diff --git a/src/python/def_classes.py b/src/python/def_classes.py
index 98c12cc0509acacee0a2a89eb2904962ae8dd73b..7c5282efffc2a28bb64ed4a3efc801154394b8d3 100644
--- a/src/python/def_classes.py
+++ b/src/python/def_classes.py
@@ -772,7 +772,7 @@ class Document:
if (flag_op) == 1 and (type(vec_in) == str):
t_aux = vec_in
- text_search = [t_aux]
+ vec_in = [t_aux]
name_tar = self.folder_database + '/' + str(self.year) + '/' + self.name_outcorrxml + '.tar.gz'
if os.path.isfile(name_tar):
@@ -787,7 +787,7 @@ class Document:
print('You need first to obtained the corrected XML file from the original pdf')
dict_text_found = dict()
- dict_all_info = dict()
+ #dict_all_info = dict()
for ind_abs, ind_page in enumerate(self.n_pages):
XML_root = ET.Element('pages')
#print(ind_abs,len(self.XML_main))
@@ -796,7 +796,7 @@ class Document:
bbox_page = np.array(XML_root[0].attrib['bbox'].split(',')).astype(np.float64)
dim_img = imarray.shape[:2]
- dict_text_found_aux = self._extract_textl_if(XML_root, text_search)
+ dict_text_found_aux = self._extract_textl_if(XML_root, vec_in, flag_op)
dict_text_found[ind_abs] = dict_text_found_aux
if flag_plot:
@@ -805,79 +805,26 @@ class Document:
aux_c = dict_text_found_aux[key][0]
coord_textline = np.concatenate((coord_textline, aux_c.reshape((4,1))), axis = 1)
- ind_c = np.argwhere([dict_text_found_aux[key][1].find(i) > -1 for i in text_search])
- colors = ['cyan','green','red','blue','yellow']
- imarray = plot_tools.highlight_text(imarray, aux_c, bbox_page,
- color_vec = colors[np.ravel(ind_c)[0]], alpha = True, filled = False, thick_line = 4)
-
- if len(text_search) > 1:
- aux_str = '_'.join(text_search)
- else:
- aux_str = text_search[0]
- self._plot_save(imarray, 'Results of search:' + aux_str, 'TextFound-' + aux_str,
- ind_page, self.path_file, flag_plot, flag_save_figs)
-
- if flag_parseall:
- all_info_fonts = self._extract_all_fonts(XML_root)
- dict_all_info[ind_abs] = all_info_fonts
-
- if flag_parseall:
- return dict_text_found, dict_all_info
- else:
- return dict_text_found
-
- def parse_text(self, suffix_xml='_data', name_outxml = '02_extractedxml', name_outcorrxml='04_correctedxml',
- text_search = '', flag_plot = 1, flag_save_figs = 0, flag_parseall = 0):
-
- if 'name_outcorrxml' not in self.__dict__.keys():
- self.name_outcorrxml = name_outcorrxml
-
- if type(text_search) == str:
- t_aux = text_search
- text_search = [t_aux]
-
- name_tar = self.folder_database + '/' + str(self.year) + '/' + self.name_outcorrxml + '.tar.gz'
- if os.path.isfile(name_tar):
- name_xml = './' + str(self.year) + '/' + str(self.id_doc) + suffix_xml + 'corr.xml'
- if name_xml in utils_proc.get_list(self.year, self.folder_database, name_outcorrxml)[0]:
- h_xml = utils_proc.get_handlerfile(name_xml, self.folder_database, name_outcorrxml)
- XML_tree = ET.parse(h_xml)
- XML_main_corr = XML_tree.getroot()
- else:
- print('You need first to obtained the corrected XML file from the original pdf')
- else:
- print('You need first to obtained the corrected XML file from the original pdf')
-
- dict_text_found = dict()
- dict_all_info = dict()
- for ind_abs, ind_page in enumerate(self.n_pages):
- XML_root = ET.Element('pages')
- #print(ind_abs,len(self.XML_main))
- XML_root.append(XML_main_corr[ind_abs])
- imarray = np.array(self.imgobj[ind_page])
- bbox_page = np.array(XML_root[0].attrib['bbox'].split(',')).astype(np.float64)
- dim_img = imarray.shape[:2]
- dict_text_found_aux = self._extract_textl_if(XML_root, text_search)
- dict_text_found[ind_abs] = dict_text_found_aux
