Pdfminer Docs
Pdfminer Docs
Release 0.0.1
Levia3
1 PDFMiner 1
1.1 What’s It? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 How to Install . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Command Line Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.5 TODO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.6 Related Projects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.7 Terms and Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
i
ii
CHAPTER 1
PDFMiner
PDFMiner is a tool for extracting information from PDF documents. Unlike other PDF-related tools, it focuses entirely
on getting and analyzing text data. PDFMiner allows one to obtain the exact location of text in a page, as well as other
information such as fonts or lines. It includes a PDF converter that can transform PDF files into other text formats
(such as HTML). It has an extensible PDF parser that can be used for other purposes than text analysis.
1.1.1 Features
1
pdfminer, Release 0.0.1
1.1.2 Download
Source distribution:
http://pypi.python.org/pypi/pdfminer/
github:
https://github.com/euske/pdfminer/
$ pdf2txt.py samples/simple1.pdf
Hello
World
Hello
World
H e l l o
W o r l d
H e l l o
W o r l d
6. Done!
2 Chapter 1. PDFMiner
pdfminer, Release 0.0.1
In order to process CJK languages, you need an additional step to take during installation:
# make cmap
python tools/conv_cmap.py pdfminer/cmap Adobe-CNS1 cmaprsrc/cid2code_Adobe_
˓→CNS1.txt
reading 'cmaprsrc/cid2code_Adobe_CNS1.txt'...
writing 'CNS1_H.py'...
...
(this may take several minutes)
On Windows machines which don’t have make command, paste the following commands on a command line prompt:
mkdir pdfminer\cmap
python tools\conv_cmap.py -c B5=cp950 -c UniCNS-UTF8=utf-8 pdfminer\cmap
˓→Adobe-CNS1 cmaprsrc\cid2code_Adobe_CNS1.txt
˓→Adobe_Korea1.txt
1.3.1 pdf2txt.py
pdf2txt.py extracts text contents from a PDF file. It extracts all the text that are to be rendered programmatically,
i.e. text represented as ASCII or Unicode strings. It cannot recognize text drawn as images that would require optical
character recognition. It also extracts the corresponding locations, font names, font sizes, writing direction (horizontal
or vertical) for each text portion. You need to provide a password for protected PDF documents when its access is
restricted. You cannot extract any text from a PDF document which does not have extraction permission.
Note: Not all characters in a PDF can be safely converted to Unicode.
Examples
Options
-o filename
Specifies the output file name. By default, it prints the extracted contents to stdout in text format.
-p pageno[,pageno,...]
Specifies the comma-separated list of the page numbers to be extracted. Page numbers start at one. By
default, it extracts text from all the pages.
-c codec
Specifies the output codec.
-t type
Specifies the output format. The following formats are currently supported.
• text : TEXT format. (Default)
• html : HTML format. Not recommended for extraction purposes because the markup is messy.
• xml : XML format. Provides the most information.
• tag : “Tagged PDF” format. A tagged PDF has its own contents annotated with HTML-like tags.
pdf2txt tries to extract its content streams rather than inferring its text locations. Tags used here are
defined in the PDF specification (See §10.7 “Tagged PDF”).
-I image_directory
Specifies the output directory for image extraction. Currently only JPEG images are supported.
-M char_margin
-L line_margin
-W word_margin
These are the parameters used for layout analysis. In an actual PDF file, text portions might be split
into several chunks in the middle of its running, depending on the authoring software. Therefore, text
extraction needs to splice text chunks. In the figure below, two text chunks whose distance is closer
than the char_margin (shown as M) is considered continuous and get grouped into one. Also, two lines
whose distance is closer than the line_margin (L) is grouped as a text box, which is a rectangular area that
contains a “cluster” of text portions. Furthermore, it may be required to insert blank characters (spaces)
as necessary if the distance between two words is greater than the word_margin (W), as a blank between
words might not be represented as a space, but indicated by the positioning of each word.
Each value is specified not as an actual length, but as a proportion of the length to the size of each character
in question. The default values are M = 2.0, L = 0.5, and W = 0.1, respectively.
4 Chapter 1. PDFMiner
pdfminer, Release 0.0.1
-F boxes_flow
Specifies how much a horizontal and vertical position of a text matters when determining a text order. The
value should be within the range of -1.0 (only horizontal position matters) to +1.0 (only vertical position
matters). The default value is 0.5.
-C
Suppress object caching. This will reduce the memory consumption but also slows down the process.
