cookie.c

harvest是一个下载html网页得机器人

        if (strcmp("TRUE", ptr) && strcmp("FALSE", ptr)) {
          /* only if the path doesn't look like a boolean option! */
          co->path = strdup(ptr);
          break;
        }
        /* this doesn't look like a path, make one up! */
        co->path = strdup("/");
        fields++; /* add a field and fall down to secure */
        /* FALLTHROUGH */
      case 3:
        co->secure = strequal(ptr, "TRUE");
        break;
      case 4:
        co->expires = atoi(ptr);
        break;
      case 5:
        co->name = strdup(ptr);
        break;
      case 6:
        co->value = strdup(ptr);
        break;
      }
    }

    if(6 == fields) {
      /* we got a cookie with blank contents, fix it */
      co->value = strdup("");
    }
    else if(7 != fields) {
      /* we did not find the sufficient number of fields to recognize this
         as a valid line, abort and go home */
      free_cookiemess(co);
      return NULL;
    }
  }

  if(!c->running &&    /* read from a file */
     c->newsession &&  /* clean session cookies */
     !co->expires) {   /* this is a session cookie since it doesn't expire! */
    free_cookiemess(co);
    return NULL;
  }

  co->livecookie = c->running;

  /* now, we have parsed the incoming line, we must now check if this
     superceeds an already existing cookie, which it may if the previous have
     the same domain and path as this */

  clist = c->cookies;
  replace_old = FALSE;
  while(clist) {
    if(strequal(clist->name, co->name)) {
      /* the names are identical */

      if(clist->domain && co->domain) {
        if(strequal(clist->domain, co->domain))
          /* The domains are identical */
          replace_old=TRUE;
      }
      else if(!clist->domain && !co->domain)
        replace_old = TRUE;

      if(replace_old) {
        /* the domains were identical */

        if(clist->path && co->path) {
          if(strequal(clist->path, co->path)) {
            replace_old = TRUE;
          }
          else
            replace_old = FALSE;
        }
        else if(!clist->path && !co->path)
          replace_old = TRUE;
        else
          replace_old = FALSE;
        
      }

      if(replace_old && !co->livecookie && clist->livecookie) {
        /* Both cookies matched fine, except that the already present
           cookie is "live", which means it was set from a header, while
           the new one isn't "live" and thus only read from a file. We let
           live cookies stay alive */

        /* Free the newcomer and get out of here! */
        if(co->domain)
          free(co->domain);
        if(co->path)
          free(co->path);
        if(co->name)
          free(co->name);
        if(co->value)
          free(co->value);

        free(co);
        return NULL;
      }

      if(replace_old) {
        co->next = clist->next; /* get the next-pointer first */

        /* then free all the old pointers */
        if(clist->name)
          free(clist->name);
        if(clist->value)
          free(clist->value);
        if(clist->domain)
          free(clist->domain);
        if(clist->path)
          free(clist->path);
        if(clist->expirestr)
          free(clist->expirestr);

        if(clist->version)
          free(clist->version);
        if(clist->maxage)
          free(clist->maxage);

        *clist = *co;  /* then store all the new data */

        free(co);   /* free the newly alloced memory */
        co = clist; /* point to the previous struct instead */

        /* We have replaced a cookie, now skip the rest of the list but
           make sure the 'lastc' pointer is properly set */
        do {
          lastc = clist;
          clist = clist->next;
        } while(clist);
        break;
      }
    }
    lastc = clist;
    clist = clist->next;
  }

  if(c->running)
    /* Only show this when NOT reading the cookies from a file */
    infof(data, "%s cookie %s=\"%s\" for domain %s, path %s, expire %d\n",
          replace_old?"Replaced":"Added", co->name, co->value,
          co->domain, co->path, co->expires);

  if(!replace_old) {
    /* then make the last item point on this new one */
    if(lastc)
      lastc->next = co;
    else
      c->cookies = co;
  }

  c->numcookies++; /* one more cookie in the jar */
  return co;
}

/*****************************************************************************
 *
 * Curl_cookie_init()
 *
 * Inits a cookie struct to read data from a local file. This is always
 * called before any cookies are set. File may be NULL.
 *
 * If 'newsession' is TRUE, discard all "session cookies" on read from file.
 *
 ****************************************************************************/
struct CookieInfo *Curl_cookie_init(struct SessionHandle *data,
                                    char *file,
                                    struct CookieInfo *inc,
                                    bool newsession)
{
  char line[MAX_COOKIE_LINE];
  struct CookieInfo *c;
  FILE *fp;
  bool fromfile=TRUE;
  
  if(NULL == inc) {
    /* we didn't get a struct, create one */
    c = (struct CookieInfo *)malloc(sizeof(struct CookieInfo));
    if(!c)
      return NULL; /* failed to get memory */
    memset(c, 0, sizeof(struct CookieInfo));
    c->filename = strdup(file?file:"none"); /* copy the name just in case */
  }
  else {
    /* we got an already existing one, use that */
    c = inc;
  }
  c->running = FALSE; /* this is not running, this is init */

  if(file && strequal(file, "-")) {
    fp = stdin;
    fromfile=FALSE;
  }
  else
    fp = file?fopen(file, "r"):NULL;

  c->newsession = newsession; /* new session? */

  if(fp) {
    char *lineptr;
    bool headerline;
    while(fgets(line, MAX_COOKIE_LINE, fp)) {
      if(checkprefix("Set-Cookie:", line)) {
        /* This is a cookie line, get it! */
        lineptr=&line[11];
        headerline=TRUE;
      }
      else {
        lineptr=line;
        headerline=FALSE;
      }
      while(*lineptr && isspace((int)*lineptr))
        lineptr++;

