142

Is there a way to search PDF files using the power of grep, without converting to text first in Ubuntu?

14 Answers 14

141

Install the package pdfgrep, then use the command:

find /path -iname '*.pdf' -exec pdfgrep pattern {} +

——————

Simpliest way is

pdfgrep 'pattern' *.pdf
pdfgrep 'pattern' file.pdf 
  • 5
    This works in mac osx (Mavericks) as well. Install it using brew. Simple. Thanks. – mikiemorales Jan 23 '14 at 1:28
  • 7
    Out of curiosity I checked the source of pdfgrep and it uses poppler to extract strings from the pdf. Almost exactly as @wag's answer only pagewise rather than, presumably, the entire document. – Andrew Martin Sep 16 '14 at 11:11
  • 4
    pdfgrep also has a recursive flag. So this answer could perhaps be reduced to: pdfgrep -R pattern /path/. Though it might be less effective if it goes through every file even if it isn't a PDF. And I notice that it has issues with international characters such as å, ä and ö. – Rovanion Jan 14 '16 at 12:11
  • 1
    Actually, the -n option is a pro for pdfgrep as it allows to include the page number in the output (might be helpful for further processing). – JepZ Nov 10 '17 at 20:18
  • 4
    This answer would be easier to use if it explained which bits of the command are meant to copied literally and which are placeholders. What's pattern? What's {}? What's up with the ` +`? I have no idea upon first reading... so off to the manpage I go, I suppose. – Mark Amery Apr 20 '18 at 14:44
57

If you have poppler-utils installed (default on Ubuntu Desktop), you could "convert" it on the fly and pipe it to grep:

pdftotext my.pdf - | grep 'pattern'

This won't create a .txt file.

  • 1
    so .. you extract the text before you grep it which means the answer is "no". – akira Jan 31 '11 at 15:18
  • 18
    @akira The OP probably meant "without opening the PDF in a viewer and exporting to text" – Michael Mrozek Jan 31 '11 at 17:36
  • 5
    @akira Where do you see "grep only"? – Michael Mrozek Jan 31 '11 at 18:55
  • 6
    @akira Well, I already said what I think he probably meant; he doesn't want to export to text before processing it. I very much doubt he has a problem with any command that converts to text in any way; there's no reason not to – Michael Mrozek Feb 1 '11 at 5:52
  • 2
    @sherrellbc The second argument of pdftotext is the filename it should write to. However, by convention, tools typically allow you to write to stdout instead of to a file by specifying a - instead. Similarly, some tools would write to stdout by default if you omit such an argument entirely (but this is not always possible without creating ambiguity). – Joost Sep 23 '16 at 14:06
12

pdfgrep was written for exactly this purpose and is available in Ubuntu.

It tries to be mostly compatible to grep and thus provides "the power of grep", only specialized for PDFs. That includes common grep options, such as --recursive, --ignore-case or --color.

In contrast to pdftotext | grep, pdfgrep can output the page number of a match in a performant way and is generally faster when it doesn't have to search the whole document (e.g. --max-count or --quiet).

The basic usage is:

pdfgrep PATTERN FILE..

where PATTERN is your search string and FILE a list of filenames (or wildcards in a shell).

See the manpage for more infos.

7

No.

A pdf consists of chunks of data, some of them text, some of them pictures and some of them really magical fancy XYZ (eg. .u3d files). Those chunks are most of the times compressed (eg. flat, check http://www.verypdf.com/pdfinfoeditor/compression.htm). In order to 'grep' a .pdf you have to reverse the compression aka extract the text.

You can do that either per file with tools such as pdf2text and grep the result, or you run an 'indexer' (look at xapian.org or lucene) which builds an searchable index out of your .pdf files and then you can use the search engine tools of that indexer to get the content of the pdf.

But no, you can not grep pdf files and hope for reliable answers without extracting the text first.

  • 5
    Considering pdfgrep exists (see above), a flat "no" is incorrect. – Jonathan Cross Aug 28 '18 at 10:18
6

Recoll can search PDFs. It doesn't support regular expressions, but it has lots of other search options, so it might fit your needs.

