lex.uri.c

在linux下的crawler程序,来自北大天网tiny search engine spider

      284,  282,    0,    0,    0,    0,  284,  282,  282,  284,
      282,  282,  283,  283,  296,  283,  283,    0,    0,    0,
      296,  283,  283,  296,  283,  283,  292,    0,    0,    0,
        0,    0,  292,  292,    0,  292,  292,  294,    0,    0,
        0,    0,    0,  294,  294,    0,  294,  294,  298,  298,
      298,  298,  298,  298,  298,  298,  298,  298,  298,  298,
      298,  298,  298,  298,  298,  298,  298,  298,  298,  298,
      298,  298,  298,  298,  299,    0,  302,    0,    0,    0,
      299,    0,  302,  299,  300,  302,    0,    0,    0,    0,
      300,  300,    0,  300,  300,  301,    0,    0,    0,    0,

        0,  301,  301,    0,  301,  301,  303,  303,  303,  303,
      303,  303,  303,  303,  303,  303,  303,  303,  303,  303,
      303,  303,  303,  303,  303,  303,  303,  303,  303,  303,
      303,  303,  304,  304,  310,  304,    0,    0,    0,    0,
      310,  304,  304,  310,  304,  304,  305,  305,  305,  305,
      305,  305,  305,  305,  305,  305,  305,  305,  305,  305,
      305,  305,  305,  305,  305,  305,  305,  305,  305,  305,
      305,  305,  306,  306,  306,  306,  306,  306,  306,  306,
      306,  306,  306,  306,  306,  306,  306,  306,  306,  306,
      306,  306,  306,  306,  306,  306,  306,  306,  307,  307,

      313,  307,  307,    0,    0,    0,  313,  307,  307,  313,
      307,  307,  316,    0,  316,  316,  316,  316,  316,  316,
      316,  316,  316,  316,  316,  316,  316,  316,  316,  316,
        0,  316,  316,  316,  316,  316,  316,  316,  317,  320,
      321,    0,    0,    0,  317,  320,  321,  317,  320,  321,
      323,    0,  323,  323,  323,  323,  323,  323,  323,  323,
      323,  323,  323,  323,  323,  323,  323,  323,    0,  323,
      323,  323,  323,  323,  323,  323,  324,    0,  327,    0,
        0,    0,  324,    0,  327,  324,  325,  327,  325,  325,
      325,  325,  325,  325,  325,  325,  325,  325,  325,  325,

      325,  325,  325,  325,    0,  325,  325,  325,  325,  325,
      325,  325,  326,  326,  326,  326,  326,  326,  326,  326,
      326,  326,  326,  326,  326,  326,  326,  326,  326,  326,
      326,  326,  326,  326,  326,  326,  326,  326,  328,  328,
      328,  328,  328,  328,  328,  328,  328,  328,  328,  328,
      328,  328,  328,  328,  328,  328,  328,  328,  328,  328,
      328,  328,  328,  328,  330,  332,  335,    0,    0,    0,
      330,  332,  335,  330,  332,  335,  336,  340,    0,    0,
        0,    0,  336,  340,    0,  336,  340,  348,    0,    0,
        0,    0,    0,  348,  348,    0,  348,  348,  349,    0,

        0,    0,    0,    0,  349,  349,    0,  349,  349,  353,
        0,  357,    0,    0,    0,  353,  353,  357,  353,  353,
      357,  358,  358,  358,  358,  358,  358,  358,  358,  358,
      358,  358,  358,  358,  358,  358,  358,  358,  358,  358,
      358,  358,  358,  358,  358,  358,  358,  360,  363,    0,
        0,    0,    0,  360,  363,    0,  360,  363,  364,  364,
      364,  364,  364,  364,  364,  364,  364,  364,  364,  364,
      364,  364,  364,  364,  364,  364,  364,  364,  364,  364,
      364,  364,  364,  364,  365,  365,  365,  365,  365,  365,
      365,  365,  365,  365,  365,  365,  365,  365,  365,  365,

