initial commit.

4 files changed, 378 insertions, 0 deletions

diff --git a/cuturl.c b/cuturl.c
new file mode 100644
index 0000000..3f09020
--- /dev/null
+++ b/cuturl.c
@@ -0,0 +1,191 @@
+#include <netdb.h>
+#include <stdio.h>
+#include <string.h>
+#include <stdlib.h>
+#include <errno.h>
+#include <unistd.h>
+#include <sys/wait.h>
+
+#include "url.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 AorB(a,b) ((a)?(a):(b))
+
+#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 *name[2];
+ char *line=0;
+ char sport[10];
+ char args[256];
+ struct servent *serv;
+ int i,j,c=0;
+ int size=1024;
+ int status;
+ char fixme=0;
+ char using_stdin=1;
+ char malloced=0;
+ struct url u;
+ if(argc > 1) {
+  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;
+  }
+ }
+ argv++;
+ argc--;
+ while(1) {
+  u.scheme=0;
+  u.username=0;
+  u.password=0;
+  u.domain=0;
+  u.port=0;
+  u.path=0;
+  u.query_string=0;
+  u.fragment_id=0;
+  if(!using_stdin) c=0;
+  if(argc >= 1) {
+   for(;argc>0;argc--,argv++) {
+    for(i=0;long_opts[i];i++) {
+     if(!strncmp(*argv,"--",2)) {
+      if(!strcmp(*argv+2,long_opts[i])) {
+       args[c]=1<<i;
+       c++;
+       break;
+      }
+     }
+    }
+    fixme=0;
+    if(**argv=='-' && argv[0][1] != '-') {
+     for(j=1;argv[0][j];j++) {
+      for(i=0;short_opts[i];i++) {
+       if(argv[0][j]==*short_opts[i]) {
+        args[c]=1<<i;
+        c++;
+        fixme=1;
+       }
+      }
+     }
+    }
+    if(fixme) continue;
+    if(long_opts[i]) continue;
+    //if we get here we are at data instead of flags. work on it.
+    line=*argv;
+    using_stdin=0;
+    argc--;
+    argv++;
+    break;
+   }
+  }
+  if(!argc && !line) {//if we are out of arguments and it didn't include data
+   using_stdin=1;
+  }
+  if(using_stdin) {
+   line=malloc(size+1);
+   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;
+
+  urlfromline(&u,line);
+
+  // printf("scheme://username:password@domain:port/path?query_string#fragment_id\n\n");
+  //let's set them to what'll get printed now...
+
+  u.scheme=AorB(u.scheme,AorB(getenv("CUTURL_SCHEME"),"DEFAULT"));
+  u.username=AorB(u.username,AorB(getenv("CUTURL_USERNAME"),"DEFAULT"));
+  u.password=AorB(u.password,AorB(getenv("CUTURL_PASSWORD"),"DEFAULT"));
+  u.domain=AorB(u.domain,AorB(getenv("CURURL_DOMAIN"),"DEFAULT"));
+  serv=getservbyname(u.scheme,strcmp(u.scheme,"udp")?"tcp":"udp");//gets default port for the scheme. http -> 80
+  if(serv) snprintf(sport,sizeof(sport)-1,"%d",ntohs(serv->s_port));
+  u.port=AorB(u.port,AorB(getenv("CUTURL_PORT"),(serv?sport:"DEFAULT")));
+  u.path=AorB(u.path,AorB(getenv("CUTURL_PATH"),"DEFAULT"));
+  u.query_string=AorB(u.query_string,AorB(getenv("CUTURL_QUERY_STRING"),"DEFAULT"));
+  u.fragment_id=AorB(u.fragment_id,AorB(getenv("CUTURL_FRAGMENT_ID"),"DEFAULT"));
+
+  if((name[0]=getenv("CUTURL__"))) {
+   setenv("CUTURL__SCHEME",u.scheme,1);
+   setenv("CUTURL__USERNAME",u.username,1);
+   setenv("CUTURL__PASSWORD",u.password,1);
+   setenv("CUTURL__DOMAIN",u.domain,1);
+   setenv("CUTURL__PORT",u.port,1);
+   setenv("CUTURL__PATH",u.path,1);
+   setenv("CUTURL__QUERY_STRING",u.query_string,1);
+   setenv("CUTURL__FRAGMENT_ID",u.fragment_id,1);
+   name[1]=0;
+   switch(fork()) {
+    case 0:
+     execv(name[0],name);
+     perror("execv");
+     return errno;
+    case -1:
+     perror("fork");
+     return errno;
+    default:
+     break;
+   }
+   wait(&status);
+  } else {
+   if(c) {
+    for(i=0;i<c;i++) {
+     if(args[i]&F_SCHEME) printf("%s\n",u.scheme);
+     if(args[i]&F_USERNAME) printf("%s\n",u.username);
+     if(args[i]&F_PASSWORD) printf("%s\n",u.password);
+     if(args[i]&F_DOMAIN) printf("%s\n",u.domain);
+     if(args[i]&F_PORT) printf("%s\n",u.port);
+     if(args[i]&F_PATH) printf("%s\n",u.path);
+     if(args[i]&F_QUERY_STRING) printf("%s\n",u.query_string);
+     if(args[i]&F_FRAGMENT_ID) printf("%s\n",u.fragment_id);
+    }
+   } else {
+    printf("scheme: %s\n",u.scheme);
+    printf("username: %s\n",u.username);
+    printf("password: %s\n",u.password);
+    printf("domain: %s\n",u.domain);
+    printf("port: %s\n",u.port);
+    printf("path: %s\n",u.path);
+    printf("query_string: %s\n",u.query_string);
+    printf("fragment_id: %s\n",u.fragment_id);
+   }
+  }
+  if(malloced) {
+   free(line);
+   malloced=0;
+   line=0;
+  } else {
+   line=0;//???
