def remake_pads(summaries, stdscr, index):
new_pads = []
for i in range(0, len(summaries)):
subject = utils.format_html(summaries[i]['subject'])
contents = utils.format_html(summaries[i]['content_snipet'])
reverse = i == index
new_pads.append(summary_pad(subject, contents, stdscr, reverse))
return new_pads
def remake_pads(summaries, stdscr, index):
new_pads = []
for i in range(0, len(summaries)):
subject = utils.format_html(summaries[i]['subject'])
contents = utils.format_html(summaries[i]['content_snipet'])
reverse = i == index
new_pads.append(summary_pad(subject, contents, stdscr, reverse))
return new_pads
def excel_gt_json_generate():
"""
This function is used to generate rr_gt.json file for calculating similarity.
Returns:
Generate rr_gt.json for metirc.py to calculate similarity
"""
count = 0
gt_gt = {}
excel_path = '/home/gita/Downloads/rr_table/excel/'
text_path = '/home/gita/Downloads/rr_table/text/'
html_path = '/home/gita/Downloads/rr_table/html/'
father_path_file = '/home/gita/Downloads/final/f.txt'
gt_json_file = './src/rr_gt.json'
with open(father_path_file, 'r') as jf:
file_list = jf.readlines()
for file in file_list:
file_name = file.split('/')[-1].split('.')[0]
excel_file = excel_path + file_name + '.xlsx'
text_file = text_path + file_name + '.json'
adjacent_matrix, cell_text_list = excel_to_matrix(
excel_file, text_file)
gt_json = matrix_to_html(cell_text_list, adjacent_matrix)
gt_html = format_html(gt_json)
with open(html_path + file_name + '.html', 'w',
encoding='utf-8') as hf:
hf.write(str(gt_html))
gt_gt[file_name + '.json'] = {"html": gt_html}
count += 1
print(count, file_name)
# if count >= 1:
# break
with open(gt_json_file, 'w') as tf:
tf.write(json.dumps(gt_gt))
def excel_pred_json_generate():
"""
Process the results of OCR and PointerNet to recovery .html and generate .json
Returns:
Generate rr_pred.json for metirc.py to calculate similarity.
"""
count = 0
error_count = 0
text_path = '/home/gita/Downloads/rr_table/text/'
uf_path = '/home/gita/Downloads/final/f1/'
df_path = '/home/gita/Downloads/final/f2/'
lm_path = '/home/gita/Downloads/final/m1/'
rm_path = '/home/gita/Downloads/final/m2/'
img_path_txt = '/home/gita/Downloads/final/f.txt'
pred_html_path = './recovered_html/rr_html/'
pred_json_file = './src/rr_pred.json'
father_dict = get_parent_dict(uf_path,
df_path,
lm_path,
rm_path,
img_path_txt,
number_per_file=2)
file_list = get_specific_files('/home/gita/Downloads/rr_table/text/',
['.json'])
pred_json = {}
for file_name in file_list:
# check_list = ['000917.json']
# if file_name not in check_list:
# continue
print(count, file_name)
text_file = text_path + file_name
# try:
preds = axis_to_rows(text_file)
# except:
# error_count += 1
# print("-------------------------------------------------------------")
preds_list, matrix = preds_to_matrix(preds, father_dict[file_name])
if matrix is None:
continue
img_json = matrix_to_html(preds_list, matrix)
pred_html = format_html(img_json)
with open(pred_html_path + file_name + '.html', 'w',
encoding='utf-8') as hf:
hf.write(str(pred_html))
pred_json[file_name] = pred_html
count += 1
# if count >= len(check_list):
# break
print(error_count)
with open(pred_json_file, 'w') as tf:
tf.write(json.dumps(pred_json).strip())
def ret_json_generate():
"""
Process the results of OCR and PointerNet to recovery .html and generate .json
Returns:
Generate html.json for metirc.py to calculate similarity.
