-/* $Id: mandocdb.c,v 1.26 2011/12/08 01:00:58 kristaps Exp $ */
+/* $Id: mandocdb.c,v 1.46 2012/03/23 06:52:17 kristaps Exp $ */
/*
* Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
* Copyright (c) 2011 Ingo Schwarze <schwarze@openbsd.org>
#include <sys/param.h>
#include <sys/types.h>
-#include <sys/stat.h>
#include <assert.h>
+#include <ctype.h>
#include <dirent.h>
+#include <errno.h>
#include <fcntl.h>
#include <getopt.h>
#include <stdio.h>
#define MANDOC_SRC 0x1
#define MANDOC_FORM 0x2
+#define WARNING(_f, _b, _fmt, _args...) \
+ do if (warnings) { \
+ fprintf(stderr, "%s: ", (_b)); \
+ fprintf(stderr, (_fmt), ##_args); \
+ if ('\0' != *(_f)) \
+ fprintf(stderr, ": %s", (_f)); \
+ fprintf(stderr, "\n"); \
+ } while (/* CONSTCOND */ 0)
+
+/* Access to the mandoc database on disk. */
+
+struct mdb {
+ char idxn[MAXPATHLEN]; /* index db filename */
+ char dbn[MAXPATHLEN]; /* keyword db filename */
+ DB *idx; /* index recno database */
+ DB *db; /* keyword btree database */
+};
+
+/* Stack of temporarily unused index records. */
+
+struct recs {
+ recno_t *stack; /* pointer to a malloc'ed array */
+ size_t size; /* number of allocated slots */
+ size_t cur; /* current number of empty records */
+ recno_t last; /* last record number in the index */
+};
+
/* Tiny list for files. No need to bring in QUEUE. */
struct of {
/* Operation we're going to perform. */
enum op {
- OP_NEW = 0, /* new database */
+ OP_DEFAULT = 0, /* new dbs from dir list or default config */
+ OP_CONFFILE, /* new databases from custom config file */
OP_UPDATE, /* delete/add entries in existing database */
- OP_DELETE /* delete entries from existing database */
+ OP_DELETE, /* delete entries from existing database */
+ OP_TEST /* change no databases, report potential problems */
};
#define MAN_ARGS DB *hash, \
static void hash_reset(DB **);
static void index_merge(const struct of *, struct mparse *,
struct buf *, struct buf *, DB *,
- DB *, const char *, DB *, const char *,
- recno_t, const recno_t *, size_t);
-static void index_prune(const struct of *, DB *,
- const char *, DB *, const char *,
- recno_t *, recno_t **, size_t *);
-static void ofile_argbuild(int, char *[], struct of **);
-static int ofile_dirbuild(const char *, const char *,
- const char *, int, struct of **);
+ struct mdb *, struct recs *,
+ const char *);
+static void index_prune(const struct of *, struct mdb *,
+ struct recs *, const char *);
+static void ofile_argbuild(int, char *[],
+ struct of **, const char *);
+static void ofile_dirbuild(const char *, const char *,
+ const char *, int, struct of **, char *);
static void ofile_free(struct of *);
-static void pformatted(DB *, struct buf *, struct buf *,
- const struct of *);
+static void pformatted(DB *, struct buf *, struct buf *,
+ const struct of *, const char *);
static int pman_node(MAN_ARGS);
static void pmdoc_node(MDOC_ARGS);
static int pmdoc_head(MDOC_ARGS);
static int pmdoc_Sh(MDOC_ARGS);
static int pmdoc_St(MDOC_ARGS);
static int pmdoc_Xr(MDOC_ARGS);
-static void usage(void);
#define MDOCF_CHILD 0x01 /* Automatically index child nodes. */
static const char *progname;
static int use_all; /* Use all directories and files. */
static int verb; /* Output verbosity level. */
+static int warnings; /* Potential problems in manuals. */
int
main(int argc, char *argv[])
{
struct mparse *mp; /* parse sequence */
struct manpaths dirs;
+ struct mdb mdb;
+ struct recs recs;
enum op op; /* current operation */
const char *dir;
- char *cp;
- char pbuf[PATH_MAX],
- ibuf[MAXPATHLEN], /* index fname */
- fbuf[MAXPATHLEN]; /* btree fname */
int ch, i, flags;
- DB *idx, /* index database */
- *db, /* keyword database */
- *hash; /* temporary keyword hashtable */
+ char dirbuf[MAXPATHLEN];
+ DB *hash; /* temporary keyword hashtable */
BTREEINFO info; /* btree configuration */
- recno_t maxrec; /* last record number in the index */
- recno_t *recs; /* the numbers of all empty records */
- size_t sz1, sz2,
- recsz, /* number of allocated slots in recs */
- reccur; /* current number of empty records */
+ size_t sz1, sz2;
struct buf buf, /* keyword buffer */
dbuf; /* description buffer */
struct of *of; /* list of files for processing */
++progname;
memset(&dirs, 0, sizeof(struct manpaths));
+ memset(&mdb, 0, sizeof(struct mdb));
+ memset(&recs, 0, sizeof(struct recs));
- verb = 0;
- use_all = 0;
of = NULL;
- db = idx = NULL;
mp = NULL;
hash = NULL;
- recs = NULL;
