| version 1.11, 2011/11/24 12:54:19 |
version 1.235, 2017/01/10 21:59:47 |
|
|
| /* $Id$ */ |
/* $Id$ */ |
| /* |
/* |
| * Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv> |
* Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv> |
| |
* Copyright (c) 2011-2017 Ingo Schwarze <schwarze@openbsd.org> |
| |
* Copyright (c) 2016 Ed Maste <emaste@freebsd.org> |
| * |
* |
| * Permission to use, copy, modify, and distribute this software for any |
* Permission to use, copy, modify, and distribute this software for any |
| * purpose with or without fee is hereby granted, provided that the above |
* purpose with or without fee is hereby granted, provided that the above |
| * copyright notice and this permission notice appear in all copies. |
* copyright notice and this permission notice appear in all copies. |
| * |
* |
| * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES |
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES |
| * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF |
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF |
| * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR |
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR |
| * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES |
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES |
| * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN |
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN |
| * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF |
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF |
| * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. |
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. |
| */ |
*/ |
| #ifdef HAVE_CONFIG_H |
|
| #include "config.h" |
#include "config.h" |
| #endif |
|
| |
|
| #include <sys/param.h> |
#include <sys/types.h> |
| |
#include <sys/stat.h> |
| |
#include <sys/wait.h> |
| |
|
| #include <assert.h> |
#include <assert.h> |
| #include <dirent.h> |
#include <ctype.h> |
| |
#if HAVE_ERR |
| |
#include <err.h> |
| |
#endif |
| |
#include <errno.h> |
| #include <fcntl.h> |
#include <fcntl.h> |
| #include <getopt.h> |
#if HAVE_FTS |
| |
#include <fts.h> |
| |
#else |
| |
#include "compat_fts.h" |
| |
#endif |
| |
#include <limits.h> |
| |
#if HAVE_SANDBOX_INIT |
| |
#include <sandbox.h> |
| |
#endif |
| |
#include <stdarg.h> |
| |
#include <stddef.h> |
| #include <stdio.h> |
#include <stdio.h> |
| #include <stdint.h> |
#include <stdint.h> |
| #include <stdlib.h> |
#include <stdlib.h> |
| #include <string.h> |
#include <string.h> |
| |
#include <unistd.h> |
| |
|
| #ifdef __linux__ |
#include "mandoc_aux.h" |
| # include <db_185.h> |
#include "mandoc_ohash.h" |
| #else |
|
| # include <db.h> |
|
| #endif |
|
| |
|
| #include "man.h" |
|
| #include "mdoc.h" |
|
| #include "mandoc.h" |
#include "mandoc.h" |
| #include "mandocdb.h" |
#include "roff.h" |
| #include "manpath.h" |
#include "mdoc.h" |
| |
#include "man.h" |
| |
#include "manconf.h" |
| |
#include "mansearch.h" |
| |
#include "dba_array.h" |
| |
#include "dba.h" |
| |
|
| #define MANDOC_BUFSZ BUFSIZ |
extern const char *const mansearch_keynames[]; |
| #define MANDOC_SLOP 1024 |
|
| |
|
| /* Tiny list for files. No need to bring in QUEUE. */ |
enum op { |
| |
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_TEST /* change no databases, report potential problems */ |
| |
}; |
| |
|
| struct of { |
struct str { |
| char *fname; /* heap-allocated */ |
const struct mpage *mpage; /* if set, the owning parse */ |
| struct of *next; /* NULL for last one */ |
uint64_t mask; /* bitmask in sequence */ |
| struct of *first; /* first in list */ |
char key[]; /* rendered text */ |
| }; |
}; |
| |
|
| /* Buffer for storing growable data. */ |
struct inodev { |
| |
ino_t st_ino; |
| |
dev_t st_dev; |
| |
}; |
| |
|
| struct buf { |
struct mpage { |
| char *cp; |
struct inodev inodev; /* used for hashing routine */ |
| size_t len; /* current length */ |
struct dba_array *dba; |
| size_t size; /* total buffer size */ |
char *sec; /* section from file content */ |
| |
char *arch; /* architecture from file content */ |
| |
char *title; /* title from file content */ |
| |
char *desc; /* description from file content */ |
| |
struct mpage *next; /* singly linked list */ |
| |
struct mlink *mlinks; /* singly linked list */ |
| |
int name_head_done; |
| |
enum form form; /* format from file content */ |
| }; |
}; |
| |
|
| /* Operation we're going to perform. */ |
struct mlink { |
| |
char file[PATH_MAX]; /* filename rel. to manpath */ |
| |
char *dsec; /* section from directory */ |
| |
char *arch; /* architecture from directory */ |
| |
char *name; /* name from file name (not empty) */ |
| |
char *fsec; /* section from file name suffix */ |
| |
struct mlink *next; /* singly linked list */ |
| |
struct mpage *mpage; /* parent */ |
| |
int gzip; /* filename has a .gz suffix */ |
| |
enum form dform; /* format from directory */ |
| |
enum form fform; /* format from file name suffix */ |
| |
}; |
| |
|
| enum op { |
typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *, |
| OP_NEW = 0, /* new database */ |
const struct roff_node *); |
| OP_UPDATE, /* delete/add entries in existing database */ |
|
| OP_DELETE /* delete entries from existing database */ |
struct mdoc_handler { |
| |
mdoc_fp fp; /* optional handler */ |
| |
uint64_t mask; /* set unless handler returns 0 */ |
| |
int taboo; /* node flags that must not be set */ |
| }; |
}; |
| |
|
| #define MAN_ARGS DB *hash, \ |
|
| struct buf *buf, \ |
|
| struct buf *dbuf, \ |
|
| const struct man_node *n |
|
| #define MDOC_ARGS DB *hash, \ |
|
| struct buf *buf, \ |
|
| struct buf *dbuf, \ |
|
| const struct mdoc_node *n, \ |
|
| const struct mdoc_meta *m |
|
| |
|
| static void buf_appendmdoc(struct buf *, |
int mandocdb(int, char *[]); |
| const struct mdoc_node *, int); |
|
| static void buf_append(struct buf *, const char *); |
|
| static void buf_appendb(struct buf *, |
|
| const void *, size_t); |
|
| static void dbt_put(DB *, const char *, DBT *, DBT *); |
|
| static void hash_put(DB *, const struct buf *, uint64_t); |
|
