#include "malloc.h"
#include "html.h"
#include "httpd.h"
+#include "fgetline.h"
#define PNAME "agedu"
#define lenof(x) (sizeof((x))/sizeof(*(x)))
+/*
+ * Path separator. This global variable affects the behaviour of
+ * various parts of the code when they need to deal with path
+ * separators. The path separator appropriate to a particular data
+ * set is encoded in the index file storing that data set; data
+ * sets generated on Unix will of course have the default '/', but
+ * foreign data sets are conceivable and must be handled correctly.
+ */
+char pathsep = '/';
+
void fatal(const char *fmt, ...)
{
va_list ap;
ino_t datafile_ino;
time_t last_output_update;
int progress, progwidth;
+ int straight_to_dump;
struct inclusion_exclusion *inex;
int ninex;
int crossfs;
};
+static void dump_line(const char *pathname, const struct trie_file *tf)
+{
+ const char *p;
+ printf("%llu %llu ", tf->size, tf->atime);
+ for (p = pathname; *p; p++) {
+ if (*p >= ' ' && *p < 127 && *p != '%')
+ putchar(*p);
+ else
+ printf("%%%02x", (unsigned char)*p);
+ }
+ putchar('\n');
+}
+
static int gotdata(void *vctx, const char *pathname, const struct stat64 *st)
{
struct ctx *ctx = (struct ctx *)vctx;
if (!ctx->crossfs && st->st_dev != ctx->filesystem_dev)
return 0;
- file.blocks = st->st_blocks;
+ file.size = (unsigned long long)512 * st->st_blocks;
file.atime = st->st_atime;
/*
* Filter based on wildcards.
*/
include = 1;
- filename = strrchr(pathname, '/');
+ filename = strrchr(pathname, pathsep);
if (!filename)
filename = pathname;
else
if (!S_ISDIR(st->st_mode))
return 0; /* just ignore */
else
- file.blocks = 0;
+ file.size = 0;
}
- triebuild_add(ctx->tb, pathname, &file);
+ if (ctx->straight_to_dump)
+ dump_line(pathname, &file);
+ else
+ triebuild_add(ctx->tb, pathname, &file);
- t = time(NULL);
- if (t != ctx->last_output_update) {
- if (ctx->progress) {
+ if (ctx->progress) {
+ t = time(NULL);
+ if (t != ctx->last_output_update) {
fprintf(stderr, "%-*.*s\r", ctx->progwidth, ctx->progwidth,
pathname);
fflush(stderr);
+ ctx->last_output_update = t;
}
- ctx->last_output_update = t;
}
return 1;
* (inclusive) and that filename with a ^A on the end
* (exclusive). So find the x indices for each.
*/
- sprintf(pathbuf, "%s\001", querydir);
+ strcpy(pathbuf, querydir);
+ make_successor(pathbuf);
xi1 = trie_before(mappedfile, querydir);
xi2 = trie_before(mappedfile, pathbuf);
if (s1 == s2)
return; /* no space taken up => no display */
- /* Display in units of 2 512-byte blocks = 1Kb */
- printf("%-11llu %s\n", (s2 - s1) / 2, querydir);
+ /* Display in units of 1Kb */
+ printf("%-11llu %s\n", (s2 - s1) / 1024, querydir);
if (depth > 0) {
/*
while (xi1 < xi2) {
trie_getpath(mappedfile, xi1, pathbuf);
text_query(mappedfile, pathbuf, t, depth-1);
- strcat(pathbuf, "\001");
+ make_successor(pathbuf);
xi1 = trie_before(mappedfile, pathbuf);
}
}
HELPPFX("actions") \
VAL(SCAN) SHORT(s) LONG(scan) \