-
- if flag_plot:
- coord_textline = np.array([]).reshape((4,0))
- for key in dict_text_found_aux.keys():
- aux_c = dict_text_found_aux[key][0]
- coord_textline = np.concatenate((coord_textline, aux_c.reshape((4,1))), axis = 1)
+ if flag_op == 1:
+ ind_c = np.argwhere([dict_text_found_aux[key][1].find(i) > -1 for i in vec_in])
+ if len(vec_in) > 1:
+ aux_str = 'string: ' + '_'.join(vec_in)
+ else:
+ aux_str = 'string: ' + vec_in[0]
+ elif flag_op == 2:
+ ind_c = dict_text_found_aux[key][2]
+ aux_str = 'size: ' + '_'.join(np.array(vec_in).astype(str))
- ind_c = np.argwhere([dict_text_found_aux[key][1].find(i) > -1 for i in text_search])
colors = ['cyan','green','red','blue','yellow']
imarray = plot_tools.highlight_text(imarray, aux_c, bbox_page,
color_vec = colors[np.ravel(ind_c)[0]], alpha = True, filled = False, thick_line = 4)
- if len(text_search) > 1:
- aux_str = '_'.join(text_search)
- else:
- aux_str = text_search[0]
- self._plot_save(imarray, 'Results of search:' + aux_str, 'TextFound-' + aux_str,
+ self._plot_save(imarray, 'Results of search ' + aux_str, 'TextFound-' + aux_str,
ind_page, self.path_file, flag_plot, flag_save_figs)
-
+
+ return dict_text_found
+
+ '''
if flag_parseall:
all_info_fonts = self._extract_all_fonts(XML_root)
dict_all_info[ind_abs] = all_info_fonts
@@ -886,86 +833,53 @@ class Document:
return dict_text_found, dict_all_info
else:
return dict_text_found
-
- def parse_fontsize(self, suffix_xml='_data', name_outxml = '02_extractedxml', name_outcorrxml='04_correctedxml',
- font_max_min = [8,10], flag_plot = 1, flag_save_figs = 0):
-
- if 'name_outcorrxml' not in self.__dict__.keys():
- self.name_outcorrxml = name_outcorrxml
+ '''
-
- name_tar = self.folder_database + '/' + str(self.year) + '/' + self.name_outcorrxml + '.tar.gz'
- if os.path.isfile(name_tar):
- name_xml = './' + str(self.year) + '/' + str(self.id_doc) + suffix_xml + 'corr.xml'
- if name_xml in utils_proc.get_list(self.year, self.folder_database, name_outcorrxml)[0]:
- h_xml = utils_proc.get_handlerfile(name_xml, self.folder_database, name_outcorrxml)
- XML_tree = ET.parse(h_xml)
- XML_main_corr = XML_tree.getroot()
- else:
- print('You need first to obtained the corrected XML file from the original pdf')
- else:
- print('You need first to obtained the corrected XML file from the original pdf')
-
- dict_text_found = dict()
- for ind_abs, ind_page in enumerate(self.n_pages):
- XML_root = ET.Element('pages')
- #print(ind_abs,len(self.XML_main))
- XML_root.append(XML_main_corr[ind_abs])
- imarray = np.array(self.imgobj[ind_page])
- bbox_page = np.array(XML_root[0].attrib['bbox'].split(',')).astype(np.float64)
- dim_img = imarray.shape[:2]
- dict_text_found_aux = self._extract_textl_ifsize(XML_root, font_max_min)
- dict_text_found[ind_abs] = dict_text_found_aux
-
- if flag_plot:
- coord_textline = np.array([]).reshape((4,0))
- for key in dict_text_found_aux.keys():
- aux_c = dict_text_found_aux[key][0]
- coord_textline = np.concatenate((coord_textline, aux_c.reshape((4,1))), axis = 1)
-
- ind_c = dict_text_found_aux[key][2]
- colors = ['cyan','green','red','blue','yellow']
- imarray = plot_tools.highlight_text(imarray, aux_c, bbox_page,
- color_vec = colors[np.ravel(ind_c)[0]], alpha = True, filled = False, thick_line = 4)
-
- aux_str = '_'.join(np.array(font_max_min).astype(str))
- self._plot_save(imarray, 'Results of search size:' + aux_str, 'TextFoundSize-' + aux_str,
- ind_page, self.path_file, flag_plot, flag_save_figs)
-
- return dict_text_found
-
