#!/bin/bash
## receives tif+box pair (from tesseract-training procedure) and crops the region described by the first box;  Eugene Reimer  2009;
## wanted this to determine whether the 12-pixel-too-low bug was in tesseract-makebox or in the viewer (tesseractTrainer.py);
## may go on to make this more general, making one crop for each row, or perhaps making a drawn-upon image instead;  better just to use tesseractTrainer:-)
##
## TURNS OUT I don't even need to fix tesseractTrainer -- 'tis tesseract-makebox that is broken--!!--

PAGEHT=$(identify -format "%h" $1)			##get height of the tif-image
read -r BCHR XLO YLO XHI YHI PAGENBR  <${1/tif/box}	##read line of boxfile;  vars get empty-string if omitted (all if empty-line; PAGENBR if simple tiff)
((H=YHI-YLO, W=XHI-XLO, T=PAGEHT-YHI))			##convert to coordinate-system used by the netpbm routines

echo "PAGEHT:$PAGEHT; pnmcut $XLO $T $W $H >tmpCharBox-${1%.*}-$XLO-$YLO-$XHI-$YHI.pbm"		##DEBUG
convert $1 tmp$$.pbm										##convert to PBM
pnmcut $XLO $T $W $H tmp$$.pbm  >tmpCharBox-${1%.*}-$XLO-$YLO-$XHI-$YHI.pbm			##crop using pnmcut
rm tmp$$.pbm