- recsz = reccur = 0;
- maxrec = 0;
- op = OP_NEW;
+ op = OP_DEFAULT;
dir = NULL;
- while (-1 != (ch = getopt(argc, argv, "ad:u:v")))
+ while (-1 != (ch = getopt(argc, argv, "aC:d:tu:vW")))
switch (ch) {
case ('a'):
use_all = 1;
break;
+ case ('C'):
+ if (op) {
+ fprintf(stderr,
+ "-C: conflicting options\n");
+ goto usage;
+ }
+ dir = optarg;
+ op = OP_CONFFILE;
+ break;
case ('d'):
+ if (op) {
+ fprintf(stderr,
+ "-d: conflicting options\n");
+ goto usage;
+ }
dir = optarg;
op = OP_UPDATE;
break;
+ case ('t'):
+ dup2(STDOUT_FILENO, STDERR_FILENO);
+ if (op) {
+ fprintf(stderr,
+ "-t: conflicting options\n");
+ goto usage;
+ }
+ op = OP_TEST;
+ use_all = 1;
+ warnings = 1;
+ break;
case ('u'):
+ if (op) {
+ fprintf(stderr,
+ "-u: conflicting options\n");
+ goto usage;
+ }
dir = optarg;
op = OP_DELETE;
break;
case ('v'):
verb++;
break;
+ case ('W'):
+ warnings = 1;
+ break;
default:
- usage();
- return((int)MANDOCLEVEL_BADARG);
+ goto usage;
}
argc -= optind;
argv += optind;
+ if (OP_CONFFILE == op && argc > 0) {
+ fprintf(stderr, "-C: too many arguments\n");
+ goto usage;
+ }
+
memset(&info, 0, sizeof(BTREEINFO));
+ info.lorder = 4321;
info.flags = R_DUP;
mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL);
buf.cp = mandoc_malloc(buf.size);
dbuf.cp = mandoc_malloc(dbuf.size);
- flags = OP_NEW == op ? O_CREAT|O_TRUNC|O_RDWR : O_CREAT|O_RDWR;
+ if (OP_TEST == op) {
+ ofile_argbuild(argc, argv, &of, ".");
+ if (NULL == of)
+ goto out;
+ index_merge(of, mp, &dbuf, &buf,
+ hash, &mdb, &recs, ".");
+ goto out;
+ }
if (OP_UPDATE == op || OP_DELETE == op) {
- ibuf[0] = fbuf[0] = '\0';
-
- strlcat(fbuf, dir, MAXPATHLEN);
- strlcat(fbuf, "/", MAXPATHLEN);
- sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN);
+ strlcat(mdb.dbn, dir, MAXPATHLEN);
+ strlcat(mdb.dbn, "/", MAXPATHLEN);
+ sz1 = strlcat(mdb.dbn, MANDOC_DB, MAXPATHLEN);
- strlcat(ibuf, dir, MAXPATHLEN);
- strlcat(ibuf, "/", MAXPATHLEN);
- sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN);
+ strlcat(mdb.idxn, dir, MAXPATHLEN);
+ strlcat(mdb.idxn, "/", MAXPATHLEN);
+ sz2 = strlcat(mdb.idxn, MANDOC_IDX, MAXPATHLEN);
if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) {
- fprintf(stderr, "%s: Path too long\n", dir);
+ fprintf(stderr, "%s: path too long\n", dir);
exit((int)MANDOCLEVEL_BADARG);
}
- db = dbopen(fbuf, flags, 0644, DB_BTREE, &info);
- idx = dbopen(ibuf, flags, 0644, DB_RECNO, NULL);
+ flags = O_CREAT | O_RDWR;
+ mdb.db = dbopen(mdb.dbn, flags, 0644, DB_BTREE, &info);
+ mdb.idx = dbopen(mdb.idxn, flags, 0644, DB_RECNO, NULL);
- if (NULL == db) {
- perror(fbuf);
+ if (NULL == mdb.db) {
+ perror(mdb.dbn);
exit((int)MANDOCLEVEL_SYSERR);
- } else if (NULL == idx) {
- perror(ibuf);
+ } else if (NULL == mdb.idx) {
+ perror(mdb.idxn);
exit((int)MANDOCLEVEL_SYSERR);
}
- if (verb > 2) {
- printf("%s: Opened\n", fbuf);
- printf("%s: Opened\n", ibuf);
- }
+ ofile_argbuild(argc, argv, &of, dir);
- ofile_argbuild(argc, argv, &of);
if (NULL == of)
goto out;
- of = of->first;
-
- index_prune(of, db, fbuf, idx, ibuf,
- &maxrec, &recs, &recsz);
+ index_prune(of, &mdb, &recs, dir);
/*
- * Go to the root of the respective manual tree
- * such that .so links work. In case of failure,
- * just prod on, even though .so links won't work.
+ * Go to the root of the respective manual tree.
+ * This must work or no manuals may be found (they're
+ * indexed relative to the root).
*/
if (OP_UPDATE == op) {
- chdir(dir);
+ if (-1 == chdir(dir)) {
+ perror(dir);
+ exit((int)MANDOCLEVEL_SYSERR);
+ }
index_merge(of, mp, &dbuf, &buf, hash,
- db, fbuf, idx, ibuf,
- maxrec, recs, reccur);
+ &mdb, &recs, dir);
}
goto out;
if (argc > 0) {
dirs.paths = mandoc_calloc(argc, sizeof(char *));
dirs.sz = argc;
- for (i = 0; i < argc; i++) {
- if (NULL == (cp = realpath(argv[i], pbuf))) {
- perror(argv[i]);
- goto out;
- }
- dirs.paths[i] = mandoc_strdup(cp);
- }
+ for (i = 0; i < argc; i++)
+ dirs.paths[i] = mandoc_strdup(argv[i]);
} else
- manpath_parse(&dirs, NULL, NULL);
+ manpath_parse(&dirs, dir, NULL, NULL);
for (i = 0; i < dirs.sz; i++) {
- ibuf[0] = fbuf[0] = '\0';
-
- strlcat(fbuf, dirs.paths[i], MAXPATHLEN);
- strlcat(fbuf, "/", MAXPATHLEN);
- sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN);
-
- strlcat(ibuf, dirs.paths[i], MAXPATHLEN);
- strlcat(ibuf, "/", MAXPATHLEN);
- sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN);
+ /*
+ * Go to the root of the respective manual tree.
