On 21-Oct-05 Marco Venanzi wrote:
Hi, I'm trying to read data from a PDF file.Is it possible to do it
with R? Thanks, Marco
Basically, No.
But you may be lucky with "copy&paste" using the mouse, from
the display generated in Acrobat Reader to a text file.
The basic procedure here is
1. Click on the "Text Select Tool" (a button usually marked with a "T");
2. Use the mouse to highlight the block of text you want to copy;
3. Depending on your operating system/graphics display: In Windows
you have (IIRC) to go to "Edit"->""Copy"; in Unlix/Linux with
X Windows do nothing;
4. "Paste" it into your text file, again as appropriate for your
operating system.
However, you may not be lucky.
PDF can store its content in stange ways, and what may look on the
screen like contiguous and consecutive text is stored internally
in separate "blocks" (what PDF calls "objects"). And this can apply
even to little bits of text in a paragraph.
When you paste the marked text, it will go in in the order that
PDF finds the blocks in the file. As a result, your text file
may contain bits of text in random order.
This especially applies to things arranged in tables. But it
very much depends on the software that generated the PDF in
the first place.
Since often the data in a PDF file which you may want to copy
in this way will be tabular, you are likely to encounter this
problem!
You can tell this is going to happen when you use the mouse to
highlight the text you intend to copy: starting with the mouse
iin say the top LH corner, move it slowly towards the lower
RH corner of the block. If the highlighting jumps all over the
screen, and/or outside the area you are trying to highlight,
then this is what's happening.
In that case I have sometimes done it by copying lots of little
blocks, too small to provoke the effect. But this is very tedious.
There are other things one can try, such as printing from the
PDF file to a PostScript file, and then using a program like
ps2ascii (which can deal directly with PDF) or pstotext; but frankly
no such program is likely to make a good job of this, because of
the way PS and PDF work.
Sorry to appear unhelpful! But you may get somewhere.