commit 094e8a216592321bd680e6705f3981d6bf78c927
parent 463c035f38af82db2a7c26b227976f28b81b452f
Author: Tomas Hlavaty <tom@logand.com>
Date: Wed, 19 Jan 2011 09:21:32 +0100
config moved to ~/.w3mail dir, tidy introduced
Diffstat:
5 files changed, 117 insertions(+), 29 deletions(-)
diff --git a/postfeed.sh b/postfeed.sh
@@ -1,3 +1,2 @@
#!/bin/sh
-
-cat ~/.postfeed | xargs -n1 ./rss2links.sh | grep -v '^$' >>~/.w3mail.in
+cat ~/.w3mail/feed | xargs -n1 ./rss2links.sh | grep -v '^$' >>~/.w3mail/in
diff --git a/utils.c b/utils.c
@@ -91,6 +91,15 @@ void dir(void *udata, dir_cb cb, char *path, ...) {
va_end(v);
}
+int systemf(char *cmd, ...) {
+ va_list v;
+ va_start(v, cmd);
+ char buf[BLEN];
+ vsnprintf(buf, BLEN, cmd, v);
+ va_end(v);
+ return system(buf);
+}
+
void wget(char *url, char *fname) {
// http://www.gnu.org/software/wget/manual/html_node/Exit-Status.html
static char *wgetmsg[] = {
diff --git a/utils.h b/utils.h
@@ -22,6 +22,7 @@ extern void out(void *udata, out_cb cb, char *cmd, ...);
extern void tmpf(void *udata, tmpf_cb cb, char *fname, ...);
extern void dir(void *udata, dir_cb cb, char *path, ...);
+extern int systemf(char *cmd, ...);
extern void wget(char *url, char *fname);
#endif /* UTILS_H */
diff --git a/w3mail.c b/w3mail.c
@@ -20,6 +20,7 @@
*/
+#include <string.h>
#include <unistd.h>
#include <time.h>
@@ -42,26 +43,68 @@ static void cb_echo(void *udata, FILE* in) {
fwrite(buf, sizeof(char), n, out);
}
+static void now(char *buf) {
+ time_t t = time(NULL);
+ struct tm *tm = localtime(&t);
+ strftime(buf, BLEN, "%a, %d %b %Y %T %z", tm);
+}
+
+#define pr(...) {fprintf(out, __VA_ARGS__); fprintf(out, "\n");}
+
+struct tidy {
+ char *fin;
+ char *fout;
+ char *url;
+};
+
+static void cb_tidy(void *udata, FILE* in) {
+ struct tidy *x = udata;
+ char name[BLEN], arg[BLEN];
+ strcpy(name, "bbc");
+ strcpy(arg, "");
+ // TODO compute filter name from url
+ /* while(!feof(in)) { */
+ /* char line[BLEN]; */
+ /* if(!fgets(line, BLEN, in)) break; */
+ /* rtrim(line); */
+ /* // TODO run filter based on url */
+ /* // if head(url) then */
+ /* // if filters/%s filter exist */
+ /* // in(???, ???, "filters/%s '%s'", filter, arg); */
+ /* } */
+ // w3m -dump /tmp/a.html
+ systemf("~/.w3mail/filter/%s '%s' <%s >%s", name, arg, x->fin, x->fout);
+}
+
+static void tidy(char *fin, char *fout, char *url) {
+ struct tidy x = {fin, fout, url};
+ in(&x, cb_tidy, "cat ~/.w3mail/tidy");
+}
+
+struct xtidy {
+ FILE *out;
+ char *fname;
+ char *url;
+};
+
+static void cb_xtidy(void *udata, char *fname) {
+ struct xtidy *x = udata;
+ tidy(x->fname, fname, x->url);
+ in(x->out, cb_echo, "base64 %s", fname);
+ //in(x->out, cb_echo, "w3m -dump -T text/html %s", fname);
+}
+
struct sendmail {
char *from;
char *to;
char *id;
char *subj;
+ char *url;
+ char *mime;
+ int html;
char *fname;
};
-#define pr(...) {fprintf(out, __VA_ARGS__); fprintf(out, "\n");}
-
-static void cb_read_first_line(void *udata, FILE* in) {
- rtrim(fgets(udata, BLEN, in));
-}
-
-static void now(char *buf) {
- time_t t = time(NULL);
- struct tm *tm = localtime(&t);
- strftime(buf, BLEN, "%a, %d %b %Y %T %z", tm);
-}
-
static void cb_sendmail(void *udata, FILE* out) {
struct sendmail *x = udata;
pr("From: %s", x->from);
@@ -70,18 +113,35 @@ static void cb_sendmail(void *udata, FILE* out) {
pr("Subject: %s", x->subj);
pr("User-Agent: w3mail");
pr("MIME-Version: 1.0");
+ //pr("Content-Type: %s", x->html ? "text/plain; charset=utf8" : x->mime);
+ pr("Content-Type: %s", x->mime);
char buf[BLEN];
- //in(buf, cb_read_first_line, "date -R");
now(buf);
- pr("Date: %s", buf);
- in(buf, cb_read_first_line, "file -bi %s", x->fname);
- // TODO fix application/xml if html
- pr("Content-Type: %s", buf);
- pr("Content-Transfer-Encoding: base64");
+ pr("Date: %s", buf); // TODO from feed?
