bool lookfor(char *tag,FILE *fp)
{
char tagname[256];
for(;;) {
if (!nexttag(tagname,fp)) return false;
if (strcmp(tag,tagname)==0) return true;
skipinfo(fp,0);
} /* forever */
return false;
} /* lookfor */
int lookfor2(char *tag,char *tag2,FILE *fp)
{
char tagname[256];
for(;;) {
if (!nexttag(tagname,fp)) return 0;
if (strcmp(tag,tagname)==0) return 1;
if (strcmp(tag2,tagname)==0) return 2;
skipinfo(fp,0);
} /* forever */
return 0;
} /* lookfor2 */
/*
* Main command processor.
* Accept and execute commands until a quit command.
*/
void
commands(void)
{
int c = 0;
int action;
char *cbuf;
int newaction;
int save_search_type;
char *extra;
char tbuf[2];
PARG parg;
IFILE old_ifile;
IFILE new_ifile;
char *tagfile;
search_type = SRCH_FORW;
wscroll = (sc_height + 1) / 2;
newaction = A_NOACTION;
for (;;) {
mca = 0;
cmd_accept();
number = 0;
curropt = NULL;
/*
* See if any signals need processing.
*/
if (sigs) {
psignals();
if (quitting)
quit(QUIT_SAVED_STATUS);
}
/*
* Display prompt and accept a character.
*/
cmd_reset();
prompt();
if (sigs)
continue;
if (newaction == A_NOACTION)
c = getcc();
again:
if (sigs)
continue;
if (newaction != A_NOACTION) {
action = newaction;
newaction = A_NOACTION;
} else {
/*
* If we are in a multicharacter command, call mca_char.
* Otherwise we call fcmd_decode to determine the
* action to be performed.
*/
if (mca)
switch (mca_char(c)) {
case MCA_MORE:
/*
* Need another character.
*/
c = getcc();
goto again;
case MCA_DONE:
/*
* Command has been handled by mca_char.
* Start clean with a prompt.
*/
continue;
case NO_MCA:
/*
* Not a multi-char command
* (at least, not anymore).
*/
break;
}
/*
* Decode the command character and decide what to do.
*/
if (mca) {
/*
* We're in a multichar command.
* Add the character to the command buffer
* and display it on the screen.
* If the user backspaces past the start
* of the line, abort the command.
*/
if (cmd_char(c) == CC_QUIT || len_cmdbuf() == 0)
continue;
cbuf = get_cmdbuf();
} else {
/*
* Don't use cmd_char if we're starting fresh
* at the beginning of a command, because we
* don't want to echo the command until we know
* it is a multichar command. We also don't
* want erase_char/kill_char to be treated
* as line editing characters.
*/
tbuf[0] = (char)c;
tbuf[1] = '\0';
cbuf = tbuf;
}
extra = NULL;
action = fcmd_decode(cbuf, &extra);
/*
* If an "extra" string was returned,
* process it as a string of command characters.
*/
if (extra != NULL)
ungetsc(extra);
}
/*
* Clear the cmdbuf string.
* (But not if we're in the prefix of a command,
* because the partial command string is kept there.)
*/
if (action != A_PREFIX)
cmd_reset();
switch (action) {
case A_DIGIT:
/*
* First digit of a number.
*/
start_mca(A_DIGIT, ":", (void*)NULL, CF_QUIT_ON_ERASE);
goto again;
case A_F_WINDOW:
/*
* Forward one window (and set the window size).
*/
if (number > 0)
swindow = (int)number;
/* FALLTHRU */
case A_F_SCREEN:
/*
* Forward one screen.
*/
if (number <= 0)
number = get_swindow();
cmd_exec();
if (show_attn)
set_attnpos(bottompos);
forward((int)number, 0, 1);
break;
case A_B_WINDOW:
/*
* Backward one window (and set the window size).
*/
if (number > 0)
swindow = (int)number;
/* FALLTHRU */
case A_B_SCREEN:
/*
* Backward one screen.