HELPARG("directory") HELPOPT("scan and index a directory") \
- NOVAL(DUMP) SHORT(d) LONG(dump) HELPOPT("dump the index file") \
+ NOVAL(DUMP) SHORT(d) LONG(dump) HELPOPT("dump the index file on stdout") \
+ VAL(SCANDUMP) SHORT(S) LONG(scan_dump) \
+ HELPARG("directory") HELPOPT("scan only, generating a dump") \
+ NOVAL(LOAD) SHORT(l) LONG(load) \
+ HELPOPT("load and index a dump file") \
VAL(TEXT) SHORT(t) LONG(text) \
HELPARG("subdir") HELPOPT("print a plain text report on a subdirectory") \
VAL(HTML) SHORT(H) LONG(html) \
HELPARG("subdir") HELPOPT("print an HTML report on a subdirectory") \
NOVAL(HTTPD) SHORT(w) LONG(web) LONG(server) LONG(httpd) \
- HELPOPT("serve reports from a temporary web server") \
+ HELPOPT("serve HTML reports from a temporary web server") \
HELPPFX("options") \
VAL(DATAFILE) SHORT(f) LONG(file) \
HELPARG("filename") HELPOPT("[all modes] specify index file") \
HELPARG("age") HELPOPT("[--text] include only files older than this") \
VAL(AGERANGE) SHORT(r) LONG(age_range) LONG(range) LONG(ages) \
HELPARG("age[-age]") HELPOPT("[--html,--web] set limits of colour coding") \
+ VAL(SERVERADDR) LONG(address) LONG(addr) LONG(server_address) \
+ LONG(server_addr) \
+ HELPARG("addr[:port]") HELPOPT("[--web] specify HTTP server address") \
VAL(AUTH) LONG(auth) LONG(http_auth) LONG(httpd_auth) \
LONG(server_auth) LONG(web_auth) \
HELPARG("type") HELPOPT("[--web] specify HTTP authentication method") \
+ VAL(AUTHFILE) LONG(auth_file) \
+ HELPARG("filename") HELPOPT("[--web] read HTTP Basic user/pass from file") \
+ VAL(AUTHFD) LONG(auth_fd) \
+ HELPARG("fd") HELPOPT("[--web] read HTTP Basic user/pass from fd") \
HELPPFX("also") \
NOVAL(HELP) SHORT(h) LONG(help) HELPOPT("display this help text") \
NOVAL(VERSION) SHORT(V) LONG(version) HELPOPT("report version number") \
char *scandir = NULL;
char *querydir = NULL;
int doing_opts = 1;
- enum { USAGE, TEXT, HTML, SCAN, DUMP, HTTPD } mode = USAGE;
+ enum { USAGE, TEXT, HTML, SCAN, DUMP, SCANDUMP, LOAD, HTTPD } mode = USAGE;
time_t now = time(NULL);
time_t textcutoff = now, htmlnewest = now, htmloldest = now;
int htmlautoagerange = 1;
+ const char *httpserveraddr = NULL;
+ int httpserverport = 0;
+ const char *httpauthdata = NULL;
int auth = HTTPD_AUTH_MAGIC | HTTPD_AUTH_BASIC;
int progress = 1;
struct inclusion_exclusion *inex = NULL;
mode = SCAN;
scandir = optval;
break;
+ case OPT_SCANDUMP:
+ mode = SCANDUMP;
+ scandir = optval;
+ break;
case OPT_DUMP:
mode = DUMP;
break;
+ case OPT_LOAD:
+ mode = LOAD;
+ break;
case OPT_TEXT:
querydir = optval;
mode = TEXT;
htmlautoagerange = 0;
}
break;
+ case OPT_SERVERADDR:
+ {
+ char *port;
+ if (optval[0] == '[' &&
+ (port = strchr(optval, ']')) != NULL)
+ port++;
+ else
+ port = optval;
+ port += strcspn(port, ":");
+ if (port)
+ *port++ = '\0';
+ httpserveraddr = optval;
+ httpserverport = atoi(port);
+ }
+ break;
case OPT_AUTH:
if (!strcmp(optval, "magic"))
auth = HTTPD_AUTH_MAGIC;