| 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 *, int, |
|
| recno_t, const recno_t *, size_t); |
|
| static void index_prune(const struct of *, DB *, |
|
| const char *, DB *, const char *, |
|
| int, recno_t *, recno_t **, size_t *); |
|
| static void ofile_argbuild(char *[], int, int, struct of **); |
|
| static int ofile_dirbuild(const char *, int, struct of **); |
|
| static void ofile_free(struct of *); |
|
| static int pman_node(MAN_ARGS); |
|
| static void pmdoc_node(MDOC_ARGS); |
|
| static void pmdoc_An(MDOC_ARGS); |
|
| static void pmdoc_Cd(MDOC_ARGS); |
|
| static void pmdoc_Er(MDOC_ARGS); |
|
| static void pmdoc_Ev(MDOC_ARGS); |
|
| static void pmdoc_Fd(MDOC_ARGS); |
|
| static void pmdoc_In(MDOC_ARGS); |
|
| static void pmdoc_Fn(MDOC_ARGS); |
|
| static void pmdoc_Fo(MDOC_ARGS); |
|
| static void pmdoc_Nd(MDOC_ARGS); |
|
| static void pmdoc_Nm(MDOC_ARGS); |
|
| static void pmdoc_Pa(MDOC_ARGS); |
|
| static void pmdoc_St(MDOC_ARGS); |
|
| static void pmdoc_Vt(MDOC_ARGS); |
|
| static void pmdoc_Xr(MDOC_ARGS); |
|
| static void usage(void); |
|
| |
|
| typedef void (*pmdoc_nf)(MDOC_ARGS); |
static void dbadd(struct dba *, struct mpage *); |
| |
static void dbadd_mlink(const struct mlink *mlink); |
| |
static void dbprune(struct dba *); |
| |
static void dbwrite(struct dba *); |
| |
static void filescan(const char *); |
| |
#if HAVE_FTS_COMPARE_CONST |
| |
static int fts_compare(const FTSENT *const *, const FTSENT *const *); |
| |
#else |
| |
static int fts_compare(const FTSENT **, const FTSENT **); |
| |
#endif |
| |
static void mlink_add(struct mlink *, const struct stat *); |
| |
static void mlink_check(struct mpage *, struct mlink *); |
| |
static void mlink_free(struct mlink *); |
| |
static void mlinks_undupe(struct mpage *); |
| |
static void mpages_free(void); |
| |
static void mpages_merge(struct dba *, struct mparse *); |
| |
static void parse_cat(struct mpage *, int); |
| |
static void parse_man(struct mpage *, const struct roff_meta *, |
| |
const struct roff_node *); |
| |
static void parse_mdoc(struct mpage *, const struct roff_meta *, |
| |
const struct roff_node *); |
| |
static int parse_mdoc_head(struct mpage *, const struct roff_meta *, |
| |
const struct roff_node *); |
| |
static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *, |
| |
const struct roff_node *); |
| |
static void parse_mdoc_fname(struct mpage *, const struct roff_node *); |
| |
static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *, |
| |
const struct roff_node *); |
| |
static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *, |
| |
const struct roff_node *); |
| |
static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *, |
| |
const struct roff_node *); |
| |
static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *, |
| |
const struct roff_node *); |
| |
static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *, |
| |
const struct roff_node *); |
| |
static int parse_mdoc_Va(struct mpage *, const struct roff_meta *, |
| |
const struct roff_node *); |
| |
static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *, |
| |
const struct roff_node *); |
| |
static void putkey(const struct mpage *, char *, uint64_t); |
| |
static void putkeys(const struct mpage *, char *, size_t, uint64_t); |
| |
static void putmdockey(const struct mpage *, |
| |
const struct roff_node *, uint64_t, int); |
| |
static int render_string(char **, size_t *); |
| |
static void say(const char *, const char *, ...) |
| |
__attribute__((__format__ (printf, 2, 3))); |
| |
static int set_basedir(const char *, int); |
| |
static int treescan(void); |
| |
static size_t utf8(unsigned int, char [7]); |
| |
|
| static const pmdoc_nf mdocs[MDOC_MAX] = { |
static int nodb; /* no database changes */ |
| NULL, /* Ap */ |
static int mparse_options; /* abort the parse early */ |
| NULL, /* Dd */ |
static int use_all; /* use all found files */ |
| NULL, /* Dt */ |
static int debug; /* print what we're doing */ |
| NULL, /* Os */ |
static int warnings; /* warn about crap */ |
| NULL, /* Sh */ |
static int write_utf8; /* write UTF-8 output; else ASCII */ |
| NULL, /* Ss */ |
static int exitcode; /* to be returned by main */ |
| NULL, /* Pp */ |
static enum op op; /* operational mode */ |
| NULL, /* D1 */ |
static char basedir[PATH_MAX]; /* current base directory */ |
| NULL, /* Dl */ |
static struct mpage *mpage_head; /* list of distinct manual pages */ |
| NULL, /* Bd */ |
static struct ohash mpages; /* table of distinct manual pages */ |
| NULL, /* Ed */ |
static struct ohash mlinks; /* table of directory entries */ |
| NULL, /* Bl */ |
static struct ohash names; /* table of all names */ |
| NULL, /* El */ |
static struct ohash strings; /* table of all strings */ |
| NULL, /* It */ |
static uint64_t name_mask; |
| NULL, /* Ad */ |
|
| pmdoc_An, /* An */ |
static const struct mdoc_handler mdocs[MDOC_MAX] = { |
| NULL, /* Ar */ |
{ NULL, 0, 0 }, /* Ap */ |
| pmdoc_Cd, /* Cd */ |
{ NULL, 0, NODE_NOPRT }, /* Dd */ |
| NULL, /* Cm */ |
{ NULL, 0, NODE_NOPRT }, /* Dt */ |
| NULL, /* Dv */ |
{ NULL, 0, NODE_NOPRT }, /* Os */ |
| pmdoc_Er, /* Er */ |
{ parse_mdoc_Sh, TYPE_Sh, 0 }, /* Sh */ |
| pmdoc_Ev, /* Ev */ |
{ parse_mdoc_head, TYPE_Ss, 0 }, /* Ss */ |
| NULL, /* Ex */ |
{ NULL, 0, 0 }, /* Pp */ |
| NULL, /* Fa */ |
{ NULL, 0, 0 }, /* D1 */ |
| pmdoc_Fd, /* Fd */ |
{ NULL, 0, 0 }, /* Dl */ |
| NULL, /* Fl */ |
{ NULL, 0, 0 }, /* Bd */ |
| pmdoc_Fn, /* Fn */ |
{ NULL, 0, 0 }, /* Ed */ |
| NULL, /* Ft */ |
{ NULL, 0, 0 }, /* Bl */ |
| NULL, /* Ic */ |
{ NULL, 0, 0 }, /* El */ |
| pmdoc_In, /* In */ |
{ NULL, 0, 0 }, /* It */ |
| NULL, /* Li */ |
{ NULL, 0, 0 }, /* Ad */ |
| pmdoc_Nd, /* Nd */ |
{ NULL, TYPE_An, 0 }, /* An */ |
| pmdoc_Nm, /* Nm */ |
{ NULL, TYPE_Ar, 0 }, /* Ar */ |
| NULL, /* Op */ |
{ NULL, TYPE_Cd, 0 }, /* Cd */ |
| NULL, /* Ot */ |
{ NULL, TYPE_Cm, 0 }, /* Cm */ |
| pmdoc_Pa, /* Pa */ |
{ NULL, TYPE_Dv, 0 }, /* Dv */ |
| NULL, /* Rv */ |
{ NULL, TYPE_Er, 0 }, /* Er */ |
| pmdoc_St, /* St */ |
{ NULL, TYPE_Ev, 0 }, /* Ev */ |
| pmdoc_Vt, /* Va */ |
{ NULL, 0, 0 }, /* Ex */ |
| pmdoc_Vt, /* Vt */ |
{ NULL, TYPE_Fa, 0 }, /* Fa */ |