      Curl_cookie_add(data, c, headerline, lineptr, NULL, NULL);
    }
    if(fromfile)
      fclose(fp);
  }

  c->running = TRUE;          /* now, we're running */

  return c;
}

/*****************************************************************************
 *
 * Curl_cookie_getlist()
 *
 * For a given host and path, return a linked list of cookies that the
 * client should send to the server if used now. The secure boolean informs
 * the cookie if a secure connection is achieved or not.
 *
 * It shall only return cookies that haven't expired.
 *
 ****************************************************************************/

struct Cookie *Curl_cookie_getlist(struct CookieInfo *c,
                                   char *host, char *path, bool secure)
{
   struct Cookie *newco;
   struct Cookie *co;
   time_t now = time(NULL);
   struct Cookie *mainco=NULL;

   if(!c || !c->cookies)
      return NULL; /* no cookie struct or no cookies in the struct */

   co = c->cookies;

   while(co) {
     /* only process this cookie if it is not expired or had no expire
        date AND that if the cookie requires we're secure we must only
        continue if we are! */
     if( (co->expires<=0 || (co->expires> now)) &&
         (co->secure?secure:TRUE) ) {
       
       /* now check if the domain is correct */
       if(!co->domain ||
          (co->tailmatch && tailmatch(co->domain, host)) ||
          (!co->tailmatch && strequal(host, co->domain)) ) {
         /* the right part of the host matches the domain stuff in the
            cookie data */
         
         /* now check the left part of the path with the cookies path
            requirement */
         if(!co->path ||
            checkprefix(co->path, path) ) {

           /* and now, we know this is a match and we should create an
              entry for the return-linked-list */
           
           newco = (struct Cookie *)malloc(sizeof(struct Cookie));
           if(newco) {
             /* first, copy the whole source cookie: */
             memcpy(newco, co, sizeof(struct Cookie));

             /* then modify our next */
             newco->next = mainco;
             
             /* point the main to us */
             mainco = newco;
           }
         }
       }
     }
     co = co->next;
   }

   return mainco; /* return the new list */
}


/*****************************************************************************
 *
 * Curl_cookie_freelist()
 *
 * Free a list of cookies previously returned by Curl_cookie_getlist();
 *
 ****************************************************************************/

void Curl_cookie_freelist(struct Cookie *co)
{
   struct Cookie *next;
   if(co) {
      while(co) {
	 next = co->next;
	 free(co); /* we only free the struct since the "members" are all
		      just copied! */
	 co = next;
      }
   }
}

/*****************************************************************************
 *
 * Curl_cookie_cleanup()
 *
 * Free a "cookie object" previous created with cookie_init().
 *
 ****************************************************************************/
void Curl_cookie_cleanup(struct CookieInfo *c)
{
   struct Cookie *co;
   struct Cookie *next;
   if(c) {
      if(c->filename)
	 free(c->filename);
      co = c->cookies;

      while(co) {
	 if(co->name)
	    free(co->name);
	 if(co->value)
	    free(co->value);
	 if(co->domain)
	    free(co->domain);
	 if(co->path)
	    free(co->path);
	 if(co->expirestr)
	    free(co->expirestr);

	 if(co->version)
	    free(co->version);
	 if(co->maxage)
	    free(co->maxage);

	 next = co->next;
	 free(co);
	 co = next;
      }
      free(c); /* free the base struct as well */
   }
}

/*
 * Curl_cookie_output()
 *
 * Writes all internally known cookies to the specified file. Specify
 * "-" as file name to write to stdout.
 *
 * The function returns non-zero on write failure.
 */
int Curl_cookie_output(struct CookieInfo *c, char *dumphere)
{
  struct Cookie *co;
  FILE *out;
  bool use_stdout=FALSE;

  if((NULL == c) || (0 == c->numcookies))
    /* If there are no known cookies, we don't write or even create any
       destination file */
    return 0;

  if(strequal("-", dumphere)) {
    /* use stdout */
    out = stdout;
    use_stdout=TRUE;
  }
  else {
    out = fopen(dumphere, "w");
    if(!out)
      return 1; /* failure */
  }

  if(c) {
    fputs("# Netscape HTTP Cookie File\n"
          "# http://www.netscape.com/newsref/std/cookie_spec.html\n"
          "# This file was generated by libcurl! Edit at your own risk.\n\n",
          out);
    co = c->cookies;
     
    while(co) {
      fprintf(out,
              "%s%s\t" /* domain */
              "%s\t" /* tailmatch */
              "%s\t" /* path */
              "%s\t" /* secure */
              "%u\t" /* expires */
              "%s\t" /* name */
              "%s\n", /* value */

              /* Make sure all domains are prefixed with a dot if they allow
                 tailmatching. This is Mozilla-style. */
              (co->tailmatch && co->domain && co->domain[0] != '.')? ".":"",
              co->domain?co->domain:"unknown",
              co->tailmatch?"TRUE":"FALSE",
              co->path?co->path:"/",
              co->secure?"TRUE":"FALSE",
              (unsigned int)co->expires,
              co->name,
              co->value?co->value:"");

      co=co->next;
    }
  }

  if(!use_stdout)
    fclose(out);

  return 0;
}

#endif /* CURL_DISABLE_HTTP */