#include <stdio.h>
#include <string.h>
#include <unistd.h>
#include <stdlib.h>
#include <ctype.h>

typedef short bool;
#define TRUE (bool)1
#define FALSE (bool)0

typedef unsigned char byte;

#define ALLOC(type) (type *)malloc(sizeof(type))
#define SALLOC(s)   (char *)malloc(strlen(s)+1)
#define NALLOC(type,n) (type *)malloc((n)*sizeof(type))
#define STRCPY(D,S) D = SALLOC(S); strcpy(D,S);

/************ lclint macros ************/

#define streq(A,B) (strcmp(A,B) == 0)
#define strneq(A,B,C) (strncmp(A,B,C) == 0)
#define memeq(A,B,C) (memcmp(A,B,C) == 0)

/*********** debug macros *********/

#define DYN_DEBUG
#ifdef DEBUG
#define DEBUG_PRINT(X)   printf X
#define DEBUG_P(X)  printf(X)
#define DEBUG_P1(X,A1)  printf(X,A1)
#define DEBUG_P2(X,A1,A2)  printf(X,A1,A2)
#define DEBUG_P3(X,A1,A2,A3)  printf(X,A1,A2,A3)
#define DEBUG_P4(X,A1,A2,A3,A4)  printf(X,A1,A2,A3,A4)
#define DEBUG_P5(X,A1,A2,A3,A4,A5)  printf(X,A1,A2,A3,A4,A5)
#define BREAK { int i = i/0; } /* sneaky trick to remain in debugger */
#else
#ifdef DYN_DEBUG
bool option_debug = FALSE;
#define DEBUG_PRINT(X)   if (option_debug) printf X
#define DEBUG_P(X)  if (option_debug) printf(X)
#define DEBUG_P1(X,A1)  if (option_debug) printf(X,A1)
#define DEBUG_P2(X,A1,A2)  if (option_debug) printf(X,A1,A2)
#define DEBUG_P3(X,A1,A2,A3)  if (option_debug) printf(X,A1,A2,A3)
#define DEBUG_P4(X,A1,A2,A3,A4)  if (option_debug) printf(X,A1,A2,A3,A4)
#define DEBUG_P5(X,A1,A2,A3,A4,A5)  if (option_debug) printf(X,A1,A2,A3,A4,A5)
#define BREAK { int i = i/0; } /* sneaky trick to remain in debugger */
#else
#define DEBUG_PRINT(X)
#define DEBUG_P(X)  
#define DEBUG_P1(X,A1)  
#define DEBUG_P2(X,A1,A2)  
#define DEBUG_P3(X,A1,A2,A3)  
#define DEBUG_P4(X,A1,A2,A3,A4) 
#define DEBUG_P5(X,A1,A2,A3,A4,A5) 
#define BREAK { printf("\nInternal error, please report\n"); abort(); }
#endif
#endif

#define DEBUG_GN(X) /* fputc(X, fout) */

#define DO_DEBUG_PRINT(X)   printf X
#define DO_DEBUG_P(X)  printf(X)
#define DO_DEBUG_P1(X,A1)  printf(X,A1)
#define DO_DEBUG_P2(X,A1,A2)  printf(X,A1,A2)
#define DO_DEBUG_P3(X,A1,A2,A3)  printf(X,A1,A2,A3)
#define DO_DEBUG_P4(X,A1,A2,A3,A4)  printf(X,A1,A2,A3,A4)

/* sizes of buffers used during reading of HTML files: */
#define MAX_SF     80  /* max size of name of source HTML file name */
#define MAX_DF    600  /* max size of name of reference HTML file name */
#define MAX_HC    100  /* max size of HTML commands */

char df_buffer[MAX_DF+1];

/* URL of document */
char *document_URL = NULL,
     *server_URL = NULL, /* server part of document_URL (without last '/'). */
     *file_URL = NULL;   /* file part of document_URL (starting with '/'). */

bool is_URL(name) char *name;
/* returns TRUE if name is an URL. */
{   return    memeq(name, "news:", 5)
           || memeq(name, "http:", 5)
           || memeq(name, "file:", 5)
           || memeq(name, "ftp:", 4)
           || memeq(name, "wais:", 5)
           || memeq(name, "gopher:", 7)
           || memeq(name, "mailto:", 7)
           || memeq(name, "telnet:", 7);
}

bool is_html(name) char *name;
/* returns TRUE if name has .html extension */
{   return    streq(name + strlen(name) - 5, ".html")
           || streq(name + strlen(name) - 4, ".htm");
}



bool norm_URL(origin, file) char *origin, *file;
/* Normalizes the file name `file' appearing in HTML file `origin', with
   the following steps:
   1.  If `file' is empty, use assign `origin' to `file'.
       Else if `file' is not an URL and does not start with '/'
       then glue it together with directories in `origin'.
   2.  If `file' is not an URL glue it together with document URL.
   3.  If `file' starts with document URL, remove it.
*/
{   /* assume that origin: ['/'](<dn> '/')* <fn> */