| pmdoc_Xr, /* Xr */ |
{ parse_mdoc_Fd, 0, 0 }, /* Fd */ |
| NULL, /* %A */ |
{ NULL, TYPE_Fl, 0 }, /* Fl */ |
| NULL, /* %B */ |
{ parse_mdoc_Fn, 0, 0 }, /* Fn */ |
| NULL, /* %D */ |
{ NULL, TYPE_Ft, 0 }, /* Ft */ |
| NULL, /* %I */ |
{ NULL, TYPE_Ic, 0 }, /* Ic */ |
| NULL, /* %J */ |
{ NULL, TYPE_In, 0 }, /* In */ |
| NULL, /* %N */ |
{ NULL, TYPE_Li, 0 }, /* Li */ |
| NULL, /* %O */ |
{ parse_mdoc_Nd, 0, 0 }, /* Nd */ |
| NULL, /* %P */ |
{ parse_mdoc_Nm, 0, 0 }, /* Nm */ |
| NULL, /* %R */ |
{ NULL, 0, 0 }, /* Op */ |
| NULL, /* %T */ |
{ NULL, 0, 0 }, /* Ot */ |
| NULL, /* %V */ |
{ NULL, TYPE_Pa, NODE_NOSRC }, /* Pa */ |
| NULL, /* Ac */ |
{ NULL, 0, 0 }, /* Rv */ |
| NULL, /* Ao */ |
{ NULL, TYPE_St, 0 }, /* St */ |
| NULL, /* Aq */ |
{ parse_mdoc_Va, TYPE_Va, 0 }, /* Va */ |
| NULL, /* At */ |
{ parse_mdoc_Va, TYPE_Vt, 0 }, /* Vt */ |
| NULL, /* Bc */ |
{ parse_mdoc_Xr, 0, 0 }, /* Xr */ |
| NULL, /* Bf */ |
{ NULL, 0, 0 }, /* %A */ |
| NULL, /* Bo */ |
{ NULL, 0, 0 }, /* %B */ |
| NULL, /* Bq */ |
{ NULL, 0, 0 }, /* %D */ |
| NULL, /* Bsx */ |
{ NULL, 0, 0 }, /* %I */ |
| NULL, /* Bx */ |
{ NULL, 0, 0 }, /* %J */ |
| NULL, /* Db */ |
{ NULL, 0, 0 }, /* %N */ |
| NULL, /* Dc */ |
{ NULL, 0, 0 }, /* %O */ |
| NULL, /* Do */ |
{ NULL, 0, 0 }, /* %P */ |
| NULL, /* Dq */ |
{ NULL, 0, 0 }, /* %R */ |
| NULL, /* Ec */ |
{ NULL, 0, 0 }, /* %T */ |
| NULL, /* Ef */ |
{ NULL, 0, 0 }, /* %V */ |
| NULL, /* Em */ |
{ NULL, 0, 0 }, /* Ac */ |
| NULL, /* Eo */ |
{ NULL, 0, 0 }, /* Ao */ |
| NULL, /* Fx */ |
{ NULL, 0, 0 }, /* Aq */ |
| NULL, /* Ms */ |
{ NULL, TYPE_At, 0 }, /* At */ |
| NULL, /* No */ |
{ NULL, 0, 0 }, /* Bc */ |
| NULL, /* Ns */ |
{ NULL, 0, 0 }, /* Bf */ |
| NULL, /* Nx */ |
{ NULL, 0, 0 }, /* Bo */ |
| NULL, /* Ox */ |
{ NULL, 0, 0 }, /* Bq */ |
| NULL, /* Pc */ |
{ NULL, TYPE_Bsx, NODE_NOSRC }, /* Bsx */ |
| NULL, /* Pf */ |
{ NULL, TYPE_Bx, 0 }, /* Bx */ |
| NULL, /* Po */ |
{ NULL, 0, 0 }, /* Db */ |
| NULL, /* Pq */ |
{ NULL, 0, 0 }, /* Dc */ |
| NULL, /* Qc */ |
{ NULL, 0, 0 }, /* Do */ |
| NULL, /* Ql */ |
{ NULL, 0, 0 }, /* Dq */ |
| NULL, /* Qo */ |
{ NULL, 0, 0 }, /* Ec */ |
| NULL, /* Qq */ |
{ NULL, 0, 0 }, /* Ef */ |
| NULL, /* Re */ |
{ NULL, TYPE_Em, 0 }, /* Em */ |
| NULL, /* Rs */ |
{ NULL, 0, 0 }, /* Eo */ |
| NULL, /* Sc */ |
{ NULL, TYPE_Fx, NODE_NOSRC }, /* Fx */ |
| NULL, /* So */ |
{ NULL, TYPE_Ms, 0 }, /* Ms */ |
| NULL, /* Sq */ |
{ NULL, 0, 0 }, /* No */ |
| NULL, /* Sm */ |
{ NULL, 0, 0 }, /* Ns */ |
| NULL, /* Sx */ |
{ NULL, TYPE_Nx, NODE_NOSRC }, /* Nx */ |
| NULL, /* Sy */ |
{ NULL, TYPE_Ox, NODE_NOSRC }, /* Ox */ |
| NULL, /* Tn */ |
{ NULL, 0, 0 }, /* Pc */ |
| NULL, /* Ux */ |
{ NULL, 0, 0 }, /* Pf */ |
| NULL, /* Xc */ |
{ NULL, 0, 0 }, /* Po */ |
| NULL, /* Xo */ |
{ NULL, 0, 0 }, /* Pq */ |
| pmdoc_Fo, /* Fo */ |
{ NULL, 0, 0 }, /* Qc */ |
| NULL, /* Fc */ |
{ NULL, 0, 0 }, /* Ql */ |
| NULL, /* Oo */ |
{ NULL, 0, 0 }, /* Qo */ |
| NULL, /* Oc */ |
{ NULL, 0, 0 }, /* Qq */ |
| NULL, /* Bk */ |
{ NULL, 0, 0 }, /* Re */ |
| NULL, /* Ek */ |
{ NULL, 0, 0 }, /* Rs */ |
| NULL, /* Bt */ |
{ NULL, 0, 0 }, /* Sc */ |
| NULL, /* Hf */ |
{ NULL, 0, 0 }, /* So */ |
| NULL, /* Fr */ |
{ NULL, 0, 0 }, /* Sq */ |
| NULL, /* Ud */ |
{ NULL, 0, 0 }, /* Sm */ |
| NULL, /* Lb */ |
{ NULL, 0, 0 }, /* Sx */ |
| NULL, /* Lp */ |
{ NULL, TYPE_Sy, 0 }, /* Sy */ |
| NULL, /* Lk */ |
{ NULL, TYPE_Tn, 0 }, /* Tn */ |
| NULL, /* Mt */ |
{ NULL, 0, NODE_NOSRC }, /* Ux */ |
| NULL, /* Brq */ |
{ NULL, 0, 0 }, /* Xc */ |
| NULL, /* Bro */ |
{ NULL, 0, 0 }, /* Xo */ |
| NULL, /* Brc */ |
{ parse_mdoc_Fo, 0, 0 }, /* Fo */ |
| NULL, /* %C */ |
{ NULL, 0, 0 }, /* Fc */ |
| NULL, /* Es */ |
{ NULL, 0, 0 }, /* Oo */ |
| NULL, /* En */ |
{ NULL, 0, 0 }, /* Oc */ |
| NULL, /* Dx */ |
{ NULL, 0, 0 }, /* Bk */ |
| NULL, /* %Q */ |
{ NULL, 0, 0 }, /* Ek */ |
| NULL, /* br */ |
{ NULL, 0, 0 }, /* Bt */ |
| NULL, /* sp */ |
{ NULL, 0, 0 }, /* Hf */ |
| NULL, /* %U */ |
{ NULL, 0, 0 }, /* Fr */ |
| NULL, /* Ta */ |
{ NULL, 0, 0 }, /* Ud */ |
| |
{ NULL, TYPE_Lb, 0 }, /* Lb */ |
| |
{ NULL, 0, 0 }, /* Lp */ |
| |
{ NULL, TYPE_Lk, 0 }, /* Lk */ |
| |
{ NULL, TYPE_Mt, NODE_NOSRC }, /* Mt */ |
| |
{ NULL, 0, 0 }, /* Brq */ |
| |
{ NULL, 0, 0 }, /* Bro */ |
| |
{ NULL, 0, 0 }, /* Brc */ |
| |
{ NULL, 0, 0 }, /* %C */ |
| |
{ NULL, 0, 0 }, /* Es */ |
| |
{ NULL, 0, 0 }, /* En */ |
| |
{ NULL, TYPE_Dx, NODE_NOSRC }, /* Dx */ |
| |
{ NULL, 0, 0 }, /* %Q */ |
| |
{ NULL, 0, 0 }, /* br */ |
| |
{ NULL, 0, 0 }, /* sp */ |
| |
{ NULL, 0, 0 }, /* %U */ |
| |
{ NULL, 0, 0 }, /* Ta */ |
| |
{ NULL, 0, 0 }, /* ll */ |
| }; |
}; |
| |
|
| static const char *progname; |
|
| |
|
| int |
int |
| main(int argc, char *argv[]) |
mandocdb(int argc, char *argv[]) |
| { |
{ |
| struct mparse *mp; /* parse sequence */ |
struct manconf conf; |
| struct manpaths dirs; |
struct mparse *mp; |
| enum op op; /* current operation */ |
struct dba *dba; |
| const char *dir; |
const char *path_arg, *progname; |
| char ibuf[MAXPATHLEN], /* index fname */ |
size_t j, sz; |
| fbuf[MAXPATHLEN]; /* btree fname */ |
int ch, i; |
| int verb, /* output verbosity */ |
|
| ch, i, flags; |
|
| DB *idx, /* index database */ |
|
| *db, /* keyword database */ |
|
| *hash; /* temporary keyword hashtable */ |
|
| BTREEINFO info; /* btree configuration */ |
|
| recno_t maxrec; /* supremum of all records */ |
|
| recno_t *recs; /* buffer of empty records */ |
|
| size_t sz1, sz2, |
|
| recsz, /* buffer size of recs */ |
|
| reccur; /* valid number of recs */ |
|
| struct buf buf, /* keyword buffer */ |
|
| dbuf; /* description buffer */ |
|
| struct of *of; /* list of files for processing */ |
|
| extern int optind; |
|
| extern char *optarg; |
|
| |
|
| progname = strrchr(argv[0], '/'); |
#if HAVE_PLEDGE |
| if (progname == NULL) |
if (pledge("stdio rpath wpath cpath fattr flock proc exec", NULL) == -1) { |
| progname = argv[0]; |
warn("pledge"); |
| else |
return (int)MANDOCLEVEL_SYSERR; |
| ++progname; |
} |
| |
#endif |
| |
|
| memset(&dirs, 0, sizeof(struct manpaths)); |
#if HAVE_SANDBOX_INIT |
| |
if (sandbox_init(kSBXProfileNoInternet, SANDBOX_NAMED, NULL) == -1) { |
