mirror of
https://github.com/xroche/httrack.git
synced 2026-06-18 16:22:55 +03:00
Compare commits
6 Commits
cleanup/de
...
feature/ap
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
0f4b2596b2 | ||
|
|
4a676bb5e1 | ||
|
|
36b4e834b8 | ||
|
|
bbb423f025 | ||
|
|
eed46e0b09 | ||
|
|
fa57f0148f |
@@ -2779,7 +2779,7 @@ void back_wait(struct_back * sback, httrackp * opt, cache_back * cache,
|
||||
if (strcmp(back[i].url_fil, "/robots.txt")) {
|
||||
if (back[i].r.statuscode == HTTP_OK) { // 'OK'
|
||||
if (!is_hypertext_mime(opt, back[i].r.contenttype, back[i].url_fil)) { // pas HTML
|
||||
if (opt->getmode & 2) { // on peut ecrire des non html
|
||||
if (opt->getmode & HTS_GETMODE_NONHTML) {
|
||||
int fcheck = 0;
|
||||
int last_errno = 0;
|
||||
|
||||
@@ -2852,7 +2852,7 @@ void back_wait(struct_back * sback, httrackp * opt, cache_back * cache,
|
||||
}
|
||||
}
|
||||
}
|
||||
} else { // on coupe tout!
|
||||
} else { // on coupe tout!
|
||||
hts_log_print(opt, LOG_DEBUG,
|
||||
"File cancelled (non HTML): %s%s",
|
||||
back[i].url_adr, back[i].url_fil);
|
||||
@@ -3661,7 +3661,7 @@ void back_wait(struct_back * sback, httrackp * opt, cache_back * cache,
|
||||
#endif
|
||||
if (sz >= 0) {
|
||||
if (!is_hypertext_mime(opt, back[i].r.contenttype, back[i].url_sav)) { // pas HTML
|
||||
if (opt->getmode & 2) { // on peut ecrire des non html **sinon ben euhh sera intercepté plus loin, donc rap sur ce qui va sortir**
|
||||
if (opt->getmode & HTS_GETMODE_NONHTML) {
|
||||
filenote(&opt->state.strc, back[i].url_sav, NULL); // noter fichier comme connu
|
||||
file_notify(opt, back[i].url_adr, back[i].url_fil,
|
||||
back[i].url_sav, 0, 1,
|
||||
|
||||
@@ -370,7 +370,7 @@ int cache_selftests(httrackp *opt, const char *dir) {
|
||||
StringCopy(opt->path_html, base);
|
||||
StringCopy(opt->path_html_utf8, base);
|
||||
}
|
||||
opt->cache = 1;
|
||||
opt->cache = HTS_CACHE_PRIORITY;
|
||||
|
||||
/* pass 1: create everything in a single write session */
|
||||
selftest_open_for_write(&cache, opt);
|
||||
@@ -547,7 +547,7 @@ static void golden_setup(httrackp *opt, const char *dir) {
|
||||
StringCopy(opt->path_log, base);
|
||||
StringCopy(opt->path_html, base);
|
||||
StringCopy(opt->path_html_utf8, base);
|
||||
opt->cache = 1;
|
||||
opt->cache = HTS_CACHE_PRIORITY;
|
||||
}
|
||||
|
||||
int cache_golden_selftest(httrackp *opt, const char *dir, int regen) {
|
||||
|
||||
@@ -135,7 +135,8 @@ HTSEXT_API T_SOC catch_url_init(int *port, /* 128 bytes */ char *adr) {
|
||||
// returns 0 if error
|
||||
// url: buffer where URL must be stored - or ip:port in case of failure
|
||||
// data: 32Kb
|
||||
HTSEXT_API int catch_url(T_SOC soc, char *url, char *method, char *data) {
|
||||
HTSEXT_API hts_boolean catch_url(T_SOC soc, char *url, char *method,
|
||||
char *data) {
|
||||
int retour = 0;
|
||||
|
||||
// connexion (accept)
|
||||
|
||||
@@ -1835,9 +1835,10 @@ int httpmirror(char *url1, httrackp * opt) {
|
||||
a++; // sauter espace(s)
|
||||
if (strnotempty(a)) {
|
||||
#ifdef IGNORE_RESTRICTIVE_ROBOTS
|
||||
if (strcmp(a, "/") != 0 || opt->robots >= 3)
|
||||
if (strcmp(a, "/") != 0 ||
|
||||
opt->robots >= HTS_ROBOTS_ALWAYS_STRICT)
|
||||
#endif
|
||||
{ /* ignoring disallow: / */
|
||||
{ /* ignoring disallow: / */
|
||||
if ((strlen(buff) + strlen(a) + 8) < sizeof(buff)) {
|
||||
strcatbuff(buff, a);
|
||||
strcatbuff(buff, "\n");
|
||||
@@ -1932,10 +1933,10 @@ int httpmirror(char *url1, httrackp * opt) {
|
||||
"Warning: store %s without scan: %s", r.contenttype,
|
||||
savename());
|
||||
} else {
|
||||
if ((opt->getmode & 2) != 0) { // ok autorisé
|
||||
if ((opt->getmode & HTS_GETMODE_NONHTML) != 0) {
|
||||
hts_log_print(opt, LOG_DEBUG, "Store %s: %s", r.contenttype,
|
||||
savename());
|
||||
} else { // lien non autorisé! (ex: cgi-bin en html)
|
||||
} else { // lien non autorisé! (ex: cgi-bin en html)
|
||||
hts_log_print(opt, LOG_DEBUG,
|
||||
"non-html file ignored after upload at %s : %s",
|
||||
urladr(), urlfil());
|
||||
@@ -2052,7 +2053,7 @@ int httpmirror(char *url1, httrackp * opt) {
|
||||
ptr++;
|
||||
|
||||
// faut-il sauter le(s) lien(s) suivant(s)? (fichiers images à passer après les html)
|
||||
if (opt->getmode & 4) { // sauver les non html après
|
||||
if (opt->getmode & HTS_GETMODE_HTML_FIRST) {
|
||||
// sauter les fichiers selon la passe
|
||||
if (!numero_passe) {
|
||||
while((ptr < opt->lien_tot) ? (heap(ptr)->pass2) : 0)
|
||||
@@ -2584,7 +2585,7 @@ static int mkdir_compat(const char *pathname) {
|
||||
|
||||
/* path must end with "/" or with the finename (/tmp/bar/ or /tmp/bar/foo.zip) */
|
||||
/* Note: preserve errno */
|
||||
HTSEXT_API int dir_exists(const char *path) {
|
||||
HTSEXT_API hts_boolean dir_exists(const char *path) {
|
||||
const int err = errno;
|
||||
STRUCT_STAT st;
|
||||
char BIGSTK file[HTS_URLMAXSIZE * 2];
|
||||
@@ -3645,7 +3646,7 @@ HTSEXT_API int hts_setpause(httrackp * opt, int p) {
|
||||
}
|
||||
|
||||
// ask for termination
|
||||
HTSEXT_API int hts_request_stop(httrackp * opt, int force) {
|
||||
HTSEXT_API int hts_request_stop(httrackp *opt, hts_boolean force) {
|
||||
if (opt != NULL) {
|
||||
hts_log_print(opt, LOG_ERROR, "Exit requested by shell or user");
|
||||
hts_mutexlock(&opt->state.lock);
|
||||
@@ -3655,7 +3656,7 @@ HTSEXT_API int hts_request_stop(httrackp * opt, int force) {
|
||||
return 0;
|
||||
}
|
||||
|
||||
HTSEXT_API int hts_has_stopped(httrackp * opt) {
|
||||
HTSEXT_API hts_boolean hts_has_stopped(httrackp *opt) {
|
||||
int ended;
|
||||
hts_mutexlock(&opt->state.lock);
|
||||
ended = opt->state.is_ended;
|
||||
@@ -3677,12 +3678,12 @@ HTSEXT_API int hts_has_stopped(httrackp * opt) {
|
||||
//}
|
||||
// ajout d'URL
|
||||
// -1 : erreur
|
||||
HTSEXT_API int hts_addurl(httrackp * opt, char **url) {
|
||||
HTSEXT_API hts_boolean hts_addurl(httrackp *opt, char **url) {
|
||||
if (url)
|
||||
opt->state._hts_addurl = url;
|
||||
return (opt->state._hts_addurl != NULL);
|
||||
}
|
||||
HTSEXT_API int hts_resetaddurl(httrackp * opt) {
|
||||
HTSEXT_API hts_boolean hts_resetaddurl(httrackp *opt) {
|
||||
opt->state._hts_addurl = NULL;
|
||||
return (opt->state._hts_addurl != NULL);
|
||||
}
|
||||
@@ -3736,10 +3737,10 @@ HTSEXT_API int copy_htsopt(const httrackp * from, httrackp * to) {
|
||||
|
||||
// test all: bit 8 de travel
|
||||
if (from->travel > -1) {
|
||||
if (from->travel & 256)
|
||||
to->travel |= 256;
|
||||
if (from->travel & HTS_TRAVEL_TEST_ALL)
|
||||
to->travel |= HTS_TRAVEL_TEST_ALL;
|
||||
else
|
||||
to->travel &= 255;
|
||||
to->travel &= HTS_TRAVEL_SCOPE_MASK;
|
||||
}
|
||||
|
||||
return 0;
|
||||
|
||||
@@ -1431,7 +1431,7 @@ static int hts_main_internal(int argc, char **argv, httrackp * opt) {
|
||||
StringBuff(opt->path_log), "hts-in_progress.lock"))) { // fichier lock?
|
||||
//char s[32];
|
||||
|
||||
opt->cache = 1; // cache prioritaire
|
||||
opt->cache = HTS_CACHE_PRIORITY; // cache prioritaire
|
||||
if (opt->quiet == 0) {
|
||||
if ((fexist
|
||||
(fconcat
|
||||
@@ -1465,7 +1465,7 @@ static int hts_main_internal(int argc, char **argv, httrackp * opt) {
|
||||
(fconcat
|
||||
(OPT_GET_BUFF(opt), OPT_GET_BUFF_SIZE(opt), StringBuff(opt->path_html), "index.html"))) {
|
||||
//char s[32];
|
||||
opt->cache = 2; // cache vient après test de validité
|
||||
opt->cache = HTS_CACHE_TEST_UPDATE;
|
||||
if (opt->quiet == 0) {
|
||||
if ((fexist
|
||||
(fconcat
|
||||
@@ -1558,25 +1558,25 @@ static int hts_main_internal(int argc, char **argv, httrackp * opt) {
|
||||
return 0; // déja fait normalement
|
||||
//
|
||||
case 'g': // récupérer un (ou plusieurs) fichiers isolés
|
||||
opt->wizard = 2; // le wizard on peut plus s'en passer..