"""
ocred_path = '/home/gita/Downloads/mini_result/mini_json/'
img_path_txt = '/home/gita/Downloads/mini_result_50/mini_father.txt'
uf_path = '/home/gita/Downloads/mini_result_50/father/'
df_path = ''
lm_path = '/home/gita/Downloads/mini_result_50/mother_p/'
rm_path = '/home/gita/Downloads/mini_result_50/mother_n/'
ret_html_path = './recovered_html/mini_50/'
ret_json_file = './src/mini_pred_50.json'
father_dict = get_parent_dict(uf_path,
df_path,
lm_path,
rm_path,
img_path_txt,
number_per_file=2)
ocred_files = get_specific_files(ocred_path)
count = 0
pred_json = {}
for file_name in ocred_files:
if file_name in [
'PMC3707453_006_00.png.json', 'PMC6022086_007_00.png.json'
]:
continue
preds = parse_ocred_json(ocred_path + file_name)
preds = merge_by_iou(preds)
count += 1
print(count, file_name)
preds_list, matrix = preds_to_matrix(preds, father_dict[file_name])
if matrix is None:
continue
img_json = matrix_to_html(preds_list, matrix)
pred_html = format_html(img_json)
pred_json[file_name[:-5]] = pred_html
with open(ret_html_path + file_name + '.html', 'w',
encoding='utf-8') as hf:
hf.write(str(pred_html))
# break
with open(ret_json_file, 'w') as tf:
tf.write(json.dumps(pred_json).strip())
def post_pad(post, network, stdscr):
lines = []
width = stdscr.getmaxyx()[1]
bold = curses.A_BOLD
norm = curses.A_NORMAL
#lines.append((" ", norm))
subject = utils.format_html(post['subject'])
for line in wrap(subject, width - 2):
lines.append((line, bold))
contents = utils.format_html(post['contents']).replace(
'___bold_start___', '').replace('___bold_end___', '')
for line in wrap(contents, width - 2):
lines.append((line, norm))
author = network.get_user_name(post['uid']).encode(
'ascii', 'ignore') if post['uid'] else 'Anonymous'
right_author = " " * (width - len(author) - 4) + '- ' + author
lines.append((right_author, norm))
if post['student_answer']:
lines.append(("Student Answer:", bold))
answer = post['student_answer']
contents = utils.format_html(answer['contents']).replace(
'___bold_start___', '').replace('___bold_end___', '')
for line in wrap(contents, width - 2):
lines.append((line, norm))
author = network.get_user_name(answer['uid']).encode(
'ascii', 'ignore') if answer['uid'] else 'Anonymous'
right_author = " " * (width - len(author) - 4) + '- ' + author
lines.append((right_author, norm))
if post['instructor_answer']:
lines.append(("Instructor Answer:", bold))
answer = post['instructor_answer']
contents = utils.format_html(answer['contents']).replace(
'___bold_start___', '').replace('___bold_end___', '')
for line in wrap(contents, width - 2):
lines.append((line, norm))
author = network.get_user_name(answer['uid']).encode(
'ascii', 'ignore') if answer['uid'] else 'Anonymous'
right_author = " " * (width - len(author) - 4) + '- ' + author
lines.append((right_author, norm))
if len(post['followups']) > 0:
lines.append(("Followup Discussions:", bold))
for followup in post['followups']:
subject = utils.format_html(followup['subject'])
hasUID = 'uid' in followup and followup['uid']
author = network.get_user_name(
followup['uid']) if hasUID else 'Anonymous'
entry = subject + " (" + author.encode('ascii', 'ignore') + ")"
if followup['plus_ones'] > 0:
entry += ' +' + str(followup['plus_ones'])
for line in wrap(entry, width - 2):
lines.append((line, norm))
for reply in followup['children']:
subject = utils.format_html(reply['subject'])
hasUID = 'uid' in reply and reply['uid']
author = network.get_user_name(
reply['uid']) if hasUID else 'Anonymous'
entry = subject + " (" + author.encode('ascii', 'ignore') + ")"
first = True
for line in wrap(entry, width - 9):
if first:
lines.append((" " + line, norm))
first = False
else:
lines.append((" " + line, norm))
lines.append(("", norm))
pad = curses.newpad(len(lines) + 1, width)
for item in lines:
contents, style = item
pad.addstr("\n " + contents, style)
return pad
def view_summaries(stdscr, feed, network):
curses.use_default_colors()
pads = []
summaries = []
global data
data = []
unsaved_post = None
height_sum = 0
window_height = stdscr.getmaxyx()[0]
window_width = stdscr.getmaxyx()[1]
while True:
post = feed.next_post()
if not post:
break
subject = utils.format_html(post['subject'])
contents = utils.format_html(post['content_snipet'])
subject_lines = textwrap.wrap(subject, window_width - 2)
lines = textwrap.wrap(contents, window_width - 2)