| |
warnx("sandbox_init"); |
| |
return (int)MANDOCLEVEL_SYSERR; |
| |
} |
| |
#endif |
| |
|
| verb = 0; |
memset(&conf, 0, sizeof(conf)); |
| of = NULL; |
|
| db = idx = NULL; |
|
| mp = NULL; |
|
| hash = NULL; |
|
| recs = NULL; |
|
| recsz = reccur = 0; |
|
| maxrec = 0; |
|
| op = OP_NEW; |
|
| dir = NULL; |
|
| |
|
| while (-1 != (ch = getopt(argc, argv, "d:u:v"))) |
/* |
| |
* We accept a few different invocations. |
| |
* The CHECKOP macro makes sure that invocation styles don't |
| |
* clobber each other. |
| |
*/ |
| |
#define CHECKOP(_op, _ch) do \ |
| |
if (OP_DEFAULT != (_op)) { \ |
| |
warnx("-%c: Conflicting option", (_ch)); \ |
| |
goto usage; \ |
| |
} while (/*CONSTCOND*/0) |
| |
|
| |
path_arg = NULL; |
| |
op = OP_DEFAULT; |
| |
|
| |
while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v"))) |
| switch (ch) { |
switch (ch) { |
| case ('d'): |
case 'a': |
| dir = optarg; |
use_all = 1; |
| |
break; |
| |
case 'C': |
| |
CHECKOP(op, ch); |
| |
path_arg = optarg; |
| |
op = OP_CONFFILE; |
| |
break; |
| |
case 'D': |
| |
debug++; |
| |
break; |
| |
case 'd': |
| |
CHECKOP(op, ch); |
| |
path_arg = optarg; |
| op = OP_UPDATE; |
op = OP_UPDATE; |
| break; |
break; |
| case ('u'): |
case 'n': |
| dir = optarg; |
nodb = 1; |
| |
break; |
| |
case 'p': |
| |
warnings = 1; |
| |
break; |
| |
case 'Q': |
| |
mparse_options |= MPARSE_QUICK; |
| |
break; |
| |
case 'T': |
| |
if (strcmp(optarg, "utf8")) { |
| |
warnx("-T%s: Unsupported output format", |
| |
optarg); |
| |
goto usage; |
| |
} |
| |
write_utf8 = 1; |
| |
break; |
| |
case 't': |
| |
CHECKOP(op, ch); |
| |
dup2(STDOUT_FILENO, STDERR_FILENO); |
| |
op = OP_TEST; |
| |
nodb = warnings = 1; |
| |
break; |
| |
case 'u': |
| |
CHECKOP(op, ch); |
| |
path_arg = optarg; |
| op = OP_DELETE; |
op = OP_DELETE; |
| break; |
break; |
| case ('v'): |
case 'v': |
| verb++; |
/* Compatibility with espie@'s makewhatis. */ |
| break; |
break; |
| default: |
default: |
| usage(); |
goto usage; |
| return((int)MANDOCLEVEL_BADARG); |
|
| } |
} |
| |
|
| argc -= optind; |
argc -= optind; |
| argv += optind; |
argv += optind; |
| |
|
| memset(&info, 0, sizeof(BTREEINFO)); |
#if HAVE_PLEDGE |
| info.flags = R_DUP; |
if (nodb) { |
| |
if (pledge("stdio rpath", NULL) == -1) { |
| |
warn("pledge"); |
| |
return (int)MANDOCLEVEL_SYSERR; |
| |
} |
| |
} |
| |
#endif |
| |
|
| mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL); |
if (OP_CONFFILE == op && argc > 0) { |
| |
warnx("-C: Too many arguments"); |
| |
goto usage; |
| |
} |
| |
|
| memset(&buf, 0, sizeof(struct buf)); |
exitcode = (int)MANDOCLEVEL_OK; |
| memset(&dbuf, 0, sizeof(struct buf)); |
mchars_alloc(); |
| |
mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL, NULL); |
| |
mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev)); |
| |
mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file)); |
| |
|
| buf.size = dbuf.size = MANDOC_BUFSZ; |
if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) { |
| |
|
| buf.cp = mandoc_malloc(buf.size); |
/* |
| dbuf.cp = mandoc_malloc(dbuf.size); |
* Most of these deal with a specific directory. |
| |
* Jump into that directory first. |
| |
*/ |
| |
if (OP_TEST != op && 0 == set_basedir(path_arg, 1)) |
| |
goto out; |
| |
|
| flags = OP_NEW == op ? O_CREAT|O_TRUNC|O_RDWR : O_CREAT|O_RDWR; |
dba = nodb ? dba_new(128) : dba_read(MANDOC_DB); |
| |
if (dba != NULL) { |
| |
/* |
| |
* The existing database is usable. Process |
| |
* all files specified on the command-line. |
| |
*/ |
| |
#if HAVE_PLEDGE |
| |
if (!nodb) { |
| |
if (pledge("stdio rpath wpath cpath fattr flock", NULL) == -1) { |
| |
warn("pledge"); |
| |
exitcode = (int)MANDOCLEVEL_SYSERR; |
| |
goto out; |
| |
} |
| |
} |
| |
#endif |
| |
use_all = 1; |
| |
for (i = 0; i < argc; i++) |
| |
filescan(argv[i]); |
| |
if (nodb == 0) |
| |
dbprune(dba); |
| |
} else { |
| |
/* Database missing or corrupt. */ |
| |
if (op != OP_UPDATE || errno != ENOENT) |
| |
say(MANDOC_DB, "%s: Automatically recreating" |
| |
" from scratch", strerror(errno)); |
| |
exitcode = (int)MANDOCLEVEL_OK; |
| |
op = OP_DEFAULT; |
| |
if (0 == treescan()) |
| |
goto out; |
| |
dba = dba_new(128); |
| |
} |
| |
if (OP_DELETE != op) |
| |
mpages_merge(dba, mp); |
| |
if (nodb == 0) |
| |
dbwrite(dba); |
| |
dba_free(dba); |
| |
} else { |
| |
/* |
| |
* If we have arguments, use them as our manpaths. |
| |
* If we don't, use man.conf(5). |
| |
*/ |
| |
if (argc > 0) { |
| |
conf.manpath.paths = mandoc_reallocarray(NULL, |
| |
argc, sizeof(char *)); |
| |
conf.manpath.sz = (size_t)argc; |
| |
for (i = 0; i < argc; i++) |
| |
conf.manpath.paths[i] = mandoc_strdup(argv[i]); |
| |
} else |
| |
manconf_parse(&conf, path_arg, NULL, NULL); |
| |
|
| if (OP_UPDATE == op || OP_DELETE == op) { |
if (conf.manpath.sz == 0) { |
| ibuf[0] = fbuf[0] = '\0'; |
exitcode = (int)MANDOCLEVEL_BADARG; |
| |
say("", "Empty manpath"); |
| |
} |
| |
|
| strlcat(fbuf, dir, MAXPATHLEN); |
/* |
| strlcat(fbuf, "/", MAXPATHLEN); |
* First scan the tree rooted at a base directory, then |
| sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN); |
* build a new database and finally move it into place. |
| |
* Ignore zero-length directories and strip trailing |
| |
* slashes. |
| |
*/ |
| |
for (j = 0; j < conf.manpath.sz; j++) { |
| |
sz = strlen(conf.manpath.paths[j]); |
| |
if (sz && conf.manpath.paths[j][sz - 1] == '/') |
| |
conf.manpath.paths[j][--sz] = '\0'; |
| |
if (0 == sz) |
| |
continue; |
| |
|
| strlcat(ibuf, dir, MAXPATHLEN); |
if (j) { |
| strlcat(ibuf, "/", MAXPATHLEN); |
mandoc_ohash_init(&mpages, 6, |
| sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN); |
offsetof(struct mpage, inodev)); |
| |
mandoc_ohash_init(&mlinks, 6, |
| |
offsetof(struct mlink, file)); |
| |
} |
| |
|
| if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) { |
if ( ! set_basedir(conf.manpath.paths[j], argc > 0)) |
| fprintf(stderr, "%s: Path too long\n", dir); |
continue; |
| exit((int)MANDOCLEVEL_BADARG); |
if (0 == treescan()) |
| |
continue; |
| |
dba = dba_new(128); |
| |
mpages_merge(dba, mp); |