   DEBUG_PRINT(("norm_URL(%s, %s) %s %s\n", 
                origin, file, server_URL, file_URL));
/* Step 1: */
    
    /* if file is empty, use origin: */
    if (file[0] == '\0')
    {   if (strlen(origin) < MAX_DF)
	    strcpy(file, origin);
	else
	    return FALSE;
    }
    /* if file is not an URL and does not start with '/'
       then glue it together with directories in origin:  */
    else if (file[0] != '/' && !is_URL(file))
    {   int i = strlen(origin);
	char *s = file;

        DEBUG_PRINT(("glue %s with %s", origin, file));
	
	/* remove file-name from origin: */
	while (i > 0 && origin[i - 1] != '/')
	    i--;
        DEBUG_PRINT((" : %s + %s\n", origin, file));
	/* cancel last directory in origin with '../': */
	while (   i > 1 && origin[i - 1] == '/'
               && s[0] == '.'
	       && (   (s[1] == '.' && (s[2] == '/' || s[2] == '\0'))
                   || s[1] == '/' || s[1] == '\0'))
            if (s[1] == '/')
                s += 2;
            else if (s[1] == '\0')
                s++;
            else 
	    {   s += s[2] == '/' ? 3 : 2;  
	        do
		    i--;
	        while (i > 0 && origin[i - 1] != '/');
	    }
	
	if (   i == 1 && origin[0] == '/'
	    && s[0] == '.' && s[1] == '.' /* && s[2] == '/'*/)
	    return FALSE;
	else if (i == 0)
	    strcpy(file, s);
	else if(i + strlen(s) < MAX_DF)
	{   memcpy(df_buffer, origin, i); 
	    strcpy(df_buffer + i, s);
	    strcpy(file, df_buffer);
	}
	else
	    return FALSE;
    }

    DEBUG_PRINT(("After step 1: %s\n", file));

/* Step 2. */
    
    if (   document_URL != NULL
	&& file[0] == '.' && file[1] == '.' && file[2] == '/') 
    {   int i = strlen(file_URL) - 1;
	char *s = file;

        DEBUG_PRINT(("glue %s + %s\n", file_URL, file));	
	/* assume that file_URL is of the form: '/' (<dn> '/')*  */
	while (i > 0 && s[0] == '.' && s[1] == '.' && s[2] == '/')
	{   s += 3;  
	    i--;
	    while (i > 0 && file_URL[i] != '/') 
	       i--;
            DEBUG_PRINT(("replace %s with %s\n", s, file_URL + i));
	}
        i += strlen(server_URL);
	if (i + 1 + strlen(s) < MAX_DF)
	{   memcpy(df_buffer, document_URL, i + 1); 
	    strcpy(df_buffer + i + 1, s);
	    strcpy(file, df_buffer);
	}
	else
	    return FALSE;
    }
    /* if file starts with '/' add server_URL: */
    else if (document_URL != NULL && file[0] == '/') 
	if (strlen(file) + strlen(server_URL) < MAX_DF)
	{   strcpy(df_buffer, server_URL);
	    strcat(df_buffer, file);
	    strcpy(file, df_buffer);
	}
	else 
	    return FALSE;

    DEBUG_PRINT(("After step 2: %s\n", file));

/* Step 3: */

    /* if URL starts with document URL, remove it: */
    if (   document_URL != NULL
	&& memeq(file, document_URL, strlen(document_URL)))
	strcpy(file, file + strlen(document_URL));

    DEBUG_PRINT(("After step 3: %s\n", file));

/* Step 4: */

    if (streq(file, "."))
        file[0] = '\0';

    /* if not URL and not .html, add index.html, when file exists */
    if (!is_URL(file) && !is_html(file) && strlen(file) + 12 < MAX_DF)
    {   strcpy(df_buffer, file);
        if (df_buffer[0] != '\0' && df_buffer[strlen(df_buffer)-1] != '/')
            strcat(df_buffer, "/");
        strcat(df_buffer, "index.html");
        DEBUG_PRINT(("Try: %s\n", df_buffer));
        if (access(df_buffer, R_OK) == 0)
            strcpy(file, df_buffer);
    }

