static int __cdecl htsshow_checkhtml(t_hts_callbackarg *carg, httrackp *opt, char* html,int len,const char* url_address,const char* url_file) {
struct hts_proj *proj = CALLBACKARG_USERDEF(carg);
if (strstr(url_address, "dianying.yisou.com")) {
fprintf(stderr, "Parsing html file: http://%s%s [%ldk]\n",url_address, url_file, sizeof(html));
}
return 1;
}
static int process(t_hts_callbackarg * carg, httrackp * opt, char *html,
int len, const char *address, const char *filename) {
t_my_userdef *userdef = (t_my_userdef *) CALLBACKARG_USERDEF(carg);
/*char * const stringfilter = userdef->stringfilter; */
char **const stringfilters = userdef->stringfilters;
/* */
int i = 0;
int getIt = 0;
char *pos;
/* Call parent functions if multiple callbacks are chained. */
if (CALLBACKARG_PREV_FUN(carg, check_html) != NULL) {
if (!CALLBACKARG_PREV_FUN(carg, check_html)
(CALLBACKARG_PREV_CARG(carg), opt, html, len, address, filename)) {
return 0; /* Abort */
}
}
/* Process */
if (strcmp(address, "primary") == 0 && strcmp(filename, "/primary") == 0) /* primary page (list of links) */
return 1;
while(stringfilters[i] != NULL && !getIt) {
if ((pos = strstr(html, stringfilters[i])) != NULL) {
int j;
getIt = 1;
fprintf(stderr,
"** callback info: found '%s' keyword in '%s%s', crawling this page!\n",
stringfilters[i], address, filename);
fprintf(stderr, "** details:\n(..)");
for(j = 0; j < 72 && pos[j]; j++) {
if (pos[j] > 32)
fprintf(stderr, "%c", pos[j]);
else
fprintf(stderr, "?");
}
fprintf(stderr, "(..)\n");
}
i++;
}
if (getIt) {
return 1; /* success */
} else {
fprintf(stderr,
"** callback info: won't parse '%s%s' (no specified keywords found)\n",
address, filename);
return 0; /* this page sucks, don't parse it */
}
}
/* local function called as "end" callback */
static int end_of_mirror(t_hts_callbackarg *carg, httrackp *opt) {
const char *arginfo = (char*) CALLBACKARG_USERDEF(carg);
fprintf(stderr, "* mirror end\n");
hts_log(opt, arginfo, "mirror ended");
/* call parent functions if multiple callbacks are chained. you can skip this part, if you don't want previous callbacks to be called. */
if (CALLBACKARG_PREV_FUN(carg, end) != NULL) {
/* status is ok on our side, return other callabck's status */
return CALLBACKARG_PREV_FUN(carg, end)(CALLBACKARG_PREV_CARG(carg), opt);
}
return 1; /* success */
}
/* local function called as "end" callback */
static int end_of_mirror(t_hts_callbackarg /*the carg structure, holding various information*/*carg, /*the option settings*/httrackp *opt) {
void *ourDummyArg = (void*) CALLBACKARG_USERDEF(carg); /*optional user-defined arg*/
/* processing */
fprintf(stderr, "That's all, folks!\n");
/* call parent functions if multiple callbacks are chained. you can skip this part, if you don't want previous callbacks to be called. */
if (CALLBACKARG_PREV_FUN(carg, end) != NULL) {
/* status is ok on our side, return other callabck's status */
return CALLBACKARG_PREV_FUN(carg, end)(CALLBACKARG_PREV_CARG(carg), opt);
}
return 1; /* success */
}
static int end(t_hts_callbackarg *carg, httrackp *opt) {
t_my_userdef *userdef = (t_my_userdef*) CALLBACKARG_USERDEF(carg);
fprintf(stderr, "** info: wrapper_exit() called!\n");
if (userdef != NULL) {
free(userdef);
userdef = NULL;
}
/* Call parent functions if multiple callbacks are chained. */
if (CALLBACKARG_PREV_FUN(carg, end) != NULL) {
return CALLBACKARG_PREV_FUN(carg, end)(CALLBACKARG_PREV_CARG(carg), opt);
}
return 1; /* success */
}
/* local function called as "check_html" callback */
static int process_file(t_hts_callbackarg /*the carg structure, holding various information*/*carg, /*the option settings*/httrackp *opt,
/*other parameters are callback-specific*/
char* html, int len, const char* url_address, const char* url_file) {
