*much* nicer cache_clean program

README

@@ -105,14 +105,15 @@ Cleaning the cache
 
 ccache tends to quickly fill up the cache directory. You may find the
 ccache_clean utility for removing old cache files. If called with no
-arguments it will remove any cache files older than a week. You can
-also pass a single argument specifying the number of days old files
-need to be to be cleaned. For example:
+arguments it will trim the cache to be less than 1 Gigabyte. You can
+also pass a single argument specifying the size limit on the cache,
+for example:
+     ccache_clean 2G
+would clear the oldest files to bring the cache below 2G in size. You
+can use 'M' for megabytes, 'G' for gigabytes or 'K' for kilobytes.
 
-     ccache_clean 3
-
-will remove any files older than 3 days. You may wish to call
-ccache_clean from a cron job to keep your disk space usage reasonable.
+You may wish to call ccache_clean from a cron job to keep your disk
+space usage reasonable.
 
 How it works
 ------------

ccache.h

@@ -41,7 +41,9 @@ int copy_file(const char *src, const char *dest);
 int create_dir(const char *dir);
 void x_asprintf(char **ptr, const char *format, ...);
 char *x_strdup(const char *s);
-void traverse(const char *dir, void (*fn)(const char *));
+void *x_realloc(void *ptr, size_t size);
+void *x_malloc(size_t size);
+void traverse(const char *dir, void (*fn)(const char *, struct stat *));
 
 int execute(char **argv, 
 	    const char *path_stdout,

ccache_clean.c

@@ -22,29 +22,72 @@
 
 #include "ccache.h"
 
-static time_t threshold;
+#define BLOCK_SIZE 1024
+
 char *cache_logfile = NULL;
 
-static void clean_fn(const char *fname)
+static struct files {
+	char *fname;
+	time_t mtime;
+	size_t size;
+} **files;
+static unsigned allocated;
+static unsigned num_files;
+static size_t total_size;
+static size_t size_threshold = 1024*1024; /* 1G default */
+
+static int files_compare(struct files *f1, struct files *f2)
 {
-	struct stat st;
-	if (stat(fname, &st) != 0 || !S_ISREG(st.st_mode)) return;
+	return f2->mtime - f1->mtime;
+}
 
-	if (st.st_mtime >= threshold) return;
+/* this builds the list of files in the cache */
+static void traverse_fn(const char *fname, struct stat *st)
+{
+	if (!S_ISREG(st->st_mode)) return;
 
-	if (unlink(fname) != 0) {
-		cc_log("unlink %s - %s\n", fname, strerror(errno));
-		return;
+	if (num_files == allocated) {
+		allocated = 10000 + num_files*2;
+		files = x_realloc(files, sizeof(struct files *)*allocated);
 	}
-	cc_log("cleaned %s\n", fname);
+
+	files[num_files] = x_malloc(sizeof(struct files *));
+	files[num_files]->fname = x_strdup(fname);
+	files[num_files]->mtime = st->st_mtime;
+	/* we deliberately overestimate by up to 1 block */
+	files[num_files]->size = 1 + (st->st_size/BLOCK_SIZE);
+	total_size += files[num_files]->size;
+	num_files++;
+}
+
+static void sort_and_clean(void)
+{
+	unsigned i;
+
+	if (num_files > 1) {
+		/* sort in ascending data order */
+		qsort(files, num_files, sizeof(struct files *), files_compare);
+	}
+
+	/* delete enough files to bring us below the threshold */
+	for (i=0;i<num_files && total_size >= size_threshold;i++) {
+		if (unlink(files[i]->fname) != 0) {
+			fprintf(stderr, "unlink %s - %s\n", 
+				files[i]->fname, strerror(errno));
+			continue;
+		}
+		total_size -= files[i]->size;
+	}
+
+	printf("cleaned %d of %d files (cache is now %.1f MByte)\n", 
+	       i, num_files, total_size/1024.0);
 }
 
 
 int main(int argc, char *argv[])
 {
 	char *cache_dir;
-	int num_days = 7;
-
+	
 	cache_dir = getenv("CCACHE_DIR");
 	if (!cache_dir) {
 		x_asprintf(&cache_dir, "%s/.ccache", getenv("HOME"));
@@ -59,13 +102,33 @@ int main(int argc, char *argv[])
 		exit(1);
 	}
 
+	/* work out what size cache they want */
 	if (argc > 1) {
-		num_days = atoi(argv[1]);
+		char *s = argv[1];
+		char m;
+		size_threshold = atoi(s);
+		m = s[strlen(s)-1];
+		switch (m) {
+		case 'G':
+		case 'g':
+			size_threshold *= 1024*1024;
+			break;
+		case 'M':
+		case 'm':
+			size_threshold *= 1024;
+			break;
+		case 'K':
+		case 'k':
+			size_threshold *= 1;
+			break;
+		}
 	}
 
-	threshold = time(NULL) - num_days*24*60*60;
+	/* build a list of files */
+	traverse(cache_dir, traverse_fn);
 
-	traverse(cache_dir, clean_fn);
+	/* clean the cache */
+	sort_and_clean();
 
 	return 0;
 }

util.c

@@ -142,12 +142,38 @@ char *x_strdup(const char *s)
 	return ret;
 }
 
+/*
+  this is like malloc() but dies if the malloc fails
+*/
+void *x_malloc(size_t size)
+{
+	void *ret;
+	ret = malloc(size);
+	if (!ret) {
+		fatal("out of memory in malloc\n");
+	}
+	return ret;
+}
+
+/*
+  this is like strdup() but dies if the malloc fails
+*/
+void *x_realloc(void *ptr, size_t size)
+{
+	if (!ptr) return x_malloc(size);
+	ptr = realloc(ptr, size);
+	if (!ptr) {
+		fatal("out of memory in x_realloc");
+	}
+	return ptr;
+}
+
 
 /* 
    revsusive directory traversal - used for cleanup
    fn() is called on all files/dirs in the tree
  */
-void traverse(const char *dir, void (*fn)(const char *))
+void traverse(const char *dir, void (*fn)(const char *, struct stat *))
 {
 	DIR *d;
 	struct dirent *de;
@@ -173,7 +199,7 @@ void traverse(const char *dir, void (*fn)(const char *))
 			traverse(fname, fn);
 		}
 
-		fn(fname);
+		fn(fname, &st);
 		free(fname);
 	}