    DEBUG_PRINT(("After step 4: %s\n", file));

/* Step 4: */

    return TRUE;
}

void accept_root_URL(char *URL)
{   int strlen_URL = strlen(URL);

    document_URL = NALLOC(char, strlen_URL + 2);
    strcpy(document_URL, URL);
    if (document_URL[strlen_URL - 1] != '/')
    {   document_URL[strlen_URL] = '/';
        document_URL[strlen_URL + 1] = '\0';
    }
    server_URL = NULL;
    {   int i;
        for (i = 0; URL[i] != '\0' && URL[i] != ':'; i++);
        if (URL[i] != '\0' && URL[i+1] == '/' && URL[i+2] == '/')
        {   i += 3;
            while (URL[i] != '\0' && URL[i] != '/')
                i++;
            if (URL[i] == '/')
                file_URL = document_URL + i;
            else
                file_URL = "/";
            URL[i] = '\0';
            server_URL = SALLOC(URL);
            strcpy(server_URL, URL);
        }
    }
    if (server_URL == NULL)
    {   
        fprintf(stderr, "%%html: illegal URL `%s'\n", document_URL);
        document_URL = NULL;
    }
}

typedef struct
{ char *from;
  char *to;
  char *state;
} rule_t;

rule_t *rules[10000];
int nrrules = 0;

void skip_spaces(FILE *fin, char *r_ch, char *command, int *r_i);
void skip_spaces(FILE *fin, char *r_ch, char *command, int *r_i)
{   char ch = *r_ch;
    int i = *r_i;

    /* skip spaces */
    while (!feof(fin) && (ch == ' ' || ch == '\n'))
    {
        command[i++] = ch;
        ch = (char)fgetc(fin);
    }

    /* process comments */
    while (!feof(fin) && ch == '-')
    {   
        command[i++] = ch;
        ch = (char)fgetc(fin);
        if (ch != '-')
            break;
        while(!feof(fin))
        {   command[i++] = ch;
            ch = (char)fgetc(fin);
            if (ch == '-' && i > 0 && command[i-1] == '-')
            {   command[i++] = ch; 
                ch = (char)fgetc(fin);
                break;
            }
        }

        /* skip spaces */
        while (!feof(fin) && (ch == ' ' || ch == '\n'))
        {
            command[i++] = ch;
            ch = (char)fgetc(fin);
        }
    }
    *r_ch = ch;
    *r_i = i;
}


bool update_extref( char *fn, FILE *fin, FILE *fout) 
{
   int body_pos = 0;
   bool forget_a_close = FALSE;
   bool file_has_changed = FALSE;

   char ch = (char)fgetc(fin);
   
   while (!feof(fin))
       if (ch == '<')
       {  int i = 0;
          char command[30];

          ch = (char)fgetc(fin);
          while( !feof(fin) && (isalpha(ch) || ch == '/' ))
          {   command[i++] = toupper(ch);
              ch = (char)fgetc(fin);
          }
          command[i] = '\0';
          if ( forget_a_close && streq( command, "/A" ) )
          {
              while( !feof(fin) && ch != '>' )
                  ch = (char)fgetc(fin);
              if ( ch == '>' )
                  ch = (char)fgetc(fin);
              forget_a_close = FALSE;
          }
          else if ( streq(command, "A") || streq(command, "IMG"))
          {
              char  a_command[1000];
              int a_i = 0;
              int attr_name_i;
              char attr_name[MAX_HC + 1];
              int attr_val_i;
              char attr_val[MAX_DF + 1];
              bool empty = TRUE;

              forget_a_close = FALSE;

              strcpy(a_command, command);
              a_i = strlen(a_command);

              while( !feof(fin) && ch != '>')
              {   bool found_is = FALSE;

                  skip_spaces(fin, &ch, a_command, &a_i);

                  if (feof(fin) || ch == '>')
                      break;

                  attr_name_i = i;
                  /* scan attribute in attr_name */
                  i = 0;
                  while (!feof(fin) && ch != '>' && ch != '=' && ch != ' '
                         && ch != '\n' && ch != '\t')
                  {   if (i < MAX_HC)
                          attr_name[i++] = tolower(ch);
                      a_command[a_i++] = ch;
                      ch = (char)fgetc(fin);
                  }
                  attr_name[i] = '\0';