*/
if (number <= 0)
number = get_swindow();
cmd_exec();
backward((int)number, 0, 1);
break;
case A_F_LINE:
/*
* Forward N (default 1) line.
*/
if (number <= 0)
number = 1;
cmd_exec();
if (show_attn == OPT_ONPLUS && number > 1)
set_attnpos(bottompos);
forward((int)number, 0, 0);
break;
case A_B_LINE:
/*
* Backward N (default 1) line.
*/
if (number <= 0)
number = 1;
cmd_exec();
backward((int)number, 0, 0);
break;
case A_F_SKIP:
/*
* Skip ahead one screen, and then number lines.
*/
if (number <= 0) {
number = get_swindow();
} else {
number += get_swindow();
}
cmd_exec();
if (show_attn == OPT_ONPLUS)
set_attnpos(bottompos);
forward((int)number, 0, 1);
break;
case A_FF_LINE:
/*
* Force forward N (default 1) line.
*/
if (number <= 0)
number = 1;
cmd_exec();
if (show_attn == OPT_ONPLUS && number > 1)
set_attnpos(bottompos);
forward((int)number, 1, 0);
break;
case A_BF_LINE:
/*
* Force backward N (default 1) line.
*/
if (number <= 0)
number = 1;
cmd_exec();
backward((int)number, 1, 0);
break;
case A_FF_SCREEN:
/*
* Force forward one screen.
*/
if (number <= 0)
number = get_swindow();
cmd_exec();
if (show_attn == OPT_ONPLUS)
set_attnpos(bottompos);
forward((int)number, 1, 0);
break;
case A_F_FOREVER:
/*
* Forward forever, ignoring EOF.
*/
newaction = forw_loop(0);
break;
case A_F_UNTIL_HILITE:
newaction = forw_loop(1);
break;
case A_F_SCROLL:
/*
* Forward N lines
* (default same as last 'd' or 'u' command).
*/
if (number > 0)
wscroll = (int)number;
cmd_exec();
if (show_attn == OPT_ONPLUS)
set_attnpos(bottompos);
forward(wscroll, 0, 0);
break;
case A_B_SCROLL:
/*
* Forward N lines
* (default same as last 'd' or 'u' command).
*/
if (number > 0)
wscroll = (int)number;
cmd_exec();
backward(wscroll, 0, 0);
break;
case A_FREPAINT:
/*
* Flush buffers, then repaint screen.
* Don't flush the buffers on a pipe!
*/
clear_buffers();
/* FALLTHRU */
case A_REPAINT:
/*
* Repaint screen.
*/
cmd_exec();
repaint();
break;
case A_GOLINE:
/*
* Go to line N, default beginning of file.
*/
if (number <= 0)
number = 1;
cmd_exec();
jump_back(number);
break;
case A_PERCENT:
/*
* Go to a specified percentage into the file.
*/
if (number < 0) {
number = 0;
fraction = 0;
}
if (number > 100) {
number = 100;
fraction = 0;
}
cmd_exec();
jump_percent((int)number, fraction);
break;
case A_GOEND:
/*
* Go to line N, default end of file.
*/
cmd_exec();
if (number <= 0)
jump_forw();
else
jump_back(number);
break;
case A_GOPOS:
/*
* Go to a specified byte position in the file.
*/
cmd_exec();
if (number < 0)
number = 0;
jump_line_loc((off_t) number, jump_sline);
break;
case A_STAT:
/*
* Print file name, etc.
*/
if (ch_getflags() & CH_HELPFILE)
break;
cmd_exec();
parg.p_string = eq_message();
error("%s", &parg);
break;
case A_VERSION:
/*
* Print version number, without the "@(#)".
*/
cmd_exec();
dispversion();
break;
case A_QUIT:
/*
* Exit.
*/
if (curr_ifile != NULL_IFILE &&
ch_getflags() & CH_HELPFILE) {
/*
* Quit while viewing the help file
* just means return to viewing the
* previous file.
*/
hshift = save_hshift;
if (edit_prev(1) == 0)
break;
}
if (extra != NULL)
quit(*extra);
quit(QUIT_OK);
break;
/*
* Define abbreviation for a commonly used sequence below.