| |
if (nodb == 0) |
| |
dbwrite(dba); |
| |
dba_free(dba); |
| |
|
| |
if (j + 1 < conf.manpath.sz) { |
| |
mpages_free(); |
| |
ohash_delete(&mpages); |
| |
ohash_delete(&mlinks); |
| |
} |
| } |
} |
| |
} |
| |
out: |
| |
manconf_free(&conf); |
| |
mparse_free(mp); |
| |
mchars_free(); |
| |
mpages_free(); |
| |
ohash_delete(&mpages); |
| |
ohash_delete(&mlinks); |
| |
return exitcode; |
| |
usage: |
| |
progname = getprogname(); |
| |
fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n" |
| |
" %s [-aDnpQ] [-Tutf8] dir ...\n" |
| |
" %s [-DnpQ] [-Tutf8] -d dir [file ...]\n" |
| |
" %s [-Dnp] -u dir [file ...]\n" |
| |
" %s [-Q] -t file ...\n", |
| |
progname, progname, progname, progname, progname); |
| |
|
| db = dbopen(fbuf, flags, 0644, DB_BTREE, &info); |
return (int)MANDOCLEVEL_BADARG; |
| idx = dbopen(ibuf, flags, 0644, DB_RECNO, NULL); |
} |
| |
|
| if (NULL == db) { |
/* |
| perror(fbuf); |
* To get a singly linked list in alpha order while inserting entries |
| exit((int)MANDOCLEVEL_SYSERR); |
* at the beginning, process directory entries in reverse alpha order. |
| } else if (NULL == db) { |
*/ |
| perror(ibuf); |
static int |
| exit((int)MANDOCLEVEL_SYSERR); |
#if HAVE_FTS_COMPARE_CONST |
| |
fts_compare(const FTSENT *const *a, const FTSENT *const *b) |
| |
#else |
| |
fts_compare(const FTSENT **a, const FTSENT **b) |
| |
#endif |
| |
{ |
| |
return -strcmp((*a)->fts_name, (*b)->fts_name); |
| |
} |
| |
|
| |
/* |
| |
* Scan a directory tree rooted at "basedir" for manpages. |
| |
* We use fts(), scanning directory parts along the way for clues to our |
| |
* section and architecture. |
| |
* |
| |
* If use_all has been specified, grok all files. |
| |
* If not, sanitise paths to the following: |
| |
* |
| |
* [./]man*[/<arch>]/<name>.<section> |
| |
* or |
| |
* [./]cat<section>[/<arch>]/<name>.0 |
| |
* |
| |
* TODO: accommodate for multi-language directories. |
| |
*/ |
| |
static int |
| |
treescan(void) |
| |
{ |
| |
char buf[PATH_MAX]; |
| |
FTS *f; |
| |
FTSENT *ff; |
| |
struct mlink *mlink; |
| |
int gzip; |
| |
enum form dform; |
| |
char *dsec, *arch, *fsec, *cp; |
| |
const char *path; |
| |
const char *argv[2]; |
| |
|
| |
argv[0] = "."; |
| |
argv[1] = (char *)NULL; |
| |
|
| |
f = fts_open((char * const *)argv, FTS_PHYSICAL | FTS_NOCHDIR, |
| |
fts_compare); |
| |
if (f == NULL) { |
| |
exitcode = (int)MANDOCLEVEL_SYSERR; |
| |
say("", "&fts_open"); |
| |
return 0; |
| |
} |
| |
|
| |
dsec = arch = NULL; |
| |
dform = FORM_NONE; |
| |
|
| |
while ((ff = fts_read(f)) != NULL) { |
| |
path = ff->fts_path + 2; |
| |
switch (ff->fts_info) { |
| |
|
| |
/* |
| |
* Symbolic links require various sanity checks, |
| |
* then get handled just like regular files. |
| |
*/ |
| |
case FTS_SL: |
| |
if (realpath(path, buf) == NULL) { |
| |
if (warnings) |
| |
say(path, "&realpath"); |
| |
continue; |
| |
} |
| |
if (strstr(buf, basedir) != buf |
| |
#ifdef HOMEBREWDIR |
| |
&& strstr(buf, HOMEBREWDIR) != buf |
| |
#endif |
| |
) { |
| |
if (warnings) say("", |
| |
"%s: outside base directory", buf); |
| |
continue; |
| |
} |
| |
/* Use logical inode to avoid mpages dupe. */ |
| |
if (stat(path, ff->fts_statp) == -1) { |
| |
if (warnings) |
| |
say(path, "&stat"); |
| |
continue; |
| |
} |
| |
/* FALLTHROUGH */ |
| |
|
| |
/* |
| |
* If we're a regular file, add an mlink by using the |
| |
* stored directory data and handling the filename. |
| |
*/ |
| |
case FTS_F: |
| |
if ( ! strcmp(path, MANDOC_DB)) |
| |
continue; |
| |
if ( ! use_all && ff->fts_level < 2) { |
| |
if (warnings) |
| |
say(path, "Extraneous file"); |
| |
continue; |
| |
} |
| |
gzip = 0; |
| |
fsec = NULL; |
| |
while (fsec == NULL) { |
| |
fsec = strrchr(ff->fts_name, '.'); |
| |
if (fsec == NULL || strcmp(fsec+1, "gz")) |
| |
break; |
| |
gzip = 1; |
| |
*fsec = '\0'; |
| |
fsec = NULL; |
| |
} |
| |
if (fsec == NULL) { |
| |
if ( ! use_all) { |
| |
if (warnings) |
| |
say(path, |
| |
"No filename suffix"); |
| |
continue; |
| |
} |
| |
} else if ( ! strcmp(++fsec, "html")) { |
| |
if (warnings) |
| |
say(path, "Skip html"); |
| |
continue; |
| |
} else if ( ! strcmp(fsec, "ps")) { |
| |
if (warnings) |
| |
say(path, "Skip ps"); |
| |
continue; |
| |
} else if ( ! strcmp(fsec, "pdf")) { |
| |
if (warnings) |
| |
say(path, "Skip pdf"); |
| |
continue; |
| |
} else if ( ! use_all && |
| |
((dform == FORM_SRC && |
| |
strncmp(fsec, dsec, strlen(dsec))) || |
| |
(dform == FORM_CAT && strcmp(fsec, "0")))) { |
| |
if (warnings) |
| |
say(path, "Wrong filename suffix"); |
| |
continue; |
| |
} else |
| |
fsec[-1] = '\0'; |
| |
|
| |
mlink = mandoc_calloc(1, sizeof(struct mlink)); |
| |
if (strlcpy(mlink->file, path, |
| |
sizeof(mlink->file)) >= |
| |
sizeof(mlink->file)) { |
| |
say(path, "Filename too long"); |
| |
free(mlink); |
| |
continue; |
| |
} |
| |
mlink->dform = dform; |
| |
mlink->dsec = dsec; |
| |
mlink->arch = arch; |
| |
mlink->name = ff->fts_name; |
| |
mlink->fsec = fsec; |
| |
mlink->gzip = gzip; |
| |
mlink_add(mlink, ff->fts_statp); |
| |
continue; |
| |
|
| |
case FTS_D: |
| |
case FTS_DP: |
| |
break; |
| |
|
| |
default: |
| |
if (warnings) |
| |
say(path, "Not a regular file"); |
| |
continue; |
| } |
} |
| |
|
| if (verb > 2) { |
switch (ff->fts_level) { |
| printf("%s: Opened\n", fbuf); |
case 0: |
| printf("%s: Opened\n", ibuf); |
/* Ignore the root directory. */ |
| |
break; |
| |
case 1: |
| |
/* |
| |
* This might contain manX/ or catX/. |
| |
* Try to infer this from the name. |
| |
* If we're not in use_all, enforce it. |
| |
*/ |
| |
cp = ff->fts_name; |
| |
if (ff->fts_info == FTS_DP) { |
| |
dform = FORM_NONE; |
| |
dsec = NULL; |
| |
break; |
| |
} |
| |
|
| |
if ( ! strncmp(cp, "man", 3)) { |
| |
dform = FORM_SRC; |
| |
dsec = cp + 3; |
| |
} else if ( ! strncmp(cp, "cat", 3)) { |
| |
dform = FORM_CAT; |
| |
dsec = cp + 3; |
| |
} else { |
| |
dform = FORM_NONE; |
| |
dsec = NULL; |
| |
} |
| |
|
| |
if (dsec != NULL || use_all) |
| |
break; |
| |
|
| |
if (warnings) |
| |
say(path, "Unknown directory part"); |
| |
fts_set(f, ff, FTS_SKIP); |
| |
break; |
| |
case 2: |
| |