+ * This must work or no manuals may be found:
+ * They are indexed relative to the root.
+ */
- if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) {
- fprintf(stderr, "%s: Path too long\n",
- dirs.paths[i]);
- exit((int)MANDOCLEVEL_BADARG);
+ if (-1 == chdir(dirs.paths[i])) {
+ perror(dirs.paths[i]);
+ exit((int)MANDOCLEVEL_SYSERR);
}
- if (db)
- (*db->close)(db);
- if (idx)
- (*idx->close)(idx);
+ strlcpy(mdb.dbn, MANDOC_DB, MAXPATHLEN);
+ strlcpy(mdb.idxn, MANDOC_IDX, MAXPATHLEN);
- db = dbopen(fbuf, flags, 0644, DB_BTREE, &info);
- idx = dbopen(ibuf, flags, 0644, DB_RECNO, NULL);
+ flags = O_CREAT | O_TRUNC | O_RDWR;
+ mdb.db = dbopen(mdb.dbn, flags, 0644, DB_BTREE, &info);
+ mdb.idx = dbopen(mdb.idxn, flags, 0644, DB_RECNO, NULL);
- if (NULL == db) {
- perror(fbuf);
+ if (NULL == mdb.db) {
+ perror(mdb.dbn);
exit((int)MANDOCLEVEL_SYSERR);
- } else if (NULL == idx) {
- perror(ibuf);
+ } else if (NULL == mdb.idx) {
+ perror(mdb.idxn);
exit((int)MANDOCLEVEL_SYSERR);
}
- if (verb > 2) {
- printf("%s: Truncated\n", fbuf);
- printf("%s: Truncated\n", ibuf);
- }
-
- ofile_free(of);
- of = NULL;
-
- if ( ! ofile_dirbuild(dirs.paths[i], NULL, NULL,
- 0, &of))
- exit((int)MANDOCLEVEL_SYSERR);
-
- if (NULL == of)
- continue;
-
- of = of->first;
-
/*
- * Go to the root of the respective manual tree
- * such that .so links work. In case of failure,
- * just prod on, even though .so links won't work.
+ * Search for manuals and fill the new database.
*/
- chdir(dirs.paths[i]);
- index_merge(of, mp, &dbuf, &buf, hash, db, fbuf,
- idx, ibuf, maxrec, recs, reccur);
+ strlcpy(dirbuf, dirs.paths[i], MAXPATHLEN);
+ ofile_dirbuild(".", "", "", 0, &of, dirbuf);
+
+ if (NULL != of) {
+ index_merge(of, mp, &dbuf, &buf, hash,
+ &mdb, &recs, dirs.paths[i]);
+ ofile_free(of);
+ of = NULL;
+ }
+
+ (*mdb.db->close)(mdb.db);
+ (*mdb.idx->close)(mdb.idx);
+ mdb.db = NULL;
+ mdb.idx = NULL;
}
out:
- if (db)
- (*db->close)(db);
- if (idx)
- (*idx->close)(idx);
+ if (mdb.db)
+ (*mdb.db->close)(mdb.db);
+ if (mdb.idx)
+ (*mdb.idx->close)(mdb.idx);
if (hash)
(*hash->close)(hash);
if (mp)
ofile_free(of);
free(buf.cp);
free(dbuf.cp);
- free(recs);
+ free(recs.stack);
return(MANDOCLEVEL_OK);
+
+usage:
+ fprintf(stderr,
+ "usage: %s [-av] [-C file] | dir ... | -t file ...\n"
+ " -d dir [file ...] | "
+ "-u dir [file ...]\n",
+ progname);
+
+ return((int)MANDOCLEVEL_BADARG);
}
void
index_merge(const struct of *of, struct mparse *mp,
struct buf *dbuf, struct buf *buf, DB *hash,
- DB *db, const char *dbf, DB *idx, const char *idxf,
- recno_t maxrec, const recno_t *recs, size_t reccur)
+ struct mdb *mdb, struct recs *recs,
+ const char *basedir)
{
recno_t rec;
- int ch;
+ int ch, skip;
DBT key, val;
+ DB *files; /* temporary file name table */
+ char emptystring[1] = {'\0'};
struct mdoc *mdoc;
struct man *man;
- const char *fn, *msec, *mtitle, *arch;
+ char *p;
+ const char *fn, *msec, *march, *mtitle;
+ uint64_t mask;
size_t sv;
unsigned seq;
- struct db_val vbuf;
+ uint64_t vbuf[2];
+ char type;
+
+ if (warnings) {
+ files = NULL;
+ hash_reset(&files);
+ }
- for (rec = 0; of; of = of->next) {
+ rec = 0;
+ for (of = of->first; of; of = of->next) {
fn = of->fname;
/*
- * Reclaim an empty index record, if available.
+ * Try interpreting the file as mdoc(7) or man(7)
+ * source code, unless it is already known to be
+ * formatted. Fall back to formatted mode.
*/
- if (reccur > 0) {
- --reccur;
- rec = recs[(int)reccur];
- } else if (maxrec > 0) {
- rec = maxrec;
- maxrec = 0;
- } else
- rec++;
-
mparse_reset(mp);
- hash_reset(&hash);
mdoc = NULL;
man = NULL;
- /*
- * Try interpreting the file as mdoc(7) or man(7)
- * source code, unless it is already known to be
- * formatted. Fall back to formatted mode.