- def _extract_textl_ifsize(self, XML_root, font_max_min = [8,10]):
+ def _extract_textl_if(self, XML_root, vec_in, flag_op):
# The key is as follows: <block_id>_<textbox_id>_<y1>_<x1>
dict_text_found = dict()
- font_max_min_mat = np.zeros((2,len(font_max_min)))
- for i in range(len(font_max_min) - 1):
- font_max_min_mat[0,i] = font_max_min[i]
- font_max_min_mat[1,i] = font_max_min[i + 1]
+
+ if flag_op == 2:
+ font_max_min_mat = np.zeros((2,len(vec_in)))
+ for i in range(len(vec_in) - 1):
+ font_max_min_mat[0,i] = vec_in[i]
+ font_max_min_mat[1,i] = vec_in[i + 1]
for ind_el in range(0, len(XML_root[0])):
for ind_line in range(0, len(XML_root[0][ind_el])):
if XML_root[0][ind_el][ind_line].tag == 'textline':
#print(XML_root[0][ind_el][ind_line].text)
str_text = str(XML_root[0][ind_el][ind_line].text)
- ind_f2 = utils_proc.find_all(str_text, 'size="')
- ind_val = list()
- for i2 in range(len(ind_f2)):
- f_size = np.array(str_text[ind_f2[i2] + len('size'):].split('"')[1]).astype(np.float64)
- ind_int = np.intersect1d(np.argwhere(font_max_min_mat[0,:] < f_size), np.argwhere(font_max_min_mat[1,:] > f_size))
- if len(ind_int):
- ind_val.append(ind_int)
+ if flag_op == 1:
+ if np.sum([str_text.find(i) > -1 for i in vec_in]):
+ #if (str_text.find(text_search)) > -1:
+ key_d = '_'.join([XML_root[0][ind_el].attrib['block'],XML_root[0][ind_el].attrib['id'],
+ str(XML_root[0][ind_el][ind_line].attrib['bbox'].split(',')[0]),
+ str(XML_root[0][ind_el][ind_line].attrib['bbox'].split(',')[1])])
+ coord_textline_aux = np.array(XML_root[0][ind_el][ind_line].attrib['bbox'].split(',')).astype(np.float64)
+ dict_text_found[key_d] = [coord_textline_aux,str_text]
- #print(ind_val, ind_val.shape)
- if np.sum(np.array(ind_val) > -1):
- #if (str_text.find(text_search)) > -1:
- key_d = '_'.join([XML_root[0][ind_el].attrib['block'],XML_root[0][ind_el].attrib['id'],
- str(XML_root[0][ind_el][ind_line].attrib['bbox'].split(',')[0]),
- str(XML_root[0][ind_el][ind_line].attrib['bbox'].split(',')[1])])
- coord_textline_aux = np.array(XML_root[0][ind_el][ind_line].attrib['bbox'].split(',')).astype(np.float64)
- dict_text_found[key_d] = [coord_textline_aux, str_text, np.max(ind_val)]
-
- return dict_text_found
+ elif flag_op == 2:
+ ind_f2 = utils_proc.find_all(str_text, 'size="')
+ ind_val = list()
+ for i2 in range(len(ind_f2)):
+ f_size = np.array(str_text[ind_f2[i2] + len('size'):].split('"')[1]).astype(np.float64)
+ ind_int = np.intersect1d(np.argwhere(font_max_min_mat[0,:] < f_size), np.argwhere(font_max_min_mat[1,:] > f_size))
+ if len(ind_int):
+ ind_val.append(ind_int)
+
+ if np.sum(np.array(ind_val) > -1):
+ #if (str_text.find(text_search)) > -1:
+ key_d = '_'.join([XML_root[0][ind_el].attrib['block'],XML_root[0][ind_el].attrib['id'],
+ str(XML_root[0][ind_el][ind_line].attrib['bbox'].split(',')[0]),
+ str(XML_root[0][ind_el][ind_line].attrib['bbox'].split(',')[1])])
+ coord_textline_aux = np.array(XML_root[0][ind_el][ind_line].attrib['bbox'].split(',')).astype(np.float64)
+ dict_text_found[key_d] = [coord_textline_aux, str_text, np.max(ind_val)]
+
+ return dict_text_found
+
+
def parse_all(self, suffix_xml='_data', name_outxml = '02_extractedxml', name_outcorrxml='04_correctedxml'):
@@ -1007,26 +921,8 @@ class Document:
return dict_all_info, mat_all_f
- def _extract_textl_if(self, XML_root, text_search = ''):
- # The key is as follows: <block_id>_<textbox_id>_<y1>_<x1>
- dict_text_found = dict()
-
- for ind_el in range(0, len(XML_root[0])):
- for ind_line in range(0, len(XML_root[0][ind_el])):