      365,  365,  365,  365,  365,  365,  365,  365,  365,  365,
      372,    0,    0,    0,    0,    0,  372,    0,    0,  372,
      373,  373,  373,  373,  373,  373,  373,  373,  373,  373,
      373,  373,  373,  373,  373,  373,  373,  373,  373,  373,
      373,  373,  373,  373,  373,  373,  374,    0,    0,    0,
        0,    0,  374,  374,    0,  374,  374,  375,    0,  376,
        0,    0,    0,  375,  375,  376,  375,  375,  376,  377,
      377,  384,  377,    0,    0,    0,    0,  384,  377,  377,
      384,  377,  377,  378,    0,    0,    0,    0,    0,  378,
      378,    0,  378,  378,  379,  379,  385,  379,  379,    0,

        0,    0,  385,  379,  379,  385,  379,  379,  383,    0,
        0,    0,    0,    0,  383,  383,    0,  383,  383,  386,
        0,  386,  386,  386,  386,  386,  386,  386,  386,  386,
      386,  386,  386,  386,  386,  386,  386,    0,  386,  386,
      386,  386,  386,  386,  386,  387,    0,  395,    0,    0,
        0,  387,    0,  395,  387,  388,  395,  388,  388,  388,
      388,  388,  388,  388,  388,  388,  388,  388,  388,  388,
      388,  388,  388,    0,  388,  388,  388,  388,  388,  388,
      388,  391,    0,  391,  391,  391,  391,  391,  391,  391,
      391,  391,  391,  391,  391,  391,  391,  391,  391,    0,

      391,  391,  391,  391,  391,  391,  391,  398,    0,    0,
        0,    0,    0,  398,  398,    0,  398,  398,  402,    0,
      405,    0,    0,    0,  402,  402,  405,  402,  402,  405,
      408,    0,  410,    0,    0,    0,  408,    0,  410,  408,
      409,  410,  413,    0,    0,    0,  409,  409,  413,  409,
      409,  413,  414,  414,  414,  414,  414,  414,  414,  414,
      414,  414,  414,  414,  414,  414,  414,  414,  414,  414,
      414,  414,  414,  414,  414,  414,  414,  414,  417,  420,
        0,    0,    0,    0,  417,  420,    0,  417,  420,  421,
        0,    0,    0,    0,    0,  421,  421,    0,  421,  421,

      422,    0,  423,    0,    0,    0,  422,  422,  423,  422,
      422,  423,  424,  424,  426,  424,    0,    0,    0,    0,
      426,  424,  424,  426,  424,  424,  425,  425,  448,  425,
      425,    0,    0,    0,  448,  425,  425,  448,  425,  425,
      434,    0,    0,    0,    0,    0,  434,  434,    0,  434,
      434,  436,    0,  450,    0,    0,    0,  436,  436,  450,
      436,  436,  450,  453,  454,  458,    0,    0,    0,  453,
      454,  458,  453,  454,  458,  466,    0,    0,    0,    0,
        0,  466,  466,    0,  466,  466,  467,    0,    0,    0,
        0,    0,  467,  467,    0,  467,  467,  471,    0,  479,

        0,    0,    0,  471,  471,  479,  471,  471,  479,  482,
        0,    0,    0,    0,    0,  482,  482,    0,  482,  482,
      486,    0,    0,    0,    0,    0,  486,  486,    0,  486,
      486,  514,  514,  515,  515,  516,  516,  517,  517,  518,
      518,  519,  519,  520,  520,  521,  521,  513,  513,  513,
      513,  513,  513,  513,  513,  513,  513,  513,  513,  513,
      513,  513,  513,  513,  513,  513,  513,  513,  513,  513,
      513,  513,  513,  513,  513,  513
    } ;

