catdoc.1

harvest是一个下载html网页得机器人

.TH catdoc 1  "Version 0.91" "MS-Word reader"
.SH NAME
catdoc \- reads MS-Word file and puts its content as plain text on standard output
.SH SYNOPSIS

.BR catdoc " [" -vlu8btawx "] [" -m " 
.IR number ] 
[
.B -s
.IR charset ] 
[
.B -d 
.IR charset ] 
[ 
.B -f
.IR output-format ]
.I file

.SH DESCRIPTION

.B catdoc
behaves much like
.BR cat (1)
but it reads MS-Word file and produces human-readable text on standard output.
Optionally it can use 
.BR latex (1)
escape sequenses for characters which have special meaning for LaTeX.
It also makes some effort to recognize MS-Word tables, although it never
tries to write correct headers for LaTeX tabular environment. Additional
output formats, such is HTML can be easily defined. 
.PP
.B catdoc
doesn't attempt to extract formatting information other than tables from
MS-Word document, so different output modes means mainly that different
charachers should be escaped and different ways used to represent characters,
missing from output charset. See CHARACTER SUBSTITUTION below 

.PP
.B catdoc
uses internal 
.BR unicode (4)
representation of text, so it is able to convert texts when charset in
source document doesn't match charset on target system.
See CHARACTER SETS below.
.PP
If no file names supplied, 
.B catdoc
processes its standard input unless it is terminal. It is unlikely that 
somebody could type Word document from keyboard, so if 
.B catdoc 
invoked without arguments and stdin is not redirected, it prints brief
usage message and exits. 
Processing of standard input (even among other files) can be forced using
dash '-' as file name.
.PP
By default, 
.B catdoc
wraps lines which are more than 72 chars long and separates paragraphs by
blank lines. This behavoir can be turned of by 
.B -w
switch. In 
.I wide
mode 
.B  catdoc prints each paragraph as one long line, suitable for import into
word processors which perform word wrapping theirselves.
 

.SH OPTIONS
.TP 8
.B -a 
- shortcut for -f ascii. Produces ASCII text as output.
Separates table columns with TAB
.TP 8
.B -b
- process broken MS-Word file. Normally,
.B catdoc checks if first 8 bytes
of file is Microsoft OLE signature. If so, it processes file, otherwise
it just copies it to stdin. It is intended to use 
.B catdoc 
as filter for viewing all files with 
.I .doc
extension.
.TP 8
.BI -d charset
- specifies destination charset name. Charset file has format described in
CHARACTER SETS below and should have 
.B .txt
extension  and reside in 
.B catdoc library directory (normally /usr/local/lib/catdoc).
.TP 8
.BI -f format
- specifies output format as described in CHARACTER SUBSTITUTION below.
.B catdoc
comes with two output formats - ascii and tex. You can add your own if you
wish.
.TP 8
.B  -l
Causes 
.B catdoc
to list names of available charsets to the stdout and exit successfully.
.TP 8
.BI -m number
Specifies right margin for text  (default 72). 
.B -m 0
is equivalent to
.B -w
.TP 8
.BI -s charset
Specifies source charset. (one used in Word document), if Word document
doesn't contain UTF-16  text.
.TP 8
.B -t
- shortcut for 
.B -f tex
 converts all printable chars, which have special meaning for 
.BR LaTeX (1)
into appropriate control sequences. Separates table columns by 
.BR &.
.TP 8
.B -u
- declares that Word  document  contain  UNICODE   (UTF-16) represntation
of text (as some Word-97 documents). If catdoc fails to correct  Word document
with  default charset,   try    this  option.
.TP 8
.B -8
- declares is Word document is 8 bit. Just in case that catdoc
 recognizes file format incorrectly.
.TP 8
.B -w
disables word wrapping. By default 
.B catdoc
output is splitted into lines not longer than 72 (or  number, specified by
-m  option)   characters and paragraphs
are separated by blank line. With this option each paragraph is one
long line. 
.TP 8
.B -x
causes catdoc to output unknown UNICODE characher as \\xNNNN, instead
of question marks.
.TP 8
.B -v
causes catdoc to print some useless information about word document
structure to stdout before actual start of text.