void *ourDummyArg = (void*) CALLBACKARG_USERDEF(carg); /*optional user-defined arg*/
/* call parent functions if multiple callbacks are chained. you can skip this part, if you don't want previous callbacks to be called. */
if (CALLBACKARG_PREV_FUN(carg, check_html) != NULL) {
if (!CALLBACKARG_PREV_FUN(carg, check_html)(CALLBACKARG_PREV_CARG(carg), opt,
html, len, url_address, url_file)) {
return 0; /* abort */
}
}
printf("file %s%s content: %s\n", url_address, url_file, html);
return 1; /* success */
}
static int check_detectedlink_end(t_hts_callbackarg *carg, httrackp *opt) {
char *base = (char*) CALLBACKARG_USERDEF(carg);
fprintf(stderr, "Unplugged ..\n");
if (base != NULL) {
free(base);
base = NULL;
}
/* Call parent functions if multiple callbacks are chained. */
if (CALLBACKARG_PREV_FUN(carg, end) != NULL) {
return CALLBACKARG_PREV_FUN(carg, end)(CALLBACKARG_PREV_CARG(carg), opt);
}
return 1; /* success */
}
static int check_detectedlink(t_hts_callbackarg *carg, httrackp *opt, char* link) {
t_my_userdef *userdef = (t_my_userdef*) CALLBACKARG_USERDEF(carg);
char * const currentURLBeingParsed = userdef->currentURLBeingParsed;
/* Call parent functions if multiple callbacks are chained. */
if (CALLBACKARG_PREV_FUN(carg, linkdetected) != NULL) {
if (!CALLBACKARG_PREV_FUN(carg, linkdetected)(CALLBACKARG_PREV_CARG(carg), opt, link)) {
return 0; /* Abort */
}
}
/* Process */
printf("[%s] -> [%s]\n", currentURLBeingParsed, link);
return 1; /* success */
}
static int process_file(t_hts_callbackarg *carg, httrackp *opt, char* html, int len, const char* url_address, const char* url_file) {
t_my_userdef *userdef = (t_my_userdef*) CALLBACKARG_USERDEF(carg);
char * const currentURLBeingParsed = userdef->currentURLBeingParsed;
/* Call parent functions if multiple callbacks are chained. */
if (CALLBACKARG_PREV_FUN(carg, check_html) != NULL) {
if (!CALLBACKARG_PREV_FUN(carg, check_html)(CALLBACKARG_PREV_CARG(carg), opt, html, len, url_address, url_file)) {
return 0; /* Abort */
}
}
/* Process */
printf("now parsing %s%s..\n", url_address, url_file);
strcpy(currentURLBeingParsed, url_address);
strcat(currentURLBeingParsed, url_file);
return 1; /* success */
}
static int check_detectedlink(t_hts_callbackarg *carg, httrackp* opt, char* link) {
const char *base = (char*) CALLBACKARG_USERDEF(carg);
/* Call parent functions if multiple callbacks are chained. */
if (CALLBACKARG_PREV_FUN(carg, linkdetected) != NULL) {
if (!CALLBACKARG_PREV_FUN(carg, linkdetected)(CALLBACKARG_PREV_CARG(carg), opt, link)) {
return 0; /* Abort */
}
}
/* The incoming (read/write) buffer is at least HTS_URLMAXSIZE bytes long */
if (strncmp(link, "http://", 7) == 0 || strncmp(link, "https://", 8) == 0) {
char temp[HTS_URLMAXSIZE * 2];
strcpy(temp, base);
strcat(temp, link);
strcpy(link, temp);
}
return 1; /* success */
}
/* local function called as "check_html" callback */
static int process_file(t_hts_callbackarg *carg, httrackp *opt,
char* html, int len, const char* url_address, const char* url_file) {
void *ourDummyArg = (void*) CALLBACKARG_USERDEF(carg); /*optional user-defined arg*/
char *fmt;
/* call parent functions if multiple callbacks are chained. you can skip this part, if you don't want previous callbacks to be called. */
if (CALLBACKARG_PREV_FUN(carg, check_html) != NULL) {
if (!CALLBACKARG_PREV_FUN(carg, check_html)(CALLBACKARG_PREV_CARG(carg), opt,
html, len, url_address, url_file)) {
return 0; /* abort */
}
}
/* log */
fprintf(stderr, "* parsing file %s%s\n", url_address, url_file);
fmt = malloc(strlen(url_address) + strlen(url_file) + 128);
sprintf(fmt, " parsing file %s%s", url_address, url_file);
hts_log(opt, "log-wrapper-info", fmt);
free(fmt);
return 1; /* success */
}