save items in a Maildir/

rssdl is back in a usable state! (for rss only)

.gitignore

@@ -2,3 +2,6 @@ rssdl
 *.o
 vgcore*
 *.core
+*/
+*.rss
+*.atom

Makefile

@@ -1,13 +1,13 @@
 include config.mk
 
-SRC = main.c fetch.c xml.c save.c hash.c
+SRC = main.c xml.c parse.c save.c
 OBJ = ${SRC:.c=.o}
 
 all: rssdl
 
-fetch.o: fetch.h xml.h
-save.o: save.h parse.h hash.h
-hash.o: hash.h
+parse.o: parse.h save.h xml.h
+save.o: save.h parse.h
+xml.o: xml.h
 
 rssdl: ${OBJ}
 	${CC} ${CPPFLAGS} ${CFLAGS} -o $@ ${OBJ} ${LDFLAGS}

main.c

@@ -1,13 +1,13 @@
 #include <stdio.h>
 #include <sys/stat.h>
+
 #include <unistd.h>
-#include "fetch.h"
-#include "save.h"
 
-int main(int argc, char **argv)
+#include "parse.h"
+
+int main(int argc, char *argv[])
 {
 	char *maildir;
-	int exit_code = 0;
 
 	if (argc != 2) {
 		printf("Usage: %s maildir\n", argv[0]);
@@ -22,11 +22,8 @@ int main(int argc, char **argv)
 	mkdir("cur", 0755);
 	mkdir("new", 0755);
 
-	if (fetch_rss() == -1)
-		exit_code = 1;
-
-	unlink("tmp/summary");
-	unlink("tmp/desc");
+	if (parse_item() != 0)
+		return 1;
 
-	return exit_code;
+	return 0;
 }

parse.c

@@ -1,8 +1,10 @@
 #include <stdio.h>
+#include <stdlib.h>
 #include <string.h>
 #include <time.h>
 
 #include "parse.h"
+#include "save.h"
 #include "xml.h"
 
 #define istag(name)	strcmp(x->tag, name) == 0
@@ -13,7 +15,6 @@ char		feedtitle[256];
 
 static void 	xmldata(XMLParser *, const char *, size_t);
 static void 	xmldataentity(XMLParser *, const char *, size_t);
-static void 	xmldatastart(XMLParser *);
 static void 	xmltagend(XMLParser *, const char *, size_t, int);
 static void 	xmltagstart(XMLParser *, const char *, size_t);
 static void	rfc822_date(char *);
@@ -60,24 +61,15 @@ xmltagend(XMLParser *x, const char *t, size_t tl, int isshort)
 		return;
 
 	if (strcmp(x->tag, "item") == 0) {
-		fclose(item.desc);
 		isitem = 0;
 
+		fseek(item.desc, 0, SEEK_SET);
 		rfc822_date(item.date);
 
-		printf("From: rssdl@localhost\n"
-		    "Subject: [%s] %s\n"
-		    "Date: %s\n"
-		    "Content-Type: text/html\n"
-		    "\nArticle Link: %s\n\n",
-		    feedtitle, item.title, item.date, item.link);
-
-		item.desc = fopen("/tmp/rssdl.abc", "r");
-		char line[8192];
-		while (fgets(line, sizeof(line), item.desc) != NULL)
-			printf("%s\n", line);
-		printf("\n");
+		save_item(item, feedtitle);
 
+		fclose(item.desc);
+		item.desc = NULL;
 		memset(&item, 0, sizeof(struct rss_item));
 	}
 }
@@ -87,10 +79,8 @@ xmltagstart(XMLParser *x, const char *t, size_t tl)
 {
 	if (strcmp(x->tag, "item") == 0) {
 		isitem = 1;
-		/*
-		 * use mkstemp or some other random temporary path name
-		 */
-		item.desc = fopen("/tmp/rssdl.abc", "w");
+		snprintf(item.descfn, sizeof(item.descfn), "tmp/desc.%6x", rand());
+		item.desc = fopen(item.descfn, "w+");
 	}
 }
 
@@ -112,7 +102,7 @@ rfc822_date(char *date)
 }
 
 int
-main(void)
+parse_item(void)
 {
 	XMLParser x = {0};
 

parse.h

@@ -1,7 +1,15 @@
+#ifndef PARSE_H_INC
+#define PARSE_H_INC
+
 struct rss_item {
-        char title[256];
-        char link[256];
-        char date[32];
+	char title[256];
+	char link[256];
+	char date[32];
 	long epochsec;
-        FILE *desc;
+	char descfn[16];
+	FILE *desc;
 };
+
+int parse_item(void);
+
+#endif

