path: root/src/bin/cuturl.c

#include <stdio.h>
#include <string.h>
#include <stdlib.h>
/*
 schemes are case sensitive but cononicals are lower case.
 domain is case insensitive. return it lowercased.
 port is optional and in decimal
 path
 scheme://username:password@domain:port/path?query_string#fragment_id
 mailto:username@domain

 optional stuff:
 scheme, username, password, port, path, query_string, fragment_id
*/

/*
#define DEFAULT_SCHEME "http"
#define DEFAULT_USERNAME "anonymous"
#define DEFAULT_PASSWORD "anonymous"
#define DEFAULT_PORT "80"
#define DEFAULT_QUERY_STRING ""
#define DEFAULT_FRAGMENT_ID ""

#define DEFAULT_SCHEME "DEFAULT"
#define DEFAULT_USERNAME "DEFAULT"
#define DEFAULT_PASSWORD "DEFAULT"
#define DEFAULT_PORT "DEFAULT"
#define DEFAULT_PATH "DEFAULT"
#define DEFAULT_QUERY_STRING "DEFAULT"
#define DEFAULT_FRAGMENT_ID "DEFAULT"
*/

#define AorB(a,b) ((a)?(a):(b))

#define DEFAULT_SCHEME AorB(getenv("CUTURL_SCHEME"),"DEFAULT")
#define DEFAULT_USERNAME AorB(getenv("CUTURL_USERNAME"),"DEFAULT")
#define DEFAULT_PASSWORD AorB(getenv("CUTURL_PASSWORD"),"DEFAULT")
#define DEFAULT_PORT AorB(getenv("CUTURL_PORT"),"DEFAULT")
#define DEFAULT_PATH AorB(getenv("CUTURL_PATH"),"DEFAULT")
#define DEFAULT_QUERY_STRING AorB(getenv("CUTURL_QUERY_STRING"),"DEFAULT")
#define DEFAULT_FRAGMENT_ID AorB(getenv("CUTURL_FRAGMENT_ID"),"DEFAULT")

#define F_SCHEME 1<<0
#define F_USERNAME 1<<1
#define F_PASSWORD 1<<2
#define F_DOMAIN 1<<3
#define F_PORT 1<<4
#define F_PATH 1<<5
#define F_QUERY_STRING 1<<6
#define F_FRAGMENT_ID 1<<7

char *long_opts[]={"scheme","username","password","domain","port","path","query_string","fragment_id",0};
char *short_opts[]={"s","u","k","d","P","p","q","f"};

int main(int argc,char *argv[]) {
 char *line;
 char *scheme=0;
 char *username=0;
 char *password=0;
 char *domain=0;
 char *port=0;
 char *path=0;
 char *query_string=0;
 char *fragment_id=0;
 //exactly 8 parts! let's store that in a byte.
 unsigned char flags=0;
 int i;
 int size=1024;
 char using_stdin=1;
 char malloced=0;
 if(argc > 0) {
  if(!strcmp(argv[1],"--help") || !strcmp(argv[1],"-h")) {
   printf("usage: echo urls | cuturl [options]\n");
   printf("usage: cuturl [options] url [options] [url]\n\n");
   printf("options: \n");
   for(i=0;long_opts[i];i++) {
    printf("        -%s|--%s\n",short_opts[i],long_opts[i]);
   }
   printf("To set default values use environment variables like: CUTURL_[OPTION]\n");
   return 2;
  }
 }
 while(1) {
  //
  // todo: add argument parsing
  //
  if(!using_stdin) flags=0;
  if(argc > 0) {
   for(argc--,argv++;argc>0;argc--,argv++) {
    for(i=0;long_opts[i];i++) {
     if(!strncmp(*argv,"--",2)) {
      if(!strcmp(*argv+2,long_opts[i])) {
       flags|=(1<<i);
       break;
      }
     }
     else if(**argv=='-') {
      if(*(*argv+1)==*short_opts[i]) {
       flags|=(1<<i);
       break;
      }
     }
    }
    if(long_opts[i]) continue;
    line=*argv;
    using_stdin=0;
    argc--;
    argv++;
    break;
   }
  }
  if(using_stdin) {
   line=malloc(size);
   malloced=1;
   if(!fgets(line,size,stdin)) {
    return 0;
   }
  }
  if(!line) return 0;
  for(i=0;line[i] && line[i] != '\n' && line[i] != '\r';i++);
  line[i]=0;