/* |
| |
* Possibly our architecture. |
| |
* If we're descending, keep tabs on it. |
| |
*/ |
| |
if (ff->fts_info != FTS_DP && dsec != NULL) |
| |
arch = ff->fts_name; |
| |
else |
| |
arch = NULL; |
| |
break; |
| |
default: |
| |
if (ff->fts_info == FTS_DP || use_all) |
| |
break; |
| |
if (warnings) |
| |
say(path, "Extraneous directory part"); |
| |
fts_set(f, ff, FTS_SKIP); |
| |
break; |
| } |
} |
| |
} |
| |
|
| ofile_argbuild(argv, argc, verb, &of); |
fts_close(f); |
| if (NULL == of) |
return 1; |
| goto out; |
} |
| |
|
| of = of->first; |
/* |
| |
* Add a file to the mlinks table. |
| |
* Do not verify that it's a "valid" looking manpage (we'll do that |
| |
* later). |
| |
* |
| |
* Try to infer the manual section, architecture, and page name from the |
| |
* path, assuming it looks like |
| |
* |
| |
* [./]man*[/<arch>]/<name>.<section> |
| |
* or |
| |
* [./]cat<section>[/<arch>]/<name>.0 |
| |
* |
| |
* See treescan() for the fts(3) version of this. |
| |
*/ |
| |
static void |
| |
filescan(const char *file) |
| |
{ |
| |
char buf[PATH_MAX]; |
| |
struct stat st; |
| |
struct mlink *mlink; |
| |
char *p, *start; |
| |
|
| index_prune(of, db, fbuf, idx, ibuf, verb, |
assert(use_all); |
| &maxrec, &recs, &recsz); |
|
| |
|
| if (OP_UPDATE == op) |
if (0 == strncmp(file, "./", 2)) |
| index_merge(of, mp, &dbuf, &buf, hash, |
file += 2; |
| db, fbuf, idx, ibuf, verb, |
|
| maxrec, recs, reccur); |
|
| |
|
| goto out; |
/* |
| |
* We have to do lstat(2) before realpath(3) loses |
| |
* the information whether this is a symbolic link. |
| |
* We need to know that because for symbolic links, |
| |
* we want to use the orginal file name, while for |
| |
* regular files, we want to use the real path. |
| |
*/ |
| |
if (-1 == lstat(file, &st)) { |
| |
exitcode = (int)MANDOCLEVEL_BADARG; |
| |
say(file, "&lstat"); |
| |
return; |
| |
} else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) { |
| |
exitcode = (int)MANDOCLEVEL_BADARG; |
| |
say(file, "Not a regular file"); |
| |
return; |
| } |
} |
| |
|
| /* |
/* |
| * Configure the directories we're going to scan. |
* We have to resolve the file name to the real path |
| * If we have command-line arguments, use them. |
* in any case for the base directory check. |
| * If not, we use man(1)'s method (see mandocdb.8). |
|
| */ |
*/ |
| |
if (NULL == realpath(file, buf)) { |
| |
exitcode = (int)MANDOCLEVEL_BADARG; |
| |
say(file, "&realpath"); |
| |
return; |
| |
} |
| |
|
| if (argc > 0) { |
if (OP_TEST == op) |
| dirs.paths = mandoc_malloc(argc * sizeof(char *)); |
start = buf; |
| dirs.sz = argc; |
else if (strstr(buf, basedir) == buf) |
| for (i = 0; i < argc; i++) |
start = buf + strlen(basedir); |
| dirs.paths[i] = mandoc_strdup(argv[i]); |
#ifdef HOMEBREWDIR |
| } else |
else if (strstr(buf, HOMEBREWDIR) == buf) |
| manpath_parse(&dirs, NULL, NULL); |
start = buf; |
| |
#endif |
| |
else { |
| |
exitcode = (int)MANDOCLEVEL_BADARG; |
| |
say("", "%s: outside base directory", buf); |
| |
return; |
| |
} |
| |
|
| for (i = 0; i < dirs.sz; i++) { |
/* |
| ibuf[0] = fbuf[0] = '\0'; |
* Now we are sure the file is inside our tree. |
| |
* If it is a symbolic link, ignore the real path |
| |
* and use the original name. |
| |
* This implies passing stuff like "cat1/../man1/foo.1" |
| |
* on the command line won't work. So don't do that. |
| |
* Note the stat(2) can still fail if the link target |
| |
* doesn't exist. |
| |
*/ |
| |
if (S_IFLNK & st.st_mode) { |
| |
if (-1 == stat(buf, &st)) { |
| |
exitcode = (int)MANDOCLEVEL_BADARG; |
| |
say(file, "&stat"); |
| |
return; |
| |
} |
| |
if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) { |
| |
say(file, "Filename too long"); |
| |
return; |
| |
} |
| |
start = buf; |
| |
if (OP_TEST != op && strstr(buf, basedir) == buf) |
| |
start += strlen(basedir); |
| |
} |
| |
|
| strlcat(fbuf, dirs.paths[i], MAXPATHLEN); |
mlink = mandoc_calloc(1, sizeof(struct mlink)); |
| strlcat(fbuf, "/", MAXPATHLEN); |
mlink->dform = FORM_NONE; |
| sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN); |
if (strlcpy(mlink->file, start, sizeof(mlink->file)) >= |
| |
sizeof(mlink->file)) { |
| |
say(start, "Filename too long"); |
| |
free(mlink); |
| |
return; |
| |
} |
| |
|
| strlcat(ibuf, dirs.paths[i], MAXPATHLEN); |
/* |
| strlcat(ibuf, "/", MAXPATHLEN); |
* First try to guess our directory structure. |
| sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN); |
* If we find a separator, try to look for man* or cat*. |
| |
* If we find one of these and what's underneath is a directory, |
| |
* assume it's an architecture. |
| |
*/ |
| |
if (NULL != (p = strchr(start, '/'))) { |
| |
*p++ = '\0'; |
| |
if (0 == strncmp(start, "man", 3)) { |
| |
mlink->dform = FORM_SRC; |
| |
mlink->dsec = start + 3; |
| |
} else if (0 == strncmp(start, "cat", 3)) { |
| |
mlink->dform = FORM_CAT; |
| |
mlink->dsec = start + 3; |
| |
} |
| |
|
| if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) { |
start = p; |
| fprintf(stderr, "%s: Path too long\n", |
if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) { |
| dirs.paths[i]); |
*p++ = '\0'; |
| exit((int)MANDOCLEVEL_BADARG); |
mlink->arch = start; |
| |
start = p; |
| } |
} |
| |
} |
| |
|
| db = dbopen(fbuf, flags, 0644, DB_BTREE, &info); |
/* |
| idx = dbopen(ibuf, flags, 0644, DB_RECNO, NULL); |
* Now check the file suffix. |
| |
* Suffix of `.0' indicates a catpage, `.1-9' is a manpage. |
| |
*/ |
| |
p = strrchr(start, '\0'); |
| |
while (p-- > start && '/' != *p && '.' != *p) |
| |
/* Loop. */ ; |
| |
|
| if (NULL == db) { |
if ('.' == *p) { |
| perror(fbuf); |
*p++ = '\0'; |
| exit((int)MANDOCLEVEL_SYSERR); |
mlink->fsec = p; |
| } else if (NULL == db) { |
} |
| perror(ibuf); |
|
| exit((int)MANDOCLEVEL_SYSERR); |
|
| } |
|
| |
|
| if (verb > 2) { |
/* |
| printf("%s: Truncated\n", fbuf); |
* Now try to parse the name. |
| printf("%s: Truncated\n", ibuf); |
* Use the filename portion of the path. |
| } |
*/ |
| |
mlink->name = start; |
| |
if (NULL != (p = strrchr(start, '/'))) { |
| |
mlink->name = p + 1; |
| |
*p = '\0'; |
| |
} |
| |
mlink_add(mlink, &st); |