- */
-
if ((MANDOC_SRC & of->src_form ||
! (MANDOC_FORM & of->src_form)) &&
MANDOCLEVEL_FATAL > mparse_readfd(mp, -1, fn))
if (NULL != mdoc) {
msec = mdoc_meta(mdoc)->msec;
- arch = mdoc_meta(mdoc)->arch;
+ march = mdoc_meta(mdoc)->arch;
+ if (NULL == march)
+ march = "";
mtitle = mdoc_meta(mdoc)->title;
} else if (NULL != man) {
msec = man_meta(man)->msec;
- arch = NULL;
+ march = "";
mtitle = man_meta(man)->title;
} else {
msec = of->sec;
- arch = of->arch;
+ march = of->arch;
mtitle = of->title;
}
/*
- * By default, skip a file if the manual section
- * and architecture given in the file disagree
- * with the directory where the file is located.
+ * Check whether the manual section given in a file
+ * agrees with the directory where the file is located.
+ * Some manuals have suffixes like (3p) on their
+ * section number either inside the file or in the
+ * directory name, some are linked into more than one
+ * section, like encrypt(1) = makekey(8). Do not skip
+ * manuals for such reasons.
*/
- if (0 == use_all) {
- assert(of->sec);
- assert(msec);
- if (strcmp(msec, of->sec))
- continue;
-
- if (NULL == arch) {
- if (NULL != of->arch)
- continue;
- } else if (NULL == of->arch ||
- strcmp(arch, of->arch))
- continue;
- }
+ skip = 0;
+ assert(of->sec);
+ assert(msec);
+ if (strcasecmp(msec, of->sec))
+ WARNING(fn, basedir, "Section \"%s\" manual "
+ "in \"%s\" directory", msec, of->sec);
+ /*
+ * Manual page directories exist for each kernel
+ * architecture as returned by machine(1).
+ * However, many manuals only depend on the
+ * application architecture as returned by arch(1).
+ * For example, some (2/ARM) manuals are shared
+ * across the "armish" and "zaurus" kernel
+ * architectures.
+ * A few manuals are even shared across completely
+ * different architectures, for example fdformat(1)
+ * on amd64, i386, sparc, and sparc64.
+ * Thus, warn about architecture mismatches,
+ * but don't skip manuals for this reason.
+ */
- if (NULL == arch)
- arch = "";
+ assert(of->arch);
+ assert(march);
+ if (strcasecmp(march, of->arch))
+ WARNING(fn, basedir, "Architecture \"%s\" "
+ "manual in \"%s\" directory",
+ march, of->arch);
- /*
+ /*
* By default, skip a file if the title given
* in the file disagrees with the file name.
- * If both agree, use the file name as the title,
- * because the one in the file usually is all caps.
+ * Do not warn, this happens for all MLINKs.
*/
assert(of->title);
assert(mtitle);
+ if (strcasecmp(mtitle, of->title))
+ skip = 1;
- if (0 == strcasecmp(mtitle, of->title))
- mtitle = of->title;
- else if (0 == use_all)
+ /*
+ * Build a title string for the file. If it matches
+ * the location of the file, remember the title as
+ * found; else, remember it as missing.
+ */
+
+ if (warnings) {
+ buf->len = 0;
+ buf_appendb(buf, mtitle, strlen(mtitle));
+ buf_appendb(buf, "(", 1);
+ buf_appendb(buf, msec, strlen(msec));
+ if ('\0' != *march) {
+ buf_appendb(buf, "/", 1);
+ buf_appendb(buf, march, strlen(march));
+ }
+ buf_appendb(buf, ")", 2);
+ for (p = buf->cp; '\0' != *p; p++)
+ *p = tolower(*p);
+ key.data = buf->cp;
+ key.size = buf->len;
+ val.data = NULL;
+ val.size = 0;
+ if (0 == skip)
+ val.data = emptystring;
+ else {
+ ch = (*files->get)(files, &key, &val, 0);
+ if (ch < 0) {
+ perror("hash");
+ exit((int)MANDOCLEVEL_SYSERR);
+ } else if (ch > 0) {
+ val.data = (void *)fn;
+ val.size = strlen(fn) + 1;
+ } else
+ val.data = NULL;
+ }
+ if (NULL != val.data &&
+ (*files->put)(files, &key, &val, 0) < 0) {
+ perror("hash");
+ exit((int)MANDOCLEVEL_SYSERR);
+ }
+ }
+
+ if (skip && !use_all)
continue;
- /*
+ /*
* The index record value consists of a nil-terminated
* filename, a nil-terminated manual section, and a
- * nil-terminated description. Since the description
- * may not be set, we set a sentinel to see if we're
- * going to write a nil byte in its place.
+ * nil-terminated description. Use the actual
+ * location of the file, such that the user can find
+ * it with man(1). Since the description may not be
+ * set, we set a sentinel to see if we're going to
+ * write a nil byte in its place.
*/
dbuf->len = 0;
- buf_append(dbuf, mdoc ? "mdoc" : (man ? "man" : "cat"));
+ type = mdoc ? 'd' : (man ? 'a' : 'c');
+ buf_appendb(dbuf, &type, 1);
buf_appendb(dbuf, fn, strlen(fn) + 1);
- buf_appendb(dbuf, msec, strlen(msec) + 1);
- buf_appendb(dbuf, mtitle, strlen(mtitle) + 1);
- buf_appendb(dbuf, arch, strlen(arch) + 1);
+ buf_appendb(dbuf, of->sec, strlen(of->sec) + 1);
+ buf_appendb(dbuf, of->title, strlen(of->title) + 1);
+ buf_appendb(dbuf, of->arch, strlen(of->arch) + 1);
sv = dbuf->len;
- /* Fix the record number in the btree value. */
+ /*
+ * Collect keyword/mask pairs.
+ * Each pair will become a new btree node.
+ */
+ hash_reset(&hash);
if (mdoc)
pmdoc_node(hash, buf, dbuf,
mdoc_node(mdoc), mdoc_meta(mdoc));
else if (man)
pman_node(hash, buf, dbuf, man_node(man));
else
- pformatted(hash, buf, dbuf, of);
+ pformatted(hash, buf, dbuf, of, basedir);
+
+ /* Test mode, do not access any database. */
+
+ if (NULL == mdb->db || NULL == mdb->idx)
+ continue;
+
+ /*
+ * Make sure the file name is always registered
+ * as an .Nm search key.