return 1;
}
break;
+ case OPT_AUTHFILE:
+ case OPT_AUTHFD:
+ {
+ int fd;
+ char namebuf[40];
+ const char *name;
+ char *authbuf;
+ int authlen, authsize;
+ int ret;
+
+ if (optid == OPT_AUTHFILE) {
+ fd = open(optval, O_RDONLY);
+ if (fd < 0) {
+ fprintf(stderr, "%s: %s: open: %s\n", PNAME,
+ optval, strerror(errno));
+ return 1;
+ }
+ name = optval;
+ } else {
+ fd = atoi(optval);
+ name = namebuf;
+ sprintf(namebuf, "fd %d", fd);
+ }
+
+ authlen = 0;
+ authsize = 256;
+ authbuf = snewn(authsize, char);
+ while ((ret = read(fd, authbuf+authlen,
+ authsize-authlen)) > 0) {
+ authlen += ret;
+ if ((authsize - authlen) < (authsize / 16)) {
+ authsize = authlen * 3 / 2 + 4096;
+ authbuf = sresize(authbuf, authsize, char);
+ }
+ }
+ if (ret < 0) {
+ fprintf(stderr, "%s: %s: read: %s\n", PNAME,
+ name, strerror(errno));
+ return 1;
+ }
+ if (optid == OPT_AUTHFILE)
+ close(fd);
+ httpauthdata = authbuf;
+ }
+ break;
case OPT_INCLUDE:
case OPT_INCLUDEPATH:
case OPT_EXCLUDE:
if (mode == USAGE) {
usage(stderr);
return 1;
- } else if (mode == SCAN) {
-
- fd = open(filename, O_RDWR | O_TRUNC | O_CREAT, S_IRWXU);
- if (fd < 0) {
- fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename,
- strerror(errno));
- return 1;
+ } else if (mode == SCAN || mode == SCANDUMP || mode == LOAD) {
+
+ if (mode == LOAD) {
+ char *buf = fgetline(stdin);
+ unsigned newpathsep;
+ buf[strcspn(buf, "\r\n")] = '\0';
+ if (1 != sscanf(buf, "agedu dump file. pathsep=%x",
+ &newpathsep)) {
+ fprintf(stderr, "%s: header in dump file not recognised\n",
+ PNAME);
+ return 1;
+ }
+ pathsep = (char)newpathsep;
+ sfree(buf);
}
- if (stat(scandir, &st) < 0) {
- fprintf(stderr, "%s: %s: stat: %s\n", PNAME, scandir,
- strerror(errno));
- return 1;
+ if (mode == SCAN || mode == LOAD) {
+ /*
+ * Prepare to write out the index file.
+ */
+ fd = open(filename, O_RDWR | O_TRUNC | O_CREAT, S_IRWXU);
+ if (fd < 0) {
+ fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename,
+ strerror(errno));
+ return 1;
+ }
+ if (fstat(fd, &st) < 0) {
+ perror("agedu: fstat");
+ return 1;
+ }
+ ctx->datafile_dev = st.st_dev;
+ ctx->datafile_ino = st.st_ino;
+ ctx->straight_to_dump = 0;
+ } else {
+ ctx->datafile_dev = -1;
+ ctx->datafile_ino = -1;
+ ctx->straight_to_dump = 1;
}
- ctx->filesystem_dev = crossfs ? 0 : st.st_dev;
- if (fstat(fd, &st) < 0) {
- perror("agedu: fstat");
- return 1;
+ if (mode == SCAN || mode == SCANDUMP) {
+ if (stat(scandir, &st) < 0) {
+ fprintf(stderr, "%s: %s: stat: %s\n", PNAME, scandir,
+ strerror(errno));
+ return 1;
+ }
+ ctx->filesystem_dev = crossfs ? 0 : st.st_dev;
}
- ctx->datafile_dev = st.st_dev;
- ctx->datafile_ino = st.st_ino;
+
ctx->inex = inex;
ctx->ninex = ninex;
ctx->crossfs = crossfs;
ctx->progwidth = 79;
}
+ if (mode == SCANDUMP)
+ printf("agedu dump file. pathsep=%02x\n", (unsigned char)pathsep);
+
/*
* Scan the directory tree, and write out the trie component
* of the data file.