+ //pr("Content-Transfer-Encoding: %s", x->html ? "8bit" : "base64");
+ pr("Content-Transfer-Encoding: %s", "base64");
pr("%s", "");
- // TODO if html > use xpath to select interesting content
- // TODO if html > filter out javascript (script tags)
- in(out, cb_echo, "base64 %s", x->fname);
+ if(x->html) {
+ struct xtidy y = {out, x->fname, x->url};
+ tmpf(&y, cb_xtidy, "/tmp/w3mail-XXXXXX");
+ }
+ else in(out, cb_echo, "base64 %s", x->fname);
+}
+
+static void cb_read_first_line(void *udata, FILE* in) {
+ rtrim(fgets(udata, BLEN, in));
+}
+
+static int head(char *sub, char *str) {
+ return 0 == strncmp(sub, str, strlen(sub));
+}
+
+static int equal(char *str1, char *str2) {
+ return 0 == strcmp(str1, str2);
+}
+
+static int is_html(char *elt) {
+ return equal("html", elt) || equal("xhtml", elt);
}
struct w3mail {
@@ -94,12 +154,30 @@ struct w3mail {
static void cb_w3mail(void *udata, char *fname) {
struct w3mail *x = udata;
- char sum[BLEN], id[BLEN], subj[BLEN];
+ char sum[BLEN], id[BLEN], subj[BLEN], mime[BLEN];
+ int html = 0;
wget(x->url, fname);
+ in(mime, cb_read_first_line, "file -bi %s", fname);
+ if(head("application/xml;", mime)) {
+ char elt[BLEN];
+ in(elt, cb_read_first_line, "xmlstarlet el %s", fname);
+ if(equal("rss", elt)) {
+ // TODO handle rss feed
+ }
+ if(equal("atom", elt)) {
+ // TODO handle atom feed
+ }
+ else if(is_html(elt)) {
+ // TODO fix mime application/xml to html
+ html = 1;
+ }
+ else html = is_html(elt);
+ }
+ else html = head("text/html;", mime);
md5sum(fname, sum);
snprintf(id, BLEN, "<%s@%s>", sum, x->host);
- snprintf(subj, BLEN, "[w3mail] %s", x->url);
- struct sendmail y = {x->from, x->to, id, subj, fname};
+ snprintf(subj, BLEN, "[w3mail] %s", x->url); // TODO from feed?
+ struct sendmail y = {x->from, x->to, id, subj, x->url, mime, html, fname};
out(&y, cb_sendmail, "%s", x->cmd);
}
@@ -119,7 +197,7 @@ int main(int argc, char *argv[]) {
if(argc == 1) {
char cmd[BLEN], from[BLEN], to[BLEN], host[BLEN], url[BLEN];
struct w3mail x = {cmd, from, to, host, url};
- in(&x, cb_config, "cat ~/.w3mail");
+ in(&x, cb_config, "cat ~/.w3mail/config");
while(!feof(stdin)) {
char *line = fgets(url, BLEN, stdin);
if(!line) break;
@@ -130,7 +208,7 @@ int main(int argc, char *argv[]) {
else if(argc == 2) {
char cmd[BLEN], from[BLEN], to[BLEN], host[BLEN];
struct w3mail x = {cmd, from, to, host, argv[1]};
- in(&x, cb_config, "cat ~/.w3mail");
+ in(&x, cb_config, "cat ~/.w3mail/config");
run(&x);
}
else if(argc == 6) {
diff --git a/w3maild.sh b/w3maild.sh
@@ -1 +1,2 @@
-echo >~/.w3mail.in; tail -f ~/.w3mail.in | xargs -n1 -P20 w3mail 2>>~/.w3mail.log
+#!/bin/sh
+echo >~/.w3mail/in; tail -f ~/.w3mail/in | xargs -n1 -P20 w3mail 2>>~/.w3mail/log