+ */
+ buf->len = 0;
+ buf_append(buf, of->title);
+ hash_put(hash, buf, TYPE_Nm);
/*
- * Copy from the in-memory hashtable of pending keywords
- * into the database.
+ * Reclaim an empty index record, if available.
+ * Use its record number for all new btree nodes.
+ */
+
+ if (recs->cur > 0) {
+ recs->cur--;
+ rec = recs->stack[(int)recs->cur];
+ } else if (recs->last > 0) {
+ rec = recs->last;
+ recs->last = 0;
+ } else
+ rec++;
+ vbuf[1] = htobe64(rec);
+
+ /*
+ * Copy from the in-memory hashtable of pending
+ * keyword/mask pairs into the database.
*/
- vbuf.rec = htobe32(rec);
seq = R_FIRST;
while (0 == (ch = (*hash->seq)(hash, &key, &val, seq))) {
seq = R_NEXT;
- vbuf.mask = htobe64(*(uint64_t *)val.data);
- val.size = sizeof(struct db_val);
+ assert(sizeof(uint64_t) == val.size);
+ memcpy(&mask, val.data, val.size);
+ vbuf[0] = htobe64(mask);
+ val.size = sizeof(vbuf);
val.data = &vbuf;
- dbt_put(db, dbf, &key, &val);
+ dbt_put(mdb->db, mdb->dbn, &key, &val);
}
if (ch < 0) {
perror("hash");
exit((int)MANDOCLEVEL_SYSERR);
}
-
+
/*
* Apply to the index. If we haven't had a description
* set, put an empty one in now.
val.size = dbuf->len;
if (verb)
- printf("%s: Added index\n", fn);
+ printf("%s: Adding to index: %s\n", basedir, fn);
- dbt_put(idx, idxf, &key, &val);
+ dbt_put(mdb->idx, mdb->idxn, &key, &val);
+ }
+
+ /*
+ * Iterate the remembered file titles and check that
+ * all files can be found by their main title.
+ */
+
+ if (warnings) {
+ seq = R_FIRST;
+ while (0 == (*files->seq)(files, &key, &val, seq)) {
+ seq = R_NEXT;
+ if (val.size)
+ WARNING((char *)val.data, basedir,
+ "Probably unreachable, title "
+ "is %s", (char *)key.data);
+ }
+ (*files->close)(files);
}
}
* in `idx' (zeroing its value size).
*/
static void
-index_prune(const struct of *ofile, DB *db, const char *dbf,
- DB *idx, const char *idxf,
- recno_t *maxrec, recno_t **recs, size_t *recsz)
+index_prune(const struct of *ofile, struct mdb *mdb,
+ struct recs *recs, const char *basedir)
{
const struct of *of;
- const char *fn, *cp;
- struct db_val *vbuf;
+ const char *fn;
+ uint64_t vbuf[2];
unsigned seq, sseq;
DBT key, val;
- size_t reccur;
int ch;
- reccur = 0;
+ recs->cur = 0;
seq = R_FIRST;
- while (0 == (ch = (*idx->seq)(idx, &key, &val, seq))) {
+ while (0 == (ch = (*mdb->idx->seq)(mdb->idx, &key, &val, seq))) {
seq = R_NEXT;
- *maxrec = *(recno_t *)key.data;
- cp = val.data;
+ assert(sizeof(recno_t) == key.size);
+ memcpy(&recs->last, key.data, key.size);
/* Deleted records are zero-sized. Skip them. */
* Failing any of these, we go into our error handler.
*/
- if (NULL == (fn = memchr(cp, '\0', val.size)))
- break;
- if (++fn - cp >= (int)val.size)
- break;
- if (NULL == memchr(fn, '\0', val.size - (fn - cp)))
+ fn = (char *)val.data + 1;
+ if (NULL == memchr(fn, '\0', val.size - 1))
break;
- /*
+ /*
* Search for the file in those we care about.
* XXX: build this into a tree. Too slow.