-n
Suppress layout analysis.
-A
Forces to perform layout analysis for all the text strings, including text contained in figures.
-V
Allows vertical writing detection.
-Y layout_mode
Specifies how the page layout should be preserved. (Currently only applies to HTML format.)
• exact : preserve the exact location of each individual character (a large and messy HTML).
• normal : preserve the location and line breaks in each text block. (Default)
• loose : preserve the overall location of each text block.
-E extractdir
Specifies the extraction directory of embedded files.
-s scale
Specifies the output scale. Can be used in HTML format only.
-m maxpages
Specifies the maximum number of pages to extract. By default, it extracts all the pages in a document.
-P password
Provides the user password to access PDF contents.
-d
Increases the debug level.
1.3.2 dumppdf.py
dumppdf.py dumps the internal contents of a PDF file in pseudo-XML format. This program is primarily for
debugging purposes, but it’s also possible to extract some meaningful contents (such as images).
Examples
$ dumppdf.py -a foo.pdf
(dump all the headers and contents, except stream objects)
$ dumppdf.py -T foo.pdf
(dump the table of contents)
Options
-a
Instructs to dump all the objects. By default, it only prints the document trailer (like a header).
-i objno,objno, ...
Specifies PDF object IDs to display. Comma-separated IDs, or multiple -i options are accepted.
-p pageno,pageno, ...
Specifies the page number to be extracted. Comma-separated page numbers, or multiple -p options are
accepted. Note that page numbers start at one, not zero.
-r (raw)
-b (binary)
-t (text)
Specifies the output format of stream contents. Because the contents of stream objects can be very large,
they are omitted when none of the options above is specified.
With -r option, the “raw” stream contents are dumped without decompression. With -b option, the
decompressed contents are dumped as a binary blob. With -t option, the decompressed contents are
dumped in a text format, similar to repr() manner. When -r or -b option is given, no stream header
is displayed for the ease of saving it to a file.
-T
Shows the table of contents.
-E directory
Extracts embedded files from the pdf into the given directory.
-P password
6 Chapter 1. PDFMiner
pdfminer, Release 0.0.1
1.4 Changes
1.4. Changes 7
pdfminer, Release 0.0.1
8 Chapter 1. PDFMiner
pdfminer, Release 0.0.1
1.5 TODO
• pyPdf
• xpdf
• pdfbox
• mupdf
1.5. TODO 9
pdfminer, Release 0.0.1
10 Chapter 1. PDFMiner
CHAPTER 2
This page explains how to use PDFMiner as a library from other applications.
2.1 Overview
PDF is evil. Although it is called a PDF “document”, it’s nothing like Word or HTML document. PDF is more like
a graphic representation. PDF contents are just a bunch of instructions that tell how to place the stuff at each exact
position on a display or paper. In most cases, it has no logical structure such as sentences or paragraphs and it cannot
adapt itself when the paper size changes. PDFMiner attempts to reconstruct some of those structures by guessing from
its positioning, but there’s nothing guaranteed to work. Ugly, I know. Again, PDF is evil.
[More technical details about the internal structure of PDF: “How to Extract Text Contents from PDF Manually”
(part1) (part2) (part3)]
Because a PDF file has such a big and complex structure, parsing a PDF file as a whole is time and memory consuming.
However, not every part is needed for most PDF processing tasks. Therefore PDFMiner takes a strategy of lazy
parsing, which is to parse the stuff only when it’s necessary. To parse PDF files, you need to use at least two classes:
PDFParser and PDFDocument. These two objects are associated with each other. PDFParser fetches data from
a file, and PDFDocument stores it. You’ll also need PDFPageInterpreter to process the page contents and
PDFDevice to translate it to whatever you need. PDFResourceManager is used to store shared resources such
as fonts or images.
Figure 1 shows the relationship between the classes in PDFMiner.
11
pdfminer, Release 0.0.1
A layout analyzer returns a LTPage object for each page in the PDF document. This object contains child objects
within the page, forming a tree structure. Figure 2 shows the relationship between these objects.
Some PDF documents use page numbers as destinations, while others use page numbers and the physical location
within the page. Since PDF does not have a logical structure, and it does not provide a way to refer to any in-page
object from the outside, there’s no way to tell exactly which part of text these destinations are referring to.
You can extend PDFPageInterpreter and PDFDevice class in order to process them differently / obtain other
information.
Yusuke Shinyama