save.c

@@ -0,0 +1,114 @@
+#include <sys/types.h>
+#include <sys/stat.h>
+
+#include <fcntl.h>
+#include <unistd.h>
+#include <dirent.h>
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "save.h"
+#include "parse.h"
+
+static unsigned int fletcher16_file(FILE *);
+static int item_file_exists(char *);
+
+static unsigned int
+fletcher16_file(FILE *fp)
+{
+	unsigned int sum1 = 0, sum2 = 0;
+	int c;
+
+	while ((c = fgetc(fp)) != EOF) {
+		sum1 += (sum1 + c) % 255;
+		sum2 += (sum2 + sum1) % 255;
+	}
+
+	return (sum2 << 8) | sum1;
+}
+
+/* check if file exists already in cur/ or new/ */
+static int
+item_file_exists(char *filename)
+{
+	DIR *dp;
+	struct dirent *dir;
+
+	if ((dp = opendir("cur/"))) {
+		while ((dir = readdir(dp))) {
+			if (strstr(dir->d_name, filename + 4)) {
+				closedir(dp);
+				return 1;
+			}
+		}
+	}
+	closedir(dp);
+	if ((dp = opendir("new/"))) {
+		while ((dir = readdir(dp))) {
+			if (strstr(dir->d_name, filename + 4)) {
+				closedir(dp);
+				return 1;
+			}
+		}
+	}
+	closedir(dp);
+
+	return 0;
+}
+
+int
+save_item(struct rss_item item, char *feedtitle)
+{
+	FILE *fp;
+
+	static char hostname[256];
+	char path[512], newpath[512], line[1024];
+	unsigned int digest;
+
+	if (gethostname(hostname, sizeof(hostname)) == -1) {
+		perror("failed to get hostname, setting to localhost");
+		strncpy(hostname, "localhost", sizeof(hostname));
+	}
+	snprintf(path, sizeof(path), "tmp/%li.rssdl-tmp.%s:2,", item.epochsec, hostname);
+
+	fp = fopen(path, "w");
+	if (!fp) {
+		fprintf(stderr, "failed to open %s for writing\n", path);
+		goto CLEANUP;
+	}
+
+	fprintf(fp, "From: rssdl@%s\n"
+	    "Subject: [%s] %s\n"
+	    "Date: %s\n"
+	    "Content-Type: text/html\n"
+	    "\nArticle Link: %s\n\n",
+	    hostname, feedtitle, item.title, item.date, item.link);
+	if (item.desc) {
+		while (fgets(line, sizeof(line), item.desc) != NULL)
+			fputs(line, fp);
+	}
+
+	fclose(fp);
+
+	fp = fopen(path, "r");
+	if (!fp) {
+		fprintf(stderr, "failed to open %s for reading\n", path);
+		goto CLEANUP;
+	}
+
+	digest = fletcher16_file(fp);
+	snprintf(newpath, sizeof(newpath), "new/%li.%x.%s:2,", item.epochsec, digest, hostname);
+
+	if (item_file_exists(newpath) == 1)
+		goto CLEANUP;
+
+	if (rename(path, newpath) == 0)
+		puts(newpath);
+
+CLEANUP:
+	fclose(fp);
+	unlink(path);
+
+	return 0;
+}

save.h

@@ -0,0 +1,8 @@
+#ifndef SAVE_H_INC
+#define SAVE_H_INC
+
+#include "parse.h"
+
+int save_item(struct rss_item item, char *feedtitle);
+
+#endif

xml.c

@@ -224,20 +224,20 @@ codepointtoutf8(long r, char *s)
 	} else if (r <= 0x07FF) {
 		/* 2 bytes: 00000aaa aabbbbbb */
 		s[0] = 0xC0 | ((r & 0x0007C0) >>  6); /* 110aaaaa */
-		s[1] = 0x80 |  (r & 0x00003F);        /* 10bbbbbb */
+		s[1] = 0x80 |  (r & 0x00003F);	      /* 10bbbbbb */
 		return 2;
 	} else if (r <= 0xFFFF) {
 		/* 3 bytes: aaaabbbb bbcccccc */
 		s[0] = 0xE0 | ((r & 0x00F000) >> 12); /* 1110aaaa */
 		s[1] = 0x80 | ((r & 0x000FC0) >>  6); /* 10bbbbbb */
-		s[2] = 0x80 |  (r & 0x00003F);        /* 10cccccc */
+		s[2] = 0x80 |  (r & 0x00003F);	      /* 10cccccc */
 		return 3;
 	} else {
 		/* 4 bytes: 000aaabb bbbbcccc ccdddddd */
 		s[0] = 0xF0 | ((r & 0x1C0000) >> 18); /* 11110aaa */
 		s[1] = 0x80 | ((r & 0x03F000) >> 12); /* 10bbbbbb */
 		s[2] = 0x80 | ((r & 0x000FC0) >>  6); /* 10cccccc */
-		s[3] = 0x80 |  (r & 0x00003F);        /* 10dddddd */
+		s[3] = 0x80 |  (r & 0x00003F);	      /* 10dddddd */
 		return 4;
 	}
 }