| |
} |
| |
|
| ofile_free(of); |
static void |
| of = NULL; |
mlink_add(struct mlink *mlink, const struct stat *st) |
| |
{ |
| |
struct inodev inodev; |
| |
struct mpage *mpage; |
| |
unsigned int slot; |
| |
|
| if ( ! ofile_dirbuild(dirs.paths[i], verb, &of)) |
assert(NULL != mlink->file); |
| exit((int)MANDOCLEVEL_SYSERR); |
|
| |
|
| if (NULL == of) |
mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : ""); |
| continue; |
mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : ""); |
| |
mlink->name = mandoc_strdup(mlink->name ? mlink->name : ""); |
| |
mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : ""); |
| |
|
| of = of->first; |
if ('0' == *mlink->fsec) { |
| |
free(mlink->fsec); |
| |
mlink->fsec = mandoc_strdup(mlink->dsec); |
| |
mlink->fform = FORM_CAT; |
| |
} else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec) |
| |
mlink->fform = FORM_SRC; |
| |
else |
| |
mlink->fform = FORM_NONE; |
| |
|
| index_merge(of, mp, &dbuf, &buf, hash, db, fbuf, |
slot = ohash_qlookup(&mlinks, mlink->file); |
| idx, ibuf, verb, maxrec, recs, reccur); |
assert(NULL == ohash_find(&mlinks, slot)); |
| } |
ohash_insert(&mlinks, slot, mlink); |
| |
|
| out: |
memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */ |
| if (db) |
inodev.st_ino = st->st_ino; |
| (*db->close)(db); |
inodev.st_dev = st->st_dev; |
| if (idx) |
slot = ohash_lookup_memory(&mpages, (char *)&inodev, |
| (*idx->close)(idx); |
sizeof(struct inodev), inodev.st_ino); |
| if (hash) |
mpage = ohash_find(&mpages, slot); |
| (*hash->close)(hash); |
if (NULL == mpage) { |
| if (mp) |
mpage = mandoc_calloc(1, sizeof(struct mpage)); |
| mparse_free(mp); |
mpage->inodev.st_ino = inodev.st_ino; |
| |
mpage->inodev.st_dev = inodev.st_dev; |
| |
mpage->form = FORM_NONE; |
| |
mpage->next = mpage_head; |
| |
mpage_head = mpage; |
| |
ohash_insert(&mpages, slot, mpage); |
| |
} else |
| |
mlink->next = mpage->mlinks; |
| |
mpage->mlinks = mlink; |
| |
mlink->mpage = mpage; |
| |
} |
| |
|
| manpath_free(&dirs); |
static void |
| ofile_free(of); |
mlink_free(struct mlink *mlink) |
| free(buf.cp); |
{ |
| free(dbuf.cp); |
|
| free(recs); |
|
| |
|
| return(MANDOCLEVEL_OK); |
free(mlink->dsec); |
| |
free(mlink->arch); |
| |
free(mlink->name); |
| |
free(mlink->fsec); |
| |
free(mlink); |
| } |
} |
| |
|
| void |
static void |
| index_merge(const struct of *of, struct mparse *mp, |
mpages_free(void) |
| struct buf *dbuf, struct buf *buf, |
|
| DB *hash, DB *db, const char *dbf, |
|
| DB *idx, const char *idxf, int verb, |
|
| recno_t maxrec, const recno_t *recs, size_t reccur) |
|
| { |
{ |
| recno_t rec; |
struct mpage *mpage; |
| int ch; |
struct mlink *mlink; |
| DBT key, val; |
|
| struct mdoc *mdoc; |
|
| struct man *man; |
|
| const char *fn, *msec, *mtitle, *arch; |
|
| size_t sv; |
|
| unsigned seq; |
|
| struct db_val vbuf; |
|
| |
|
| for (rec = 0; of; of = of->next) { |
while ((mpage = mpage_head) != NULL) { |
| fn = of->fname; |
while ((mlink = mpage->mlinks) != NULL) { |
| if (reccur > 0) { |
mpage->mlinks = mlink->next; |
| --reccur; |
mlink_free(mlink); |
| rec = recs[(int)reccur]; |
} |
| } else if (maxrec > 0) { |
mpage_head = mpage->next; |
| rec = maxrec; |
free(mpage->sec); |
| maxrec = 0; |
free(mpage->arch); |
| } else |
free(mpage->title); |
| rec++; |
free(mpage->desc); |
| |
free(mpage); |
| |
} |
| |
} |
| |
|
| mparse_reset(mp); |
/* |
| hash_reset(&hash); |
* For each mlink to the mpage, check whether the path looks like |
| |
* it is formatted, and if it does, check whether a source manual |
| |
* exists by the same name, ignoring the suffix. |
| |
* If both conditions hold, drop the mlink. |
| |
*/ |
| |
static void |
| |
mlinks_undupe(struct mpage *mpage) |
| |
{ |
| |
char buf[PATH_MAX]; |
| |
struct mlink **prev; |
| |
struct mlink *mlink; |
| |
char *bufp; |
| |
|
| if (mparse_readfd(mp, -1, fn) >= MANDOCLEVEL_FATAL) { |
mpage->form = FORM_CAT; |
| fprintf(stderr, "%s: Parse failure\n", fn); |
prev = &mpage->mlinks; |
| continue; |
while (NULL != (mlink = *prev)) { |
| |
if (FORM_CAT != mlink->dform) { |
| |
mpage->form = FORM_NONE; |
| |
goto nextlink; |
| } |
} |
| |
(void)strlcpy(buf, mlink->file, sizeof(buf)); |
| |
bufp = strstr(buf, "cat"); |
| |
assert(NULL != bufp); |
| |
memcpy(bufp, "man", 3); |
| |
if (NULL != (bufp = strrchr(buf, '.'))) |
| |
*++bufp = '\0'; |
| |
(void)strlcat(buf, mlink->dsec, sizeof(buf)); |
| |
if (NULL == ohash_find(&mlinks, |
| |
ohash_qlookup(&mlinks, buf))) |
| |
goto nextlink; |
| |
if (warnings) |
| |
say(mlink->file, "Man source exists: %s", buf); |
| |
if (use_all) |
| |
goto nextlink; |
| |
*prev = mlink->next; |
| |
mlink_free(mlink); |
| |
continue; |
| |
nextlink: |
| |
prev = &(*prev)->next; |
| |
} |
| |
} |
| |
|
| mparse_result(mp, &mdoc, &man); |
static void |
| if (NULL == mdoc && NULL == man) |
mlink_check(struct mpage *mpage, struct mlink *mlink) |
| continue; |
{ |
| |
struct str *str; |
| |
unsigned int slot; |
| |
|
| msec = NULL != mdoc ? |
/* |
| mdoc_meta(mdoc)->msec : man_meta(man)->msec; |
* Check whether the manual section given in a file |
| mtitle = NULL != mdoc ? |
* agrees with the directory where the file is located. |
| mdoc_meta(mdoc)->title : man_meta(man)->title; |
* Some manuals have suffixes like (3p) on their |
| arch = NULL != mdoc ? |
* section number either inside the file or in the |
| mdoc_meta(mdoc)->arch : NULL; |
* directory name, some are linked into more than one |
| |
* section, like encrypt(1) = makekey(8). |
| |
*/ |
| |
|
| if (NULL == arch) |
if (FORM_SRC == mpage->form && |
| arch = ""; |
strcasecmp(mpage->sec, mlink->dsec)) |
| |
say(mlink->file, "Section \"%s\" manual in %s directory", |
| |
mpage->sec, mlink->dsec); |
| |
|
| /* |
/* |
| * The index record value consists of a nil-terminated |
* Manual page directories exist for each kernel |
| * filename, a nil-terminated manual section, and a |
* architecture as returned by machine(1). |
| * nil-terminated description. Since the description |
* However, many manuals only depend on the |
| * may not be set, we set a sentinel to see if we're |