|
||||
opt->wizard = HTS_WIZARD_AUTO;
|
||||
//opt->wizard=0; // pas de wizard
|
||||
opt->cache = 0; // ni de cache
|
||||
opt->cache = HTS_CACHE_NONE; // ni de cache
|
||||
opt->makeindex = 0; // ni d'index
|
||||
httrack_logmode = 1; // erreurs à l'écran
|
||||
opt->savename_type = 1003; // mettre dans le répertoire courant
|
||||
opt->depth = 0; // ne pas explorer la page
|
||||
opt->accept_cookie = 0; // pas de cookies
|
||||
opt->robots = 0; // pas de robots
|
||||
opt->robots = HTS_ROBOTS_NEVER; // pas de robots
|
||||
break;
|
||||
case 'w':
|
||||
opt->wizard = 2; // wizard 'soft' (ne pose pas de questions)
|
||||
opt->travel = 0;
|
||||
opt->seeker = 1;
|
||||
opt->wizard = HTS_WIZARD_AUTO;
|
||||
opt->travel = HTS_TRAVEL_SAME_ADDRESS;
|
||||
opt->seeker = HTS_SEEKER_DOWN;
|
||||
break;
|
||||
case 'W':
|
||||
opt->wizard = 1; // Wizard-Help (pose des questions)
|
||||
opt->travel = 0;
|
||||
opt->seeker = 1;
|
||||
opt->wizard = HTS_WIZARD_ASK; // Wizard-Help (pose des questions)
|
||||
opt->travel = HTS_TRAVEL_SAME_ADDRESS;
|
||||
opt->seeker = HTS_SEEKER_DOWN;
|
||||
break;
|
||||
case 'r': // n'est plus le recurse get bestial mais wizard itou!
|
||||
if (isdigit((unsigned char) *(com + 1))) {
|
||||
@@ -1598,19 +1598,23 @@ static int hts_main_internal(int argc, char **argv, httrackp * opt) {
|
||||
// note: les tests opt->depth sont pour éviter de faire
|
||||
// un miroir du web (:-O) accidentelement ;-)
|
||||
case 'a': /*if (opt->depth==9999) opt->depth=3; */
|
||||
opt->travel = 0 + (opt->travel & 256);
|
||||
opt->travel =
|
||||
HTS_TRAVEL_SAME_ADDRESS + (opt->travel & HTS_TRAVEL_TEST_ALL);
|
||||
break;
|
||||
case 'd': /*if (opt->depth==9999) opt->depth=3; */
|
||||
opt->travel = 1 + (opt->travel & 256);
|
||||
opt->travel =
|
||||
HTS_TRAVEL_SAME_DOMAIN + (opt->travel & HTS_TRAVEL_TEST_ALL);
|
||||
break;
|
||||
case 'l': /*if (opt->depth==9999) opt->depth=3; */
|
||||
opt->travel = 2 + (opt->travel & 256);
|
||||
opt->travel =
|
||||
HTS_TRAVEL_SAME_TLD + (opt->travel & HTS_TRAVEL_TEST_ALL);
|
||||
break;
|
||||
case 'e': /*if (opt->depth==9999) opt->depth=3; */
|
||||
opt->travel = 7 + (opt->travel & 256);
|
||||
opt->travel =
|
||||
HTS_TRAVEL_EVERYWHERE + (opt->travel & HTS_TRAVEL_TEST_ALL);
|
||||
break;
|
||||
case 't':
|
||||
opt->travel |= 256;
|
||||
opt->travel |= HTS_TRAVEL_TEST_ALL;
|
||||
break;
|
||||
case 'n':
|
||||
opt->nearlink = 1;
|
||||
@@ -1620,16 +1624,16 @@ static int hts_main_internal(int argc, char **argv, httrackp * opt) {
|
||||
break;
|
||||
//
|
||||
case 'U':
|
||||
opt->seeker = 2;
|
||||
opt->seeker = HTS_SEEKER_UP;
|
||||
break;
|
||||
case 'D':
|
||||
opt->seeker = 1;
|
||||
opt->seeker = HTS_SEEKER_DOWN;
|
||||
break;
|
||||
case 'S':
|
||||
opt->seeker = 0;
|
||||
break;
|
||||
case 'B':
|
||||
opt->seeker = 3;
|
||||
opt->seeker = HTS_SEEKER_DOWN | HTS_SEEKER_UP;
|
||||
break;
|
||||
//
|
||||
case 'Y':
|
||||
@@ -1659,12 +1663,12 @@ static int hts_main_internal(int argc, char **argv, httrackp * opt) {
|
||||
//case 'A': opt->urlmode=1; break;
|
||||
//case 'R': opt->urlmode=2; break;
|
||||
case 'K':
|
||||
opt->urlmode = 0;
|
||||
opt->urlmode = HTS_URLMODE_ABSOLUTE;
|
||||
if (isdigit((unsigned char) *(com + 1))) {
|
||||
sscanf(com + 1, "%d", &opt->urlmode);
|
||||
if (opt->urlmode == 0) { // in fact K0 ==> K2
|
||||
sscanf(com + 1, "%d", (int *) &opt->urlmode);
|
||||
if (opt->urlmode == HTS_URLMODE_ABSOLUTE) { // in fact K0 ==> K2
|
||||
// and K ==> K0
|
||||
opt->urlmode = 2;
|
||||
opt->urlmode = HTS_URLMODE_RELATIVE;
|
||||
}
|
||||
while(isdigit((unsigned char) *(com + 1)))
|
||||
com++;
|
||||
@@ -1779,7 +1783,7 @@ static int hts_main_internal(int argc, char **argv, httrackp * opt) {
|
||||
break;
|
||||
//
|
||||
case 'b':
|
||||
sscanf(com + 1, "%d", &opt->accept_cookie);
|
||||
sscanf(com + 1, "%d", (int *) &opt->accept_cookie);
|
||||
while(isdigit((unsigned char) *(com + 1)))
|
||||
com++;
|
||||
break;
|
||||
@@ -1831,33 +1835,33 @@ static int hts_main_internal(int argc, char **argv, httrackp * opt) {
|
||||
break;
|
||||
case 's':
|
||||
if (isdigit((unsigned char) *(com + 1))) {
|
||||
sscanf(com + 1, "%d", &opt->robots);
|
||||
sscanf(com + 1, "%d", (int *) &opt->robots);
|
||||
while(isdigit((unsigned char) *(com + 1)))
|
||||
com++;
|
||||
} else
|
||||
opt->robots = 1;
|
||||
opt->robots = HTS_ROBOTS_SOMETIMES;
|
||||
#if DEBUG_ROBOTS
|
||||
printf("robots.txt mode set to %d\n", opt->robots);
|
||||
#endif
|
||||
break;
|
||||
case 'o':
|
||||
sscanf(com + 1, "%d", &opt->errpage);
|
||||
sscanf(com + 1, "%d", (int *) &opt->errpage);
|
||||
while(isdigit((unsigned char) *(com + 1)))
|
||||
com++;
|
||||
break;
|
||||
case 'u':
|
||||
sscanf(com + 1, "%d", &opt->check_type);
|
||||
sscanf(com + 1, "%d", (int *) &opt->check_type);
|
||||
while(isdigit((unsigned char) *(com + 1)))
|
||||
com++;
|
||||
break;
|
||||
//
|
||||
case 'C':
|
||||
if (isdigit((unsigned char) *(com + 1))) {
|
||||
sscanf(com + 1, "%d", &opt->cache);
|
||||
sscanf(com + 1, "%d", (int *) &opt->cache);
|
||||
while(isdigit((unsigned char) *(com + 1)))
|
||||
com++;
|
||||
} else
|
||||
opt->cache = 1;
|
||||
opt->cache = HTS_CACHE_PRIORITY;
|
||||
break;
|
||||
case 'k':
|
||||
opt->all_in_cache = 1;
|
||||
@@ -1913,7 +1917,7 @@ static int hts_main_internal(int argc, char **argv, httrackp * opt) {
|
||||
case 'I':
|
||||
opt->kindex = 1;
|
||||
if (isdigit((unsigned char) *(com + 1))) {
|
||||
sscanf(com + 1, "%d", &opt->kindex);
|
||||
sscanf(com + 1, "%d", (int *) &opt->kindex);
|
||||
while(isdigit((unsigned char) *(com + 1)))
|
||||
com++;
|
||||
}
|
||||
@@ -2045,7 +2049,7 @@ static int hts_main_internal(int argc, char **argv, httrackp * opt) {
|
||||
// preserve: no footer, original links
|
||||
case 'p':
|
||||
StringClear(opt->footer);
|
||||
opt->urlmode = 4;
|
||||
opt->urlmode = HTS_URLMODE_KEEP_ORIGINAL;
|
||||
break;
|
||||
case 'L': // URL list
|
||||
if ((na + 1 >= argc) || (argv[na + 1][0] == '-')) {
|
||||
@@ -3610,12 +3614,12 @@ static int hts_main_internal(int argc, char **argv, httrackp * opt) {
|
||||
printf("Mirror launched on %s by HTTrack Website Copier/"
|
||||
HTTRACK_VERSION "%s " HTTRACK_AFF_AUTHORS "" LF, t,
|
||||
hts_get_version_info(opt));
|
||||
if (opt->wizard == 0) {
|
||||
if (opt->wizard == HTS_WIZARD_NONE) {
|
||||
printf
|
||||
("mirroring %s with %d levels, %d sockets,t=%d,s=%d,logm=%d,lnk=%d,mdg=%d\n",
|
||||
url, opt->depth, opt->maxsoc, opt->travel, opt->seeker,
|
||||
httrack_logmode, opt->urlmode, opt->getmode);
|
||||
} else { // the magic wizard
|
||||
} else { // the magic wizard
|
||||
printf("mirroring %s with the wizard help..\n", url);
|
||||
}
|
||||
}
|
||||
|
||||
@@ -242,6 +242,14 @@ Please visit our Website: http://www.httrack.com
|
||||
#define HTS_NOPARAM "(none)"
|
||||
#define HTS_NOPARAM2 "\"(none)\""
|
||||
|
||||
/* Boolean flag for option fields and API yes/no returns. An enum (not C bool)
|
||||
so it stays int-sized: option fields keep the httrackp layout/ABI, and a
|
||||
return type stays compatible with the int it replaces. */
|
||||
#ifndef HTS_DEF_DEFSTRUCT_hts_boolean
|
||||
#define HTS_DEF_DEFSTRUCT_hts_boolean
|
||||
typedef enum hts_boolean { HTS_FALSE = 0, HTS_TRUE = 1 } hts_boolean;
|
||||
#endif
|
||||
|
||||
/* Larger/smaller of two values. Macros: arguments are evaluated twice. */
|
||||
#define maximum(A,B) ( (A) > (B) ? (A) : (B) )
|
||||
|
||||
|
||||
114
src/htslib.c
114
src/htslib.c
@@ -3646,8 +3646,9 @@ HTSEXT_API char *unescape_http(char *const catbuff, const size_t size, const cha
|
||||
// DOES NOT DECODE %25 (part of CHAR_DELIM)
|
||||
// no_high & 1: decode high chars
|
||||
// no_high & 2: decode space
|
||||
HTSEXT_API char *unescape_http_unharm(char *const catbuff, const size_t size,
|
||||
const char *s, const int no_high) {
|
||||
HTSEXT_API char *unescape_http_unharm(char *const catbuff, const size_t size,
|
||||
const char *s,
|
||||
const hts_boolean no_high) {
|
||||
size_t i, j;
|
||||
|
||||
RUNTIME_TIME_CHECK_SIZE(size);
|
||||
@@ -3931,8 +3932,8 @@ void hts_replace(char *s, char from, char to) {
|
||||
|
||||
// guess a local file's mime type (e.g. fil="toto.gif" -> s="image/gif")
|
||||
// returns 1 if a type was written to s, 0 otherwise
|
||||
int guess_httptype_sized(httrackp *opt, char *s, size_t ssize,
|
||||
const char *fil) {
|
||||
hts_boolean guess_httptype_sized(httrackp *opt, char *s, size_t ssize,
|
||||
const char *fil) {
|
||||
return get_httptype_sized(opt, s, ssize, fil, 1);
|
||||
}
|
||||
|
||||
@@ -3945,8 +3946,8 @@ void guess_httptype(httrackp * opt, char *s, const char *fil) {
|
||||
// write the mime type for fil into s (capacity ssize)
|
||||
// flag: 1 to always return a type (the "application/..." / octet-stream
|
||||
// fallback) returns 1 if a type was written to s, 0 otherwise
|
||||
HTSEXT_API int get_httptype_sized(httrackp *opt, char *s, size_t ssize,
|
||||
const char *fil, int flag) {
|
||||
HTSEXT_API hts_boolean get_httptype_sized(httrackp *opt, char *s, size_t ssize,
|
||||
const char *fil, hts_boolean flag) {
|
||||
// userdef overrides get_httptype (a rule with an empty value, e.g. "--assume
|
||||
// cgi=", matches but writes nothing: report it as "no type" like the old
|
||||
// code, whose callers tested strnotempty(s))
|
||||
@@ -4196,7 +4197,7 @@ HTSEXT_API int is_userknowntype(httrackp * opt, const char *fil) {
|
||||
|
||||
// page dynamique?