*/
#define DO_SEARCH() \
if (number <= 0) number = 1; \
mca_search(); \
cmd_exec(); \
multi_search(NULL, (int)number);
case A_F_SEARCH:
/*
* Search forward for a pattern.
* Get the first char of the pattern.
*/
search_type = SRCH_FORW;
if (number <= 0)
number = 1;
mca_search();
c = getcc();
goto again;
case A_B_SEARCH:
/*
* Search backward for a pattern.
* Get the first char of the pattern.
*/
search_type = SRCH_BACK;
if (number <= 0)
number = 1;
mca_search();
c = getcc();
goto again;
case A_FILTER:
search_type = SRCH_FORW | SRCH_FILTER;
mca_search();
c = getcc();
goto again;
case A_AGAIN_SEARCH:
/*
* Repeat previous search.
*/
DO_SEARCH();
break;
case A_T_AGAIN_SEARCH:
/*
* Repeat previous search, multiple files.
*/
search_type |= SRCH_PAST_EOF;
DO_SEARCH();
break;
case A_REVERSE_SEARCH:
/*
* Repeat previous search, in reverse direction.
*/
save_search_type = search_type;
search_type = SRCH_REVERSE(search_type);
DO_SEARCH();
search_type = save_search_type;
break;
case A_T_REVERSE_SEARCH:
/*
* Repeat previous search,
* multiple files in reverse direction.
*/
save_search_type = search_type;
search_type = SRCH_REVERSE(search_type);
search_type |= SRCH_PAST_EOF;
DO_SEARCH();
search_type = save_search_type;
break;
case A_UNDO_SEARCH:
undo_search();
break;
case A_HELP:
/*
* Help.
*/
if (ch_getflags() & CH_HELPFILE)
break;
if (ungot != NULL || unget_end) {
error(less_is_more
? "Invalid option -p h"
: "Invalid option ++h",
NULL_PARG);
break;
}
cmd_exec();
save_hshift = hshift;
hshift = 0;
(void) edit(FAKE_HELPFILE);
break;
case A_EXAMINE:
/*
* Edit a new file. Get the filename.
*/
if (secure) {
error("Command not available", NULL_PARG);
break;
}
start_mca(A_EXAMINE, "Examine: ", ml_examine, 0);
c = getcc();
goto again;
case A_VISUAL:
/*
* Invoke an editor on the input file.
*/
if (secure) {
error("Command not available", NULL_PARG);
break;
}
if (ch_getflags() & CH_HELPFILE)
break;
if (strcmp(get_filename(curr_ifile), "-") == 0) {
error("Cannot edit standard input", NULL_PARG);
break;
}
if (curr_altfilename != NULL) {
error("WARNING: This file was viewed via "
"LESSOPEN", NULL_PARG);
}
start_mca(A_SHELL, "!", ml_shell, 0);
/*
* Expand the editor prototype string
* and pass it to the system to execute.
* (Make sure the screen is displayed so the
* expansion of "+%lm" works.)
*/
make_display();
cmd_exec();
lsystem(pr_expand(editproto, 0), NULL);
break;
case A_NEXT_FILE:
/*
* Examine next file.
*/
if (ntags()) {
error("No next file", NULL_PARG);
break;
}
if (number <= 0)
number = 1;
if (edit_next((int)number)) {
if (get_quit_at_eof() && eof_displayed() &&
!(ch_getflags() & CH_HELPFILE))
quit(QUIT_OK);
parg.p_string = (number > 1) ? "(N-th) " : "";
error("No %snext file", &parg);
}
break;
case A_PREV_FILE:
/*
* Examine previous file.