.SH CHARACTER SETS
When processing MS-Word file 
.B catdoc
uses information about two character sets, typically different
 -  input and output. They are stored in plain text files in 
.B catdoc
library directory. Character set files should contain two whitespace-separated
hexadecimal numbers - 8-bit code in character set and 16-bit unicode code.
Anything from hash mark to end of line is ignored, as well as blank lines.

.B catdoc 
distribution includes some of these character sets. Additional character set
definitions, directly usable by 
.B catdoc 
can be obtained from ftp.unicode.org. Charset files have
.B .txt
suffix, which shouldn't be specified in command-line or configuration
files.  
.PP
Note that
.B catdoc 
is distributed with Cyrillic charsets as default. If you are not
Russian, you probably don't want it, an should reconfigure catdoc at 
compile time or in runtime configuration file.
.PP
When dealing with documents with charsets other than default, remember
that Microsoft never uses ISO charsets. While letters in, say cp1252 are
at the same position as in ISO-8859-1, some punctuation signs would be
lost, if you specify ISO-8859-1 as input charset. If you use cp1252,
catdoc would deal with those signs as described in CHARACTER
SUBSTITUTION below.

.SH CHARACTER SUBSTITUTION 
.B catdoc
converts  MS-Word file into following internal unicode representation:
.TP 4
1. Paragraphs are separated by ASCII Line Feed symbol (0x000A)
.TP 4
2. Table cells within row are separated by ASCII Field Separator symbol
(0x001C)
.TP 4
3. Table rows are separated by ASCII Record Separator (0x001E) 
.TP 4
4. All printable characters, including whitespace are represented with their
respective UNICODE codes.
.PP 
This UNICODE representation is subsequentely converted into 8-bit text in
target character set using following four-step algorithm:
.TP 4
1. List of special characters is searched for given unicode character.
If found, then appropriate multi-character sequence is output instead of
character. 
.TP 4
2. If there is an equivalent in target character set, it is output.
.TP 4
3. Otherwise, replacement list is searched and, if there is multi-character
substitution for this UNICODE char, it is output.
.TP 4
4. If all above fails, "Unknown char" symbol (question mark) is output.
.PP
Lists of special characters and list of substitution are character
set-independent, becouse special chars should be escaped regardless of their
existense in target character set  (usially, they are parts of US-ASCII, and
therefore exist in any character set) and replacement list is searched only
for those characters, which are not found in target character set.
.PP
These lists are stored in
.B catdoc 
library directory in files with prefix of format name. These files have
following format:
.PP
Each line can be either comment (starting with hash mark) or contain
hexadecimal UNICODE value, separated by whitespace from string, which
would be substituted instead of it. If string contain no whitespace it 
can be used as is, otherwise it should be enclosed in single or double
quotes. Usial backslash sequences like 
.IR '\en' , '\et'
can be used in these string.


.SH RUNTIME CONFIGURATION
Upon startup catdoc reads its system-wide configuration file (
.B catdocrc in 
.B catdoc
library directory) and then
user-specific configuration file
.BR ${HOME}/.catdocrc.
.PP
These files can contain following directives:
.TP 8
.BI "source_charset = " charset-name
Sets default source charset, which would be used if no 
.B -s
option specified. Consult configuration of nearby windows
workstation to find one you need.
.TP 8
.BI "target_charset = "  charset-name
 Sets default output charset. You probably know, which one you use.
.TP 8
.BI "charset_path = "  directory-list
colon-separated list of directories, which are searched for charset files.
This allows you to install additional charsets in your home directory.
.TP 8
.BI "map_path = " directory-list
colon-separated list of directories, which are searched for special character
map and replacement map.
.TP 8
.BI "format = " "format name"
Output format which would be used by default.
.B catdoc
comes with two formats - 
.BR ascii " and " tex
but nothing prevents you from writing your own format (set two map files -
special character map and replacement map).
.TP 8
.BI "unknown_char = " "character specification"
sets characher to output instead of unknown unicode character (default '?')
Character specification can have one of two form - character enclosed in
single quotes or hexadecimal code.

.SH BUGS

Doesn't handle
fast-saves properly. Prints footnotes as separate paragraphs at the end of
file, instead of producing correct LaTeX commands. Cannot distinguish
between empty table cell and end of table row.



.SH "SEE ALSO"

.BR xls2csv (1),
.BR cat (1),
.BR strings (1),
.BR utf (4),
.BR unicode (4)

.SH AUTHOR

V.B.Wagner <[e-mail]>