def get_event_sequence(action_str, seq_type, bitmap_dim):
if seq_type == "absolute":
event_sequence = AbsoluteEventSequence.from_string(action_str)
elif seq_type == "cursor":
relative_sequence = RelativeEventSequence.from_eval_str(action_str)
abs_sequence = AbsoluteEventSequence.from_relative(relative_sequence, bitmap_dim, bitmap_dim)
event_sequence = CursorEventSequence.from_absolute(abs_sequence)
else:
event_sequence = RelativeEventSequence.from_eval_str(action_str)
return event_sequence.events
def get_event_sequence(action_str, seq_type, bitmap_dim):
if seq_type == "absolute":
event_sequence = AbsoluteEventSequence.from_string(action_str)
elif seq_type == "cursor":
relative_sequence = RelativeEventSequence.from_eval_str(action_str)
abs_sequence = AbsoluteEventSequence.from_relative(
relative_sequence, bitmap_dim, bitmap_dim)
event_sequence = CursorEventSequence.from_absolute(abs_sequence)
else:
event_sequence = RelativeEventSequence.from_eval_str(action_str)
return event_sequence.events
def write_data(rows, key_attrs, image_idx, commands_idx, actions_idx, output_file, include_bitmaps=True, grid_dims=(25,25)):
tokenizer = TreebankWordTokenizer()
bmpmaker = BitmapMaker(grid_dims[0],grid_dims[1])
with open(output_file, "w") as fout:
objs = []
for i,row in enumerate(rows):
if i % 100 == 0:
print " {} ".format(i)
image_url = row[image_idx]
if include_bitmaps:
bitmap = retrieve_bitmap(image_url_to_text_bitmap_url(image_url))
else:
bitmap = None
image_key = image_url_to_key(image_url)
commands = tokenize_description(tokenizer, row[commands_idx])
actions = row[actions_idx]
abs_seq = AbsoluteEventSequence.from_mturk_string(actions).canonicalize()
rel_seq = RelativeEventSequence.from_absolute(abs_seq)
cur_seq = CursorEventSequence.from_absolute(abs_seq)
abs_str = str(abs_seq)
rel_str = str(rel_seq)
cur_str = str(cur_seq)
raw_txt = actions.replace("\r", "").replace("\n"," ")
rel_alignments,rel_align_info = align_strings(commands, rel_str, "relative", "clever", grid_dims[0], backup_using_naive=True)
abs_alignments,abs_align_info = align_strings(commands, abs_str, "absolute", "clever", grid_dims[0], backup_using_naive=True)
join_list_of_lists = lambda l: " ".join(str(i) for i in itertools.chain(*l))
str_rel_alignments = [(c,join_list_of_lists(r)) for c,r in rel_alignments]
str_abs_alignments = [(c,join_list_of_lists(r)) for c,r in abs_alignments]
obj = {}
obj["image_url"] = image_url
obj["image_id"] = image_key
obj["sample_id"] = i
obj["commands"] = commands
obj["actions.absolute"] = abs_str
obj["actions.relative"] = rel_str
obj["actions.cursor"] = cur_str
obj["alignments.relative"] = str_rel_alignments
obj["alignments.relative.info"] = rel_align_info
obj["alignments.absolute"] = str_abs_alignments
obj["alignments.absolute.info"] = abs_align_info
for key,val in key_attrs[image_key].iteritems():
obj[key] = val
if bitmap is not None:
abs_norm_seq = AbsoluteEventSequence.from_relative(rel_seq, grid_dims[0], grid_dims[1])
try:
bmpmaker.clear()
bmpmaker.process_commands(abs_norm_seq.events)
bitmap_mturk_norm = bmpmaker.bitmap.copy()
except:
bitmap_mturk_norm = None
try:
bmpmaker.clear()
bmpmaker.process_commands(abs_seq.events)
bitmap_mturk = bmpmaker.bitmap.copy()
except:
bitmap_mturk = None
obj["actions.absolute.normalized"] = str(abs_norm_seq)
obj["bitmap.orig_generated"] = bitmap_to_list(bitmap.astype(np.int_))
obj["bitmap.from_mturk"] = bitmap_to_list(bitmap_mturk.astype(np.int_))
obj["bitmap.normalized"] = bitmap_to_list(bitmap_mturk_norm)
objs.append(obj)
json.dump(objs, fout, indent=4, sort_keys=True)
print "Created dataset at %s" % output_file