                  /* skip = and spaces */
                  while (   !feof(fin)
                         && (   ch == ' ' || ch == '\n' || ch == '='
                             || ch == '\t'))
                  {   if (ch == '=')
                          found_is = TRUE;
                      a_command[a_i++] = ch;
                      ch = (char)fgetc(fin);
                  }

                  attr_val_i = a_i ;
                  /* scan string into name */
                  i = 0;

                  if (found_is)
                  {   bool is_quoted = ch == '"';

                      if (is_quoted)
                      {
                         a_command[a_i++] = ch; 
                         ch = (char)fgetc(fin);
                      }
                      while(   !feof(fin) && ch != '>' && ch != '"'
                            && (is_quoted || ch != ' '))
                      {   if (i < MAX_DF)
                              attr_val[i++] = ch;
                          a_command[a_i++] = ch;
                          ch = (char)fgetc(fin);
                      }
                      if (ch == '\"')
                      {
                         a_command[a_i++] = ch;
                         ch = (char)fgetc(fin);
                      }
                  }
                  attr_val[i] = '\0';

                  if (streq(command, "A") && streq(attr_name, "name"))
                  {
                     empty = FALSE;
                     a_i = attr_val_i;
                     a_command[a_i++] = '"';
                     strcpy(a_command + a_i, attr_val);
                     a_i += strlen(attr_val);
                     a_command[a_i++] = '"';
                  }
                  else if(  streq(command, "A") 
                          ? streq(attr_name, "href")
                          : streq(attr_name, "src"))
                  {
                     int j;
                     char *broken = "Broken.html?";

                     if (is_URL(attr_val) || !strncmp(attr_val, broken, strlen(broken)))
                     {
                     	char *from = attr_val;
                     	char new_attr_val[1000];

						strcpy(new_attr_val, attr_val);
						
                     	if (!strncmp(from, broken, strlen(broken)))
                     	{
                     		char *s = from + strlen(broken);
							char *t = strstr(s, "|");
							if (t != NULL)
								from = t+1;
							else
								from = s;
                     	}
                     	 
                        for (j = 0; j < nrrules; j++)
                            if (streq(rules[j]->from, from))
                            {
								if (!strcmp(rules[j]->state, "ok"))
								{
									if (rules[j]->to != NULL)
										strcpy(new_attr_val, rules[j]->to); 
									else
										strcpy(new_attr_val, from);
								}
								else
								{
									sprintf(new_attr_val, "Broken.html?%s|%s",
											rules[j]->state, from);
								}
								break;
							}
						if (strcmp(attr_val, new_attr_val))
						{
							printf("In file %s changed %s into %s\n", 
								   fn, attr_val, new_attr_val);
							file_has_changed = TRUE;
						}
                        a_i = attr_val_i;
                        a_command[a_i++] = '"';
                        strcpy(a_command + a_i, new_attr_val);
                        a_i += strlen(new_attr_val);
                        a_command[a_i++] = '"';
                        empty = FALSE;
                     }
                     else
                     {
                        a_i = attr_val_i;
                        a_command[a_i++] = '"';
                        strcpy(a_command + a_i, attr_val);
                        a_i += strlen(attr_val);
                        a_command[a_i++] = '"';
                        empty = FALSE;
                     }
                 } 
              }
              a_command[a_i] = '\0';
              if (!empty)
                  fprintf(fout, "<%s>", a_command);
              else if (streq(command, "A"))
                  forget_a_close = TRUE;
              if ( ch == '>' )
                  ch = (char)fgetc(fin);
          }
          else
          {
              if ( body_pos == 1 )
              {
                  body_pos = 0;
              }
              if ( streq( command, "BODY" ) )
                 body_pos = 1; 
              fprintf(fout, "<%s", command);
              while( !feof(fin) && ch != '>' )
              {
                  fputc(ch, fout);
                  ch = (char)fgetc(fin);
              }
              fputc('>', fout);
              if ( ch == '>' )
                  ch = (char)fgetc(fin);
          }
       }
       else
       {
           fputc(ch, fout);
           ch = (char)fgetc(fin);
       }
	return file_has_changed;       
}

/* Information of a file: */
typedef struct file_T
{   struct file_T *next;
    char *name;
} file_t;

