sfeed

Simple RSS and Atom feed parser
git clone https://git.sinitax.com/codemadness/sfeed
Log | Files | Refs | README | LICENSE | Upstream | sfeed.txt

sfeed_mbox.c (4890B)


      1#include <stdio.h>
      2#include <stdlib.h>
      3#include <string.h>
      4#include <time.h>
      5#include <unistd.h>
      6
      7#include "util.h"
      8
      9static char *line;
     10static size_t linesize;
     11static char host[256], *user, dtimebuf[32], mtimebuf[32];
     12static int usecontent = 0; /* env variable: $SFEED_MBOX_CONTENT */
     13
     14static unsigned long long
     15djb2(unsigned char *s, unsigned long long hash)
     16{
     17	int c;
     18
     19	while ((c = *s++))
     20		hash = ((hash << 5) + hash) + c; /* hash * 33 + c */
     21	return hash;
     22}
     23
     24/* Unescape / decode fields printed by string_print_encoded()
     25 * "\\" to "\", "\t", to TAB, "\n" to newline. Other escape sequences are
     26 * ignored: "\z" etc. Mangle "From " in mboxrd style (always prefix >). */
     27static void
     28printcontent(const char *s, FILE *fp)
     29{
     30escapefrom:
     31	for (; *s == '>'; s++)
     32		putc('>', fp);
     33	/* escape "From ", mboxrd-style. */
     34	if (!strncmp(s, "From ", 5))
     35		putc('>', fp);
     36
     37	for (; *s; s++) {
     38		switch (*s) {
     39		case '\\':
     40			if (*(s + 1) == '\0')
     41				break;
     42			s++;
     43			switch (*s) {
     44			case 'n':
     45				putc('\n', fp);
     46				s++;
     47				goto escapefrom;
     48			case '\\': putc('\\', fp); break;
     49			case 't':  putc('\t', fp); break;
     50			}
     51			break;
     52		default:
     53			putc(*s, fp); break;
     54		}
     55	}
     56}
     57
     58static void
     59printfeed(FILE *fp, const char *feedname)
     60{
     61	char *fields[FieldLast], timebuf[32];
     62	struct tm parsedtm, *tm;
     63	time_t parsedtime;
     64	unsigned long long hash;
     65	ssize_t linelen;
     66	int ishtml;
     67
     68	while ((linelen = getline(&line, &linesize, fp)) > 0 &&
     69	       !ferror(stdout)) {
     70		if (line[linelen - 1] == '\n')
     71			line[--linelen] = '\0';
     72		hash = djb2((unsigned char *)line, 5381ULL);
     73		parseline(line, fields);
     74
     75		/* mbox + mail header */
     76		printf("From MAILER-DAEMON %s\n", mtimebuf);
     77
     78		parsedtime = 0;
     79		if (!strtotime(fields[FieldUnixTimestamp], &parsedtime) &&
     80		    (tm = gmtime_r(&parsedtime, &parsedtm)) &&
     81		    strftime(timebuf, sizeof(timebuf), "%a, %d %b %Y %H:%M:%S +0000", tm)) {
     82			printf("Date: %s\n", timebuf);
     83		} else {
     84			printf("Date: %s\n", dtimebuf); /* invalid/missing: use current time */
     85		}
     86
     87		printf("From: %s <anonymous@>\n", fields[FieldAuthor][0] ? fields[FieldAuthor] : feedname);
     88		printf("To: %s <%s@%s>\n", user, user, host);
     89		printf("Subject: %s\n", fields[FieldTitle]);
     90		printf("Message-ID: <%s%s%llu@%s>\n",
     91		       fields[FieldUnixTimestamp],
     92		       fields[FieldUnixTimestamp][0] ? "." : "",
     93		       hash, feedname);
     94
     95		ishtml = usecontent && !strcmp(fields[FieldContentType], "html");
     96		if (ishtml)
     97			fputs("Content-Type: text/html; charset=\"utf-8\"\n", stdout);
     98		else
     99			fputs("Content-Type: text/plain; charset=\"utf-8\"\n", stdout);
    100		fputs("Content-Transfer-Encoding: binary\n", stdout);
    101		printf("X-Feedname: %s\n", feedname);
    102		fputs("\n", stdout);
    103
    104		if (ishtml) {
    105			fputs("<p>\n", stdout);
    106			if (fields[FieldLink][0]) {
    107				fputs("Link: <a href=\"", stdout);
    108				xmlencode(fields[FieldLink], stdout);
    109				fputs("\">", stdout);
    110				xmlencode(fields[FieldLink], stdout);
    111				fputs("</a><br/>\n", stdout);
    112			}
    113			if (fields[FieldEnclosure][0]) {
    114				fputs("Enclosure: <a href=\"", stdout);
    115				xmlencode(fields[FieldEnclosure], stdout);
    116				fputs("\">", stdout);
    117				xmlencode(fields[FieldEnclosure], stdout);
    118				fputs("</a><br/>\n", stdout);
    119			}
    120			fputs("</p>\n", stdout);
    121		} else {
    122			if (fields[FieldLink][0])
    123				printf("Link:      %s\n", fields[FieldLink]);
    124			if (fields[FieldEnclosure][0])
    125				printf("Enclosure: %s\n", fields[FieldEnclosure]);
    126		}
    127		if (usecontent) {
    128			fputs("\n", stdout);
    129			if (ishtml && fields[FieldLink][0]) {
    130				fputs("<base href=\"", stdout);
    131				xmlencode(fields[FieldLink], stdout);
    132				fputs("\"/>\n", stdout);
    133			}
    134			printcontent(fields[FieldContent], stdout);
    135		}
    136		fputs("\n\n", stdout);
    137	}
    138}
    139
    140int
    141main(int argc, char *argv[])
    142{
    143	struct tm tmnow;
    144	time_t now;
    145	FILE *fp;
    146	char *name, *tmp;
    147	int i;
    148
    149	if (pledge(argc == 1 ? "stdio" : "stdio rpath", NULL) == -1)
    150		err(1, "pledge");
    151
    152	if ((tmp = getenv("SFEED_MBOX_CONTENT")))
    153		usecontent = !strcmp(tmp, "1");
    154	if (!(user = getenv("USER")))
    155		user = "you";
    156	if (gethostname(host, sizeof(host)) == -1)
    157		err(1, "gethostname");
    158	if ((now = time(NULL)) == (time_t)-1)
    159		errx(1, "time");
    160	if (!gmtime_r(&now, &tmnow))
    161		err(1, "gmtime_r: can't get current time");
    162	if (!strftime(mtimebuf, sizeof(mtimebuf), "%a %b %d %H:%M:%S %Y", &tmnow))
    163		errx(1, "strftime: can't format current time");
    164	if (!strftime(dtimebuf, sizeof(dtimebuf), "%a, %d %b %Y %H:%M:%S +0000", &tmnow))
    165		errx(1, "strftime: can't format current time");
    166
    167	if (argc == 1) {
    168		printfeed(stdin, "");
    169		checkfileerror(stdin, "<stdin>", 'r');
    170	} else {
    171		for (i = 1; i < argc; i++) {
    172			if (!(fp = fopen(argv[i], "r")))
    173				err(1, "fopen: %s", argv[i]);
    174			name = ((name = strrchr(argv[i], '/'))) ? name + 1 : argv[i];
    175			printfeed(fp, name);
    176			checkfileerror(fp, argv[i], 'r');
    177			checkfileerror(stdout, "<stdout>", 'w');
    178			fclose(fp);
    179		}
    180	}
    181
    182	checkfileerror(stdout, "<stdout>", 'w');
    183
    184	return 0;
    185}