|
kusano |
fc6ab3 |
A Fast Method for Identifying Plain Text Files
|
|
kusano |
fc6ab3 |
==============================================
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
Introduction
|
|
kusano |
fc6ab3 |
------------
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
Given a file coming from an unknown source, it is sometimes desirable
|
|
kusano |
fc6ab3 |
to find out whether the format of that file is plain text. Although
|
|
kusano |
fc6ab3 |
this may appear like a simple task, a fully accurate detection of the
|
|
kusano |
fc6ab3 |
file type requires heavy-duty semantic analysis on the file contents.
|
|
kusano |
fc6ab3 |
It is, however, possible to obtain satisfactory results by employing
|
|
kusano |
fc6ab3 |
various heuristics.
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
Previous versions of PKZip and other zip-compatible compression tools
|
|
kusano |
fc6ab3 |
were using a crude detection scheme: if more than 80% (4/5) of the bytes
|
|
kusano |
fc6ab3 |
found in a certain buffer are within the range [7..127], the file is
|
|
kusano |
fc6ab3 |
labeled as plain text, otherwise it is labeled as binary. A prominent
|
|
kusano |
fc6ab3 |
limitation of this scheme is the restriction to Latin-based alphabets.
|
|
kusano |
fc6ab3 |
Other alphabets, like Greek, Cyrillic or Asian, make extensive use of
|
|
kusano |
fc6ab3 |
the bytes within the range [128..255], and texts using these alphabets
|
|
kusano |
fc6ab3 |
are most often misidentified by this scheme; in other words, the rate
|
|
kusano |
fc6ab3 |
of false negatives is sometimes too high, which means that the recall
|
|
kusano |
fc6ab3 |
is low. Another weakness of this scheme is a reduced precision, due to
|
|
kusano |
fc6ab3 |
the false positives that may occur when binary files containing large
|
|
kusano |
fc6ab3 |
amounts of textual characters are misidentified as plain text.
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
In this article we propose a new, simple detection scheme that features
|
|
kusano |
fc6ab3 |
a much increased precision and a near-100% recall. This scheme is
|
|
kusano |
fc6ab3 |
designed to work on ASCII, Unicode and other ASCII-derived alphabets,
|
|
kusano |
fc6ab3 |
and it handles single-byte encodings (ISO-8859, MacRoman, KOI8, etc.)
|
|
kusano |
fc6ab3 |
and variable-sized encodings (ISO-2022, UTF-8, etc.). Wider encodings
|
|
kusano |
fc6ab3 |
(UCS-2/UTF-16 and UCS-4/UTF-32) are not handled, however.
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
The Algorithm
|
|
kusano |
fc6ab3 |
-------------
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
The algorithm works by dividing the set of bytecodes [0..255] into three
|
|
kusano |
fc6ab3 |
categories:
|
|
kusano |
fc6ab3 |
- The white list of textual bytecodes:
|
|
kusano |
fc6ab3 |
9 (TAB), 10 (LF), 13 (CR), 32 (SPACE) to 255.
|
|
kusano |
fc6ab3 |
- The gray list of tolerated bytecodes:
|
|
kusano |
fc6ab3 |
7 (BEL), 8 (BS), 11 (VT), 12 (FF), 26 (SUB), 27 (ESC).
|
|
kusano |
fc6ab3 |
- The black list of undesired, non-textual bytecodes:
|
|
kusano |
fc6ab3 |
0 (NUL) to 6, 14 to 31.
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
If a file contains at least one byte that belongs to the white list and
|
|
kusano |
fc6ab3 |
no byte that belongs to the black list, then the file is categorized as
|
|
kusano |
fc6ab3 |
plain text; otherwise, it is categorized as binary. (The boundary case,
|
|
kusano |
fc6ab3 |
when the file is empty, automatically falls into the latter category.)
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
Rationale
|
|
kusano |
fc6ab3 |
---------
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
The idea behind this algorithm relies on two observations.