  //split at first single / into line and path
  for(i=0;line[i];i++) {
   if(line[i] == '/' && line[i+1] == '/') {
    i++;
    continue;
   }
   if(line[i] == '/') {
    line[i]=0;
    path=line+i+1;
    break;
   }
  }

  if(path) {
   if(strchr(path,'?')) {
    query_string=strchr(path,'?');
    *query_string=0;
    query_string++;
   }
  }

  if(query_string) {
   if(strchr(query_string,'#')) {
    fragment_id=strchr(query_string,'#');
    *fragment_id=0;
    fragment_id++;
   }
  }

  if(strstr(line,"://")) {
   scheme=line;
   domain=strstr(line,"://");
   *domain=0;
   domain+=3;
  } else {
   domain=line;
  }

  if(domain) {
   if(strchr(domain,'@')) {
    username=domain;
    domain=strchr(domain,'@');
    *domain=0;
    domain++;
   }
  }

  if(username) {
   if(strchr(username,':')) {
    password=strchr(username,':');
    *password=0;
    password++;
   }
  }

  if(domain) {
   if(strchr(domain,']')) {
    if(strchr(strchr(domain,']'),':')) {
     port=strchr(strchr(domain,']'),':');
     *port=0;
     port++;
    }
   } else {
    if(strchr(domain,':')) {
     port=strchr(domain,':');
     *port=0;
     port++;
    }
   }
  }

  if(domain) {
   if(strchr(domain,'?')) {
    query_string=strchr(domain,'?');
    *query_string=0;
    query_string++;
   }
  }

  if(!scheme && username) {
   scheme=username;
   username=password;
   password=0;
  }

  if(!domain) {
   fprintf(stderr,"how the fuck is this supposed to be a URL?\n");
   return 1;
  }
  // printf("scheme://username:password@domain:port/path?query_string#fragment_id\n\n");
  //let's set them to what'll get printed now...

  scheme=AorB(scheme,DEFAULT_SCHEME);
  username=AorB(username,DEFAULT_USERNAME);
  password=AorB(password,DEFAULT_PASSWORD);
  //domain=domain; doesn't change. heh.
  port=AorB(port,DEFAULT_PORT);
  path=AorB(path,DEFAULT_PATH);
  query_string=AorB(query_string,DEFAULT_QUERY_STRING);
  fragment_id=AorB(fragment_id,DEFAULT_FRAGMENT_ID);

  if(flags) {
   if(flags&F_SCHEME) printf("%s\n",scheme);
   if(flags&F_USERNAME) printf("%s\n",username);
   if(flags&F_PASSWORD) printf("%s\n",password);
   if(flags&F_DOMAIN) printf("%s\n",domain);
   if(flags&F_PORT) printf("%s\n",port);
   if(flags&F_PATH) printf("%s\n",path);
   if(flags&F_QUERY_STRING) printf("%s\n",query_string);
   if(flags&F_FRAGMENT_ID) printf("%s\n",fragment_id);
  } else {
   printf("scheme: %s\n",scheme);
   printf("username: %s\n",username);
   printf("password: %s\n",password);
   printf("domain: %s\n",domain);
   printf("port: %s\n",port);
   printf("path: %s\n",path);
   printf("query_string: %s\n",query_string);
   printf("fragment_id: %s\n",fragment_id);
  }
  if(malloced) {
   free(line);
   malloced=0;
  } else {
   line=0;
  }
 }
 return 0;
}