static yy_state_type yy_state_buf[YY_BUF_SIZE + 2], *yy_state_ptr;
static char *yy_full_match;
static int yy_lp;
static int yy_looking_for_trail_begin = 0;
static int yy_full_lp;
static int *yy_full_state;
#define YY_TRAILING_MASK 0x2000
#define YY_TRAILING_HEAD_MASK 0x4000
#define REJECT \
{ \
*yy_cp = yy_hold_char; /* undo effects of setting up yytext */ \
yy_cp = yy_full_match; /* restore poss. backed-over text */ \
yy_lp = yy_full_lp; /* restore orig. accepting pos. */ \
yy_state_ptr = yy_full_state; /* restore orig. state */ \
yy_current_state = *yy_state_ptr; /* restore curr. state */ \
++yy_lp; \
goto find_rule; \
}
#define yymore() yymore_used_but_not_detected
#define YY_MORE_ADJ 0
#define YY_RESTORE_YY_MORE_OFFSET
char *yytext;
#line 1 "uri.l"
#define INITIAL 0
/*
 * uri.l -- Routines dealing with URI, mainly parsing and merging.
 * Created: Xie Han, OS lab of Peking University. <[e-mail]>
 *
 * The grammar of URI is quite simply so yacc is not needed.
 *
 * Created: Sep 25 04:15am 2003. version 0.1.1
 *		# Can read URI from standard input, parse it and print every
 *		  part of the URI.
 *
 * Updated: Sep 25 05:10pm 2003. version 0.1.2
 *
 * Updated: Sep 25 10:30pm 2003. version 0.1.3
 *		# A lot of changes in the program structure, much more clear.
 *		# Now the result is written into a uri structure.
 *		# A little bug. String "?" is accepted as a valid URI.
 *
 * Updated: Sep 26 02:16am 2003. version 0.1.4
 *		# 2 states are added to solve the bug in version 0.1.3.
 *		# The uri structure was redesigned.
 *
 * Updated: Sep 27 05:21am 2003. version 1.0.0!!!!
 *		# All states are changed from exclusive to shared to make
 *		  the lex program more compatible because some lexer do not
 *		  support exclusive start condition.
 *		# "unput" is substituted by "yyless" when putting back excess
 *		  text, because the latter cost less.
 *		# 4 parsing functions are completed, each of which has different
 *		  characteristic.
 *		# Function "uri_rally_parse" is added, which cooperates with
 *		  other lex programs.
 *
 * Updated: Sep 28 1:00pm 2003. version 1.1.0
 *		# The function "uri_rel_to_abs" is under construction.
 *		# The URI structure is changed again: "path_type" field is remove,
 *		  and "abs_path", "rel_path", "opaque_path" are substituted by
 *		  "path" because the original design is too sophisticated.
 *
 * Updated: National Day 5:06am 2003. version 1.1.1
 *		# The function "__uri_path_merge" which merges the base path and
 *		  the relative path is finished and tested and proved to be a
 *		  nice job! This function is the core of merging a base URI and
 *		  a relative URI into an absolute URI.
 *		# The previous function "uri_rel_to_abs" is commented out. It's
 *		  terrible.
 *
 * Updated: National Day 5:00pm 2003. version 1.1.2
 *		# Now we distinguish between "no" and "empty". Related fields
 *		  include "userinfo", "port", "query", and "fragment". The previous
 *		  version treats "no" and "empty" equally, which is not right.
 *
 * Updated: Oct 2 2:42am 2003. version 1.1.3
 *		# Solve the conflict of sever and reg_name by adding a start
 *		  condition and some ugly codes that are normally not reached.
 *		# The "uri_rel_to_abs" is rewritten but has not been tested.
 *
 * Updated: Oct 4 6:33pm 2003. version 1.2.0
 *		# Function "uri_path_merge" now returns the length of the result
 *		  path while the result path is returned by a pointer.
 *		# Function "uri_rel_to_abs" is renamed "uri_merge", and it will
