| Revision | dcc972c4aa27a52f4caa7382fcb5b816f47b1fea (tree) |
|---|---|
| Time | 2008-04-11 21:32:10 |
| Author | iselllo |
| Commiter | iselllo |
I added the file search_pdf_create_txt.sh which is a simple script to
search pdf files by keywords.
| @@ -0,0 +1,104 @@ | ||
| 1 | +#!/bin/bash | |
| 2 | +# place this file in the directory of pdf files | |
| 3 | +# you want to search | |
| 4 | + | |
| 5 | +echo "Do you want to: " | |
| 6 | +echo "1- create/update index" | |
| 7 | +echo "2- search biblio" | |
| 8 | +echo "_________ " | |
| 9 | +echo "NOTE: index only need to be created once" | |
| 10 | +read choice | |
| 11 | +echo "your choice [$choice]" | |
| 12 | + | |
| 13 | +if [ $choice -eq 1 ] | |
| 14 | +then | |
| 15 | +if [ -f ./tmp_menu_file ] | |
| 16 | +then | |
| 17 | +rm ./tmp_menu_file | |
| 18 | +fi | |
| 19 | +if [ -f ./log_treat ] | |
| 20 | +then | |
| 21 | +rm ./log_treat | |
| 22 | +fi | |
| 23 | +if [ -d ./text_format ] | |
| 24 | +then | |
| 25 | +echo "an index has already been created" | |
| 26 | +echo "update in progress" | |
| 27 | +else | |
| 28 | +mkdir ./text_format | |
| 29 | +fi | |
| 30 | +count=0 | |
| 31 | +echo "reading directory" | |
| 32 | +for file in `ls -1 *.pdf` | |
| 33 | +do | |
| 34 | +count=`expr $count + 1` | |
| 35 | +echo "$file" >>tmp_menu_file | |
| 36 | +done | |
| 37 | +echo "reading done" | |
| 38 | +echo "[$count] pdf files" | |
| 39 | +filename=`cut -d"." -f 1 tmp_menu_file` | |
| 40 | +count=0 | |
| 41 | +for file in `echo $filename` | |
| 42 | +do | |
| 43 | +count=`expr $count + 1` | |
| 44 | +echo "loading $file" >>log_treat | |
| 45 | +echo "loading $file" | |
| 46 | +if [ -f ./text_format/$file.txt ] | |
| 47 | +then | |
| 48 | +echo "file [$file] already treated " | |
| 49 | +else | |
| 50 | +pdftotext $file.pdf ./text_format/$file.txt | |
| 51 | +echo "file [$count,$file] treated ">>log_treat | |
| 52 | +echo "file [$count,$file] treated " | |
| 53 | +fi | |
| 54 | +done | |
| 55 | +fi | |
| 56 | + | |
| 57 | +if [ $choice -eq 2 ] | |
| 58 | +then | |
| 59 | +cd ./text_format/ | |
| 60 | +fin=0 | |
| 61 | +key1="" | |
| 62 | +key2="" | |
| 63 | +key3="" | |
| 64 | +while [ $fin -eq 0 ] | |
| 65 | +do | |
| 66 | +echo "enter up to 3 keyword(s) [AND]" | |
| 67 | +echo "_____________[crtl-c for end search]" | |
| 68 | +read key1 key2 key3 | |
| 69 | +echo "keywords: [$key1,$key2,$key3]" | |
| 70 | +if [ -f ./tmp ] | |
| 71 | +then | |
| 72 | +rm ./tmp | |
| 73 | +fi | |
| 74 | +echo `egrep -i -l "$key1" *.txt` >>tmp | |
| 75 | +echo "results for [$key1] only:" | |
| 76 | +echo `egrep -i -l "$key1" *.txt` | |
| 77 | +if [ -n "$key2" ] | |
| 78 | +then | |
| 79 | +echo "results for [$key2] only:" | |
| 80 | +echo `egrep -i -l "$key2" *.txt` | |
| 81 | +for file in `cat tmp` | |
| 82 | +do | |
| 83 | +echo `egrep -i -l "$key2" $file`>>tmp2 | |
| 84 | +done | |
| 85 | +echo "results for [$key1] and [$key2]:" | |
| 86 | +echo `cat tmp2` | |
| 87 | +rm ./tmp | |
| 88 | +if [ -n "$key3" ] | |
| 89 | +then | |
| 90 | +echo "results for [$key3] only:" | |
| 91 | +echo `egrep -i -l "$key3" *.txt` | |
| 92 | +for file in `cat tmp2` | |
| 93 | +do | |
| 94 | +echo `egrep -i -l "$key3" $file`>>tmp | |
| 95 | +done | |
| 96 | +rm ./tmp2 | |
| 97 | +echo "results for [$key1] and [$key2] and [$key3]:" | |
| 98 | +echo `cat tmp` | |
| 99 | +rm ./tmp | |
| 100 | +fi | |
| 101 | +fi | |
| 102 | +done | |
| 103 | +exit | |
| 104 | +fi |
Fork
Bash-scripts/search_pdf_create_txt.sh