/*  PATH_INFO_routines.		WJS  Jul 97				*/
/*	Purpose of these routines is to extract portions of information	*/
/*  from the "PATH_INFO" environment variable as formatted by the JGOFS	*/
/*  system, and to create a properly formatted string that can be used	*/
/*  to created that environment variable.				*/
/*									*/
/*	To get information, use the functions				*/
/*		int get_level()						*/
/*		char *get_object()					*/
/*		char *get_options()					*/
/*		char *get_protocol()					*/
/*  Each takes an argument, which is usually NULL (details in the	*/
/*  per-function comments, below)					*/
/*      The package is missing get_dir, get_info, etc functions at	*/
/*  present.  See some comments at bottom				*/
/*									*/
/*	To create a string that can be used by putenv, use the function	*/
/*		char *make_PATH_INFO_putenv_string()			*/
/*  See its documentation, below					*/
/*									*/
/*									*/
/*    This implementation derived from code in outer from JGOFS		*/
/*  release 1.5, but does not match it exactly.  It will correctly	*/
/*  deal with strings that JFOFS 1.5 will read, and hopefully will	*/
/*  be flexible enough to deal with future strings.  The get_protocol	*/
/*  and get_level routines will validate things to more closely match	*/
/*  the outer code (& therefore will probably need more modification	*/
/*  in the future)							*/
/*									*/
/*    At present, the environment variable PATH_INFO contains various	*/
/*  information.  Approximately, its layout is				*/
/*	object_location,object_name,protocol,level,options		*/
/*  options, if any, are contained in curly brackets.  The rest of the	*/
/*  string is formatted as a unix file spec, with the object name in	*/
/*  the position of the file name, and the protocol/level string in	*/
/*  the position of the file type.					*/
/*  (Note that the URL precedes this string with the cgi-bin program	*/
/*  that will process the string.  Generally, it is not possible to	*/
/*  tell where the object_location begins, but because the program is	*/
/*  usually of the form "/jg/name" (eg, /jg/serv, /jg/dir), one can	*/
/*  guess.)								*/
/*									*/
/*    If PATH_INFO contains the strings .html or .flat (anywhere),	*/
/*  those strings are considered "protocol" strings by outer.  They	*/
/*  must be followed by at least a 1 digit level.  Other parts of the	*/
/*  JGOFS system limit the level to 1 digit, but outer does not.  In	*/
/*  the absence of a protocol string, outer produces a default protocol	*/
/*									*/

#define PATH_INFO_ROUTINES_VERSION "path_info_routines  version 1.1  9 Aug 1997"
/*   9 Aug 97.			   WJS					*/
/*	get_object & get_options entries				*/
/*	  [Begin version 1.1]						*/
/*  25 Jul 97.			   WJS					*/
/*	  [Begin version 1.0]						*/

  /*  Force version string into object files				*/
char *path_info_routines_version = PATH_INFO_ROUTINES_VERSION;

#include "path_info_routines.h"
#include <ctype.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

  /*  All these functions are local; defined here so that they can	*/
  /*  be arbitrarily ordered in the source file				*/
char *parse_PATH_INFO_string();
char *make_PATH_INFO_putenv_string();
char *get_object();
char *get_params();
char *get_options();
int get_level();
char *copy_substring();

/************************************************************************/
/*  get_info routines.							*/
/************************************************************************/

/*    All depend on order of fields in PATH_INFO			*/
/*    All take 1 input argument, an "info_source" string		*/
/*    If info_source is NULL, default source of info is used.  If	*/
/*  non-NULL, it is a pointer to a PATH_INFO string.  Note that		*/
/*  logically, the info we want has nothing to do with PATH_INFO,	*/
/*  allowing us to change the mechanism of specifying the various	*/
/*  pieces of information						*/
/*    Routines that return a string pointer return a pointer to a	*/
/*  dynamically allocated, null-terminated string.  They return		*/
/*  a pointer to an empty string if no info is found.  They		*/
/*  returns a NULL pointer if the string is bad somehow (see each	*/
/*  routine) or if there's a memory allocation failure			*/

