def update_children(self, reg1, reg2):
page_config = Page_Config()
if reg1['children'] != None and len(reg1['children']) > 0:
if reg2['children'] != None and len(reg2['children']) > 0:
agg_children = reg1['children'] + reg2['children']
agg_children.sort(
key=lambda x: x['boundingBox']['vertices'][0]['y'])
children = sort_regions(agg_children, [])
if len(children) > 1:
avg__region_height, avg__region_ver_dist, avg__region_width = page_config.avg_line_info(
[{
'children': children
}])
avrage_region_ver_ratio = avg__region_ver_dist / max(
1, avg__region_height)
return horzontal_merging(children, avrage_region_ver_ratio)
#v_list[idx] =v_block
else:
return children
else:
return reg1['children']
else:
if reg2['children'] != None and len(reg2['children']) > 0:
return reg2['children']
else:
return []
def update_children(reg1,reg2):
if reg1['children']!=None and len(reg1['children']) > 0 :
if reg2['children']!=None and len(reg2['children']) > 0 :
agg_children = reg1['children'] + reg2['children']
agg_children.sort(key=lambda x: x['boundingBox']['vertices'][0]['y'])
children = sort_regions(agg_children , [])
if len(children) > 1 :
return children #horzontal_merging(children)
#v_list[idx] =v_block
else:
return children
else :
return reg1['children']
else :
if reg2['children']!=None and len(reg2['children']) > 0 :
return reg2['children']
else :
return []
def region_unifier(self,file,page_g_words, page_lines,page_regions,page_c_words,path):
try:
#sort regions
page_lines = add_font(page_lines)
page_regions = filterd_regions(page_regions)
if len(page_regions) > 0 :
page_regions.sort(key=lambda x:x['boundingBox']['vertices'][0]['y'])
sorted_page_regions = sort_regions(page_regions,[])
else:
sorted_page_regions = page_regions
page_words = collate_text(file,page_c_words, page_g_words)
text_region,n_text_table_regions,tabel_region,image_region,head_foot_region = self.get_text_tabel_region(sorted_page_regions)
tabel_region = remvoe_regions(copy.deepcopy(image_region), copy.deepcopy(tabel_region))
filtered_words = remvoe_regions(copy.deepcopy(image_region), copy.deepcopy(page_words))
filtered_lines = remvoe_regions(copy.deepcopy(image_region), copy.deepcopy(page_lines))
t_list = []
for idx,table in enumerate(tabel_region):
if 'regions' in table.keys():
filtered_words = remvoe_regions(copy.deepcopy(table['regions']), copy.deepcopy(filtered_words))
filtered_lines = remvoe_regions(copy.deepcopy(table['regions']), copy.deepcopy(filtered_lines))
tabel_region[idx]['regions'] = collate_regions(regions = copy.deepcopy(table['regions']),lines = copy.deepcopy(page_words),child_class='WORD',grand_children=False,region_flag = False)
page_words = filtered_words
page_lines = filtered_lines
t_list.append(tabel_region[idx])
t_list = collate_cell_regions(copy.deepcopy(t_list),copy.deepcopy(page_words),child_class='CELL_TEXT',grand_children=True,region_flag = False)
page_words = remvoe_regions(copy.deepcopy(t_list), copy.deepcopy(page_words))
filtered_lines = remvoe_regions(copy.deepcopy(t_list), copy.deepcopy(page_lines))
filtered_words = copy.deepcopy(page_words)
text_region = remvoe_regions(copy.deepcopy(t_list) ,copy.deepcopy(text_region))
line_list = collate_regions(copy.deepcopy( filtered_lines), copy.deepcopy( filtered_words),child_class='WORD',add_font=True)
head_foot_list = collate_regions(copy.deepcopy(head_foot_region),copy.deepcopy(line_list),child_class='LINE',grand_children=True,region_flag = False)
filtered_lines = remvoe_regions(copy.deepcopy(head_foot_list), copy.deepcopy(line_list))
v_list = collate_regions( copy.deepcopy( text_region),copy.deepcopy( filtered_lines ),child_class='LINE' ,grand_children=True,add_font=True )
i_list = collate_regions(copy.deepcopy( image_region),copy.deepcopy(page_words),grand_children=True,region_flag = False,skip_enpty_children=True)
page_config = Page_Config()
avg_height, avg_ver_dist, avg_width = page_config.avg_line_info(v_list)
if avg_height == 0:
avg_height = 1
self.avg_ver_ratio = avg_ver_dist /avg_height
v_list.extend(head_foot_list)
for idx,v_block in enumerate(v_list):
if 'class' in v_list[idx].keys():
if v_list[idx]['class'] == 'TEXT':
v_list[idx]['class']= "PARA"
if 'regions' in v_block.keys():
if v_block['regions'] != None and len(v_block['regions']) > 1 :
avg__region_height, avg__region_ver_dist, avg__region_width = page_config.avg_line_info([v_block])
v_block['avg_ver_dist'] = avg__region_ver_dist
avrage_region_ver_ratio= avg__region_ver_dist / max(1,avg__region_height)
#v_block['regions'] = horzontal_merging(v_block['regions'],avrage_region_ver_ratio)
else:
log_info('region key not found for {} in page {}'.format(v_block, path),app_context.application_context )
if 'children' in v_block.keys():
v_block.pop('children')
v_list[idx] =copy.deepcopy(v_block)
for idx,t_block in enumerate(t_list):
t_list[idx]['class'] = 'TABLE'
if t_block['regions'] != None and len(t_block['regions']) > 1 :
avg__region_height, avg__region_ver_dist, avg__region_width = page_config.avg_line_info([t_block])
t_block['avg_ver_dist'] = avg__region_ver_dist
avrage_region_ver_ratio= avg__region_ver_dist / max(1,avg__region_height)
t_list[idx] =copy.deepcopy(t_block)
avg_word_sepc = page_config.avg_word_sep(line_list)
v_list.extend(t_list)
if self.check_double_column(v_list,avg_height):
print("this document is double columnssssssss")
return v_list, n_text_table_regions
flag = False
while flag==True:
v_list, flag = self.merge_remove_overlap(v_list,avg_height, avg_ver_dist, avg_width,avg_word_sepc)
except Exception as e:
log_exception("Error occured during block unifier", app_context.application_context, e)
return None ,None
return v_list, n_text_table_regions