5

You could pipe it through strings first:-

cat file.pdf | strings | grep <...etc...>
  • 8
    Just use strings file.pdf | grep <...>, you don't need cat – phunehehe Jan 31 '11 at 14:31
  • Yeah - my mind seems to work better with streams... :-) – Andy Smith Jan 31 '11 at 14:57
  • 12
    wont work if text is compressed, which it is most of the times. – akira Jan 31 '11 at 15:18
  • 6
    Even if the text is uncompressed, it's generally small pieces of sentences (not even necessarily whole words!) finely intermixed with formatting information. Not very friendly for strings or grep. – Jander Jan 31 '11 at 16:08
  • Can you think of another reason why using strings for this wouldn't work? I found that using strings works on some PDFs but not others. – hourback Nov 24 '15 at 19:58
3

Take a look at the common resource grep tool crgrep which supports searching within PDF files.

It also allows searching other resources like content nested in archives, database tables, image meta-data, POM file dependencies and web resources - and combinations of these including recursive search.

2

try this

find /path -iname *.pdf -print0 | for i in `xargs 0`; do echo $i; \
    pdftotext "$i" - | grep pattern; done

for printing the lines the pattern occurs inside the pdf

2

cd to your folder containing your pdf-file and then..

pdfgrep 'pattern' your.pdf

or if you want to search in more than just one pdf-file (e.g. in all pdf-files in your folder)

pdfgrep 'pattern'  `ls *.pdf`

or

pdfgrep 'pattern' $(ls *.pdf)
1

There is a duplicate question on StackOverflow. The people there suggest a variation of harish.venkarts answer:

find /path -name '*.pdf' -exec sh -c 'pdftotext "{}" - | grep --with-filename --label="{}" --color "your pattern"' \;

The advantage over the similar answer here is the --with-filename flag for grep. This is somewhat superior to pdfgrep as well, because the standard grep has more features.

https://stackoverflow.com/questions/4643438/how-to-search-contents-of-multiple-pdf-files

  • I think it would have been better to leave this as a comment (or edit) in the similar answer you are referring to. – Bernhard May 9 '14 at 12:07
0

Here is a quick script for search pdf in the current directory :

#!/bin/bash

if [ $# -ne 1 ]; then
  echo "usage $0 VALUE" 1>&2
  exit 1
fi

echo 'SEARCH IS CASE SENSITIVE' 1>&2

find . -name '*.pdf' -exec /bin/bash -c 'pdftotext "{}" - | grep --with-filename --label="{}" --color "$0"' $1 \;
0

I assume you mean tp not convert it on the disk, you can convert them to stdout and then grep it with pdftotext. Grepping the pdf without any sort of conversion is not a practical approach since PDF is mostly a binary format.

In the directory:

ls -1 ./*.pdf | xargs -L1 -I {} pdftotext {}  - | grep "keyword"

or in the directory and its subdirectories:

tree -fai . | grep -P ".pdf$" | xargs -L1 -I {} pdftotext {}  - | grep "keyword"

Also because some pdf are scans they need to be OCRed first. I wrote a pretty simple way to search all pdfs that cannot be greped and OCR them.

I noticed if a pdf file doesn't have any font it is usually not searchable. So knowing this we can use pdffonts.

First 2 lines of the pdffonts are the table header, so when a file is searchable has more than two line output, knowing this we can create:

gedit check_pdf_searchable.sh

then paste this

#!/bin/bash 
#set -vx
if ((`pdffonts "$1" | wc -l` < 3 )); then
echo $1
pypdfocr "$1"
fi

then make it executable

chmod +x check_pdf_searchable.sh

then list all non-searchable pdfs in the directory:

ls -1 ./*.pdf | xargs -L1 -I {} ./check_pdf_searchable.sh {}

or in the directory and its subdirectories:

tree -fai . | grep -P ".pdf$" | xargs -L1 -I {} ./check_pdf_searchable.sh {}
0

If you just want to search for pdf names/properties... or simple strings that are not compressed or encoded then instead of strings you can use the below

grep -a STRING file.pdf
cat -v file.pdf | grep STRING

From grep --help:

      --binary-files=TYPE   assume that binary files are TYPE;
                            TYPE is 'binary', 'text', or 'without-match'
  -a, --text                equivalent to --binary-files=text

and cat --help:

  -v, --show-nonprinting   use ^ and M- notation, except for LFD and TAB
0

gpdf might be what you need if you're using Gnome! Check this in case you're not using Gnome. It's got a list of CLI pdf viewers. Then you can use grep to find some pattern.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.