24
24
FONT_CONFIG_CACHE=$( mktemp -d --tmpdir font_tmp.XXXXXXXXXX)
25
25
fi
26
26
MAX_PAGES=0
27
+ SAVE_BOX_TIFF=0
27
28
OUTPUT_DIR=" /tmp/tesstrain/tessdata"
28
29
OVERWRITE=0
29
30
LINEDATA=0
@@ -139,6 +140,8 @@ parse_flags() {
139
140
i=$j ;;
140
141
--overwrite)
141
142
OVERWRITE=1 ;;
143
+ --save_box_tiff)
144
+ SAVE_BOX_TIFF=1 ;;
142
145
--linedata_only)
143
146
LINEDATA=1 ;;
144
147
--extract_font_properties)
@@ -182,7 +185,9 @@ parse_flags() {
182
185
fi
183
186
184
187
# Location where intermediate files will be created.
185
- TRAINING_DIR=${WORKSPACE_DIR} /${LANG_CODE}
188
+ TIMESTAMP=` date +%Y-%m-%d`
189
+ TMP_DIR=$( mktemp -d --tmpdir ${LANG_CODE} -${TIMESTAMP} .XXX )
190
+ TRAINING_DIR=${TMP_DIR}
186
191
# Location of log file for the whole run.
187
192
LOG_FILE=${TRAINING_DIR} /tesstrain.log
188
193
@@ -530,6 +535,9 @@ make__lstmdata() {
530
535
--puncs " ${lang_prefix} .punc" \
531
536
--output_dir " ${OUTPUT_DIR} " --lang " ${LANG_CODE} " \
532
537
" ${pass_through} " " ${lang_is_rtl} "
538
+
539
+ if (( SAVE_BOX_TIFF)) ; then
540
+ tlog " \n=== Saving box/tiff pairs for training data ==="
533
541
for f in " ${TRAINING_DIR} /${LANG_CODE} " .* .box; do
534
542
tlog " Moving ${f} to ${OUTPUT_DIR} "
535
543
mv " ${f} " " ${OUTPUT_DIR} "
@@ -538,6 +546,8 @@ make__lstmdata() {
538
546
tlog " Moving ${f} to ${OUTPUT_DIR} "
539
547
mv " ${f} " " ${OUTPUT_DIR} "
540
548
done
549
+ fi
550
+ tlog " \n=== Moving lstmf files for training data ==="
541
551
for f in " ${TRAINING_DIR} /${LANG_CODE} " .* .lstmf; do
542
552
tlog " Moving ${f} to ${OUTPUT_DIR} "
543
553
mv " ${f} " " ${OUTPUT_DIR} "
0 commit comments