diff options
Diffstat (limited to '')
| -rw-r--r-- | cpukit/zlib/doc/txtvsbin.txt | 107 |
1 files changed, 0 insertions, 107 deletions
diff --git a/cpukit/zlib/doc/txtvsbin.txt b/cpukit/zlib/doc/txtvsbin.txt deleted file mode 100644 index 3d0f0634f7..0000000000 --- a/cpukit/zlib/doc/txtvsbin.txt +++ /dev/null @@ -1,107 +0,0 @@ -A Fast Method for Identifying Plain Text Files -============================================== - - -Introduction ------------- - -Given a file coming from an unknown source, it is sometimes desirable -to find out whether the format of that file is plain text. Although -this may appear like a simple task, a fully accurate detection of the -file type requires heavy-duty semantic analysis on the file contents. -It is, however, possible to obtain satisfactory results by employing -various heuristics. - -Previous versions of PKZip and other zip-compatible compression tools -were using a crude detection scheme: if more than 80% (4/5) of the bytes -found in a certain buffer are within the range [7..127], the file is -labeled as plain text, otherwise it is labeled as binary. A prominent -limitation of this scheme is the restriction to Latin-based alphabets. -Other alphabets, like Greek, Cyrillic or Asian, make extensive use of -the bytes within the range [128..255], and texts using these alphabets -are most often misidentified by this scheme; in other words, the rate -of false negatives is sometimes too high, which means that the recall -is low. Another weakness of this scheme is a reduced precision, due to -the false positives that may occur when binary files containing large -amounts of textual characters are misidentified as plain text. - -In this article we propose a new, simple detection scheme that features -a much increased precision and a near-100% recall. This scheme is -designed to work on ASCII, Unicode and other ASCII-derived alphabets, -and it handles single-byte encodings (ISO-8859, MacRoman, KOI8, etc.) -and variable-sized encodings (ISO-2022, UTF-8, etc.). Wider encodings -(UCS-2/UTF-16 and UCS-4/UTF-32) are not handled, however. - - -The Algorithm -------------- - -The algorithm works by dividing the set of bytecodes [0..255] into three -categories: -- The white list of textual bytecodes: - 9 (TAB), 10 (LF), 13 (CR), 32 (SPACE) to 255. -- The gray list of tolerated bytecodes: - 7 (BEL), 8 (BS), 11 (VT), 12 (FF), 26 (SUB), 27 (ESC). -- The black list of undesired, non-textual bytecodes: - 0 (NUL) to 6, 14 to 31. - -If a file contains at least one byte that belongs to the white list and -no byte that belongs to the black list, then the file is categorized as -plain text; otherwise, it is categorized as binary. (The boundary case, -when the file is empty, automatically falls into the latter category.) - - -Rationale ---------- - -The idea behind this algorithm relies on two observations. - -The first observation is that, although the full range of 7-bit codes -[0..127] is properly specified by the ASCII standard, most control -characters in the range [0..31] are not used in practice. The only -widely-used, almost universally-portable control codes are 9 (TAB), -10 (LF) and 13 (CR). There are a few more control codes that are -recognized on a reduced range of platforms and text viewers/editors: -7 (BEL), 8 (BS), 11 (VT), 12 (FF), 26 (SUB) and 27 (ESC); but these -codes are rarely (if ever) used alone, without being accompanied by -some printable text. Even the newer, portable text formats such as -XML avoid using control characters outside the list mentioned here. - -The second observation is that most of the binary files tend to contain -control characters, especially 0 (NUL). Even though the older text -detection schemes observe the presence of non-ASCII codes from the range -[128..255], the precision rarely has to suffer if this upper range is -labeled as textual, because the files that are genuinely binary tend to -contain both control characters and codes from the upper range. On the -other hand, the upper range needs to be labeled as textual, because it -is used by virtually all ASCII extensions. In particular, this range is -used for encoding non-Latin scripts. - -Since there is no counting involved, other than simply observing the -presence or the absence of some byte values, the algorithm produces -consistent results, regardless what alphabet encoding is being used. -(If counting were involved, it could be possible to obtain different -results on a text encoded, say, using ISO-8859-16 versus UTF-8.) - -There is an extra category of plain text files that are "polluted" with -one or more black-listed codes, either by mistake or by peculiar design -considerations. In such cases, a scheme that tolerates a small fraction -of black-listed codes would provide an increased recall (i.e. more true -positives). This, however, incurs a reduced precision overall, since -false positives are more likely to appear in binary files that contain -large chunks of textual data. Furthermore, "polluted" plain text should -be regarded as binary by general-purpose text detection schemes, because -general-purpose text processing algorithms might not be applicable. -Under this premise, it is safe to say that our detection method provides -a near-100% recall. - -Experiments have been run on many files coming from various platforms -and applications. We tried plain text files, system logs, source code, -formatted office documents, compiled object code, etc. The results -confirm the optimistic assumptions about the capabilities of this -algorithm. - - --- -Cosmin Truta -Last updated: 2006-May-28 |