- if XML_root[0][ind_el][ind_line].tag == 'textline':
- #print(XML_root[0][ind_el][ind_line].text)
- str_text = str(XML_root[0][ind_el][ind_line].text)
- if np.sum([str_text.find(i) > -1 for i in text_search]):
- #if (str_text.find(text_search)) > -1:
- key_d = '_'.join([XML_root[0][ind_el].attrib['block'],XML_root[0][ind_el].attrib['id'],
- str(XML_root[0][ind_el][ind_line].attrib['bbox'].split(',')[0]),
- str(XML_root[0][ind_el][ind_line].attrib['bbox'].split(',')[1])])
- coord_textline_aux = np.array(XML_root[0][ind_el][ind_line].attrib['bbox'].split(',')).astype(np.float64)
- dict_text_found[key_d] = [coord_textline_aux,str_text]
-
- return dict_text_found
-
def _extract_all_fonts(self, XML_root):
+
all_info_fonts = dict()
for ind_el in range(0, len(XML_root[0])):
for ind_line in range(0, len(XML_root[0][ind_el])):
@@ -1262,3 +1158,146 @@ class Document:
utils_proc.call_with_out(command)
print("End of file %s - %s seconds -" % (self.input_file, (time.time() - start_time)))
+
+'''
+ def _extract_textl_ifsize(self, XML_root, font_max_min = [8,10]):
+ # The key is as follows: <block_id>_<textbox_id>_<y1>_<x1>
+ dict_text_found = dict()
+ font_max_min_mat = np.zeros((2,len(font_max_min)))
+ for i in range(len(font_max_min) - 1):
+ font_max_min_mat[0,i] = font_max_min[i]
+ font_max_min_mat[1,i] = font_max_min[i + 1]
+
+ for ind_el in range(0, len(XML_root[0])):
+ for ind_line in range(0, len(XML_root[0][ind_el])):
+ if XML_root[0][ind_el][ind_line].tag == 'textline':
+ #print(XML_root[0][ind_el][ind_line].text)
+ str_text = str(XML_root[0][ind_el][ind_line].text)
+
+ ind_f2 = utils_proc.find_all(str_text, 'size="')
+ ind_val = list()
+ for i2 in range(len(ind_f2)):
+ f_size = np.array(str_text[ind_f2[i2] + len('size'):].split('"')[1]).astype(np.float64)
+ ind_int = np.intersect1d(np.argwhere(font_max_min_mat[0,:] < f_size), np.argwhere(font_max_min_mat[1,:] > f_size))
+ if len(ind_int):
+ ind_val.append(ind_int)
+
+ #print(ind_val, ind_val.shape)
+ if np.sum(np.array(ind_val) > -1):
+ #if (str_text.find(text_search)) > -1:
+ key_d = '_'.join([XML_root[0][ind_el].attrib['block'],XML_root[0][ind_el].attrib['id'],
+ str(XML_root[0][ind_el][ind_line].attrib['bbox'].split(',')[0]),
+ str(XML_root[0][ind_el][ind_line].attrib['bbox'].split(',')[1])])
+ coord_textline_aux = np.array(XML_root[0][ind_el][ind_line].attrib['bbox'].split(',')).astype(np.float64)
+ dict_text_found[key_d] = [coord_textline_aux, str_text, np.max(ind_val)]
+
+ return dict_text_found
+
+ def parse_text(self, suffix_xml='_data', name_outxml = '02_extractedxml', name_outcorrxml='04_correctedxml',
+ text_search = '', flag_plot = 1, flag_save_figs = 0, flag_parseall = 0):
+
+ if 'name_outcorrxml' not in self.__dict__.keys():
+ self.name_outcorrxml = name_outcorrxml
+
+ if type(text_search) == str:
+ t_aux = text_search
+ text_search = [t_aux]
+
+ name_tar = self.folder_database + '/' + str(self.year) + '/' + self.name_outcorrxml + '.tar.gz'
+ if os.path.isfile(name_tar):
+ name_xml = './' + str(self.year) + '/' + str(self.id_doc) + suffix_xml + 'corr.xml'
+ if name_xml in utils_proc.get_list(self.year, self.folder_database, name_outcorrxml)[0]:
+ h_xml = utils_proc.get_handlerfile(name_xml, self.folder_database, name_outcorrxml)
+ XML_tree = ET.parse(h_xml)
+ XML_main_corr = XML_tree.getroot()
+ else:
+ print('You need first to obtained the corrected XML file from the original pdf')
+ else:
+ print('You need first to obtained the corrected XML file from the original pdf')
+
+ dict_text_found = dict()
+ dict_all_info = dict()