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
The first observation is that, although the full range of 7-bit codes
|
|
kusano |
fc6ab3 |
[0..127] is properly specified by the ASCII standard, most control
|
|
kusano |
fc6ab3 |
characters in the range [0..31] are not used in practice. The only
|
|
kusano |
fc6ab3 |
widely-used, almost universally-portable control codes are 9 (TAB),
|
|
kusano |
fc6ab3 |
10 (LF) and 13 (CR). There are a few more control codes that are
|
|
kusano |
fc6ab3 |
recognized on a reduced range of platforms and text viewers/editors:
|
|
kusano |
fc6ab3 |
7 (BEL), 8 (BS), 11 (VT), 12 (FF), 26 (SUB) and 27 (ESC); but these
|
|
kusano |
fc6ab3 |
codes are rarely (if ever) used alone, without being accompanied by
|
|
kusano |
fc6ab3 |
some printable text. Even the newer, portable text formats such as
|
|
kusano |
fc6ab3 |
XML avoid using control characters outside the list mentioned here.
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
The second observation is that most of the binary files tend to contain
|
|
kusano |
fc6ab3 |
control characters, especially 0 (NUL). Even though the older text
|
|
kusano |
fc6ab3 |
detection schemes observe the presence of non-ASCII codes from the range
|
|
kusano |
fc6ab3 |
[128..255], the precision rarely has to suffer if this upper range is
|
|
kusano |
fc6ab3 |
labeled as textual, because the files that are genuinely binary tend to
|
|
kusano |
fc6ab3 |
contain both control characters and codes from the upper range. On the
|
|
kusano |
fc6ab3 |
other hand, the upper range needs to be labeled as textual, because it
|
|
kusano |
fc6ab3 |
is used by virtually all ASCII extensions. In particular, this range is
|
|
kusano |
fc6ab3 |
used for encoding non-Latin scripts.
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
Since there is no counting involved, other than simply observing the
|
|
kusano |
fc6ab3 |
presence or the absence of some byte values, the algorithm produces
|
|
kusano |
fc6ab3 |
consistent results, regardless what alphabet encoding is being used.
|
|
kusano |
fc6ab3 |
(If counting were involved, it could be possible to obtain different
|
|
kusano |
fc6ab3 |
results on a text encoded, say, using ISO-8859-16 versus UTF-8.)
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
There is an extra category of plain text files that are "polluted" with
|
|
kusano |
fc6ab3 |
one or more black-listed codes, either by mistake or by peculiar design
|
|
kusano |
fc6ab3 |
considerations. In such cases, a scheme that tolerates a small fraction
|
|
kusano |
fc6ab3 |
of black-listed codes would provide an increased recall (i.e. more true
|
|
kusano |
fc6ab3 |
positives). This, however, incurs a reduced precision overall, since
|
|
kusano |
fc6ab3 |
false positives are more likely to appear in binary files that contain
|
|
kusano |
fc6ab3 |
large chunks of textual data. Furthermore, "polluted" plain text should
|
|
kusano |
fc6ab3 |
be regarded as binary by general-purpose text detection schemes, because
|
|
kusano |
fc6ab3 |
general-purpose text processing algorithms might not be applicable.
|
|
kusano |
fc6ab3 |
Under this premise, it is safe to say that our detection method provides
|
|
kusano |
fc6ab3 |
a near-100% recall.
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
Experiments have been run on many files coming from various platforms
|
|
kusano |
fc6ab3 |
and applications. We tried plain text files, system logs, source code,
|
|
kusano |
fc6ab3 |
formatted office documents, compiled object code, etc. The results
|
|
kusano |
fc6ab3 |
confirm the optimistic assumptions about the capabilities of this
|
|
kusano |
fc6ab3 |
algorithm.
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
|
|
kusano |
fc6ab3 |
--
|
|
kusano |
fc6ab3 |
Cosmin Truta
|
|
kusano |
fc6ab3 |
Last updated: 2006-May-28
|