char *copy_substring(start_ptr,end_ptr)
char *start_ptr,*end_ptr;
/*    Copy substring defined by start_ptr & end_ptr into dynamically	*/
/*  allocated string.							*/
/*    end_ptr = NULL means "copy to end of string"			*/
/*    Return pointer to dynamically allocated string (empty if no	*/
/*  characters in substring; NULL if allocation failed.			*/
{
  char *ret;
  char save1;

  if (  (start_ptr == NULL) ||  (start_ptr == end_ptr)  )  {
    ret = (char *) malloc(1);
    *ret = '\0';
  } else {
    if (end_ptr != NULL) {
      save1 = *end_ptr;
      *end_ptr = '\0';
    }
      /*  Emulate strdup function.  +1 is for terminating '\0'		*/
    ret = (char *) malloc(strlen(start_ptr)+1);
    if (ret != NULL) strcpy(ret,start_ptr);

        /*  Restore original string, if necessary			*/
    if (end_ptr != NULL) *end_ptr = save1;
  }

  return ret;
}

int get_level(info_source)
char *info_source;
/*    See parse_PATH_INFO_string & info at top.				*/
/*    See get_info argument doc, above					*/
/*    This routine returns the level of output method has been re-	*/
/*  quested to produce.  If no level was provided, returns		*/
/*  LEVEL_NOT_SPECIFIED, representing "all levels".  Returns		*/
/*  LEVEL_ERROR if format was bad (right now, this means first non-	*/
/*  numeric character after level is not start of options string ({)	*/
/*  or end-of-string (if no options).					*/
{
  char *proto_ptr,*lev_ptr,*options_ptr;
  char *term_char;
  int lev;

  parse_PATH_INFO_string(info_source,&proto_ptr,&lev_ptr,&options_ptr);

  if (lev_ptr == NULL)
      /*  Could also return error if proto_ptr non-NULL if "rules"	*/
      /*  say protocol must have level					*/
    lev = LEVEL_NOT_SPECIFIED;
  else {
    lev = strtol(lev_ptr,&term_char,10);
      /*  Assume that level is terminated by options or end of string	*/
      /*  Further assumes no white space between level and options	*/
    if (  (*term_char == '\0') || (*term_char == PATH_INFO_OPTIONS_DELIM[0])  )
        /*  Could validate level here (0-9, for example)		*/
      ;
    else lev = LEVEL_ERROR;
  }

  return lev;
}

char *get_protocol(info_source)
char *info_source;
/*    See parse_PATH_INFO_string & info at top.				*/
/*    See get_info argument doc, above					*/
/*    Returns a NULL pointer if the string is bad somehow (right now	*/
/*  only happens if string is empty, which might not be considered an	*/
/*  error.  Oh, well)							*/
{
  char *proto_ptr,*lev_ptr,*options_ptr;
  char *end_proto_ptr,*orig_proto_ptr;

  parse_PATH_INFO_string(info_source,&proto_ptr,&lev_ptr,&options_ptr);

  orig_proto_ptr = proto_ptr;
  if (proto_ptr != NULL) {
      /*  Advance beyond .						*/
    proto_ptr++;
      /*  Assume options follow level (which follows protocol)		*/
    end_proto_ptr = (lev_ptr == NULL) ? options_ptr : lev_ptr;
  }

