diff --git a/src/python/def_classes.py b/src/python/def_classes.py
index de0c2c8bdf8bbe482c5fdd726fd7de4a724d849e..ba30e380767cdca52ddcbcf21f4aecf2ea829a8b 100644
--- a/src/python/def_classes.py
+++ b/src/python/def_classes.py
@@ -545,10 +545,39 @@ class Document:
if flag_error:
print(str(ind_page) + ': non existing page!')
else:
- im_met = plot_tools.plot_labelled_boxes(imarray,label_textlines, list_allcoords_textlines)
+ im_met, groups, colors = plot_tools.plot_labelled_boxes(imarray,label_textlines, list_allcoords_textlines)
+
+ in_coord = 0
+ coords = in_coord + np.array([0, 0, 10, 10])
+ inc_page = 20
+ flag_notinto = 1
+ for ind_g, i_g in enumerate(groups):
+ if ind_g >= int(len(groups)/2) and flag_notinto:
+ flag_notinto = 0
+ coords[0] = in_coord
+ coords[1] += int(im_met.shape[1]/1.5)
+ coords[2] = in_coord + 10
+ coords[3] += int(im_met.shape[1]/1.5)
+ im_met = plot_tools.lines_box(im_met, coords, colors[ind_g], thick_line = 6)
+ coords[0] += inc_page
+ coords[2] += inc_page
+
self._plot_save(im_met, 'Textboxes labelled', 'TextbLabel', ind_page, self.path_file,
- flag_plot, flag_save_figs)
+ flag_plot, flag_save_figs)
+ coords = in_coord + np.array([0, 0, 10, 10])
+ flag_notinto = 1
+ for ind_g, i_g in enumerate(groups):
+ if ind_g >= int(len(groups)/2) and flag_notinto:
+ flag_notinto = 0
+ coords[0] = in_coord
+ coords[1] += int(im_met.shape[1]/1.5)
+ coords[2] = in_coord + 10
+ coords[3] += int(im_met.shape[1]/1.5)
+ plt.text(coords[1] + 10, coords[2], i_g, fontsize = 10, va = 'bottom', ha = 'left')
+ coords[0] += inc_page
+ coords[2] += inc_page
+
def plot_textl_ordered(self, range_pages = range(1), suffix_xml = '_data',
flag_plot = 1, flag_save_figs = 0, name_outxml = '02_extractedxml',
diff --git a/src/python/plot_tools.py b/src/python/plot_tools.py
index a2079b6ac74551a9597e0e26408b817ed4045670..da5b6c68155f4b8b3e5c324ddf7ce7fc6e861cb4 100644
--- a/src/python/plot_tools.py
+++ b/src/python/plot_tools.py
@@ -13,6 +13,7 @@ import numpy as np
import os
import pickle
import re
+from PIL import Image, ImageDraw, ImageFont
import matplotlib.pyplot as plt
@@ -24,7 +25,7 @@ def lines_box(img, coord, color_vec, thick_line):
# corners in topleftbottomright, and rows and colums
if isinstance(color_vec,str):
color_vec = np.uint8(np.array(Color(color_vec).rgb)*255)
-
+ #print(type(color_vec[0]))
img[(coord[0]):(coord[0]+thick_line),coord[1]:coord[3],:] = color_vec
img[(coord[2]-thick_line):(coord[2]),coord[1]:coord[3],:] = color_vec
img[coord[0]:coord[2],(coord[1]):(coord[1]+thick_line),:] = color_vec
@@ -165,22 +166,18 @@ def plot_labelled_boxes(img,label_textlines, list_allcoords_textlines, thick_l =
for ind_t in label_textlines['text_inheader']:
img_boxes = lines_box(img_boxes, list_allcoords_textlines[:,ind_t].astype(np.uint32),
color_vec = 'yellow', thick_line = thick_l)
+ color_nid = np.uint8(np.array([127, 0, 180]))
if 'notidentified' in label_textlines.keys():
for ind_t in label_textlines['notidentified']:
img_boxes = lines_box(img_boxes, list_allcoords_textlines[:,ind_t].astype(np.uint32),
- color_vec = [0.5, 0, 0.7], thick_line = thick_l)
+ color_vec = color_nid, thick_line = thick_l)
groups = ['Footnote', 'Header', 'Header of single col', 'Page number', '1st column',
'2nd column', 'Text in header', 'Not Identified']
- colors = ['red', 'blue', 'black', 'green', 'magenta', 'cyan', 'yellow', [0.5, 0, 0.7]]
- '''
- coords = [30, 30, 40, 40]
- inc_page = img_boxes.shape[1]/len(groups)
- for ind_g, i_g in enumerate(groups):
- legend_toppage(i_g, colors[ind_g], xin, yin)
- coords[0] += inc_page
- coords[1] += inc_page
- '''
- return img_boxes
+ colors = ['red', 'blue', 'black', 'green', 'magenta', 'cyan', 'yellow', color_nid]
+
+
+
+ return img_boxes, groups, colors
def plot_margins(img, side_margins, topb_margins, gap_line = 2, thick_l = 6):
@@ -266,9 +263,3 @@ def plot_save_parallel(folder_pickles):
name_fig = full_filename[:-3] + format_fig
fig.savefig(name_fig, format = format_fig, dpi = 200)
plt.close(fig)
-
-def legend_toppage(text_l, color_l, coord, img):
- lines_box(img, coord, color_l, thick_line = 6)
- #plt.Circle((x, y), 3, color = color_l)
- plt.text(coord[0] + 4, coord[1], text_l, fontsize = 7, va = 'bottom', ha = 'left')
- pass
\ No newline at end of file