*/
- for (of = ofile; of; of = of->next)
+ for (of = ofile->first; of; of = of->next)
if (0 == strcmp(fn, of->fname))
break;
*/
sseq = R_FIRST;
- while (0 == (ch = (*db->seq)(db, &key, &val, sseq))) {
+ while (0 == (ch = (*mdb->db->seq)(mdb->db,
+ &key, &val, sseq))) {
sseq = R_NEXT;
- if (sizeof(struct db_val) != val.size)
+ if (sizeof(vbuf) != val.size)
break;
- vbuf = val.data;
- if (*maxrec != betoh32(vbuf->rec))
+ memcpy(vbuf, val.data, val.size);
+ if (recs->last != betoh64(vbuf[1]))
continue;
- if ((ch = (*db->del)(db, &key, R_CURSOR)) < 0)
+ if ((ch = (*mdb->db->del)(mdb->db,
+ &key, R_CURSOR)) < 0)
break;
}
if (ch < 0) {
- perror(dbf);
+ perror(mdb->dbn);
exit((int)MANDOCLEVEL_SYSERR);
} else if (1 != ch) {
- fprintf(stderr, "%s: Corrupt database\n", dbf);
+ fprintf(stderr, "%s: corrupt database\n",
+ mdb->dbn);
exit((int)MANDOCLEVEL_SYSERR);
}
if (verb)
- printf("%s: Deleted index\n", fn);
+ printf("%s: Deleting from index: %s\n",
+ basedir, fn);
val.size = 0;
- ch = (*idx->put)(idx, &key, &val, R_CURSOR);
+ ch = (*mdb->idx->put)(mdb->idx, &key, &val, R_CURSOR);
if (ch < 0)
break;
cont:
- if (reccur >= *recsz) {
- *recsz += MANDOC_SLOP;
- *recs = mandoc_realloc
- (*recs, *recsz * sizeof(recno_t));
+ if (recs->cur >= recs->size) {
+ recs->size += MANDOC_SLOP;
+ recs->stack = mandoc_realloc(recs->stack,
+ recs->size * sizeof(recno_t));
}
- (*recs)[(int)reccur] = *maxrec;
- reccur++;
+ recs->stack[(int)recs->cur] = recs->last;
+ recs->cur++;
}
if (ch < 0) {
- perror(idxf);
+ perror(mdb->idxn);
exit((int)MANDOCLEVEL_SYSERR);
} else if (1 != ch) {
- fprintf(stderr, "%s: Corrupt index\n", idxf);
+ fprintf(stderr, "%s: corrupt index\n", mdb->idxn);
exit((int)MANDOCLEVEL_SYSERR);
}
- (*maxrec)++;
+ recs->last++;
}
/*
static void
hash_put(DB *db, const struct buf *buf, uint64_t mask)
{
+ uint64_t oldmask;
DBT key, val;
int rc;
if ((rc = (*db->get)(db, &key, &val, 0)) < 0) {
perror("hash");
exit((int)MANDOCLEVEL_SYSERR);
- } else if (0 == rc)
- mask |= *(uint64_t *)val.data;
+ } else if (0 == rc) {
+ assert(sizeof(uint64_t) == val.size);
+ memcpy(&oldmask, val.data, val.size);
+ mask |= oldmask;
+ }
val.data = &mask;
val.size = sizeof(uint64_t);
pman_node(MAN_ARGS)
{
const struct man_node *head, *body;
- const char *start, *sv;
- size_t sz;
+ char *start, *sv, *title;
+ size_t sz, titlesz;
if (NULL == n)
return(0);
NULL != (body = body->child) &&
MAN_TEXT == body->type) {
- assert(body->string);
- start = sv = body->string;
+ title = NULL;
+ titlesz = 0;
+ /*
+ * Suck the entire NAME section into memory.
+ * Yes, we might run away.
+ * But too many manuals have big, spread-out
+ * NAME sections over many lines.
+ */
+ for ( ; NULL != body; body = body->next) {
+ if (MAN_TEXT != body->type)
+ break;
+ if (0 == (sz = strlen(body->string)))
+ continue;
+ title = mandoc_realloc
+ (title, titlesz + sz + 1);
+ memcpy(title + titlesz, body->string, sz);
+ titlesz += sz + 1;
+ title[(int)titlesz - 1] = ' ';
+ }
+ if (NULL == title)
+ return(0);
+
+ title = mandoc_realloc(title, titlesz + 1);
+ title[(int)titlesz] = '\0';
+
+ /* Skip leading space. */
+
+ sv = title;
+ while (isspace((unsigned char)*sv))
+ sv++;
+
+ if (0 == (sz = strlen(sv))) {
+ free(title);
+ return(0);
+ }
+
+ /* Erase trailing space. */
+
+ start = &sv[sz - 1];
+ while (start > sv && isspace((unsigned char)*start))
+ *start-- = '\0';
+
+ if (start == sv) {
+ free(title);
+ return(0);
+ }
+
+ start = sv;
/*
* Go through a special heuristic dance here.
if (sv == start) {
buf_append(buf, start);
+ free(title);
return(1);
}
- while (' ' == *start)
+ while (isspace((unsigned char)*start))
start++;
if (0 == strncmp(start, "-", 1))
start += 1;
+ else if (0 == strncmp(start, "\\-\\-", 4))
+ start += 4;
else if (0 == strncmp(start, "\\-", 2))
start += 2;
else if (0 == strncmp(start, "\\(en", 4))
buf_appendb(buf, start, sz);
hash_put(hash, buf, TYPE_Nd);
+ free(title);
}
}
* By necessity, this involves rather crude guesswork.
*/
static void
-pformatted(DB *hash, struct buf *buf, struct buf *dbuf,
- const struct of *of)
+pformatted(DB *hash, struct buf *buf, struct buf *dbuf,
+ const struct of *of, const char *basedir)
{
FILE *stream;
- char *line, *p;
- size_t len, plen;
+ char *line, *p, *title;
+ size_t len, plen, titlesz;
if (NULL == (stream = fopen(of->fname, "r"))) {
- perror(of->fname);
+ WARNING(of->fname, basedir, "%s", strerror(errno));
return;
}
buf_append(buf, of->title);
hash_put(hash, buf, TYPE_Nm);
- while (NULL != (line = fgetln(stream, &len)) && '\n' != *line)
- /* Skip to first blank line. */ ;
+ /* Skip to first blank line. */
+
+ while (NULL != (line = fgetln(stream, &len)))
+ if ('\n' == *line)
+ break;
+
+ /*
+ * Assume the first line that is not indented
+ * is the first section header. Skip to it.
+ */
+
+ while (NULL != (line = fgetln(stream, &len)))
+ if ('\n' != *line && ' ' != *line)
+ break;
+
+ /*
+ * Read up until the next section into a buffer.
+ * Strip the leading and trailing newline from each read line,
+ * appending a trailing space.
+ * Ignore empty (whitespace-only) lines.
+ */
+
+ titlesz = 0;
+ title = NULL;
- while (NULL != (line = fgetln(stream, &len)) &&
- ('\n' == *line || ' ' == *line))
- /* Skip to first section header. */ ;
+ while (NULL != (line = fgetln(stream, &len))) {
+ if (' ' != *line || '\n' != line[(int)len - 1])
+ break;
+ while (len > 0 && isspace((unsigned char)*line)) {
+ line++;
+ len--;
+ }
+ if (1 == len)
+ continue;
+ title = mandoc_realloc(title, titlesz + len);
+ memcpy(title + titlesz, line, len);
+ titlesz += len;
+ title[(int)titlesz - 1] = ' ';
+ }
/*
- * If no page content can be found,
- * reuse the page title as the page description.
