return
# Output text will be partial html to be inserted inside foreignObject tag
output_text = []
# Add an initial spacing span between the panel's control bar and the body
output_text.append("<xhtml:p class=\"text_p\"><xhtml:span class=\"text_edgespan\"> </xhtml:span></xhtml:p>")
# Build up HTML lines that will be inserted as a foreignObject client-side
#print "FILENAME: {0}".format(text.GetFilename())
#print "L&C LEN: {0}".format(len(json_data["document"]["lines_and_colors"]))
#for lc_index in range(len(json_data["document"]["lines_and_colors"])):
# print json_data["document"]["lines_and_colors"][lc_index]
for lc_index in range(len(json_data["document"]["lines_and_colors"])):
entry = json_data["document"]["lines_and_colors"][lc_index]
output_text.append("<xhtml:p class=\"text_p\">")
output_text.append("<xhtml:span class=\"text_edgespan\"> </xhtml:span>")
for index in range(len(entry[0])):
if str(index) in entry[1]:
output_text.append("<xhtml:span class=\"text_coloredword\" style=\"color:{0}\">{1} </xhtml:span>".format(\
color_list[int(entry[1][str(index)])], entry[0][index]))
else:
output_text.append("<xhtml:span class=\"text_word\">{0} </xhtml:span>".format(entry[0][index]))
output_text.append("</xhtml:p>")
#output_text.append("<xhtml:br></xhtml:br>")
# Add the foreignObject HTML to the JSON
json_data["document"]["full_text"] = ''.join([str(output_text_line) for output_text_line in output_text])
# Save the number of lines in the text for panel height size client-side
json_data["document"]["line_count"] = len(json_data["document"]["lines_and_colors"])
# Dereference the lines and colors array for garbage collection
#json_data["document"].pop("lines_and_colors", None)
# Write the JSON file for this text
with open(output_dir + "json" + os.sep + "texts" + os.sep + text.GetFilename() + ".json", 'w') as fileptr:
#print "Writing {0}".format(output_dir + "json/texts/" + text.GetFilename() + ".json")
fileptr.write(json.dumps(json_data))
def main(args):
Utils_MalletInterpret.TimeAndRun(Corpus2Vis, args)
def InterpretMalletOutput(mallet_script):
print 'Interpreting MALLET output for visualization...'
# myoutput_dir = '/Users/PeregrinePickle/Documents/Programming/Corpora/Dickinson/output/myviz-output/'
myoutput_dir = os.path.join("..", "..", "..", "data", "dickinson",
"input" + os.sep)
print '\tReading in MALLET output...'
####### 1. Reading dickinson.topics.tsv
tp_collection = mallet_script.GetTopicsFileData("2.0.9")
###### 2. Reading dickinson.keys.tsv
topic_keys = mallet_script.GetKeysFileData()
###### 3. Reading dickinson.topic-state.tsv
fwt_collection = mallet_script.GetStateFileData()
###### 4. Reading dickinson.wordweights.tsv
ww_table = mallet_script.GetTopicWordWeights()
###### 5. Build a text object for each text
print '\tBuilding text objects...'
# start_time = int(round(time.time() * 1000))
# textobj_collection = TWiC_MalletInterpret.Build_TextObjects(TWiC_Poem, mallet_script, tp_collection)
# end_time = int(round(time.time() * 1000))
# print 'Unoptimized: {0}ms'.format(end_time - start_time)
# start_time = int(round(time.time() * 1000))
# text_obj_collection_opt = TWiC_MalletInterpret.Build_TextObjects_Opt(TWiC_Poem, mallet_script, tp_collection)
textobj_collection = TWiC_MalletInterpret.Build_TextObjects_Opt(
TWiC_Poem, mallet_script, tp_collection)
# end_time = int(round(time.time() * 1000))
# print 'Optimized: {0}ms'.format(end_time - start_time)
###### 6. Generate a list of unique colors for each topic
print '\tCreating color list...'
color_list = Utils_Color.Get_UniqueColorList(
len(topic_keys.corpus_topic_proportions.keys()))
###### 7. Build HTML and JSON files for each text for low and mid level TWiC representations
print '\tCreating JSON files for TWiC views of texts...'
for text in textobj_collection:
current_tp = None
for tp in tp_collection:
if text.GetFilename() == Utils_MalletInterpret.GetFilename(
tp.filename):
current_tp = tp
break
TWiC_MalletInterpret.Build_JSONForTextwithForeignObject(text, myoutput_dir, '{0}.css'.format(mallet_script.corpus_name), \
current_tp, fwt_collection, topic_keys, color_list, mallet_script)
###### 8. Build JSON files for visualization
print '\tBuilding JSON map files for TWiC visualization...'
# Build a json that shows the hierarchy of Corpus -> Text clusters -> Texts based on Jensen-Shannon Distance
TWiC_MalletInterpret.Build_CorpusMapJSON_Avg(
mallet_script.corpus_title, topic_keys.corpus_topic_proportions,
tp_collection, myoutput_dir + "json" + os.sep)
# Output a JSON of the topic-color list
# TWiC_MalletInterpret.Build_TopicColorMapJSON(color_list, myoutput_dir + "json/")
# Generate topic list JSON based on the used_topics_list
# TWiC_MalletInterpret.Build_TopicWordsJSON(topic_keys, myoutput_dir + "json/")
# New JSON format for client side
TWiC_MalletInterpret.Build_CorpusInfoJSON(
mallet_script.corpus_title, textobj_collection, tp_collection,
topic_keys, color_list, myoutput_dir + "json" + os.sep)
# Build a json that lists the distribution weights of words likely to appear in each topic
TWiC_MalletInterpret.Build_WordWeightJSON(
ww_table, myoutput_dir + "json" + os.sep)