*/
- ctx->tb = triebuild_new(fd);
- du(scandir, gotdata, ctx);
- count = triebuild_finish(ctx->tb);
- triebuild_free(ctx->tb);
-
- if (ctx->progress) {
- fprintf(stderr, "%-*s\r", ctx->progwidth, "");
- fflush(stderr);
+ if (mode != SCANDUMP) {
+ ctx->tb = triebuild_new(fd);
}
-
- /*
- * Work out how much space the cumulative index trees will
- * take; enlarge the file, and memory-map it.
- */
- if (fstat(fd, &st) < 0) {
- perror("agedu: fstat");
- return 1;
+ if (mode == LOAD) {
+ char *buf;
+ int line = 2;
+ while ((buf = fgetline(stdin)) != NULL) {
+ struct trie_file tf;
+ char *p, *q;
+
+ buf[strcspn(buf, "\r\n")] = '\0';
+
+ p = buf;
+ q = p;
+ while (*p && *p != ' ') p++;
+ if (!*p) {
+ fprintf(stderr, "%s: dump file line %d: expected at least"
+ " three fields\n", PNAME, line);
+ return 1;
+ }
+ *p++ = '\0';
+ tf.size = strtoull(q, NULL, 10);
+ q = p;
+ while (*p && *p != ' ') p++;
+ if (!*p) {
+ fprintf(stderr, "%s: dump file line %d: expected at least"
+ " three fields\n", PNAME, line);
+ return 1;
+ }
+ *p++ = '\0';
+ tf.atime = strtoull(q, NULL, 10);
+ q = buf;
+ while (*p) {
+ int c = *p;
+ if (*p == '%') {
+ int i;
+ p++;
+ c = 0;
+ for (i = 0; i < 2; i++) {
+ if (*p >= '0' && *p <= '9')
+ c += *p - '0';
+ else if (*p >= 'A' && *p <= 'F')
+ c += *p - ('A' - 10);
+ else if (*p >= 'a' && *p <= 'f')
+ c += *p - ('a' - 10);
+ else {
+ fprintf(stderr, "%s: dump file line %d: unable"
+ " to parse hex escape\n", PNAME, line);
+ }
+ p++;
+ }
+ }
+ *q++ = c;
+ p++;
+ }
+ *q = '\0';
+ triebuild_add(ctx->tb, buf, &tf);
+ sfree(buf);
+ }
+ } else {
+ du(scandir, gotdata, ctx);
}
+ if (mode != SCANDUMP) {
+ count = triebuild_finish(ctx->tb);
+ triebuild_free(ctx->tb);
- printf("Built pathname index, %d entries, %ju bytes\n", count,
- (intmax_t)st.st_size);
+ if (ctx->progress) {
+ fprintf(stderr, "%-*s\r", ctx->progwidth, "");
+ fflush(stderr);
+ }
- totalsize = index_compute_size(st.st_size, count);
+ /*
+ * Work out how much space the cumulative index trees
+ * will take; enlarge the file, and memory-map it.