+  }
+ }
+ return 0;
+}
diff --git a/matchurl.c b/matchurl.c
new file mode 100644
index 0000000..aeb4aa4
--- /dev/null
+++ b/matchurl.c
@@ -0,0 +1,41 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include "url.h"
+
+#define LINE_LENGTH 1024
+
+int main(int argc,char *argv[]) {
+ int i;
+ struct url u;
+ char *line=malloc(LINE_LENGTH);
+ char copy[LINE_LENGTH];
+ if(argc < 3) {
+  printf("usage: matchurl [s|u|k|d|P|p|q|f] [string]\n");
+  return 1;
+ }
+ while(fgets(line,LINE_LENGTH-1,stdin)) {
+  if(strchr(line,'\r')) *strchr(line,'\r')=0;
+  if(strchr(line,'\n')) *strchr(line,'\n')=0;
+  strcpy(copy,line);
+  urlfromline(&u,line);
+  //use the character in argv[1] to match stdin against argv[2]. if match print whole line.
+  for(i=1;i<argc;i+=2) {
+   switch(argv[i][0]) {
+    case 's': if(u.scheme       && !strcmp(u.scheme,argv[i+1]))       printf("%s\n",copy);
+    case 'u': if(u.username     && !strcmp(u.username,argv[i+1]))     printf("%s\n",copy);
+    case 'k': if(u.password     && !strcmp(u.password,argv[i+1]))     printf("%s\n",copy);
+    case 'd': if(u.domain       && !strcmp(u.domain,argv[i+1]))       printf("%s\n",copy);
+    case 'P': if(u.port         && !strcmp(u.port,argv[i+1]))         printf("%s\n",copy);
+    case 'p': if(u.path         && !strcmp(u.path,argv[i+1]))         printf("%s\n",copy);
+    case 'q': if(u.query_string && !strcmp(u.query_string,argv[i+1])) printf("%s\n",copy);
+    case 'f': if(u.fragment_id  && !strcmp(u.fragment_id,argv[i+1]))  printf("%s\n",copy);
+     break;
+    default:
+     printf("unknown url part letter! %c\n",argv[i][0]);
+     return 0;
+   }
+  }
+ }
+ return 0;
+}
diff --git a/url.c b/url.c
new file mode 100644
index 0000000..3fd2dae
--- /dev/null
+++ b/url.c
@@ -0,0 +1,131 @@
+//#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
+*/
+
+#include "url.h"
+
+/*
+struct url {
+ char *scheme;
+ char *username;
+ char *password;
+ char *domain;
+ char *port;
+ char *path;
+ char *query_string;
+ char *fragment_id;
+}
+*/
+
+void urlfromline(struct url *u,char *line) {
+ int i;
+  //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;
+    u->path=line+i+1;
+    break;
+   }
+  }
+  if(u->path) {
+   if(strchr(u->path,'?')) {
+    u->query_string=strchr(u->path,'?');
+    *u->query_string=0;
+    u->query_string++;
+   }
+  }
+
+  if(u->query_string) {
+   if(strchr(u->query_string,'#')) {
+    u->fragment_id=strchr(u->query_string,'#');
+    *u->fragment_id=0;
+    u->fragment_id++;
+   }
+  }
+
+  if(strstr(line,"://")) {
+   u->scheme=line;
+   u->domain=strstr(line,"://");
+   *u->domain=0;
+   u->domain+=3;
+  } else {
+   u->domain=line;
+  }
+
+  if(u->domain) {
+   if(strchr(u->domain,'@')) {
+    u->username=u->domain;
+    u->domain=strchr(u->domain,'@');
+    *u->domain=0;
+    u->domain++;
+   }
+  }
+
+  if(u->username) {
+   if(strchr(u->username,':')) {
+    u->password=strchr(u->username,':');
+    *u->password=0;
+    u->password++;
+   }
+  }
+
+  if(u->domain) {
+   if(strchr(u->domain,']')) {//the end of an IPv6 address
+    if(strchr(strchr(u->domain,']'),':')) {
+     u->port=strchr(strchr(u->domain,']'),':');
+     if(u->port[1] == '?') {//for magnet links
+      u->port=0;
+     } else {
+      *u->port=0;
+      u->port++;
+     }
+    }
+   } else {
+    if(strchr(u->domain,':')) {
+     u->port=strchr(u->domain,':');
+     if(u->port[1] == '?') {//for magnet links
+      u->port=0;
+     } else {
+      *u->port=0;
+      u->port++;
+     }
+    }
+   }
+  }
+
+  if(u->domain) {//for magnet links.
+   if(strchr(u->domain,'?')) {
+    u->query_string=strchr(u->domain,'?');
+    *u->query_string=0;
+    u->query_string++;
+   }
+  }
+
+  if(strchr(u->domain,':') && !strchr(u->domain,']')) {//for scheme:?query_string
+   u->scheme=u->domain;
+   *strchr(u->scheme,':')=0;
+   u->domain=0;
+  }
+
+  if(!u->scheme && u->username) {//for mailto:
+   u->scheme=u->username;
+   u->username=u->password;
+   u->password=0;
+  }
+}
diff --git a/url.h b/url.h
new file mode 100644
index 0000000..5e9bea8
--- /dev/null
+++ b/url.h
@@ -0,0 +1,15 @@
+#ifndef URL_H
+#define URL_H
+struct url {
+ char *scheme;
+ char *username;
+ char *password;
+ char *domain;
+ char *port;
+ char *path;
+ char *query_string;
+ char *fragment_id;
+};
+
+void urlfromline(struct url *u,char *line);
+#endif