*/
if (ntags()) {
error("No previous file", NULL_PARG);
break;
}
if (number <= 0)
number = 1;
if (edit_prev((int)number)) {
parg.p_string = (number > 1) ? "(N-th) " : "";
error("No %sprevious file", &parg);
}
break;
case A_NEXT_TAG:
if (number <= 0)
number = 1;
tagfile = nexttag((int)number);
if (tagfile == NULL) {
error("No next tag", NULL_PARG);
break;
}
if (edit(tagfile) == 0) {
off_t pos = tagsearch();
if (pos != -1)
jump_loc(pos, jump_sline);
}
break;
case A_PREV_TAG:
if (number <= 0)
number = 1;
tagfile = prevtag((int)number);
if (tagfile == NULL) {
error("No previous tag", NULL_PARG);
break;
}
if (edit(tagfile) == 0) {
off_t pos = tagsearch();
if (pos != -1)
jump_loc(pos, jump_sline);
}
break;
case A_INDEX_FILE:
/*
* Examine a particular file.
*/
if (number <= 0)
number = 1;
if (edit_index((int)number))
error("No such file", NULL_PARG);
break;
case A_REMOVE_FILE:
if (ch_getflags() & CH_HELPFILE)
break;
old_ifile = curr_ifile;
new_ifile = getoff_ifile(curr_ifile);
if (new_ifile == NULL_IFILE) {
ring_bell();
break;
}
if (edit_ifile(new_ifile) != 0) {
reedit_ifile(old_ifile);
break;
}
del_ifile(old_ifile);
break;
case A_OPT_TOGGLE:
optflag = OPT_TOGGLE;
optgetname = FALSE;
mca_opt_toggle();
c = getcc();
goto again;
case A_DISP_OPTION:
/*
* Report a flag setting.
*/
optflag = OPT_NO_TOGGLE;
optgetname = FALSE;
mca_opt_toggle();
c = getcc();
goto again;
case A_FIRSTCMD:
/*
* Set an initial command for new files.
*/
start_mca(A_FIRSTCMD, "+", NULL, 0);
c = getcc();
goto again;
case A_SHELL:
/*
* Shell escape.
*/
if (secure) {
error("Command not available", NULL_PARG);
break;
}
start_mca(A_SHELL, "!", ml_shell, 0);
c = getcc();
goto again;
case A_SETMARK:
/*
* Set a mark.
*/
if (ch_getflags() & CH_HELPFILE)
break;
start_mca(A_SETMARK, "mark: ", (void*)NULL, 0);
c = getcc();
if (c == erase_char || c == erase2_char ||
c == kill_char || c == '\n' || c == '\r')
break;
setmark(c);
break;
case A_GOMARK:
/*
* Go to a mark.
*/
start_mca(A_GOMARK, "goto mark: ", (void*)NULL, 0);
c = getcc();
if (c == erase_char || c == erase2_char ||
c == kill_char || c == '\n' || c == '\r')
break;
cmd_exec();
gomark(c);
break;
case A_PIPE:
if (secure) {
error("Command not available", NULL_PARG);
break;
}
start_mca(A_PIPE, "|mark: ", (void*)NULL, 0);
c = getcc();
if (c == erase_char || c == erase2_char ||
c == kill_char)
break;
if (c == '\n' || c == '\r')
c = '.';
if (badmark(c))
break;
pipec = c;
start_mca(A_PIPE, "!", ml_shell, 0);
c = getcc();
goto again;
case A_B_BRACKET:
case A_F_BRACKET:
start_mca(action, "Brackets: ", (void*)NULL, 0);
c = getcc();
goto again;
case A_LSHIFT:
if (number > 0)
shift_count = number;
else
number = (shift_count > 0) ?
shift_count : sc_width / 2;
if (number > hshift)
number = hshift;
hshift -= number;
screen_trashed = 1;
break;
case A_RSHIFT:
if (number > 0)
shift_count = number;
else
number = (shift_count > 0) ?
shift_count : sc_width / 2;
hshift += number;
screen_trashed = 1;
break;
case A_PREFIX:
/*
* The command is incomplete (more chars are needed).
* Display the current char, so the user knows
* what's going on, and get another character.