+ */
+ if (fstat(fd, &st) < 0) {
+ perror("agedu: fstat");
+ return 1;
+ }
- if (lseek(fd, totalsize-1, SEEK_SET) < 0) {
- perror("agedu: lseek");
- return 1;
- }
- if (write(fd, "\0", 1) < 1) {
- perror("agedu: write");
- return 1;
- }
+ printf("Built pathname index, %d entries, %ju bytes\n", count,
+ (intmax_t)st.st_size);
- printf("Upper bound on index file size = %ju bytes\n",
- (intmax_t)totalsize);
+ totalsize = index_compute_size(st.st_size, count);
- mappedfile = mmap(NULL, totalsize, PROT_READ|PROT_WRITE,MAP_SHARED, fd, 0);
- if (!mappedfile) {
- perror("agedu: mmap");
- return 1;
- }
+ if (lseek(fd, totalsize-1, SEEK_SET) < 0) {
+ perror("agedu: lseek");
+ return 1;
+ }
+ if (write(fd, "\0", 1) < 1) {
+ perror("agedu: write");
+ return 1;
+ }
- ib = indexbuild_new(mappedfile, st.st_size, count);
- tw = triewalk_new(mappedfile);
- while ((tf = triewalk_next(tw, NULL)) != NULL)
- indexbuild_add(ib, tf);
- triewalk_free(tw);
- realsize = indexbuild_realsize(ib);
- indexbuild_free(ib);
+ printf("Upper bound on index file size = %ju bytes\n",
+ (intmax_t)totalsize);
- munmap(mappedfile, totalsize);
- ftruncate(fd, realsize);
- close(fd);
- printf("Actual index file size = %ju bytes\n", (intmax_t)realsize);
+ mappedfile = mmap(NULL, totalsize, PROT_READ|PROT_WRITE,MAP_SHARED, fd, 0);
+ if (!mappedfile) {
+ perror("agedu: mmap");
+ return 1;
+ }
+
+ ib = indexbuild_new(mappedfile, st.st_size, count);
+ tw = triewalk_new(mappedfile);
+ while ((tf = triewalk_next(tw, NULL)) != NULL)
+ indexbuild_add(ib, tf);
+ triewalk_free(tw);
+ realsize = indexbuild_realsize(ib);
+ indexbuild_free(ib);
+
+ munmap(mappedfile, totalsize);
+ ftruncate(fd, realsize);
+ close(fd);
+ printf("Actual index file size = %ju bytes\n", (intmax_t)realsize);
+ }
} else if (mode == TEXT) {
size_t pathlen;
perror("agedu: mmap");
return 1;
}
+ pathsep = trie_pathsep(mappedfile);
/*
* Trim trailing slash, just in case.
*/
pathlen = strlen(querydir);
- if (pathlen > 0 && querydir[pathlen-1] == '/')
+ if (pathlen > 0 && querydir[pathlen-1] == pathsep)
querydir[--pathlen] = '\0';
text_query(mappedfile, querydir, textcutoff, 1);
perror("agedu: mmap");
return 1;
}
+ pathsep = trie_pathsep(mappedfile);
/*
* Trim trailing slash, just in case.
*/
pathlen = strlen(querydir);
- if (pathlen > 0 && querydir[pathlen-1] == '/')
+ if (pathlen > 0 && querydir[pathlen-1] == pathsep)
querydir[--pathlen] = '\0';
xi = trie_before(mappedfile, querydir);
perror("agedu: mmap");
return 1;
}
+ pathsep = trie_pathsep(mappedfile);
maxpathlen = trie_maxpathlen(mappedfile);
buf = snewn(maxpathlen, char);
+ printf("agedu dump file. pathsep=%02x\n", (unsigned char)pathsep);
tw = triewalk_new(mappedfile);
- while ((tf = triewalk_next(tw, buf)) != NULL) {
- printf("%s: %llu %llu\n", buf, tf->blocks, tf->atime);
- }
+ while ((tf = triewalk_next(tw, buf)) != NULL)
+ dump_line(buf, tf);
triewalk_free(tw);
} else if (mode == HTTPD) {
- struct html_config cfg;
+ struct html_config pcfg;
+ struct httpd_config dcfg;
fd = open(filename, O_RDONLY);
if (fd < 0) {
perror("agedu: mmap");
return 1;
}
-
- cfg.format = NULL;
- cfg.autoage = htmlautoagerange;
- cfg.oldest = htmloldest;
- cfg.newest = htmlnewest;
- run_httpd(mappedfile, auth, &cfg);
+ pathsep = trie_pathsep(mappedfile);
+
+ dcfg.address = httpserveraddr;
+ dcfg.port = httpserverport;
+ dcfg.basicauthdata = httpauthdata;
+ pcfg.format = NULL;
+ pcfg.autoage = htmlautoagerange;
+ pcfg.oldest = htmloldest;
+ pcfg.newest = htmlnewest;
+ run_httpd(mappedfile, auth, &dcfg, &pcfg);
}
return 0;