diff options
author | Nick White <git@njw.me.uk> | 2011-12-03 18:18:12 +0000 |
---|---|---|
committer | Nick White <git@njw.me.uk> | 2011-12-03 18:18:12 +0000 |
commit | 9239c8747aa79b95ae87b61b5758d089040d0616 (patch) | |
tree | b037a36a58c38f278a1bfd57938b48e61276a04a /extras | |
parent | aa4c9f8df45a9a6881e0a90e50a7f1359ac5d1e1 (diff) |
Have extras scripts take jpgs into account, and change their usage slightly
Diffstat (limited to 'extras')
-rw-r--r-- | extras/mkocrpdf.sh | 9 | ||||
-rw-r--r-- | extras/mkocrtxt.sh | 7 | ||||
-rwxr-xr-x | extras/mkpdf.sh | 5 |
3 files changed, 15 insertions, 6 deletions
diff --git a/extras/mkocrpdf.sh b/extras/mkocrpdf.sh index 1ec9a06..9d281db 100644 --- a/extras/mkocrpdf.sh +++ b/extras/mkocrpdf.sh @@ -9,9 +9,12 @@ # script does that, then reduces the results back down # to create a reasonable size PDF. +test $# -ne 1 && echo "Usage: $0 bookdir" && exit 1 +cd "$1" || exit 1 + echo 'tessedit_create_hocr 1' > hocr -for i in `ls *png` +for i in `ls` do echo "$i" @@ -42,8 +45,8 @@ do rm -f "$i.big.tif" "$i.big.tif.html" "$i.html" done -echo book.pdf - # cat the pdf pages together pdftk *pdf cat output book.pdf rm -f [0-9]*pdf hocr + +echo "$1/book.pdf" diff --git a/extras/mkocrtxt.sh b/extras/mkocrtxt.sh index 497f8a7..62f95a4 100644 --- a/extras/mkocrtxt.sh +++ b/extras/mkocrtxt.sh @@ -11,7 +11,10 @@ # converted to tiff format, so that they're readable by # any version of tesseract. -for i in `ls *png` +test $# -ne 1 && echo "Usage: $0 bookdir" && exit 1 +cd "$1" || exit 1 + +for i in `ls` do echo "$i" @@ -30,4 +33,4 @@ do rm -f "$i.big.tif" "$i.txt" done -echo book.txt +echo "$1/book.txt" diff --git a/extras/mkpdf.sh b/extras/mkpdf.sh index cbfcd1b..a2ba2c7 100755 --- a/extras/mkpdf.sh +++ b/extras/mkpdf.sh @@ -3,4 +3,7 @@ # # Requires imagemagick -convert *.png book.pdf +test $# -ne 1 && echo "Usage: $0 bookdir" && exit 1 + +cd "$1" && convert * book.pdf +echo "$1/book.pdf" |