    /*  copy_substring should handle NULL end_proto_ptr ...		*/
  proto_ptr = copy_substring(proto_ptr,end_proto_ptr);

  if (proto_ptr != NULL) {
      /*  Validate protocol here.  If bad, free proto_ptr then set to	*/
      /*  NULL								*/
	/*  Can't have . without protocol				*/
    if (  (*proto_ptr == '\0') && (orig_proto_ptr != NULL)  ) {
      free (proto_ptr);
      proto_ptr = NULL;
    }
  }

  return proto_ptr;
}

char *get_object(info_source)
char *info_source;
/*    See parse_PATH_INFO_string & info at top.				*/
/*    See get_info argument doc, above					*/
{
  char *object_ptr,*proto_ptr,*lev_ptr,*options_ptr;
  char *end_object_ptr;

  object_ptr =
    parse_PATH_INFO_string(info_source,&proto_ptr,&lev_ptr,&options_ptr);

  if (object_ptr != NULL) {
      /*  Find end of object string					*/
      /*  Assume order of things is object, protocol, level, options	*/
    if (  (end_object_ptr = proto_ptr) == NULL  )
      if (  (end_object_ptr = lev_ptr) == NULL  )
        end_object_ptr = options_ptr;
  }

    /*  copy_substring should handle NULL end_object_ptr ...		*/
  object_ptr = copy_substring(object_ptr,end_object_ptr);

  if (object_ptr != NULL) {
       /*  Validate object here.  If bad, free object_ptr then set to	*/
       /*  NULL								*/
    ;
  }

  return object_ptr;
}

char *get_options(info_source)
char *info_source;
/*    See parse_PATH_INFO_string & info at top.				*/
/*    See get_info argument doc, above					*/
/*    Returns NULL if no terminating } is found				*/
{
  char *proto_ptr,*lev_ptr,*options_ptr;
  char *end_options_ptr;

  parse_PATH_INFO_string(info_source,&proto_ptr,&lev_ptr,&options_ptr);

  if (options_ptr != NULL) {
      /*  Advance beyond {						*/
    options_ptr++;
      /*  Find {.  Must be there if options exist, or we have an	*/
      /*  improperly formatted PATH_INFO				*/
    end_options_ptr = strrchr(options_ptr,PATH_INFO_OPTIONS_DELIM[1]);
    if (end_options_ptr == NULL) return NULL;
  }

    options_ptr = copy_substring(options_ptr,end_options_ptr);

  if (options_ptr != NULL) {
      /*  Validate options here.  If bad, free options_ptr then set to	*/
      /*  NULL								*/
    ;
  }

  return options_ptr;
}

/************************************************************************/
/*  PATH_INFO analysis & formatting routines.				*/
/************************************************************************/

char *parse_PATH_INFO_string(PATH_INFO_string,proto_ptr,level_ptr,options_ptr)
char *PATH_INFO_string;
char **proto_ptr;
char **level_ptr;
char **options_ptr;
/*    This routine assumes that the final { in PATH_INFO marks the	*/
/*  beginning of the options string.  The final ., if any, before the	*/
/*  options, if any, marks the beginning of a protocol/level string.	*/
/*  The protocol consists entirely of non-numeric characters and is	*/
/*  immediately followed by an all-numeric level, the options, or the	*/
/*  end of the string							*/
/*									*/
/*    This routine accepts PATH_INFO_string as input.  If the input is	*/
/*  NULL, it uses the value of environment variable PATH_INFO.		*/
/*    It returns a pointer to the beginning of the string or NULL if	*/
/*  the environment variable is undefined.  It returns a pointer to	*/
/*  the appropriate . in *proto_ptr (NULL if none), and a pointer to	*/
/*  the first non-alphanumeric character after the . in *level_ptr	*/
/*  (NULL if none).  It returns a pointer to the appropriate { in	*/
/*  *options_ptr (NULL if none).  Note that since we do not return	*/
/*  string lengths, routines that call this routine are sensitive	*/
/*  to the order of fields in PATH_INFO	and will probably break if	*/
/*  the order is changed						*/
{
  char *env_ptr;
  char *ptr;