+ * If no page content can be found, or the input line
+ * is already the next section header, or there is no
+ * trailing newline, reuse the page title as the page
+ * description.
*/
- if (NULL == (line = fgetln(stream, &len))) {
+ if (NULL == title || '\0' == *title) {
+ WARNING(of->fname, basedir,
+ "Cannot find NAME section");
buf_appendb(dbuf, buf->cp, buf->size);
hash_put(hash, buf, TYPE_Nd);
fclose(stream);
+ free(title);
return;
}
- fclose(stream);
+
+ title = mandoc_realloc(title, titlesz + 1);
+ title[(int)titlesz] = '\0';
/*
- * If there is a dash, skip to the text following it.
+ * Skip to the first dash.
+ * Use the remaining line as the description (no more than 70
+ * bytes).
*/
- for (p = line, plen = len; plen; p++, plen--)
- if ('-' == *p)
- break;
- for ( ; plen; p++, plen--)
- if ('-' != *p && ' ' != *p && 8 != *p)
- break;
- if (0 == plen) {
- p = line;
- plen = len;
+ if (NULL != (p = strstr(title, "- "))) {
+ for (p += 2; ' ' == *p || '\b' == *p; p++)
+ /* Skip to next word. */ ;
+ } else {
+ WARNING(of->fname, basedir,
+ "No dash in title line");
+ p = title;
}
- /*
- * Copy the rest of the line, but no more than 70 bytes.
- */
+ plen = strlen(p);
- if (70 < plen)
- plen = 70;
- p[plen-1] = '\0';
- buf_appendb(dbuf, p, plen);
+ /* Strip backspace-encoding from line. */
+
+ while (NULL != (line = memchr(p, '\b', plen))) {
+ len = line - p;
+ if (0 == len) {
+ memmove(line, line + 1, plen--);
+ continue;
+ }
+ memmove(line - 1, line + 1, plen - len);
+ plen -= 2;
+ }
+
+ buf_appendb(dbuf, p, plen + 1);
buf->len = 0;
- buf_appendb(buf, p, plen);
+ buf_appendb(buf, p, plen + 1);
hash_put(hash, buf, TYPE_Nd);
+ fclose(stream);
+ free(title);
}
static void
-ofile_argbuild(int argc, char *argv[], struct of **of)
+ofile_argbuild(int argc, char *argv[],
+ struct of **of, const char *basedir)
{
char buf[MAXPATHLEN];
- char *sec, *arch, *title, *p;
+ const char *sec, *arch, *title;
+ char *p;
int i, src_form;
struct of *nof;
fprintf(stderr, "%s: Path too long\n", argv[i]);
continue;
}
- sec = arch = title = NULL;
+ sec = arch = title = "";
src_form = 0;
p = strrchr(buf, '\0');
while (p-- > buf) {
- if (NULL == sec && '.' == *p) {
+ if ('\0' == *sec && '.' == *p) {
sec = p + 1;
*p = '\0';
if ('0' == *sec)
}
if ('/' != *p)
continue;
- if (NULL == title) {
+ if ('\0' == *title) {
title = p + 1;
*p = '\0';
continue;
arch = p + 1;
break;
}
- if (NULL == title)
+ if ('\0' == *title) {
+ WARNING(argv[i], basedir,
+ "Cannot deduce title from filename");
title = buf;
+ }
/*
* Build the file structure.
nof = mandoc_calloc(1, sizeof(struct of));
nof->fname = mandoc_strdup(argv[i]);
- if (NULL != sec)
- nof->sec = mandoc_strdup(sec);
- if (NULL != arch)
- nof->arch = mandoc_strdup(arch);
+ nof->sec = mandoc_strdup(sec);
+ nof->arch = mandoc_strdup(arch);
nof->title = mandoc_strdup(title);
nof->src_form = src_form;
* Add the structure to the list.
*/
- if (verb > 2)
- printf("%s: Scheduling\n", argv[i]);
if (NULL == *of) {
*of = nof;
(*of)->first = nof;
* Recursively build up a list of files to parse.
* We use this instead of ftw() and so on because I don't want global
* variables hanging around.
- * This ignores the mandoc.db and mandoc.index files, but assumes that
+ * This ignores the mandocdb.db and mandocdb.index files, but assumes that
* everything else is a manual.
* Pass in a pointer to a NULL structure for the first invocation.
*/
-static int
+static void
ofile_dirbuild(const char *dir, const char* psec, const char *parch,
- int p_src_form, struct of **of)
+ int p_src_form, struct of **of, char *basedir)
{
char buf[MAXPATHLEN];
- struct stat sb;
size_t sz;
DIR *d;
const char *fn, *sec, *arch;
int src_form;
if (NULL == (d = opendir(dir))) {
- perror(dir);
- return(0);
+ WARNING("", dir, "%s", strerror(errno));
+ return;
}
while (NULL != (dp = readdir(d))) {
* cat<section>/[<arch>/]
*/
- if (NULL == sec) {
+ if ('\0' == *sec) {
if(0 == strncmp("man", fn, 3)) {
src_form |= MANDOC_SRC;
sec = fn + 3;
} else if (0 == strncmp("cat", fn, 3)) {
src_form |= MANDOC_FORM;
sec = fn + 3;
- } else if (use_all)
- sec = fn;
- else
- continue;
- } else if (NULL == arch && (use_all ||
- NULL == strchr(fn, '.')))