*/
if (mca != A_PREFIX) {
cmd_reset();
start_mca(A_PREFIX, " ", (void*)NULL,
CF_QUIT_ON_ERASE);
(void) cmd_char(c);
}
c = getcc();
goto again;
case A_NOACTION:
break;
default:
ring_bell();
break;
}
}
}
int main(int argc, char *argv[]) {
char logfile[BUFFSIZE], filesdir[BUFFSIZE], collectcmd[BUFFSIZE], lscmd[BUFFSIZE], datfile[BUFFSIZE], imgfile[BUFFSIZE];
FILE *master, *page;
FILE *sysout;
char pageline[BUFFSIZE], buff[BUFFSIZE], scmd[BUFFSIZE], pagefile[BUFFSIZE];
char pageurl[BUFFSIZE], fullurl[BUFFSIZE], *url;
char urldone;
pid_t wgetchild;
time_t waitstart;
char *data;
unsigned masterlength, masterloc;
unsigned pageadds;
long tester;
int p;
int firstentry;
if (argc != 3 && argc != 4) {
printf("Usage: web-collage <file> <program>\n");
exit(-2);
}
/* Fill out the commands with name */
sprintf(logfile, LOG_FILE, argv[1]);
sprintf(filesdir, FILES_DIR, argv[1]);
sprintf(collectcmd, COLLECT_CMD, argv[1], argv[1]);
sprintf(lscmd, LS_CMD, argv[1]);
sprintf(datfile, DAT_FILE, argv[1]);
sprintf(imgfile, IMG_FILE, argv[1]);
/* does the temp directory already exist? */
mkdir(filesdir, S_IRWXU);
/* Open File */
if (!(master = fopen(datfile, "r"))) {
printf("Creating File...\n");
master = fopen(datfile, "w+");
if (!master) {
perror("creating master file");
exit(-1);
}
fprintf(master, "%s\n", argv[3]); /* First URL */
fclose(master);
} else
fclose(master);
printf("Reading File...\n");
srand48(time(NULL));
master = waitreadmaster(datfile, 0, 0, 0); /* READ Lock! */
fseek(master, 0, SEEK_END);
masterlength = ftell(master);
freemaster(datfile, master, 0, 0); /* Unlock! */
int iter = 0;
while (iter++ < 3) {
/* Get random URL */
masterloc = lrand48() % masterlength;
firstentry = 0;
/* READ Lock! */
master = waitreadmaster(datfile, masterloc, masterloc, BUFFSIZE);
fgets(pageurl, BUFFSIZE, master); /* skip to line after random char */
do
if (!fgets(pageurl, BUFFSIZE - strlen(collectcmd) - 1, master)) {
rewind(master);
firstentry = 1;
fgets(pageurl, BUFFSIZE - strlen(collectcmd) - 1, master);
}
while (pageurl[0] == '\t' || pageurl[0] == '\n');
if (!firstentry) {
/* Invalidate URL */
/* READ->WRITE Lock! */
master = waitreadtowrite(datfile, master, masterloc, BUFFSIZE);
fseek(master, -(strlen(pageurl) + 0), SEEK_CUR);
voidline(master);
}
freemaster(datfile, master, masterloc, BUFFSIZE);
/* Unlock! */
pageurl[strlen(pageurl) - 1] = '\0';
printf("\nGetting %s\n", pageurl);
/* Get webpage, checking for any redirection */
sprintf(scmd, "%s \"%s\"", collectcmd, pageurl);
waitstart = time(NULL);
if (!(wgetchild = fork())) {
system(scmd);
exit(0);
}
while (validpid(wgetchild) && waitstart + 5 > time(NULL));
if (validpid(wgetchild))
kill(wgetchild, 9);
wait(NULL);
sysout = fopen(logfile, "r");
if (!sysout)
continue; /* File does not exist */
while (fgets(buff, BUFFSIZE, sysout)) {
printf("%s", buff);
if (!strncmp(buff, "Location:", 9)) {
if (strchr(buff + 10, ' '))
*strchr(buff + 10, ' ') = '\0';
strcpy(pageurl, buff + 10);
}
}
fclose(sysout);
unlink(logfile);
/* Collect new links */