+ for ind_abs, ind_page in enumerate(self.n_pages):
+ XML_root = ET.Element('pages')
+ #print(ind_abs,len(self.XML_main))
+ XML_root.append(XML_main_corr[ind_abs])
+ imarray = np.array(self.imgobj[ind_page])
+ bbox_page = np.array(XML_root[0].attrib['bbox'].split(',')).astype(np.float64)
+ dim_img = imarray.shape[:2]
+ dict_text_found_aux = self._extract_textl_if(XML_root, text_search)
+ dict_text_found[ind_abs] = dict_text_found_aux
+
+ if flag_plot:
+ coord_textline = np.array([]).reshape((4,0))
+ for key in dict_text_found_aux.keys():
+ aux_c = dict_text_found_aux[key][0]
+ coord_textline = np.concatenate((coord_textline, aux_c.reshape((4,1))), axis = 1)
+
+ ind_c = np.argwhere([dict_text_found_aux[key][1].find(i) > -1 for i in text_search])
+ colors = ['cyan','green','red','blue','yellow']
+ imarray = plot_tools.highlight_text(imarray, aux_c, bbox_page,
+ color_vec = colors[np.ravel(ind_c)[0]], alpha = True, filled = False, thick_line = 4)
+
+ if len(text_search) > 1:
+ aux_str = '_'.join(text_search)
+ else:
+ aux_str = text_search[0]
+ self._plot_save(imarray, 'Results of search:' + aux_str, 'TextFound-' + aux_str,
+ ind_page, self.path_file, flag_plot, flag_save_figs)
+
+ if flag_parseall:
+ all_info_fonts = self._extract_all_fonts(XML_root)
+ dict_all_info[ind_abs] = all_info_fonts
+
+ if flag_parseall:
+ return dict_text_found, dict_all_info
+ else:
+ return dict_text_found
+
+ def parse_fontsize(self, suffix_xml='_data', name_outxml = '02_extractedxml', name_outcorrxml='04_correctedxml',
+ font_max_min = [8,10], flag_plot = 1, flag_save_figs = 0):
+
+ if 'name_outcorrxml' not in self.__dict__.keys():
+ self.name_outcorrxml = name_outcorrxml
+
+
+ name_tar = self.folder_database + '/' + str(self.year) + '/' + self.name_outcorrxml + '.tar.gz'
+ if os.path.isfile(name_tar):
+ name_xml = './' + str(self.year) + '/' + str(self.id_doc) + suffix_xml + 'corr.xml'
+ if name_xml in utils_proc.get_list(self.year, self.folder_database, name_outcorrxml)[0]:
+ h_xml = utils_proc.get_handlerfile(name_xml, self.folder_database, name_outcorrxml)
+ XML_tree = ET.parse(h_xml)
+ XML_main_corr = XML_tree.getroot()
+ else:
+ print('You need first to obtained the corrected XML file from the original pdf')
+ else:
+ print('You need first to obtained the corrected XML file from the original pdf')
+
+ dict_text_found = dict()
+ for ind_abs, ind_page in enumerate(self.n_pages):
+ XML_root = ET.Element('pages')
+ #print(ind_abs,len(self.XML_main))
+ XML_root.append(XML_main_corr[ind_abs])
+ imarray = np.array(self.imgobj[ind_page])
+ bbox_page = np.array(XML_root[0].attrib['bbox'].split(',')).astype(np.float64)
+ dim_img = imarray.shape[:2]
+ dict_text_found_aux = self._extract_textl_ifsize(XML_root, font_max_min)
+ dict_text_found[ind_abs] = dict_text_found_aux
+
+ if flag_plot:
+ coord_textline = np.array([]).reshape((4,0))
+ for key in dict_text_found_aux.keys():
+ aux_c = dict_text_found_aux[key][0]
+ coord_textline = np.concatenate((coord_textline, aux_c.reshape((4,1))), axis = 1)
+
+ ind_c = dict_text_found_aux[key][2]
+ colors = ['cyan','green','red','blue','yellow']
+ imarray = plot_tools.highlight_text(imarray, aux_c, bbox_page,
+ color_vec = colors[np.ravel(ind_c)[0]], alpha = True, filled = False, thick_line = 4)
+
+ aux_str = '_'.join(np.array(font_max_min).astype(str))
+ self._plot_save(imarray, 'Results of search size:' + aux_str, 'TextFoundSize-' + aux_str,
+ ind_page, self.path_file, flag_plot, flag_save_figs)
+
+ return dict_text_found
+'''
\ No newline at end of file