article-analysis/ocr at master · AAAI/article-analysis · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
#!/bin/bash

if [ "$1" != "" ]
then
    echo "$1" | grep -q '\.pdf$'
    if [ $? -eq 0 ]
    then
        pdfs=$1
        dir=`dirname $1`
    else
        pdfs=`ls $1/*.pdf`
        dir=$1
    fi
else
    pdfs=`ls *.pdf`
    dir=`pwd`
fi

for pdf in $pdfs
do
    doc=`basename $pdf .pdf`

    if grep -q FontName $dir/$doc.pdf
    then
        echo "Extracting with pdftotext on $doc.pdf..."
        pdftotext $dir/$doc.pdf $doc.txt
    fi

    filesize=$(stat -c%s "$doc.txt" 2> /dev/null)
    if [ "$filesize" -le "1000" ]
    then
        echo "Performing OCR on $doc.pdf..."

        pdfimages $dir/$doc.pdf $doc

        for image in `ls $doc-*[[:digit:]].ppm 2> /dev/null`
        do
            f=`basename $image .ppm`
            convert $f.ppm $f.pbm
            rm $f.ppm
        done

        # try using "unpaper" first
        for image in `ls $doc-*[[:digit:]].pbm`
        do
            f=`basename $image .pbm`
            unpaper $image $f-fixed.pbm
            nice tesseract $f-fixed.pbm $f
        done

        cat $doc-*.txt > $doc.txt

        # if that fails, try again without "unpaper"
        filesize=$(stat -c%s "$doc.txt")
        if [ "$filesize" -le "1000" ]
        then
            for image in `ls $doc-*[[:digit:]].pbm`
            do
                f=`basename $image .pbm`
                echo "OCR on $f.pbm"
                nice tesseract $f.pbm $f
            done
        fi

        rm $doc-*.pbm $doc-*.txt
    fi
done