/* List of all files: */
file_t *the_files = NULL;

file_t *find_file(file) char *file;
/* Returns pointer to HTML file record with the name `file'.
   If such a record did not exist in the list, it is added
   alphabetically on the file name.
*/
{   file_t **p_file = &the_files;

    while (*p_file != NULL && stricmp((*p_file)->name, file) < 0)
	p_file = &(*p_file)->next;

    if (*p_file == NULL || stricmp((*p_file)->name, file))
    {   file_t *n = ALLOC(file_t);
        DEBUG_PRINT(("FILEADDED\n"));
		n->next = *p_file;
		n->name = SALLOC(file);
		strcpy(n->name, file);
		*p_file = n;
    }

    return *p_file;
}


void read_dir(void)
{
  FILE *f;
  char fn[400];

  /*fprintf(stderr, "Scan directory\n");
    system("dir \\www >\\www\\compare\\dir.txt");
    fprintf(stderr, "Ready\n");*/

  f = fopen("compare\\dir.txt", "r");
  if (f==NULL)
    return;

  while (fgets(fn, 399, f))
  {
    file_t *file;	  
    if (fn[strlen(fn)-1]=='\n')
      fn[strlen(fn)-1] = '\0';
    if (fn[16] == ' ' && strlen(fn) >= 45)  	  
    {
      file = find_file(fn+44);
    }  
  }

  fclose(f);
}    

void read_states()
{
	FILE *f = fopen("compare\\links.txt", "r");
	char buffer[10000];
	
	if (f == NULL)
	{	printf("file compare\\links.txt is missing\n");
		return;
	}
		
	fgets(buffer, 9999, f);
	fgets(buffer, 9999, f);
	while (!feof(f) && *buffer != '\0')
	{
		char *s = strstr(buffer, "\t");
		char *t = strstr(s+1, "\t");
		char *fn, *state;
		
		*s = '\0';
		if (t)
			*t = '\0';
		
		fn = buffer;
		state = s+1;
		
		rules[nrrules] = ALLOC(rule_t);
		STRCPY(rules[nrrules]->from, fn);
		STRCPY(rules[nrrules]->state, state);
		nrrules++;
		
		fgets(buffer, 9999, f);
	}

	fclose(f);
}

void read_redirects()
{
	FILE *f = fopen("compare\\redirects.txt", "r");
	char buffer[10000];
	
	if (f == NULL)
	{	printf("File compare\\redirects.txt is missing\n");
		return;
	}
	
	fgets(buffer, 9999, f);
	while (!feof(f))
	{
		char *s = strstr(buffer, "\t");
		char *t = strstr(buffer, "\n");
		char *from, *to;
		int i;
		
		*s = '\0';
		if (t)
			*t = '\0';
		
		from = buffer;
		to = s+1;
		
		for (i = 0; i < nrrules; i++)
		{
			if (!strcmp(rules[i]->from, from))
			{
				if (*to == '#' && !strcmp(rules[i]->state, "ok"))
				{
					STRCPY(rules[i]->state, to + 1);
				}
				else
				{
					STRCPY(rules[i]->to, to);
				}
				break;
			}
		}
		if (i == nrrules)
			printf("redirect %s not found\n", from);
			
		fgets(buffer, 9999, f);
	}

	fclose(f);
}

int main (int argc, char *argv[])
{
	read_dir();
	read_states();
	read_redirects();
   
/*   
	{
   		int i;
   		
   		for (i = 0; i < nrrules; i++)
   			printf("%s %s %s\n", rules[i]->from, rules[i]->to ? rules[i]->to : "",
   					rules[i]->state);
   	}
*/   	
   	{
   		file_t *file;
   		
   		for (file = the_files; file != NULL; file = file->next)
   		    if (is_html(file->name))
   			{
   				FILE *f = fopen(file->name, "r");
   				char new_name[100];
   				FILE *g = fopen("compare\\tempconverted.html", "w");
   				sprintf(new_name, "converted\\%s", file->name);
   			
   				if (f != NULL && g != NULL)
   				{
   					bool update = update_extref(file->name, f, g);
   					
   					fclose(f);
   					fclose(g);
   					if (update)
					{ 	f = fopen("compare\\tempconverted.html", "r");
 						g = fopen(new_name, "w");
 						if (f != NULL && g != NULL)  					
   						{
   							char ch = (char)fgetc(f);
   							while (!feof(f))
   							{
   								fputc(ch, g);
   								ch = (char)fgetc(f);
   							}
   						}
   					}
	   			}
   				else
   					printf("Failed to open %s\n", file->name);
   					
   				if (f != NULL)
   					fclose(f);
   				if (g != NULL)
   					fclose(g);	
   		}
   	}
}