|
||||
// is_dyntype(get_ext("foo.asp"))
|
||||
HTSEXT_API int is_dyntype(const char *fil) {
|
||||
HTSEXT_API hts_boolean is_dyntype(const char *fil) {
|
||||
int j = 0;
|
||||
|
||||
if (!fil)
|
||||
@@ -4214,7 +4215,7 @@ HTSEXT_API int is_dyntype(const char *fil) {
|
||||
|
||||
// types critiques qui ne doivent pas être changés car renvoyés par des serveurs qui ne
|
||||
// connaissent pas le type
|
||||
int may_unknown(httrackp * opt, const char *st) {
|
||||
hts_boolean may_unknown(httrackp *opt, const char *st) {
|
||||
int j = 0;
|
||||
|
||||
// types média
|
||||
@@ -5236,7 +5237,8 @@ HTSEXT_API int hts_uninit_module(void) {
|
||||
}
|
||||
|
||||
// legacy. do not use
|
||||
HTSEXT_API int hts_log(httrackp * opt, const char *prefix, const char *msg) {
|
||||
HTSEXT_API hts_boolean hts_log(httrackp *opt, const char *prefix,
|
||||
const char *msg) {
|
||||
if (opt->log != NULL) {
|
||||
fspc(opt, opt->log, prefix);
|
||||
fprintf(opt->log, "%s" LF, msg);
|
||||
@@ -5434,34 +5436,34 @@ HTSEXT_API httrackp *hts_create_opt(void) {
|
||||
|
||||
/* default settings */
|
||||
|
||||
opt->wizard = 2; // wizard automatique
|
||||
opt->quiet = 0; // questions
|
||||
//
|
||||
opt->travel = 0; // même adresse
|
||||
opt->wizard = HTS_WIZARD_AUTO; // wizard automatique
|
||||
opt->quiet = HTS_FALSE;
|
||||
//
|
||||
opt->travel = HTS_TRAVEL_SAME_ADDRESS; // même adresse
|
||||
opt->depth = 9999; // mirror total par défaut
|
||||
opt->extdepth = 0; // mais pas à l'extérieur
|
||||
opt->seeker = 1; // down
|
||||
opt->urlmode = 2; // relatif par défaut
|
||||
opt->no_type_change = 0; // change file types
|
||||
opt->seeker = HTS_SEEKER_DOWN; // down
|
||||
opt->urlmode = HTS_URLMODE_RELATIVE; // relatif par défaut
|
||||
opt->no_type_change = HTS_FALSE;
|
||||
opt->debug = LOG_NOTICE; // small log
|
||||
opt->getmode = 3; // linear scan
|
||||
opt->getmode = HTS_GETMODE_HTML | HTS_GETMODE_NONHTML;
|
||||
opt->maxsite = -1; // taille max site (aucune)
|
||||
opt->maxfile_nonhtml = -1; // taille max fichier non html
|
||||
opt->maxfile_html = -1; // idem pour html
|
||||
opt->maxsoc = 4; // nbre socket max
|
||||
opt->fragment = -1; // pas de fragmentation
|
||||
opt->nearlink = 0; // ne pas prendre les liens non-html "adjacents"
|
||||
opt->makeindex = 1; // faire un index
|
||||
opt->kindex = 0; // index 'keyword'
|
||||
opt->delete_old = 1; // effacer anciens fichiers
|
||||
opt->background_on_suspend = 1; // Background the process if Control Z calls signal suspend.
|
||||
opt->makestat = 0; // pas de fichier de stats
|
||||
opt->maketrack = 0; // ni de tracking
|
||||
opt->nearlink = HTS_FALSE;
|
||||
opt->makeindex = HTS_TRUE;
|
||||
opt->kindex = HTS_FALSE;
|
||||
opt->delete_old = HTS_TRUE;
|
||||
opt->background_on_suspend = HTS_TRUE;
|
||||
opt->makestat = HTS_FALSE;
|
||||
opt->maketrack = HTS_FALSE;
|
||||
opt->timeout = 120; // timeout par défaut (2 minutes)
|
||||
opt->cache = 1; // cache prioritaire
|
||||
opt->shell = 0; // pas de shell par defaut
|
||||
opt->cache = HTS_CACHE_PRIORITY; // cache prioritaire
|
||||
opt->shell = HTS_FALSE;
|
||||
opt->proxy.active = 0; // pas de proxy
|
||||
opt->user_agent_send = 1; // envoyer un user-agent
|
||||
opt->user_agent_send = HTS_TRUE;
|
||||
StringCopy(opt->user_agent,
|
||||
"Mozilla/4.5 (compatible; HTTrack 3.0x; Windows 98)");
|
||||
StringCopy(opt->referer, "");
|
||||
@@ -5469,34 +5471,36 @@ HTSEXT_API httrackp *hts_create_opt(void) {
|
||||
opt->savename_83 = 0; // noms longs par défaut
|
||||
opt->savename_type = 0; // avec structure originale
|
||||
opt->savename_delayed = 2; // hard delayed type (default)
|
||||
opt->delayed_cached = 1; // cached delayed type (default)
|
||||
opt->mimehtml = 0; // pas MIME-html
|
||||
opt->delayed_cached = HTS_TRUE;
|
||||
opt->mimehtml = HTS_FALSE;
|
||||
opt->parsejava = HTSPARSE_DEFAULT; // parser classes
|
||||
opt->hostcontrol = 0; // PAS de control host pour timeout et traffic jammer
|
||||
opt->retry = 2; // 2 retry par défaut
|
||||
opt->errpage = 1; // copier ou générer une page d'erreur en cas d'erreur (404 etc.)
|
||||
opt->check_type = 1; // vérifier type si inconnu (cgi,asp..) SAUF / considéré comme html
|
||||
opt->all_in_cache = 0; // ne pas tout stocker en cache
|
||||
opt->robots = 2; // traiter les robots.txt
|
||||
opt->external = 0; // liens externes normaux
|
||||
opt->passprivacy = 0; // mots de passe dans les fichiers
|
||||
opt->includequery = 1; // include query-string par défaut
|
||||
opt->mirror_first_page = 0; // pas mode mirror links
|
||||
opt->accept_cookie = 1; // gérer les cookies
|
||||
opt->errpage = HTS_TRUE;
|
||||
// d'erreur (404 etc.)
|
||||
opt->check_type = HTS_TRUE;
|
||||
// considéré comme html
|
||||
opt->all_in_cache = HTS_FALSE;
|
||||
opt->robots = HTS_ROBOTS_ALWAYS; // traiter les robots.txt
|
||||
opt->external = HTS_FALSE;
|
||||
opt->passprivacy = HTS_FALSE;
|
||||
opt->includequery = HTS_TRUE;
|
||||
opt->mirror_first_page = HTS_FALSE;
|
||||
opt->accept_cookie = HTS_TRUE;
|
||||
opt->cookie = NULL;
|
||||
opt->http10 = 0; // laisser http/1.1
|
||||
opt->nokeepalive = 0; // pas keep-alive
|
||||
opt->nocompression = 0; // pas de compression
|
||||
opt->tolerant = 0; // ne pas accepter content-length incorrect
|
||||
opt->parseall = 1; // tout parser (tags inconnus, par exemple)
|
||||
opt->parsedebug = 0; // pas de mode débuggage
|
||||
opt->norecatch = 0; // ne pas reprendre les fichiers effacés par l'utilisateur
|
||||
opt->http10 = HTS_FALSE;
|
||||
opt->nokeepalive = HTS_FALSE;
|
||||
opt->nocompression = HTS_FALSE;
|
||||
opt->tolerant = HTS_FALSE;
|
||||
opt->parseall = HTS_TRUE;
|
||||
opt->parsedebug = HTS_FALSE;
|
||||
opt->norecatch = HTS_FALSE;
|
||||
opt->verbosedisplay = 0; // pas d'animation texte
|
||||
opt->sizehack = 0; // size hack
|
||||
opt->urlhack = 1; // url hack (normalizer)
|
||||
opt->sizehack = HTS_FALSE;
|
||||
opt->urlhack = HTS_TRUE;
|
||||
StringCopy(opt->footer, HTS_DEFAULT_FOOTER);
|
||||
opt->ftp_proxy = 1; // proxy http pour ftp
|
||||
opt->convert_utf8 = 1; // convert html to UTF-8
|
||||
opt->ftp_proxy = HTS_TRUE;
|
||||
opt->convert_utf8 = HTS_TRUE;
|
||||
StringCopy(opt->filelist, "");
|
||||
StringCopy(opt->lang_iso, "en, *");
|
||||
StringCopy(opt->accept,
|
||||
@@ -5507,9 +5511,9 @@ HTSEXT_API httrackp *hts_create_opt(void) {
|
||||
//
|
||||
opt->log = stdout;
|
||||
opt->errlog = stderr;
|
||||
opt->flush = 1; // flush sur les fichiers log
|
||||
//opt->aff_progress=0;
|
||||
opt->keyboard = 0;
|
||||
opt->flush = HTS_TRUE;
|
||||
// opt->aff_progress=0;
|
||||
opt->keyboard = HTS_FALSE;
|
||||
//
|
||||
StringCopy(opt->path_html, "");
|
||||
StringCopy(opt->path_html_utf8, "");
|
||||
@@ -5526,10 +5530,10 @@ HTSEXT_API httrackp *hts_create_opt(void) {
|
||||
opt->waittime = -1; // wait until.. hh*3600+mm*60+ss
|
||||
//
|
||||
opt->exec = "";
|
||||
opt->is_update = 0; // not an update (yet)
|
||||
opt->dir_topindex = 0; // do not built top index (yet)
|
||||
opt->is_update = HTS_FALSE;
|
||||
opt->dir_topindex = HTS_FALSE;
|
||||
//
|
||||
opt->bypass_limits = 0; // enforce limits by default
|
||||
opt->bypass_limits = HTS_FALSE;
|
||||
opt->state.stop = 0; // stopper
|
||||
opt->state.exit_xh = 0; // abort
|
||||
//
|
||||
|
||||
160
src/htsopt.h
160
src/htsopt.h
@@ -285,6 +285,75 @@ typedef enum htsparsejava_flags {
|
||||
HTSPARSE_NO_AGGRESSIVE = 8 // don't aggressively parse .js or .java
|
||||
} htsparsejava_flags;
|
||||
|
||||
/* Link-rewriting style for saved pages (opt->urlmode). */
|
||||
#ifndef HTS_DEF_DEFSTRUCT_hts_urlmode
|
||||
#define HTS_DEF_DEFSTRUCT_hts_urlmode
|
||||
typedef enum hts_urlmode {
|
||||
HTS_URLMODE_ABSOLUTE = 0, /**< absolute URL (http://host/path) everywhere */
|
||||
HTS_URLMODE_ABSOLUTE_FILE = 1, /**< legacy file: form, unused */
|
||||
HTS_URLMODE_RELATIVE = 2, /**< relative link (default) */
|
||||
HTS_URLMODE_ABSOLUTE_URI = 3, /**< absolute URI from root (/path) */
|
||||
HTS_URLMODE_KEEP_ORIGINAL = 4, /**< keep the original link, do not rewrite */
|
||||
HTS_URLMODE_TRANSPARENT_PROXY = 5 /**< transparent-proxy URL */
|
||||
} hts_urlmode;
|
||||
#endif
|
||||
|
||||
/* Cache policy for updates and retries (opt->cache). */
|
||||
#ifndef HTS_DEF_DEFSTRUCT_hts_cachemode
|
||||
#define HTS_DEF_DEFSTRUCT_hts_cachemode
|
||||
typedef enum hts_cachemode {
|
||||
HTS_CACHE_NONE = 0, /**< no cache */
|
||||
HTS_CACHE_PRIORITY = 1, /**< cache takes priority over the network */
|
||||
HTS_CACHE_TEST_UPDATE = 2 /**< check for update before reuse (default) */
|
||||
} hts_cachemode;
|
||||
#endif
|
||||
|
||||
/* Interactive wizard level (opt->wizard). */
|
||||
#ifndef HTS_DEF_DEFSTRUCT_hts_wizard
|
||||
#define HTS_DEF_DEFSTRUCT_hts_wizard
|
||||
typedef enum hts_wizard {
|
||||
HTS_WIZARD_NONE = 0, /**< no wizard */
|
||||
HTS_WIZARD_ASK = 1, /**< wizard asks questions */
|
||||
HTS_WIZARD_AUTO = 2 /**< wizard runs without asking */
|
||||
} hts_wizard;
|
||||
#endif
|
||||
|
||||
/* robots.txt / meta-robots obedience level (opt->robots). */
|
||||
#ifndef HTS_DEF_DEFSTRUCT_hts_robots
|
||||
#define HTS_DEF_DEFSTRUCT_hts_robots
|
||||
typedef enum hts_robots {
|
||||
HTS_ROBOTS_NEVER = 0, /**< ignore robots rules */
|
||||
HTS_ROBOTS_SOMETIMES = 1, /**< partial obedience (default) */
|
||||
HTS_ROBOTS_ALWAYS = 2, /**< obey robots rules */
|
||||