height = len(subject_lines) + len(lines) + 2
if height_sum + height <= window_height * 2:
summaries.append(post)
data.append((subject, contents))
height_sum += height
pads.append(summary_pad(subject, contents, stdscr))
else:
unsaved_post = post
break
# Main loop
i = 0
stdscr.erase()
stdscr.refresh()
if len(pads) > 0:
pads[i] = summary_pad(data[i][0], data[i][1], stdscr, True)
stdscr.nodelay(False) # getch is non-blocking
render(pads, i, stdscr)
while True:
c = stdscr.getch()
# Quit the program
if c == ord('q') or c == curses.KEY_BACKSPACE:
break
elif c == ord('i'):
url = "https://piazza.com/class/" + network._nid + "?cid=" + summaries[
i]['id']
plat = platform.system()
if plat == "Windows":
subprocess.Popen(["explorer", url])
elif plat == "Darwin":
subprocess.Popen(["open", url])
elif plat == "Linux":
subprocess.Popen(["x-www-browser", url])
# Check for ENTER
elif c == ord('\n'):
stdscr.clear()
stdscr.refresh()
post_summary_obj = summaries[i]
id_num = post_summary_obj['id']
post_obj = network.get_post(id_num)
post_viewer.view_post(post_obj, network, stdscr)
pads = remake_pads(summaries, stdscr, i)
# Scroll down
elif c == ord('j') or c == curses.KEY_DOWN:
post = unsaved_post if unsaved_post else feed.next_post()
unsaved_post = None
if post:
subject = utils.format_html(post['subject'])
contents = utils.format_html(post['content_snipet'])
summaries.append(post)
data.append((subject, contents))
pads.append(summary_pad(subject, contents, stdscr))
if i < len(pads) - 1:
pads[i] = summary_pad(data[i][0], data[i][1], stdscr)
i += 1
pads[i] = summary_pad(data[i][0], data[i][1], stdscr, True)
# Scroll up
elif (c == ord('k') or c == curses.KEY_UP) and i > 0:
pads[i] = summary_pad(data[i][0], data[i][1], stdscr)
i -= 1
pads[i] = summary_pad(data[i][0], data[i][1], stdscr, True)
# Check for window resize
if c == curses.KEY_RESIZE:
stdscr.erase()
stdscr.refresh()
pads = remake_pads(summaries, stdscr, i)
render(pads, i, stdscr)
def view_summaries(stdscr, feed, network):
curses.use_default_colors()
pads = []
summaries = []
global data
data = []
unsaved_post = None
height_sum = 0
window_height = stdscr.getmaxyx()[0]
window_width = stdscr.getmaxyx()[1]
while True:
post = feed.next_post()
if not post:
break
subject = utils.format_html(post['subject'])
contents = utils.format_html(post['content_snipet'])
subject_lines = textwrap.wrap(subject, window_width - 2)
lines = textwrap.wrap(contents, window_width - 2)
height = len(subject_lines) + len(lines) + 2
if height_sum + height <= window_height*2:
summaries.append(post)
data.append((subject, contents))
height_sum += height
pads.append(summary_pad(subject, contents, stdscr))
else:
unsaved_post = post
break
# Main loop
i = 0
stdscr.erase()
stdscr.refresh()
if len(pads) > 0:
pads[i] = summary_pad(data[i][0], data[i][1], stdscr, True)
stdscr.nodelay(False) # getch is non-blocking
render(pads, i, stdscr)
while True:
c = stdscr.getch()
# Quit the program
if c == ord('q') or c == curses.KEY_BACKSPACE:
break
elif c == ord('i'):
url = "https://piazza.com/class/" + network._nid + "?cid=" + summaries[i]['id']
plat = platform.system()
if plat == "Windows":
subprocess.Popen(["explorer", url])
elif plat == "Darwin":
subprocess.Popen(["open", url])
elif plat == "Linux":
subprocess.Popen(["x-www-browser", url])
# Check for ENTER
elif c == ord('\n'):
stdscr.clear()
stdscr.refresh()
post_summary_obj = summaries[i]
id_num = post_summary_obj['id']
post_obj = network.get_post(id_num)
post_viewer.view_post(post_obj, network, stdscr)
pads = remake_pads(summaries, stdscr, i)
# Scroll down
elif c == ord('j') or c == curses.KEY_DOWN:
post = unsaved_post if unsaved_post else feed.next_post()
unsaved_post = None
if post:
subject = utils.format_html(post['subject'])
contents = utils.format_html(post['content_snipet'])
summaries.append(post)
data.append((subject, contents))
pads.append(summary_pad(subject, contents, stdscr))
if i < len(pads) - 1:
pads[i] = summary_pad(data[i][0], data[i][1], stdscr)
i += 1
pads[i] = summary_pad(data[i][0], data[i][1], stdscr, True)
# Scroll up
elif (c == ord('k') or c == curses.KEY_UP) and i > 0:
pads[i] = summary_pad(data[i][0], data[i][1], stdscr)
i -= 1
pads[i] = summary_pad(data[i][0], data[i][1], stdscr, True)
# Check for window resize
if c == curses.KEY_RESIZE:
stdscr.erase()
stdscr.refresh()
pads = remake_pads(summaries, stdscr, i)
render(pads, i, stdscr)