sysout = popen(lscmd, "r");
if (!sysout) {
perror("searching files directory");
exit(-3);
}
strcpy(pagefile, filesdir);
if (fgets(pagefile + strlen(pagefile), BUFFSIZE, sysout) != NULL) {
pagefile[strlen(pagefile) - 1] = '\0'; /* remove newline */
page = fopen(pagefile, "r");
addfilename(pageurl, BUFFSIZE, pagefile);
pageadds = 0;
/* Write Additional URLs */
rewind(master);
if (findlikes(page) || (rand() % KEEP_PROB)) { /* possibly just throw away */
while (nexttag(pageline, BUFFSIZE, page))
if (url = geturl(pageline)) {
if (strncasecmp(url, "http://", 7)) {
if (!getpath(fullurl, BUFFSIZE, url, pageurl))
continue;
} else
strcpy(fullurl, url);
/* search for place in file to place */
urldone = 0;
if (strlen(fullurl) > BUFFSIZE / 2)
continue; /* too big, don't add */
if (rand() % (pageadds + 1) > KEEP_PROB)
continue; /* add fewer urls as more on page */
if (strstr(fullurl, ".com") && rand() % DROP_PROB)
continue; /* only chance to add a .com link */
if ((strstr(fullurl, "yahoo.com") ||
strstr(fullurl, "www.google.com")) && rand() % DROP_PROB)
continue; /* almost no yahoo.com's, www.google.com's */
for (p = 0; p < DISLIKE_CNT; p++)
if (strstr(fullurl, dislikes[p])) /* just skip */
continue;
/* READ Lock! */
master = waitreadmaster(datfile, 0, 0, 0);
while (fgets(buff, BUFFSIZE, master)) {
if (!strncmp(buff, fullurl, strlen(fullurl)) ||
(!strncmp(buff, fullurl, strchr(buff + strlen("http://"), '/')
- buff) && !(rand() % HOST_CROWD))) {
urldone = 1;
freemaster(datfile, master, 0, 0);
/* Unlock! */
break;
}
if (buff[0] == '\t' || buff[0] == '\n')
if (strlen(buff) >= strlen(fullurl) + 1) {
masterloc = ftell(master);
freemaster(datfile, master, 0, 0);
/* Unlock! */
printf("Adding %s\n", fullurl);
pageadds++;
/* WRITE Lock! */
waitwritemaster(datfile, masterloc - strlen(buff),
masterloc - strlen(buff), BUFFSIZE);
fprintf(master, "%s\n", fullurl);
if (strlen(buff) >= strlen(fullurl) + 2)
fputc('\t', master);
fseek(master, 0, SEEK_END);
masterlength = ftell(master);
freemaster(datfile, master,
masterloc - strlen(buff), BUFFSIZE);
/* Unlock! */
urldone = 1;
break;
}
}
if (!urldone) {
if (ftell(master) > MAX_SIZE) { /* too big, just remove entries */
fseek(master, lrand48() % ftell(master), SEEK_SET);
fgets(buff, BUFFSIZE, master);
urldone = 1; /* flag that was greater than MAX_SIZE */
}
masterloc = ftell(master);
freemaster(datfile, master, 0, 0);
/* Unlock! */
printf("Adding %s\n", fullurl);
pageadds++;
/* WRITE Lock! */
waitwritemaster(datfile, masterloc, masterloc - 2, BUFFSIZE);
fprintf(master, "%s\n", fullurl);
if (urldone)
voidline(master);
fseek(master, 0, SEEK_END);
masterlength = ftell(master);
freemaster(datfile, master, masterloc - 2, BUFFSIZE);
/* Unlock! */
}
}
}
fclose(page);
/* Move file to final destination */
sprintf(scmd, "%s \"%s\" \"%s\"", argv[2], argv[1], pagefile);
system(scmd);
sleep(WAIT_TIME);
sprintf(scmd, "touch %sREMOVE.tmp", filesdir);
system(scmd);
sprintf(scmd, "rm %s*", filesdir);
system(scmd);
}
pclose(sysout);
}
}