HTS_ROBOTS_ALWAYS_STRICT = 3 /**< obey even strict rules */
|
||||
} hts_robots;
|
||||
#endif
|
||||
|
||||
/* What to fetch (opt->getmode bitmask). */
|
||||
typedef enum hts_getmode {
|
||||
HTS_GETMODE_HTML = 1 << 0, /**< save HTML files */
|
||||
HTS_GETMODE_NONHTML = 1 << 1, /**< save non-HTML files */
|
||||
HTS_GETMODE_HTML_FIRST = 1 << 2 /**< fetch HTML first, then the other files */
|
||||
} hts_getmode;
|
||||
|
||||
/* Allowed directions in the directory tree (opt->seeker bitmask). */
|
||||
typedef enum hts_seeker {
|
||||
HTS_SEEKER_DOWN = 1 << 0, /**< may descend into subdirectories */
|
||||
HTS_SEEKER_UP = 1 << 1 /**< may ascend to parent directories */
|
||||
} hts_seeker;
|
||||
|
||||
/* Link-following scope, stored in the low byte of opt->travel. */
|
||||
typedef enum hts_travel_scope {
|
||||
HTS_TRAVEL_SAME_ADDRESS = 0, /**< stay on the same address (host) */
|
||||
HTS_TRAVEL_SAME_DOMAIN = 1, /**< stay on the same principal domain */
|
||||
HTS_TRAVEL_SAME_TLD = 2, /**< stay on the same TLD (e.g. .com) */
|
||||
HTS_TRAVEL_EVERYWHERE = 7 /**< follow links anywhere on the web */
|
||||
} hts_travel_scope;
|
||||
|
||||
/* Flags OR'd into opt->travel above the scope value. */
|
||||
#define HTS_TRAVEL_SCOPE_MASK 0xff /**< mask selecting the scope value */
|
||||
#define HTS_TRAVEL_TEST_ALL (1 << 8) /**< also test forbidden URLs (-t) */
|
||||
|
||||
#ifndef HTS_DEF_FWSTRUCT_lien_buffers
|
||||
#define HTS_DEF_FWSTRUCT_lien_buffers
|
||||
typedef struct lien_buffers lien_buffers;
|
||||
@@ -308,14 +377,15 @@ typedef struct httrackp httrackp;
|
||||
struct httrackp {
|
||||
size_t size_httrackp; /**< size of this structure (version/ABI guard) */
|
||||
/* */
|
||||
int wizard; /**< interactive wizard level (none/full/light) */
|
||||
int flush; /**< fflush() log files after each write */
|
||||
hts_wizard wizard; /**< interactive wizard level (none/ask/auto) */
|
||||
hts_boolean flush; /**< fflush() log files after each write */
|
||||
int travel; /**< link-following scope (same domain, etc.) */
|
||||
int seeker; /**< allowed direction: go up and/or down the tree */
|
||||
int depth; /**< maximum recursion depth (-rN) */
|
||||
int extdepth; /**< maximum recursion depth outside the start domain */
|
||||
int urlmode; /**< saved-link rewriting style (relative, absolute, etc.) */
|
||||
int no_type_change; // do not change file type according to MIME
|
||||
hts_urlmode
|
||||
urlmode; /**< saved-link rewriting style (relative, absolute, etc.) */
|
||||
hts_boolean no_type_change; // do not change file type according to MIME
|
||||
int debug; /**< debug logging level */
|
||||
int getmode; /**< what to fetch (HTML, images, ...) bitmask */
|
||||
FILE *log; /**< informational log stream; NULL mutes it */
|
||||
@@ -325,28 +395,30 @@ struct httrackp {
|
||||
LLint maxfile_html; /**< max bytes per HTML file */
|
||||
int maxsoc; /**< max simultaneous sockets (-cN) */
|
||||
LLint fragment; /**< split site after this many bytes */
|
||||
int nearlink; /**< also fetch images/data adjacent to a page but off-site */
|
||||
int makeindex; /**< build a top-level index.html */
|
||||
int kindex; /**< build a keyword index */
|
||||
int delete_old; /**< delete locally obsolete files after update */
|
||||
hts_boolean
|
||||
nearlink; /**< also fetch images/data adjacent to a page but off-site */
|
||||
hts_boolean makeindex; /**< build a top-level index.html */
|
||||
hts_boolean kindex; /**< build a keyword index */
|
||||
hts_boolean delete_old; /**< delete locally obsolete files after update */
|
||||
int timeout; /**< connection timeout in seconds */
|
||||
int rateout; /**< minimum transfer rate (bytes/s) before abort */
|
||||
int maxtime; /**< max total mirror duration in seconds */
|
||||
int maxrate; /**< max transfer rate cap (bytes/s) */
|
||||
float maxconn; /**< max connections per second */
|
||||
int waittime; /**< scheduled start time (wall-clock seconds) */
|
||||
int cache; /**< cache generation mode */
|
||||
hts_cachemode cache; /**< cache generation mode */
|
||||
// int aff_progress; // progress bar
|
||||
int shell; /**< driven by a shell over stdin/stdout pipes */
|
||||
hts_boolean shell; /**< driven by a shell over stdin/stdout pipes */
|
||||
t_proxy proxy; /**< proxy configuration */
|
||||
int savename_83; /**< force 8.3 (DOS) file names */
|
||||
int savename_type; /**< saved-name layout (original tree, flat, ...) */
|
||||
String
|
||||
savename_userdef; /**< user-defined name template (e.g. %h%p/%n%q.%t) */
|
||||
int savename_delayed; // delayed type check
|
||||
int delayed_cached; // delayed type check can be cached to speedup updates
|
||||
int mimehtml; /**< produce a single MIME/MHTML archive */
|
||||
int user_agent_send; /**< send a User-Agent header */
|
||||
hts_boolean
|
||||
delayed_cached; // delayed type check can be cached to speedup updates
|
||||
hts_boolean mimehtml; /**< produce a single MIME/MHTML archive */
|
||||
hts_boolean user_agent_send; /**< send a User-Agent header */
|
||||
String user_agent; /**< User-Agent value (e.g. httrack/1.0) */
|
||||
String referer; /**< Referer value to send */
|
||||
String from; /**< From value to send */
|
||||
@@ -355,37 +427,39 @@ struct httrackp {
|
||||
String path_html_utf8; /**< output directory for the mirror, UTF-8 form */
|
||||
String path_bin; /**< directory for HTML templates */
|
||||
int retry; /**< extra retries on a failed transfer */
|
||||
int makestat; /**< maintain a transfer-statistics log */
|
||||
int maketrack; /**< maintain an operations-statistics log */
|
||||
hts_boolean makestat; /**< maintain a transfer-statistics log */
|
||||
hts_boolean maketrack; /**< maintain an operations-statistics log */
|
||||
int parsejava; /**< Java/JS parsing mode; see htsparsejava_flags */
|
||||
int hostcontrol; /**< drop hosts that are too slow, etc. */
|
||||
int errpage; /**< generate an error page on 404 and similar */
|
||||
int check_type; /**< probe unknown-type links (cgi/asp/dir) and follow moves
|
||||
*/
|
||||
int all_in_cache; /**< keep all retrieved data in the cache */
|
||||
int robots; /**< robots.txt handling level */
|
||||
int external; /**< render external links as error pages */
|
||||
int passprivacy; /**< strip passwords from external links */
|
||||
int includequery; /**< include the query string in saved names */
|
||||
int mirror_first_page; /**< only mirror the links of the first page */
|
||||
hts_boolean errpage; /**< generate an error page on 404 and similar */
|
||||
hts_boolean
|
||||
check_type; /**< probe unknown-type links (cgi/asp/dir) and follow moves
|
||||
*/
|
||||
hts_boolean all_in_cache; /**< keep all retrieved data in the cache */
|
||||
hts_robots robots; /**< robots.txt handling level */
|
||||
hts_boolean external; /**< render external links as error pages */
|
||||
hts_boolean passprivacy; /**< strip passwords from external links */
|
||||
hts_boolean includequery; /**< include the query string in saved names */
|
||||
hts_boolean mirror_first_page; /**< only mirror the links of the first page */
|
||||
String sys_com; /**< system command to run */
|
||||
int sys_com_exec; /**< actually execute sys_com */
|
||||
int accept_cookie; /**< accept and send cookies */
|
||||
hts_boolean sys_com_exec; /**< actually execute sys_com */
|
||||
hts_boolean accept_cookie; /**< accept and send cookies */
|
||||
t_cookie *cookie; /**< cookie store */
|
||||
int http10; /**< force HTTP/1.0 */
|
||||
int nokeepalive; /**< disable keep-alive */
|
||||
int nocompression; /**< disable content compression */
|
||||
int sizehack; /**< treat same-size response as "updated" */
|
||||
int urlhack; // force "url normalization" to avoid loops
|
||||
int tolerant; /**< accept an incorrect Content-Length */
|
||||
int parseall; /**< parse aggressively, including unknown tags with links */
|
||||
int parsedebug; /**< parser debug mode */
|
||||
int norecatch; /**< do not re-fetch files the user deleted locally */
|
||||
hts_boolean http10; /**< force HTTP/1.0 */
|
||||
hts_boolean nokeepalive; /**< disable keep-alive */
|
||||
hts_boolean nocompression; /**< disable content compression */
|
||||
hts_boolean sizehack; /**< treat same-size response as "updated" */
|
||||
hts_boolean urlhack; // force "url normalization" to avoid loops
|
||||
hts_boolean tolerant; /**< accept an incorrect Content-Length */
|
||||
hts_boolean
|
||||
parseall; /**< parse aggressively, including unknown tags with links */
|
||||
hts_boolean parsedebug; /**< parser debug mode */
|
||||
hts_boolean norecatch; /**< do not re-fetch files the user deleted locally */
|
||||
int verbosedisplay; /**< animated text progress display */
|
||||
String footer; /**< footer/info line injected into pages */
|
||||
int maxcache; /**< in-memory cache backing limit (bytes) */
|
||||
// int maxcache_anticipate; // maximum links to anticipate (upper bound)
|
||||
int ftp_proxy; /**< use the HTTP proxy for FTP too */
|
||||
hts_boolean ftp_proxy; /**< use the HTTP proxy for FTP too */
|
||||
String filelist; /**< file listing URLs to include */
|
||||
String urllist; /**< file listing filters to include */
|
||||
htsfilters filters; /**< filter pointers (+/-pattern rules) */
|
||||
@@ -399,20 +473,20 @@ struct httrackp {
|
||||
String headers; // Additional headers
|
||||
String mimedefs; // ext1=mimetype1\next2=mimetype2..