+ } else {
+ WARNING(fn, basedir, "Bad section");
+ if (use_all)
+ sec = fn;
+ else
+ continue;
+ }
+ } else if ('\0' == *arch) {
+ if (NULL != strchr(fn, '.')) {
+ WARNING(fn, basedir, "Bad architecture");
+ if (0 == use_all)
+ continue;
+ }
arch = fn;
- else if (0 == use_all)
- continue;
+ } else {
+ WARNING(fn, basedir, "Excessive subdirectory");
+ if (0 == use_all)
+ continue;
+ }
buf[0] = '\0';
strlcat(buf, dir, MAXPATHLEN);
strlcat(buf, "/", MAXPATHLEN);
+ strlcat(basedir, "/", MAXPATHLEN);
+ strlcat(basedir, fn, MAXPATHLEN);
sz = strlcat(buf, fn, MAXPATHLEN);
if (MAXPATHLEN <= sz) {
- fprintf(stderr, "%s: Path too long\n", dir);
- return(0);
+ WARNING(fn, basedir, "Path too long");
+ continue;
}
-
- if (verb > 2)
- printf("%s: Scanning\n", buf);
- if ( ! ofile_dirbuild(buf, sec, arch,
- src_form, of))
- return(0);
+ ofile_dirbuild(buf, sec, arch,
+ src_form, of, basedir);
+
+ p = strrchr(basedir, '/');
+ *p = '\0';
+ continue;
}
- if (DT_REG != dp->d_type ||
- (NULL == psec && !use_all) ||
- !strcmp(MANDOC_DB, fn) ||
- !strcmp(MANDOC_IDX, fn))
+
+ if (DT_REG != dp->d_type) {
+ WARNING(fn, basedir, "Not a regular file");
continue;
+ }
+ if (!strcmp(MANDOC_DB, fn) || !strcmp(MANDOC_IDX, fn))
+ continue;
+ if ('\0' == *psec) {
+ WARNING(fn, basedir, "File outside section");
+ if (0 == use_all)
+ continue;
+ }
/*
* By default, skip files where the file name suffix
*/
suffix = strrchr(fn, '.');
- if (0 == use_all) {
- if (NULL == suffix)
+ if (NULL == suffix) {
+ WARNING(fn, basedir, "No filename suffix");
+ if (0 == use_all)
continue;
- if ((MANDOC_SRC & src_form &&
- strcmp(suffix + 1, psec)) ||
+ } else if ((MANDOC_SRC & src_form &&
+ strcmp(suffix + 1, psec)) ||
(MANDOC_FORM & src_form &&
- strcmp(suffix + 1, "0")))
- continue;
- }
- if (NULL != suffix) {
+ strcmp(suffix + 1, "0"))) {
+ WARNING(fn, basedir, "Wrong filename suffix");
+ if (0 == use_all)
+ continue;
if ('0' == suffix[1])
src_form |= MANDOC_FORM;
else if ('1' <= suffix[1] && '9' >= suffix[1])
src_form |= MANDOC_SRC;
}
-
/*
* Skip formatted manuals if a source version is
* available. Ignore the age: it is very unlikely
* and in ports, old manuals get removed on update.
*/
if (0 == use_all && MANDOC_FORM & src_form &&
- NULL != psec) {
+ '\0' != *psec) {
buf[0] = '\0';
strlcat(buf, dir, MAXPATHLEN);
p = strrchr(buf, '/');
+ if ('\0' != *parch && NULL != p)
+ for (p--; p > buf; p--)
+ if ('/' == *p)
+ break;
if (NULL == p)
p = buf;
else
strlcat(buf, "/", MAXPATHLEN);
sz = strlcat(buf, fn, MAXPATHLEN);
if (sz >= MAXPATHLEN) {
- fprintf(stderr, "%s: Path too long\n", buf);
+ WARNING(fn, basedir, "Path too long");
continue;
}
q = strrchr(buf, '.');
*q = '\0';
sz = strlcat(buf, psec, MAXPATHLEN);
if (sz >= MAXPATHLEN) {
- fprintf(stderr,
- "%s: Path too long\n", buf);
+ WARNING(fn, basedir, "Path too long");
continue;
}
- if (0 == stat(buf, &sb))
+ if (0 == access(buf, R_OK))
continue;
}
}
buf[0] = '\0';
- strlcat(buf, dir, MAXPATHLEN);
- strlcat(buf, "/", MAXPATHLEN);
+ assert('.' == dir[0]);
+ if ('/' == dir[1]) {
+ strlcat(buf, dir + 2, MAXPATHLEN);
+ strlcat(buf, "/", MAXPATHLEN);
+ }
sz = strlcat(buf, fn, MAXPATHLEN);
if (sz >= MAXPATHLEN) {
- fprintf(stderr, "%s: Path too long\n", dir);
+ WARNING(fn, basedir, "Path too long");
continue;
}
nof = mandoc_calloc(1, sizeof(struct of));
nof->fname = mandoc_strdup(buf);
- if (NULL != psec)
- nof->sec = mandoc_strdup(psec);
- if (NULL != parch)
- nof->arch = mandoc_strdup(parch);
+ nof->sec = mandoc_strdup(psec);
+ nof->arch = mandoc_strdup(parch);
nof->src_form = src_form;
/*
* Add the structure to the list.
*/
- if (verb > 2)
- printf("%s: Scheduling\n", buf);
if (NULL == *of) {
*of = nof;
(*of)->first = nof;
}
closedir(d);
- return(1);
}
static void
{
struct of *nof;
- while (of) {
+ if (NULL != of)
+ of = of->first;
+
+ while (NULL != of) {
nof = of->next;
free(of->fname);
free(of->sec);
of = nof;
}
}
-
-static void
-usage(void)
-{
-
- fprintf(stderr, "usage: %s [-v] "
- "[-d dir [files...] |"
- " -u dir [files...] |"
- " dir...]\n", progname);
-}
git.ckatri.com / mandoc.git / blobdiff