magic

harvest是一个下载html网页得机器人

#! file
# Magic data for file(1) command.
# Machine-genererated from src/cmd/file/magdir/*; edit there only!
# Format is described in magic(files), where:
# files is 4 on V7 and BSD, 4 on SV, and ?? in the SVID.
22	short	023000		core dump data
0	short		070707		cpio archive
0	string		070707		cpio archive
0	long		0177555		very old archive
0	short		0177555		very old PDP-11 archive
0	long		0177545		old archive
0	short		0177545		old PDP-11 archive
0	long		0100554		apl workspace
0	string		=<ar>		archive
0	string		!<arch>		archive
>8	string		__.SYMDEF	random library
0	string		-h-		archive (software tools format) text
0	string	%A			bibliographic
0	string	%D			bibliographic
0	string	%T			bibliographic
0	string		/*		c program text
53	string		yyprevious	c program text (from lex)
0	string		#!\ /bin/perl	perl program
0	string		#!/bin/perl	perl program
0	string		#!\ /usr/bin/perl	perl program
0	string		#!/usr/bin/perl	perl program
0	string		#!\ /usr/local/bin/perl	perl program
0	string		#!/usr/local/bin/perl	perl program
0	string		#!\ /bin/sh	commands text
0	string		#!/bin/sh	commands text
0	string		#!\ /bin/csh	c-shell commands text
0	string		#!/bin/csh	c-shell commands text
0	string		#!\ /bin/awk	awk commands text
0	string		#!/bin/awk	awk commands text
0	string		#!\ /		some kinda commands text
0	string		#!/		some kinda commands text
0	string		#!\ 		commands text
>3	string		>\0		 for %s
0	short		017037		packed data
0	short		017436		packed data
0	short		0145405		huf output
0	string		\037\235	compressed data
# non block compressed
>2	byte		12		- with 12 bits
>2	byte		13		- with 13 bits
>2	byte		14		- with 14 bits
>2	byte		15		- with 15 bits
>2	byte		16		- with 16 bits
# block compressed
>2	byte		140		- with 12 bits
>2	byte		141		- with 13 bits
>2	byte		142		- with 14 bits
>2	byte		143		- with 15 bits
>2	byte		144		- with 16 bits
# for big endian like Suns
0	short		0x8301		DEC OSF/1 binary
# for little endian like DECs
0	short		0x0183		DEC OSF/1 binary
0	string		diff\ 	'diff' output text
0	string		***\ 		'diff' output text
0	string		Only\ in\ 	'diff' output text
0	string		Common\ subdirectories:\ 	'diff' output text
# Magic numbers for ditroff intermediate language
0	string		x\ T\ cat	titroff output for the c/a/t text
0	string		x\ T\ ps	titroff output for postscript
0	string		x\ T 		titroff output text
0	string		FirstName	faces project image
0	string	FONT	vfont text
0	short	0436	Berkeley vfont data
0	short	017001	byte-swapped Berkeley vfont data
0	string		\<MakerFile	framemaker document
0	string		GIF87a		gif image
0	string		GIF89a		gif image
0       string          \037\012        gnu gzip
0	string		\<html\> 	HTML source
0	string		\<HTML\> 	HTML source
0	string		\<TITLE\> 	HTML source
0	string		\<title\> 	HTML source
0	string		\<head\> 	HTML source
0	string		\<HEAD\> 	HTML source
0	string		\<header\> 	HTML source
0	string		\<HEADER\> 	HTML source
0	string		#magic		magic text file for file(1) cmd
# Unfortunately, saved netnews also has From line added in some news software.
0	string		From 		mail text
0	string		Forwarded 	mail text
# There are tests to ascmagic.c to cope with mail and news.
0	string		Relay-Version: 	old news text
0	string		#!\ rnews	batched news text
0	string		N#!\ rnews	mailed, batched news text
0	string		Forward\ to 	mail forwarding text
0	string		Pipe\ to 	mail piping text
0	string		Return-Path:	smtp mail text
0	string		Path:		news text
0	string		Xref:		news text
0	string		From:		news or mail text
0	string		Article 	saved news text
0	string		begin 		uuencoded mail text
#	derived empirically, your offsets may vary!
53	string		yyprevious	c program text (from lex)
0	long		0xfeedface	NeXT binary
#
# Let us not forget PostScript
0	string	%!			PostScript image
>2	string	PS-Adobe-		conforming
>11	string	1.0			at level %s
# Sun rasterfiles
0	string	\x59\xa6\x6a\x95	rasterfile
>4	long	>0		%d
>8	long	>0		x %d
>12	long	>0		x %d
>20	long	0		old format
>20	long	2		compressed
>24	long	1		with color map
# SCCS archive structure:
# \001h01207
# \001s 00276/00000/00000
# \001d D 1.1 87/09/23 08:09:20 ian 1 0
# \001c date and time created 87/09/23 08:09:20 by ian
# \001e
# \001u
# \001U
# ... etc.
# Now '\001h' happens to be the same as the 3B20's a.out magic number (0550).
# *Sigh*. And these both came from various parts of the USG.
# Maybe we should just switch everybody from SCCS to RCS!
# Further, you can't just say '\001h0', because the five-digit number
# is a checksum that could (presumably) have any leading digit,
# and we don't have regular expression matching yet. 
# Hence the following official kludge:
8	string		\001s\ 			SCCS archive.
# for big endian like Suns
0       long            0x81030108      sparc binary
0       long            0x8103010b      sparc binary
0       long            0x80020108      Sun-3 binary
0       long            0x8002010b      Sun-3 binary
# for little endian like DECs
0	short		0x0381		sparc binary
0	short		0x0280		Sun-3 binary
0	long	0x4d4d002a		tiff image
0	string		.so\ man 	troff source for unix manual page
0	string		.soman 		troff source for unix manual page
0	string		.TH 		troff source for unix manual page
0	string		.EQ 		troff source with eqn
0	string		.TS 		troff source with tbl
0	string		'\\" 		troff source
0	string		.\\" 		troff source
0	string		\\" 		troff source
0	string		.po 		troff source
0	string		.nr 		troff source
0	string		.so 		troff source
0	string		.de 		troff source
0	string		.if 		troff source