    /*  Set up for no protocol, level, or options info			*/
  *proto_ptr = NULL;
  *options_ptr = NULL;
  ptr = NULL;

  env_ptr = (PATH_INFO_string == NULL) ? 
			getenv(PATH_INFO_ENV_VAR) : PATH_INFO_string;

  if (env_ptr != NULL) {
      /*  Temporarily lop off options, if any				*/
    *options_ptr = strrchr(env_ptr,PATH_INFO_OPTIONS_DELIM[0]);
    if (*options_ptr != NULL) **options_ptr = '\0';
      /*  Protocol string, if any, is at end of option-less PATH_INFO	*/
    *proto_ptr = strrchr(env_ptr,PATH_INFO_PROTOCOL_SEP);
    if (*proto_ptr != NULL) {
      ptr = *proto_ptr;
	/*  Find first non-alpha character after .			*/
      while (isalpha(*(++ptr)))
	;
      if (*ptr == '\0') ptr = NULL;
    }
  }
  *level_ptr = ptr;

    /*  Restore options if necessary					*/
  if (*options_ptr != NULL) **options_ptr = PATH_INFO_OPTIONS_DELIM[0];
  return env_ptr;
}

char *make_PATH_INFO_putenv_string
		(env_var,PATH_INFO_default,object,protocol,level,options)
char *env_var,*object,*protocol,*options,*PATH_INFO_default;
int level;
/*    See parse_PATH_INFO_string & info at top.				*/
/*    env_var is environment variable to set up.  If *env_var = '\0',	*/
/*  the string is constructed without the leading ENV_VAR=.  NULL	*/
/*  means set up the default environment variable (PATH_INFO)		*/
/*    PATH_INFO_default is the "base" PATH_INFO string whose mod	*/
/*  will be put into the env_var string.  NULL means value of present	*/
/*  PATH_INFO								*/
/*    object, protocol, level, and options are pieces of the string to	*/
/*  be inserted into the PATH_INFO string				*/
/*  *object='\0'    means put no object string in output		*/
/*  object = NULL   means use the existing object string		*/
/*  *protocol='\0'  means put no protocol string in output		*/
/*  protocol = NULL means use the existing protocol			*/
/*  level = LEVEL_NOT_SPECIFIED means put in no level information	*/
/*  level = USE_EXISTING_LEVEL  means use the existing level string	*/
/*  *options='\0'   means put no options string in output		*/
/*  options = NULL  means use the existing options string		*/
/*  									*/
/*    This routine returns a pointer to a dynamically allocated string	*/
/*  suitable for putenv.  (Do not free this string after calling	*/
/*  putenv until you've called another putenv for the same environment	*/
/*  variable.)  Its value is a string of the form			*/
/*		ENV_VAR=good-PATH_INFO-string				*/
/*  (or just good-PATH_INFO-string - see env_var input argument)	*/
/*    Returns NULL in case of trouble.  Cause 1: inability to get	*/
/*  some dynamic memory or other.  Cause 2: finished string, when sent	*/
/*  through the get_info routines, returned error.  Note that we do	*/
/*  NOT diagnose a properly formatted final string that is too long	*/
/*  (for now, anyway)							*/
{
  char *object_ptr,*proto_ptr,*lev_ptr,*options_ptr;
  char *buf,*env;
  char *inptr,*outptr,*start_env_val;
  char levbuf[10];
  char empty = '\0';

    /*  Get pieces from user arguments or PATH_INFO string		*/
    /*  (Note that valid string pieces can be empty).			*/
    /*  Failure if can't						*/
      /*  Do level first so we don't have to worry about free'ing if	*/
      /*  bad level							*/
  if (level == USE_EXISTING_LEVEL) level = get_level(PATH_INFO_default);
  if (level == LEVEL_NOT_SPECIFIED) lev_ptr = &empty;
  else if (level < 0) return NULL;
  else   sprintf  ( (lev_ptr=levbuf), "%d", level );

  object_ptr = (object == NULL) ? get_object(PATH_INFO_default) : object;
  if (object_ptr == NULL) return NULL;

  proto_ptr = (protocol == NULL) ? get_protocol(PATH_INFO_default) : protocol;
  if (proto_ptr == NULL) return NULL;

  options_ptr = (options == NULL) ? get_options(PATH_INFO_default) : options;
  if (options_ptr == NULL) return NULL;

  env = (env_var == NULL) ? PATH_INFO_ENV_VAR : env_var;