|
||||
String mod_blacklist; /**< blacklisted modules */
|
||||
int convert_utf8; // filenames UTF-8 conversion (3.46)
|
||||
hts_boolean convert_utf8; // filenames UTF-8 conversion (3.46)
|
||||
//
|
||||
int maxlink; /**< max number of links */
|
||||
int maxfilter; /**< max number of filters */
|
||||
//
|
||||
const char *exec; /**< path of the running executable */
|
||||
//
|
||||
int quiet; /**< suppress non-wizard questions */
|
||||
int keyboard; /**< poll stdin for keyboard input */
|
||||
int bypass_limits; // bypass built-in limits
|
||||
int background_on_suspend; // background process on suspend signal
|
||||
hts_boolean quiet; /**< suppress non-wizard questions */
|
||||
hts_boolean keyboard; /**< poll stdin for keyboard input */
|
||||
hts_boolean bypass_limits; // bypass built-in limits
|
||||
hts_boolean background_on_suspend; // background process on suspend signal
|
||||
//
|
||||
int is_update; /**< this run is an update (show "File updated...") */
|
||||
int dir_topindex; /**< rebuild the top index afterwards */
|
||||
hts_boolean is_update; /**< this run is an update (show "File updated...") */
|
||||
hts_boolean dir_topindex; /**< rebuild the top index afterwards */
|
||||
//
|
||||
// callbacks
|
||||
t_hts_htmlcheck_callbacks
|
||||
|
||||
@@ -349,7 +349,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
#endif
|
||||
|
||||
// Now, parsing
|
||||
if ((opt->getmode & 1) && (ptr > 0)) { // récupérer les html sur disque
|
||||
if ((opt->getmode & HTS_GETMODE_HTML) && (ptr > 0)) {
|
||||
// créer le fichier html local
|
||||
HT_ADD_FOP; // écrire peu à peu le fichier
|
||||
}
|
||||
@@ -553,7 +553,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
if (opt->depth == heap(ptr)->depth) { // on note toujours les premiers liens
|
||||
if (!in_media) {
|
||||
if (opt->makeindex && (ptr > 0)) {
|
||||
if (opt->getmode & 1) { // autorisation d'écrire
|
||||
if (opt->getmode & HTS_GETMODE_HTML) {
|
||||
p = strfield(html, "title");
|
||||
if (p) {
|
||||
if (*(html - 1) == '/')
|
||||
@@ -704,7 +704,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
}
|
||||
}
|
||||
|
||||
if (opt->getmode & 1) { // sauver html
|
||||
if (opt->getmode & HTS_GETMODE_HTML) { // sauver html
|
||||
p = 0;
|
||||
switch (emited_footer) {
|
||||
case 0:
|
||||
@@ -740,7 +740,8 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
|
||||
if (strchr(r->adr, '\r'))
|
||||
eol = "\r\n";
|
||||
if (StringNotEmpty(opt->footer) || opt->urlmode != 4) { /* != preserve */
|
||||
if (StringNotEmpty(opt->footer) ||
|
||||
opt->urlmode != HTS_URLMODE_KEEP_ORIGINAL) {
|
||||
if (StringNotEmpty(opt->footer)) {
|
||||
char BIGSTK tempo[1024 + HTS_URLMAXSIZE * 2];
|
||||
char gmttime[256];
|
||||
@@ -1746,7 +1747,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
|
||||
// écrire codebase avant, flusher avant code
|
||||
if ((p_type == -1) || (p_type == -2)) {
|
||||
if ((opt->getmode & 1) && (ptr > 0)) {
|
||||
if ((opt->getmode & HTS_GETMODE_HTML) && (ptr > 0)) {
|
||||
HT_add_adr; // refresh
|
||||
}
|
||||
lastsaved = html; // dernier écrit+1
|
||||
@@ -1837,7 +1838,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
|
||||
// ne pas flusher après code si on doit écrire le codebase avant!
|
||||
if ((p_type != -1) && (p_type != 2) && (p_type != -2)) {
|
||||
if ((opt->getmode & 1) && (ptr > 0)) {
|
||||
if ((opt->getmode & HTS_GETMODE_HTML) && (ptr > 0)) {
|
||||
HT_add_adr; // refresh
|
||||
}
|
||||
lastsaved = html; // dernier écrit+1
|
||||
@@ -1914,7 +1915,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
if (*html != '#') { // Not empty+unique #
|
||||
if (eadr - html == 1) { // 1=link empty with delim (end_adr-start_adr)
|
||||
if (quote) {
|
||||
if ((opt->getmode & 1) && (ptr > 0)) {
|
||||
if ((opt->getmode & HTS_GETMODE_HTML) && (ptr > 0)) {
|
||||
HT_ADD("#"); // We add this for a <href="">
|
||||
}
|
||||
}
|
||||
@@ -2569,7 +2570,8 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
if ((p_type == 2) || (p_type == -2)) { // base href ou codebase, pas un lien
|
||||
hts_log_print(opt, LOG_DEBUG, "Code/Codebase: %s%s",
|
||||
afs.af.adr, afs.af.fil);
|
||||
} else if ((opt->getmode & 4) == 0) {
|
||||
} else if ((opt->getmode & HTS_GETMODE_HTML_FIRST) ==
|
||||
0) {
|
||||
hts_log_print(opt, LOG_DEBUG, "Record: %s%s -> %s",
|
||||
afs.af.adr, afs.af.fil, afs.save);
|
||||
} else {
|
||||
@@ -2592,8 +2594,8 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
lastsaved = eadr - 1 + 1; // sauter "
|
||||
}
|
||||
/* */
|
||||
else if (opt->urlmode == 0) { // URL absolue dans tous les cas
|
||||
if ((opt->getmode & 1) && (ptr > 0)) { // ecrire les html
|
||||
else if (opt->urlmode == HTS_URLMODE_ABSOLUTE) {
|
||||
if ((opt->getmode & HTS_GETMODE_HTML) && (ptr > 0)) {
|
||||
if (!link_has_authority(afs.af.adr)) {
|
||||
HT_ADD("http://");
|
||||
} else {
|
||||
@@ -2620,12 +2622,14 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
}
|
||||
lastsaved = eadr - 1; // dernier écrit+1 (enfin euh apres on fait un ++ alors hein)
|
||||
/* */
|
||||
} else if (opt->urlmode == 4) { // ne rien faire!
|
||||
} else if (opt->urlmode == HTS_URLMODE_KEEP_ORIGINAL) {
|
||||
/* */
|
||||
/* leave the link 'as is' */
|
||||
/* Sinon, dépend de interne/externe */
|
||||
} else if (forbidden_url == 1) { // le lien ne sera pas chargé, référence externe!
|
||||
if ((opt->getmode & 1) && (ptr > 0)) {
|
||||
} else if (forbidden_url ==
|
||||
1) { // le lien ne sera pas chargé, référence
|
||||
// externe!
|
||||
if ((opt->getmode & HTS_GETMODE_HTML) && (ptr > 0)) {
|
||||
if (p_type != -1) { // pas que le nom de fichier (pas classe java)
|
||||
if (!opt->external) {
|
||||
if (!link_has_authority(afs.af.adr)) {
|
||||
@@ -2674,7 +2678,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
'/') ? 1 : (ishtml(opt, afs.af.fil)))) {
|
||||
case 1:
|
||||
case -2: // html ou répertoire
|
||||
if (opt->getmode & 1) { // sauver html
|
||||
if (opt->getmode & HTS_GETMODE_HTML) {
|
||||
patch_it = 1; // redirect
|
||||
add_url = 1; // avec link?
