def check_word_subscript_exist(word):
"""
:param word:
:return:
"""
found_subscript = False
for i in range(len(word) - 1):
# avoid over doing on the in accurate.
c1 = word[i].get_text()
c2 = word[i + 1].get_text()
if not (c1.isdigit() or c1.isalpha()):
continue
if not (c2.isdigit() or c2.isalpha()):
continue
from pdfxml.pdf_util.layout_util import get_height
if get_height(word[i].bbox) == 0:
continue
hr_fea_val = height_ratio_AB(word[i].bbox, word[i + 1].bbox)
nvcd_fea_val = normalized_ycenter_diff_AB(word[i].bbox,
word[i + 1].bbox)
if hr_ratio_range[0] <= hr_fea_val <= hr_ratio_range[1] and \
nvcd_range[0] <= nvcd_fea_val <= nvcd_range[1]:
found_subscript = True
me_extraction_logger.debug("{} Bbox1:{}, {} Bbox2:{}".format(
word[i].get_text().encode('utf-8'), word[i].bbox,
word[i + 1].get_text().encode('utf-8'), word[i + 1].bbox))
return found_subscript
def adjust_bbox_h_latex_vertically(bbox, latex):
"""
:param bbox:
:param latex:
:return:
"""
# adjust vertically.
ver_latex2ur, ver_latex2lr, hor_latex2ur, hor_latex2lr = get_latex2adjustment_ratio(
)
#upper_ratio = ver_latex2ur[latex]
#lower_ratio = ver_latex2lr[latex]
upper_ratio, lower_ratio = get_upper_lower_ratio(latex, ver_latex2ur,
ver_latex2lr)
new_bbox = copy.copy(bbox)
if isinstance(new_bbox, BBox):
new_bbox = new_bbox.to_list()
else:
new_bbox = list(new_bbox)
height = get_height(bbox)
new_bbox[1] = new_bbox[1] - height * lower_ratio
new_bbox[3] = new_bbox[3] + height * upper_ratio
return new_bbox
def adjust_bbox_h_gt(bbox, gt, debug=False):
"""
width stable character.
:param bbox:
:param gt: String of glyph type of xyz
:return:
"""
gt2ur, gt2lr = get_gt2adjust()
upper_ratio = gt2ur[gt]
lower_ratio = gt2lr[gt]
if debug:
print gt, upper_ratio, lower_ratio
new_bbox = copy.copy(bbox)
new_bbox = list(new_bbox)
height = get_height(bbox)
new_bbox[1] = new_bbox[1] - height * lower_ratio
new_bbox[3] = new_bbox[3] + height * upper_ratio
return new_bbox
def normalized_At_Bb_diff_by_height_AB(b1, b2):
d = b2[1] - b1[3]
return d / get_height(b1)
def normalized_Ab_Bt_diff_by_height_AB(b1, b2):
d = b2[3] - b1[1]
return d / get_height(b1)
def normalized_Ar_Br_diff_by_height_AB(b1, b2):
d = b2[2] - b1[2]
return d / get_height(b1)
def normalized_Al_Bl_diff_by_height_AB(b1, b2):
d = b2[0] - b1[0]
return d / get_height(b1)
def normalized_xcenter_diff_by_height_AB(b1, b2):
"""
horizontal center difference normalized by height of first bbox
"""
x_center_diff = get_x_center(b1) - get_x_center(b2)
return x_center_diff / get_height(b1)
def normalized_ycenter_diff_AB(b1, b2):
"""
vertical center difference normalized by height of first bbox
"""
return (get_y_center(b2) - get_y_center(b1)) / get_height(b1)
def height_ratio_AB(b1, b2):
"""
height ratio
"""
return get_height(b2) / get_height(b1)
def normalized_ycenter_diff_by_merge_AB(b1, b2):
"""
symmetric vertical difference
"""
mb = merge_bbox(b1, b2)
return abs(get_y_center(b1) - get_y_center(b2)) / get_height(mb)