    /*  +5 is 1 for =, 1 for ., 2 for {}, and 1 for terminating '\0'	*/
  buf =	(char *) malloc (  strlen(env) 
			 + strlen(object_ptr)
			 + strlen(proto_ptr) 
			 + strlen(levbuf) 
			 + strlen(options_ptr) 
			 + 5			);
  if (buf == NULL) {
    if (object_ptr != object) free (object_ptr);
    if (proto_ptr != protocol) free (proto_ptr);
    if (options_ptr != options) free (options_ptr);
    return NULL;
  }


    /*  Build string from component pieces				*/
  outptr = buf;


  if (*env != '\0') {
      /*  strcpy (buf,env) & add =					*/
    inptr = env;
    while (*inptr != '\0') *(outptr++) = *(inptr++);
    *(outptr++) = ENV_VAR_DEFN_CHAR;
  }
    /*  Save this location, which is start of new PATH_INFO		*/
  start_env_val = outptr;

      /*  strcat (buf,object_ptr)					*/
    inptr = object_ptr;
    while (*inptr != '\0') *(outptr++) = *(inptr++);
    if (object_ptr != object) free (object_ptr);

  if (  (*proto_ptr != '\0') || (*lev_ptr != '\0')  )   {
      /*  Add ., then strcat protocol & level info			*/
    *(outptr++) = PATH_INFO_PROTOCOL_SEP;
    inptr = proto_ptr;
    while (*inptr != '\0') *(outptr++) = *(inptr++);
    if (proto_ptr != protocol) free (proto_ptr);
    inptr = lev_ptr;
    while (*inptr != '\0') *(outptr++) = *(inptr++);
  }

  if (*options_ptr != '\0') {
      /*  strcat options; prepend { and add }				*/
    *(outptr++) = PATH_INFO_OPTIONS_DELIM[0];
    inptr = options_ptr;
    while (*inptr != '\0') *(outptr++) = *(inptr++);
    *(outptr++) = PATH_INFO_OPTIONS_DELIM[1];
    if (options_ptr != options) free (options_ptr);
  }

  *outptr = '\0';

    /*  Validate input args by passing them through get_info routines	*/
    /*  That way any checks need only be coded in those routines	*/
  if (  (proto_ptr = get_protocol(start_env_val)) == NULL  ) return NULL;
  else free(proto_ptr);
  if (  (object_ptr = get_object(start_env_val)) == NULL  ) return NULL;
  else free(object_ptr);
  if (  (options_ptr = get_options(start_env_val)) == NULL  ) return NULL;
  else free(options_ptr);
  if (get_level(start_env_val) == LEVEL_ERROR) return NULL;

  return buf;
}

/************************************************************************/

/*	Options package.  Need something that verifies structure,	*/
/*	something that picks out value of desired "left sides"; eg	*/
/*	get_option("dir") returns xxx if PATH_INFO has dir=xxx;		*/
/*	and something that builds a valid options string		*/
/*		Much of this code already exists.  def's attribute	*/
/*	handling is quite similar to this.  Building the attribute	*/
/*	string after adding width=nn is the same as building		*/
/*	the string here (which is actually adding an option, which is	*/
/*	another function we'd like here).  Looking for an existing	*/
/*	width= is the same as looking up dir=, and has the same		*/
/*	problems (what if there's an iowidth= attribute?).		*/
/*		The defgb wjstbl package will parse & pick the		*/
/*	a=b; c=d;... format, with some generality (eg, allowing white	*/
/*	space).  Structurally, there is a problem in that you only	*/
/*	want to parse the string once, and then do all your picks.	*/
/*	However, that would get away from allowing many PATH_INFO	*/
/*	strings to be parsed by this package.  Another problem is that	*/
/*	wjstbl routines use others which might not be modular.  In	*/
/*      particular, uses defgb err() function.				*/