|
||||
cat_name = "external.html";
|
||||
@@ -2847,7 +2851,8 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
}
|
||||
|
||||
// érire codebase="chemin"
|
||||
if ((opt->getmode & 1) && (ptr > 0)) {
|
||||
if ((opt->getmode & HTS_GETMODE_HTML) &&
|
||||
(ptr > 0)) {
|
||||
char BIGSTK tempo4[HTS_URLMAXSIZE * 2];
|
||||
|
||||
tempo4[0] = '\0';
|
||||
@@ -2875,9 +2880,11 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
lastsaved = eadr - 1;
|
||||
}
|
||||
/*
|
||||
else if (opt->urlmode==1) { // ABSOLU, c'est le cas le moins courant
|
||||
else if (opt->urlmode==1) { // ABSOLU, c'est le cas le
|
||||
moins courant
|
||||
// NE FONCTIONNE PAS!! (et est inutile)
|
||||
if ((opt->getmode & 1) && (ptr>0)) { // ecrire les html
|
||||
if ((opt->getmode & 1) && (ptr>0)) { // ecrire les
|
||||
html
|
||||
// écrire le lien modifié, absolu
|
||||
HT_ADD("file:");
|
||||
if (*save=='/')
|
||||
@@ -2885,7 +2892,8 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
else
|
||||
HT_ADD(save)
|
||||
}
|
||||
lastsaved=eadr-1; // dernier écrit+1 (enfin euh apres on fait un ++ alors hein)
|
||||
lastsaved=eadr-1; // dernier écrit+1 (enfin euh apres
|
||||
on fait un ++ alors hein)
|
||||
}
|
||||
*/
|
||||
else if (opt->mimehtml) {
|
||||
@@ -2895,18 +2903,18 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
make_content_id(afs.af.adr, afs.af.fil, cid, sizeof(cid));
|
||||
HT_ADD_HTMLESCAPED(cid);
|
||||
lastsaved = eadr - 1; // dernier écrit+1 (enfin euh apres on fait un ++ alors hein)
|
||||
} else if (opt->urlmode == 3) { // URI absolue /
|
||||
if ((opt->getmode & 1) && (ptr > 0)) { // ecrire les html
|
||||
} else if (opt->urlmode == HTS_URLMODE_ABSOLUTE_URI) {
|
||||
if ((opt->getmode & HTS_GETMODE_HTML) && (ptr > 0)) {
|
||||
HT_ADD_HTMLESCAPED(afs.af.fil);
|
||||
}
|
||||
lastsaved = eadr - 1; // dernier écrit+1 (enfin euh apres on fait un ++ alors hein)
|
||||
} else if (opt->urlmode == 5) { // transparent proxy URL
|
||||
} else if (opt->urlmode == HTS_URLMODE_TRANSPARENT_PROXY) {
|
||||
char BIGSTK tempo[HTS_URLMAXSIZE * 2];
|
||||
const char *uri;
|
||||
int i;
|
||||
char *pos;
|
||||
|
||||
if ((opt->getmode & 1) && (ptr > 0)) { // ecrire les html
|
||||
if ((opt->getmode & HTS_GETMODE_HTML) && (ptr > 0)) {
|
||||
if (!link_has_authority(afs.af.adr)) {
|
||||
HT_ADD("http://");
|
||||
} else {
|
||||
@@ -2947,7 +2955,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
HT_ADD_HTMLESCAPED(tempo);
|
||||
}
|
||||
lastsaved = eadr - 1; // dernier écrit+1 (enfin euh apres on fait un ++ alors hein)
|
||||
} else if (opt->urlmode == 2) { // RELATIF
|
||||
} else if (opt->urlmode == HTS_URLMODE_RELATIVE) {
|
||||
char BIGSTK tempo[HTS_URLMAXSIZE * 2];
|
||||
|
||||
tempo[0] = '\0';
|
||||
@@ -3009,7 +3017,8 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
}
|
||||
|
||||
// érire codebase="chemin"
|
||||
if ((opt->getmode & 1) && (ptr > 0)) {
|
||||
if ((opt->getmode & HTS_GETMODE_HTML) &&
|
||||
(ptr > 0)) {
|
||||
char BIGSTK tempo4[HTS_URLMAXSIZE * 2];
|
||||
|
||||
tempo4[0] = '\0';
|
||||
@@ -3027,7 +3036,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
//lastsaved=adr; // dernier écrit+1
|
||||
}
|
||||
|
||||
if ((opt->getmode & 1) && (ptr > 0)) {
|
||||
if ((opt->getmode & HTS_GETMODE_HTML) && (ptr > 0)) {
|
||||
// convert to local codepage - NOT, already converted into %NN, and passed to the remote server so we do not have anything to do
|
||||
//if (str->page_charset_ != NULL && *str->page_charset_ != '\0') {
|
||||
// char *const local_save = hts_convertStringFromUTF8(tempo, strlen(tempo), str->page_charset_);
|
||||
@@ -3061,7 +3070,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
"Error building relative link %s and %s",
|
||||
afs.save, relativesavename());
|
||||
}
|
||||
} // sinon le lien sera écrit normalement
|
||||
} // sinon le lien sera écrit normalement
|
||||
|
||||
#if 0
|
||||
if (fexist(save)) { // le fichier existe..
|
||||
@@ -3089,7 +3098,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
opt->maxlink);
|
||||
hts_log_print(opt, LOG_INFO,
|
||||
"To avoid that: use #L option for more links (example: -#L1000000)");
|
||||
if ((opt->getmode & 1) && (ptr > 0)) {
|
||||
if ((opt->getmode & HTS_GETMODE_HTML) && (ptr > 0)) {
|
||||
if (fp) {
|
||||
fclose(fp);
|
||||
fp = NULL;
|
||||
@@ -3101,9 +3110,9 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
int pass_fix, dejafait = 0;
|
||||
|
||||
// Calculer la priorité de ce lien
|
||||
if ((opt->getmode & 4) == 0) { // traiter html après
|
||||
if ((opt->getmode & HTS_GETMODE_HTML_FIRST) == 0) {
|
||||
pass_fix = 0;
|
||||
} else { // vérifier que ce n'est pas un !html
|
||||
} else { // vérifier que ce n'est pas un !html
|
||||
if (!ishtml(opt, afs.af.fil))
|
||||
pass_fix = 1; // priorité inférieure (traiter après)
|
||||
else
|
||||
@@ -3167,7 +3176,8 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
if (checkrobots(_ROBOTS, afs.af.adr, "") == -1) { // robots.txt ?
|
||||
// enregistrer robots.txt (MACRO)
|
||||
if (!hts_record_link(opt, afs.af.adr, "/robots.txt", "", "", "", NULL)) {
|
||||
if ((opt->getmode & 1) && (ptr > 0)) {
|
||||
if ((opt->getmode & HTS_GETMODE_HTML) &&
|
||||
(ptr > 0)) {
|
||||
if (fp) {
|
||||
fclose(fp);
|
||||
fp = NULL;
|
||||
@@ -3206,7 +3216,8 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
// enregistrer
|
||||
if (!hts_record_link(opt, afs.af.adr, afs.af.fil, afs.save,
|
||||
former.adr, former.fil, codebase)) {
|
||||
if ((opt->getmode & 1) && (ptr > 0)) {
|
||||
if ((opt->getmode & HTS_GETMODE_HTML) &&
|
||||
(ptr > 0)) {
|
||||
if (fp) {
|
||||
fclose(fp);
|
||||
fp = NULL;
|
||||
@@ -3351,7 +3362,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
}
|
||||
// ----------
|
||||
// écrire peu à peu
|
||||
if ((opt->getmode & 1) && (ptr > 0))
|
||||
if ((opt->getmode & HTS_GETMODE_HTML) && (ptr > 0))
|
||||
HT_add_adr;
|
||||
lastsaved = html; // dernier écrit+1
|
||||
// ----------
|
||||
@@ -3411,7 +3422,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
opt->state._hts_in_html_parsing = 0; // flag
|
||||
opt->state._hts_cancel = 0; // pas de cancel
|
||||
|
||||
if ((opt->getmode & 1) && (ptr > 0)) {
|
||||
if ((opt->getmode & HTS_GETMODE_HTML) && (ptr > 0)) {
|
||||
{
|
||||
char *cAddr = TypedArrayElts(output_buffer);
|
||||
int cSize = (int) TypedArraySize(output_buffer);
|
||||
@@ -3443,7 +3454,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
|
||||
//
|
||||
} // if !error
|
||||
|
||||
if (opt->getmode & 1) {
|
||||
if (opt->getmode & HTS_GETMODE_HTML) {
|
||||
if (fp) {
|
||||
fclose(fp);
|
||||
fp = NULL;
|
||||
|
||||
@@ -1213,7 +1213,7 @@ HTSEXT_API find_handle hts_findfirst(char *path) {
|
||||
return NULL;
|
||||
}
|
||||
|
||||
HTSEXT_API int hts_findnext(find_handle find) {
|
||||
HTSEXT_API hts_boolean hts_findnext(find_handle find) {
|
||||
if (find) {
|
||||
#ifdef _WIN32
|
||||
if ((FindNextFileA(find->handle, &find->hdata)))
|
||||
@@ -1273,7 +1273,7 @@ HTSEXT_API int hts_findgetsize(find_handle find) {
|
||||
return -1;
|
||||
}
|
||||
|
||||
HTSEXT_API int hts_findisdir(find_handle find) {
|
||||
HTSEXT_API hts_boolean hts_findisdir(find_handle find) {
|
||||
if (find) {
|
||||
if (!hts_findissystem(find)) {
|
||||
#ifdef _WIN32
|
||||
@@ -1287,7 +1287,7 @@ HTSEXT_API int hts_findisdir(find_handle find) {
|
||||
}
|
||||
return 0;
|
||||
}
|
||||
HTSEXT_API int hts_findisfile(find_handle find) {
|
||||
HTSEXT_API hts_boolean hts_findisfile(find_handle find) {
|
||||
if (find) {
|
||||
if (!hts_findissystem(find)) {
|
||||
#ifdef _WIN32
|
||||
@@ -1301,7 +1301,7 @@ HTSEXT_API int hts_findisfile(find_handle find) {
|
||||
}
|
||||
return 0;
|
||||
}
|
||||
HTSEXT_API int hts_findissystem(find_handle find) {
|
||||
HTSEXT_API hts_boolean hts_findissystem(find_handle find) {
|
||||
if (find) {
|
||||
#ifdef _WIN32
|
||||
if (find->hdata.