 *		  return the string length of the result URI.
 *		# Function "uri_recombine" is coded, which recombine the a URI
 *		  structure into a URI string, and the string length is returned.
 *		# The URI merging algorithm in RFC 2396 is found not flawless, and
 *		  the problem is fixed in "uri_merge" with not very graceful means.
 *
 * Updated: Oct 5 2:09am 2003. version 1.2.1
 *		# From this version function "uri_parse_string" is implemented by
 *		  "uri_parse_bytes" which from now on is dependent on
 *		  "uri_parse_buffer".
 *		# "uri_validate_string" and "uri_validate_bytes" were added. Their
 *		  names explain what they do.
 *		# Key word "static" was taken off from the front of function
 *		  "__uri_path_merge" while it is not declared in the .h file. I
 *		  mean to make it a secret interface ^_^
 *
 * Updated: Oct 5 4:43am 2003. version 1.2.2
 *		# Cross the first change log of version 1.2.1!
 *		# "uri_validate_string" is no longer implemented by
 *		  "uri_validate_bytes".
 *
 * Updated: Oct 9 4:49am 2003. version 1.2.3
 *		# Function "uri_rally_parse" no longer exists, because it cannot
 *		  work. The original desire that it can cooperate with other lex
 *		  programs is indeed out of the question.
 *		# Macro URI_INIT changed slightly.
 *
 * Updated: Oct 10 4:28am 2003. version 1.2.4
 *		# The arguments sequence of "rel_xxx" and "base_xxx" of functions
 *		  "__uri_path_merge" and "uri_merge" are exchanged. "rel_xxx" from
 *		  now on precedes "base_xxx".
 *
 * Updated: Oct 11 3:31am 2003. version 1.2.5
 *		# URI merging algorithm is changed again. We check if the path of
 *		  the relative URI is defined before merging path, and if it's not
 *		  defined, we take the base URI's path directly other than merging
 *		  the paths. I thinks that the URI merging algorithm is now at the
 *		  edge of perfect. (The recommended merging algorithm of RFC 2396
 *		  has a lot of problems.)
 *		# The rule of "<REG_NAME>{reg_name}" is optimized, because "strcat"
 *		  is slow.
 *
 * Updated: Oct 12 4:49am 2003. version 1.2.6
 *		# "goto" statements are substituted by "do ... while (0)" and
 *		  "break" in function "uri_merge" and "uri_recombine". No real
 *		  improvement is made but showing off continues.
 *
 * Updated: Oct 12 4:55pm 2003. version 1.2.7
 *		# URIs with chinese character are supported in an obviously
 *		  not perfect way: the "unreserved" BNF allows non-US-ASCII
 *		  characters, i.e., characters with value that ranges from
 *		  128 to 255.
 *
 * Updated: Oct 13 1:06am 2003. version 1.2.8
 *		# Some codes of "__uri_path_merge" was rewritten because the
 *		  interface of stack module changed.
 *		# NOTE!!! NULL can no long be passed to "__uri_path_merge" as
 *		  a relative path or a base path, and the function from now on
 *		  won't set NULL as the result path. If the result path is
 *		  empty, the "abs_path" will point to a pointer that points to
 *		  an empty string ("\0"). So you can always "free(*abs_path)"
 *		  if "__uri_path_merge" succeeded.
 *
 * Updated: Oct 15 1:13am 2003. version 1.4.0
 *		# The update of 1.x.x version continues because version 2.0.x
 *		  runs slower than 1.x.x (half speed), though 2.0.x is written
 *		  in much shorter and very nice codes.
 *		# The new uri structure is derived from 1.3.x. 1.3.x is a test
 *		  version which has been updated into 2.0.0. Indeed, 1.4.x is
 *		  compasible with 2.0.x completely, but 2.0.x is NOT compasible
 *		  with 1.4.x because 2.0.x has only one p