|
||||
|
||||
@@ -108,15 +108,15 @@ HTSEXT_API int hts_buildtopindex(httrackp * opt, const char *path,
|
||||
// Portable directory find functions
|
||||
// Directory find functions
|
||||
HTSEXT_API find_handle hts_findfirst(char *path);
|
||||
HTSEXT_API int hts_findnext(find_handle find);
|
||||
HTSEXT_API hts_boolean hts_findnext(find_handle find);
|
||||
HTSEXT_API int hts_findclose(find_handle find);
|
||||
|
||||
//
|
||||
HTSEXT_API char *hts_findgetname(find_handle find);
|
||||
HTSEXT_API int hts_findgetsize(find_handle find);
|
||||
HTSEXT_API int hts_findisdir(find_handle find);
|
||||
HTSEXT_API int hts_findisfile(find_handle find);
|
||||
HTSEXT_API int hts_findissystem(find_handle find);
|
||||
HTSEXT_API hts_boolean hts_findisdir(find_handle find);
|
||||
HTSEXT_API hts_boolean hts_findisfile(find_handle find);
|
||||
HTSEXT_API hts_boolean hts_findissystem(find_handle find);
|
||||
|
||||
#endif
|
||||
|
||||
|
||||
158
src/htswizard.c
158
src/htswizard.c
@@ -178,7 +178,7 @@ static int hts_acceptlink_(httrackp * opt, int ptr,
|
||||
// -------------------- PHASE 1 --------------------
|
||||
|
||||
/* Doit-on traiter les non html? */
|
||||
if ((opt->getmode & 2) == 0) { // non on ne doit pas
|
||||
if ((opt->getmode & HTS_GETMODE_NONHTML) == 0) { // non on ne doit pas
|
||||
if (!ishtml(opt, fil)) { // non il ne faut pas
|
||||
//adr[0]='\0'; // ne pas traiter ce lien, pas traiter
|
||||
forbidden_url = 1; // interdire récupération du lien
|
||||
@@ -266,11 +266,11 @@ static int hts_acceptlink_(httrackp * opt, int ptr,
|
||||
test2 =
|
||||
(strchr(tempo2 + ((*tempo2 == '/') ? 1 : 0), '/') != NULL);
|
||||
if ((test1) && (test2)) { // on ne peut que descendre
|
||||
if ((opt->seeker & 1) == 0) { // interdiction de descendre
|
||||
if ((opt->seeker & HTS_SEEKER_DOWN) == 0) {
|
||||
forbidden_url = 1;
|
||||
hts_log_print(opt, LOG_DEBUG, "lower link canceled: %s%s", adr,
|
||||
fil);
|
||||
} else { // autorisé à priori - NEW
|
||||
} else { // autorisé à priori - NEW
|
||||
if (!heap(ptr)->link_import) { // ne résulte pas d'un 'moved'
|
||||
forbidden_url = 0;
|
||||
hts_log_print(opt, LOG_DEBUG, "lower link authorized: %s%s",
|
||||
@@ -278,7 +278,7 @@ static int hts_acceptlink_(httrackp * opt, int ptr,
|
||||
}
|
||||
}
|
||||
} else if ((test1) || (test2)) { // on peut descendre pour accéder au lien
|
||||
if ((opt->seeker & 1) != 0) { // on peut descendre - NEW
|
||||
if ((opt->seeker & HTS_SEEKER_DOWN) != 0) {
|
||||
if (!heap(ptr)->link_import) { // ne résulte pas d'un 'moved'
|
||||
forbidden_url = 0;
|
||||
hts_log_print(opt, LOG_DEBUG, "lower link authorized: %s%s",
|
||||
@@ -290,11 +290,11 @@ static int hts_acceptlink_(httrackp * opt, int ptr,
|
||||
|
||||
// up
|
||||
if ((!strncmp(tempo, "../", 3)) && (!strncmp(tempo2, "../", 3))) { // impossible sans monter
|
||||
if ((opt->seeker & 2) == 0) { // interdiction de monter
|
||||
if ((opt->seeker & HTS_SEEKER_UP) == 0) {
|
||||
forbidden_url = 1;
|
||||
hts_log_print(opt, LOG_DEBUG, "upper link canceled: %s%s", adr,
|
||||
fil);
|
||||
} else { // autorisé à monter - NEW
|
||||
} else { // autorisé à monter - NEW
|
||||
if (!heap(ptr)->link_import) { // ne résulte pas d'un 'moved'
|
||||
forbidden_url = 0;
|
||||
hts_log_print(opt, LOG_DEBUG, "upper link authorized: %s%s",
|
||||
@@ -302,13 +302,13 @@ static int hts_acceptlink_(httrackp * opt, int ptr,
|
||||
}
|
||||
}
|
||||
} else if ((!strncmp(tempo, "../", 3)) || (!strncmp(tempo2, "../", 3))) { // Possible en montant
|
||||
if ((opt->seeker & 2) != 0) { // autorisé à monter - NEW
|
||||
if ((opt->seeker & HTS_SEEKER_UP) != 0) {
|
||||
if (!heap(ptr)->link_import) { // ne résulte pas d'un 'moved'
|
||||
forbidden_url = 0;
|
||||
hts_log_print(opt, LOG_DEBUG, "upper link authorized: %s%s",
|
||||
adr, fil);
|
||||
}
|
||||
} // sinon autorisé en descente
|
||||
} // sinon autorisé en descente
|
||||
}
|
||||
|
||||
} else {
|
||||
@@ -345,83 +345,81 @@ static int hts_acceptlink_(httrackp * opt, int ptr,
|
||||
|
||||
//if (!opt->wizard) { // mode non wizard
|
||||
// doit-on traiter ce lien?.. vérifier droits de sortie
|
||||
switch ((opt->travel & 255)) {
|
||||
case 0:
|
||||
switch ((opt->travel & HTS_TRAVEL_SCOPE_MASK)) {
|
||||
case HTS_TRAVEL_SAME_ADDRESS:
|
||||
if (!opt->wizard) // mode non wizard
|
||||
forbidden_url = 1;
|
||||
break; // interdicton de sortir au dela de l'adresse
|
||||
case 1:{ // sortie sur le même dom.xxx
|
||||
size_t i = strlen(adr) - 1;
|
||||
size_t j = strlen(urladr()) - 1;
|
||||
case HTS_TRAVEL_SAME_DOMAIN: {
|
||||
size_t i = strlen(adr) - 1;
|
||||
size_t j = strlen(urladr()) - 1;
|
||||
|
||||
if ((i > 0) && (j > 0)) {
|
||||
while((i > 0) && (adr[i] != '.'))
|
||||
i--;
|
||||
while((j > 0) && (urladr()[j] != '.'))
|
||||
j--;
|
||||
if ((i > 0) && (j > 0)) {
|
||||
i--;
|
||||
j--;
|
||||
while((i > 0) && (adr[i] != '.'))
|
||||
i--;
|
||||
while((j > 0) && (urladr()[j] != '.'))
|
||||
j--;
|
||||
}
|
||||
}
|
||||
if ((i > 0) && (j > 0)) {
|
||||
if (!strfield2(adr + i, urladr() + j)) { // !=
|
||||
if (!opt->wizard) { // mode non wizard
|
||||
//printf("refused: %s\n",adr);
|
||||
forbidden_url = 1; // pas même domaine
|
||||
hts_log_print(opt, LOG_DEBUG,
|
||||
"foreign domain link canceled: %s%s", adr, fil);
|
||||
}
|
||||
|
||||
} else {
|
||||
if (opt->wizard) { // mode wizard
|
||||
forbidden_url = 0; // même domaine
|
||||
hts_log_print(opt, LOG_DEBUG, "same domain link authorized: %s%s",
|
||||
adr, fil);
|
||||
}
|
||||
}
|
||||
|
||||
} else
|
||||
forbidden_url = 1;
|
||||
}
|
||||
break;
|
||||
case 2:{ // sortie sur le même .xxx
|
||||
size_t i = strlen(adr) - 1;
|
||||
size_t j = strlen(urladr()) - 1;
|
||||
|
||||
while((i > 0) && (adr[i] != '.'))
|
||||
if ((i > 0) && (j > 0)) {
|
||||
while ((i > 0) && (adr[i] != '.'))
|
||||
i--;
|
||||
while((j > 0) && (urladr()[j] != '.'))
|
||||
while ((j > 0) && (urladr()[j] != '.'))
|
||||
j--;
|
||||
if ((i > 0) && (j > 0)) {
|
||||
if (!strfield2(adr + i, urladr() + j)) { // !-
|
||||
if (!opt->wizard) { // mode non wizard
|
||||
//printf("refused: %s\n",adr);
|
||||
forbidden_url = 1; // pas même .xx
|
||||
hts_log_print(opt, LOG_DEBUG,
|
||||
"foreign location link canceled: %s%s", adr, fil);
|
||||
}
|
||||
} else {
|
||||
if (opt->wizard) { // mode wizard
|
||||
forbidden_url = 0; // même domaine
|
||||
hts_log_print(opt, LOG_DEBUG,
|
||||
"same location link authorized: %s%s", adr, fil);
|
||||
}
|
||||
}
|
||||
} else
|
||||
forbidden_url = 1;
|
||||
i--;
|
||||
j--;
|
||||
while ((i > 0) && (adr[i] != '.'))
|
||||
i--;
|
||||
while ((j > 0) && (urladr()[j] != '.'))
|
||||
j--;
|
||||
}
|
||||
}
|
||||
break;
|
||||
case 7: // everywhere!!
|
||||
if ((i > 0) && (j > 0)) {
|
||||
if (!strfield2(adr + i, urladr() + j)) { // !=
|
||||
if (!opt->wizard) { // mode non wizard
|
||||
// printf("refused: %s\n",adr);
|
||||
forbidden_url = 1; // pas même domaine
|
||||
hts_log_print(opt, LOG_DEBUG, "foreign domain link canceled: %s%s",
|
||||
adr, fil);
|
||||
}
|
||||
|
||||
} else {
|
||||
if (opt->wizard) { // mode wizard
|
||||
forbidden_url = 0; // même domaine
|
||||
hts_log_print(opt, LOG_DEBUG, "same domain link authorized: %s%s",
|
||||
adr, fil);
|
||||
}
|
||||
}
|
||||
|
||||
} else
|
||||
forbidden_url = 1;
|
||||
} break;
|
||||
case HTS_TRAVEL_SAME_TLD: {
|
||||
size_t i = strlen(adr) - 1;
|
||||
size_t j = strlen(urladr()) - 1;
|
||||
|
||||
while ((i > 0) && (adr[i] != '.'))
|
||||
i--;
|
||||
while ((j > 0) && (urladr()[j] != '.'))
|
||||
j--;
|
||||
if ((i > 0) && (j > 0)) {
|
||||
if (!strfield2(adr + i, urladr() + j)) { // !-
|
||||
if (!opt->wizard) { // mode non wizard
|
||||
// printf("refused: %s\n",adr);
|
||||
forbidden_url = 1; // pas même .xx
|
||||
hts_log_print(opt, LOG_DEBUG,
|
||||
"foreign location link canceled: %s%s", adr, fil);
|
||||
}
|
||||
} else {
|
||||
if (opt->wizard) { // mode wizard
|
||||
forbidden_url = 0; // même domaine
|
||||
hts_log_print(opt, LOG_DEBUG, "same location link authorized: %s%s",
|
||||
adr, fil);
|
||||
}
|
||||
}
|
||||
} else
|
||||
forbidden_url = 1;
|
||||
} break;
|
||||
case HTS_TRAVEL_EVERYWHERE:
|
||||
if (opt->wizard) { // mode wizard
|
||||
forbidden_url = 0;
|
||||
break;
|
||||
}
|
||||
} // switch
|
||||
} // switch
|
||||
|
||||
// ANCIENNE POS -- récupérer les liens à côtés d'un lien (nearlink)
|
||||
|
||||
@@ -583,7 +581,7 @@ static int hts_acceptlink_(httrackp * opt, int ptr,
|
||||
// on doit poser la question.. peut on la poser?
|
||||
// (oui je sais quel preuve de délicatesse, merci merci)
|
||||
if ((question) && (ptr > 0) && (!force_mirror)) {
|
||||
if (opt->wizard == 2) { // éliminer tous les liens non répertoriés comme autorisés (ou inconnus)
|
||||
if (opt->wizard == HTS_WIZARD_AUTO) {
|
||||
question = 0;
|
||||
forbidden_url = 1;
|
||||
hts_log_print(opt, LOG_DEBUG,
|
||||
@@ -600,8 +598,8 @@ static int hts_acceptlink_(httrackp * opt, int ptr,
|
||||
printf("robots.txt forbidden: %s%s\n", adr, fil);
|
||||
#endif
|
||||
// question résolue, par les filtres, et mode robot non strict
|
||||
if ((!question) && (filters_answer) && (opt->robots == 1)
|
||||
&& (forbidden_url != 1)) {
|
||||
if ((!question) && (filters_answer) &&
|
||||
(opt->robots == HTS_ROBOTS_SOMETIMES) && (forbidden_url != 1)) {
|
||||
r = 0; // annuler interdiction des robots
|
||||
if (!forbidden_url) {
|
||||
hts_log_print(opt, LOG_DEBUG,
|
||||
@@ -685,7 +683,7 @@ static int hts_acceptlink_(httrackp * opt, int ptr,
|
||||
io_flush;
|
||||
} else { // lien primaire: autoriser répertoire entier
|
||||
if (!force_mirror) {
|
||||
if ((opt->seeker & 1) == 0) { // interdiction de descendre
|
||||
if ((opt->seeker & HTS_SEEKER_DOWN) == 0) {
|
||||
n = 7;
|
||||
} else {
|
||||
n = 5; // autoriser miroir répertoires descendants (lien primaire)
|
||||
@@ -712,7 +710,7 @@ static int hts_acceptlink_(httrackp * opt, int ptr,
|
||||
switch (n) {
|
||||
case -1: // sauter tout le reste
|
||||
forbidden_url = 1;
|
||||
opt->wizard = 2; // sauter tout le reste
|
||||
opt->wizard = HTS_WIZARD_AUTO; // sauter tout le reste
|
||||
break;
|
||||
case 0: // forbid the same link: adr/fil
|
||||
forbidden_url = 1;
|
||||
@@ -796,7 +794,7 @@ static int hts_acceptlink_(httrackp * opt, int ptr,
|
||||
break;
|
||||
|
||||
case 5: // allow the whole directory and its children
|
||||
if ((opt->seeker & 2) == 0) { // not allowed to go up
|
||||
if ((opt->seeker & HTS_SEEKER_UP) == 0) { // not allowed to go up
|
||||
size_t i = strlen(fil) - 1;
|
||||
|
||||
while((fil[i] != '/') && (i > 0))
|
||||
@@ -872,7 +870,7 @@ static int hts_acceptlink_(httrackp * opt, int ptr,
|
||||
// lien non autorisé, peut-on juste le tester?
|
||||
if (just_test_it) {
|
||||
if (forbidden_url == 1) {
|
||||
if (opt->travel & 256) { // tester tout de même
|
||||
if (opt->travel & HTS_TRAVEL_TEST_ALL) { // tester tout de même
|
||||
if (strfield(adr, "ftp://") == 0) { // PAS ftp!
|
||||
forbidden_url = 1; // oui oui toujours interdit (note: sert à rien car ==1 mais c pour comprendre)
|
||||
*just_test_it = 1; // mais on teste
|
||||
|
||||
@@ -206,7 +206,8 @@ HTSEXT_API htsErrorCallback hts_get_error_callback(void);
|
||||
/* Logging */
|
||||
/** Legacy: write prefix then msg to opt->log. Returns 0 if written, 1 if
|
||||
opt->log is NULL. Prefer hts_log_print(). */
|
||||
HTSEXT_API int hts_log(httrackp * opt, const char *prefix, const char *msg);
|
||||
HTSEXT_API hts_boolean hts_log(httrackp *opt, const char *prefix,
|
||||
const char *msg);
|
||||
|
||||
/** printf-style log at level @p type (an hts_log_type, optionally |LOG_ERRNO).
|
||||
Forwards to the registered log callback, and when the level is <= opt->debug
|
||||
@@ -313,7 +314,8 @@ HTSEXT_API T_SOC catch_url_init(int *port, char *adr);
|
||||
"ip:port". The buffers are caller-allocated and not bounds-checked: @p data
|
||||
must be CATCH_URL_DATA_SIZE bytes, and @p url / @p method must fit the
|
||||
captured request line. */
|
||||
HTSEXT_API int catch_url(T_SOC soc, char *url, char *method, char *data);
|
||||
HTSEXT_API hts_boolean catch_url(T_SOC soc, char *url, char *method,
|
||||
char *data);
|
||||
|
||||
/* State */
|
||||
/** Whether the engine is parsing HTML. Returns 0 if not, otherwise the percent
|
||||
@@ -334,10 +336,10 @@ HTSEXT_API int hts_is_exiting(httrackp * opt);
|
||||
caller-owned, NULL-terminated array of strings; the engine stores the
|
||||
pointer without copying, so the array and its strings must stay valid until
|
||||
the engine consumes them. @return nonzero if a list is now set. */
|
||||
HTSEXT_API int hts_addurl(httrackp * opt, char **url);
|
||||
HTSEXT_API hts_boolean hts_addurl(httrackp *opt, char **url);
|
||||
|
||||
/** Clear any pending add-URL list set by hts_addurl(). Always returns 0. */
|
||||
HTSEXT_API int hts_resetaddurl(httrackp * opt);
|
||||
HTSEXT_API hts_boolean hts_resetaddurl(httrackp *opt);
|
||||
|
||||
/** Apply the runtime-tunable options from @p from onto @p to, to adjust a live
|
||||
mirror. Only fields set to a non-sentinel value are copied; the rest of @p
|
||||
@@ -356,7 +358,7 @@ HTSEXT_API int hts_setpause(httrackp * opt, int);
|
||||
lock, so it is safe to call from another thread). @p force is currently
|
||||
ignored.
|
||||
@return 0; no-op if @p opt is NULL. */
|
||||
HTSEXT_API int hts_request_stop(httrackp * opt, int force);
|
||||
HTSEXT_API int hts_request_stop(httrackp *opt, hts_boolean force);
|
||||
|
||||
/** Queue a single in-progress file, by URL, to be cancelled by the engine.
|
||||
@p url is copied internally. Takes the state lock, so it is thread-safe.
|
||||
@@ -373,7 +375,7 @@ HTSEXT_API void hts_cancel_parsing(httrackp * opt);
|
||||
|
||||
/** Nonzero once the mirror has fully ended. Read under the engine state lock,
|
||||
so safe to poll from another thread. Wait for this before hts_free_opt(). */
|
||||
HTSEXT_API int hts_has_stopped(httrackp * opt);
|
||||
HTSEXT_API hts_boolean hts_has_stopped(httrackp *opt);
|
||||
|
||||
/* Tools */
|
||||
/** Ensure the directory chain leading to @p path exists, creating missing
|
||||
@@ -390,7 +392,7 @@ HTSEXT_API int structcheck_utf8(const char *path);
|
||||
/** Whether the directory containing @p path exists. The basename is stripped
|
||||
first, so passing a file path tests its parent directory. @return 1 if it is
|
||||
a directory, 0 otherwise. */
|
||||
HTSEXT_API int dir_exists(const char *path);
|
||||
HTSEXT_API hts_boolean dir_exists(const char *path);
|
||||
|
||||
/** Write the HTTP reason phrase for @p statuscode into @p msg, a caller buffer
|
||||
of at least 64 bytes. For an unknown code a non-empty @p msg is kept,
|
||||
@@ -573,14 +575,15 @@ HTSEXT_API char *unescape_http(char *const catbuff, const size_t size, const cha
|
||||
must-avoid escapes are kept encoded, and %25 is never decoded). @p no_high &
|
||||
1 also decodes high (>= 128) bytes; @p no_high & 2 also decodes an escaped
|
||||
space. Returns @p catbuff. */
|
||||
HTSEXT_API char *unescape_http_unharm(char *const catbuff, const size_t size, const char *s, const int no_high);
|
||||
HTSEXT_API char *unescape_http_unharm(char *const catbuff, const size_t size,
|
||||
const char *s, const hts_boolean no_high);
|
||||
|
||||
/** Determine the MIME type of local file name @p fil into @p s (capacity
|
||||
@p ssize): user --assume rules, then ".html", then the built-in extension
|
||||
table. @p flag != 0 forces a fallback type. @return 1 if a type was written,
|
||||
0 otherwise. */
|
||||
HTSEXT_API int get_httptype_sized(httrackp *opt, char *s, size_t ssize,
|
||||
const char *fil, int flag);
|
||||
HTSEXT_API hts_boolean get_httptype_sized(httrackp *opt, char *s, size_t ssize,
|
||||
const char *fil, hts_boolean flag);
|
||||
|
||||
/** @deprecated Use get_httptype_sized(). Assumes @p s has at least
|
||||
HTS_MIMETYPE_SIZE capacity. */
|
||||
@@ -600,7 +603,7 @@ HTSEXT_API int is_userknowntype(httrackp * opt, const char *fil);
|
||||
|
||||
/** 1 if @p fil, an extension such as "asp" or "php" (not a full filename), is a
|
||||
known dynamic-page type, else 0. */
|
||||
HTSEXT_API int is_dyntype(const char *fil);
|
||||
HTSEXT_API hts_boolean is_dyntype(const char *fil);
|
||||
|
||||
/** Extract the extension of @p fil (text after the last '.', stopping at '?')
|
||||
into caller scratch @p catbuff (capacity @p size) and return it. Returns ""
|
||||
@@ -610,12 +613,12 @@ HTSEXT_API const char *get_ext(char *catbuff, size_t size, const char *fil);
|
||||
|
||||
/** 1 if MIME type @p st must not be reclassified or renamed (hypertext types
|
||||
and a built-in keep-list of commonly mislabeled types), else 0. */
|
||||
HTSEXT_API int may_unknown(httrackp * opt, const char *st);
|
||||
HTSEXT_API hts_boolean may_unknown(httrackp *opt, const char *st);
|
||||
|
||||
/** Guess the MIME type of local file @p fil into @p s (capacity @p ssize),
|
||||
always producing a type. @return 1 if a type was written. */
|
||||
HTSEXT_API int guess_httptype_sized(httrackp *opt, char *s, size_t ssize,
|
||||
const char *fil);
|
||||
HTSEXT_API hts_boolean guess_httptype_sized(httrackp *opt, char *s,
|
||||
size_t ssize, const char *fil);
|
||||
|
||||
/** @deprecated Use guess_httptype_sized(). Assumes @p s has at least
|
||||
HTS_MIMETYPE_SIZE capacity. */
|
||||
@@ -677,7 +680,7 @@ HTSEXT_API find_handle hts_findfirst(char *path);
|
||||
|
||||
/** Advance to the next directory entry. Returns 1 if an entry is available, 0
|
||||
at end of directory. */
|
||||
HTSEXT_API int hts_findnext(find_handle find);
|
||||
HTSEXT_API hts_boolean hts_findnext(find_handle find);
|
||||
|
||||
/** Close the iteration and free @p find. Always returns 0; NULL is accepted. */
|
||||
HTSEXT_API int hts_findclose(find_handle find);
|
||||
@@ -692,16 +695,16 @@ HTSEXT_API int hts_findgetsize(find_handle find);
|
||||
|
||||
/** 1 if the current entry is a directory, else 0 (a system/special entry, see
|
||||
hts_findissystem(), reports 0). */
|
||||
HTSEXT_API int hts_findisdir(find_handle find);
|
||||
HTSEXT_API hts_boolean hts_findisdir(find_handle find);
|
||||
|
||||
/** 1 if the current entry is a regular file, else 0 (a system/special entry,
|
||||
see hts_findissystem(), reports 0). */
|
||||
HTSEXT_API int hts_findisfile(find_handle find);
|
||||
HTSEXT_API hts_boolean hts_findisfile(find_handle find);
|
||||
|
||||
/** 1 if the current entry is a special/system entry to skip: "." or "..", on
|
||||
POSIX also device/fifo/socket nodes, on Windows also system, hidden or
|
||||
temporary entries. Else 0. */
|
||||
HTSEXT_API int hts_findissystem(find_handle find);
|
||||
HTSEXT_API hts_boolean hts_findissystem(find_handle find);
|
||||
|
||||
/* UTF-8 aware FILE API */
|
||||
/* On non-Windows these macros resolve directly to the POSIX calls. On Windows
|
||||
|
||||
Reference in New Issue
Block a user