* application architecture as returned by arch(1). |
| * going to write a nil byte in its place. |
* 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, and sparc64. |
| |
*/ |
| |
|
| dbuf->len = 0; |
if (strcasecmp(mpage->arch, mlink->arch)) |
| buf_appendb(dbuf, fn, strlen(fn) + 1); |
say(mlink->file, "Architecture \"%s\" manual in " |
| buf_appendb(dbuf, msec, strlen(msec) + 1); |
"\"%s\" directory", mpage->arch, mlink->arch); |
| buf_appendb(dbuf, mtitle, strlen(mtitle) + 1); |
|
| buf_appendb(dbuf, arch, strlen(arch) + 1); |
|
| |
|
| sv = dbuf->len; |
/* |
| |
* XXX |
| |
* parse_cat() doesn't set NAME_TITLE yet. |
| |
*/ |
| |
|
| /* Fix the record number in the btree value. */ |
if (FORM_CAT == mpage->form) |
| |
return; |
| |
|
| if (mdoc) |
/* |
| pmdoc_node(hash, buf, dbuf, |
* Check whether this mlink |
| mdoc_node(mdoc), mdoc_meta(mdoc)); |
* appears as a name in the NAME section. |
| else |
*/ |
| pman_node(hash, buf, dbuf, man_node(man)); |
|
| |
|
| /* |
slot = ohash_qlookup(&names, mlink->name); |
| * Copy from the in-memory hashtable of pending keywords |
str = ohash_find(&names, slot); |
| * into the database. |
assert(NULL != str); |
| */ |
if ( ! (NAME_TITLE & str->mask)) |
| |
say(mlink->file, "Name missing in NAME section"); |
| |
} |
| |
|
| vbuf.rec = rec; |
/* |
| seq = R_FIRST; |
* Run through the files in the global vector "mpages" |
| while (0 == (ch = (*hash->seq)(hash, &key, &val, seq))) { |
* and add them to the database specified in "basedir". |
| seq = R_NEXT; |
* |
| |
* This handles the parsing scheme itself, using the cues of directory |
| |
* and filename to determine whether the file is parsable or not. |
| |
*/ |
| |
static void |
| |
mpages_merge(struct dba *dba, struct mparse *mp) |
| |
{ |
| |
struct mpage *mpage, *mpage_dest; |
| |
struct mlink *mlink, *mlink_dest; |
| |
struct roff_man *man; |
| |
char *sodest; |
| |
char *cp; |
| |
int fd; |
| |
|
| vbuf.mask = *(uint64_t *)val.data; |
for (mpage = mpage_head; mpage != NULL; mpage = mpage->next) { |
| val.size = sizeof(struct db_val); |
mlinks_undupe(mpage); |
| val.data = &vbuf; |
if ((mlink = mpage->mlinks) == NULL) |
| |
continue; |
| |
|
| if (verb > 1) |
name_mask = NAME_MASK; |
| printf("%s: Added keyword: %s\n", |
mandoc_ohash_init(&names, 4, offsetof(struct str, key)); |
| fn, (char *)key.data); |
mandoc_ohash_init(&strings, 6, offsetof(struct str, key)); |
| dbt_put(db, dbf, &key, &val); |
mparse_reset(mp); |
| |
man = NULL; |
| |
sodest = NULL; |
| |
|
| |
if ((fd = mparse_open(mp, mlink->file)) == -1) { |
| |
say(mlink->file, "&open"); |
| |
goto nextpage; |
| } |
} |
| if (ch < 0) { |
|
| perror("hash"); |
|
| exit((int)MANDOCLEVEL_SYSERR); |
|
| } |
|
| |
|
| /* |
/* |
| * Apply to the index. If we haven't had a description |
* Interpret the file as mdoc(7) or man(7) source |
| * set, put an empty one in now. |
* code, unless it is known to be formatted. |
| */ |
*/ |
| |
if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) { |
| |
mparse_readfd(mp, fd, mlink->file); |
| |
close(fd); |
| |
mparse_result(mp, &man, &sodest); |
| |
} |
| |
|
| if (dbuf->len == sv) |
if (sodest != NULL) { |
| buf_appendb(dbuf, "", 1); |
mlink_dest = ohash_find(&mlinks, |
| |
ohash_qlookup(&mlinks, sodest)); |
| |
if (mlink_dest == NULL) { |
| |
mandoc_asprintf(&cp, "%s.gz", sodest); |
| |
mlink_dest = ohash_find(&mlinks, |
| |
ohash_qlookup(&mlinks, cp)); |
| |
free(cp); |
| |
} |
| |
if (mlink_dest != NULL) { |
| |
|
| key.data = &rec; |
/* The .so target exists. */ |
| key.size = sizeof(recno_t); |
|
| |
|
| val.data = dbuf->cp; |
mpage_dest = mlink_dest->mpage; |
| val.size = dbuf->len; |
while (1) { |
| |
mlink->mpage = mpage_dest; |
| |
|
| if (verb) |
/* |
| printf("%s: Added index\n", fn); |
* If the target was already |
| dbt_put(idx, idxf, &key, &val); |
* processed, add the links |
| |
* to the database now. |
| |
* Otherwise, this will |
| |
* happen when we come |
| |
* to the target. |
| |
*/ |
| |
|
| |
if (mpage_dest->dba != NULL) |
| |
dbadd_mlink(mlink); |
| |
|
| |
if (mlink->next == NULL) |
| |
break; |
| |
mlink = mlink->next; |
| |
} |
| |
|
| |
/* Move all links to the target. */ |
| |
|
| |
mlink->next = mlink_dest->next; |
| |
mlink_dest->next = mpage->mlinks; |
| |
mpage->mlinks = NULL; |
| |
} |
| |
goto nextpage; |
| |
} else if (man != NULL && man->macroset == MACROSET_MDOC) { |
| |
mdoc_validate(man); |
| |
mpage->form = FORM_SRC; |
| |
mpage->sec = man->meta.msec; |
| |
mpage->sec = mandoc_strdup( |
| |
mpage->sec == NULL ? "" : mpage->sec); |
| |
mpage->arch = man->meta.arch; |
| |
mpage->arch = mandoc_strdup( |
| |
mpage->arch == NULL ? "" : mpage->arch); |
| |
mpage->title = mandoc_strdup(man->meta.title); |
| |
} else if (man != NULL && man->macroset == MACROSET_MAN) { |
| |
man_validate(man); |
| |
mpage->form = FORM_SRC; |
| |
mpage->sec = mandoc_strdup(man->meta.msec); |
| |
mpage->arch = mandoc_strdup(mlink->arch); |
| |
mpage->title = mandoc_strdup(man->meta.title); |
| |
} else { |
| |
mpage->form = FORM_CAT; |
| |
mpage->sec = mandoc_strdup(mlink->dsec); |
| |
mpage->arch = mandoc_strdup(mlink->arch); |
| |
mpage->title = mandoc_strdup(mlink->name); |
| |
} |
| |
|
| |
assert(mpage->desc == NULL); |
| |
if (man != NULL && man->macroset == MACROSET_MDOC) |
| |
parse_mdoc(mpage, &man->meta, man->first); |
| |
else if (man != NULL) |
| |
parse_man(mpage, &man->meta, man->first); |
| |
else |
| |
parse_cat(mpage, fd); |
| |
if (mpage->desc == NULL) |
| |
mpage->desc = mandoc_strdup(mpage->mlinks->name); |
| |
|
| |
if (warnings && !use_all) |
| |
for (mlink = mpage->mlinks; mlink; |
| |
mlink = mlink->next) |
| |
mlink_check(mpage, mlink); |
| |
|
| |
dbadd(dba, mpage); |
| |
mlink = mpage->mlinks; |
| |
|
| |
nextpage: |
| |
ohash_delete(&strings); |
| |
ohash_delete(&names); |
| } |
} |
| } |
} |
| |
|
| /* |
|
| * Scan through all entries in the index file `idx' and prune those |
|
| * entries in `ofile'. |
|
| * Pruning consists of removing from `db', then invalidating the entry |
|
| * in `idx' (zeroing its value size). |
|
| */ |
|
| static void |
static void |
| index_prune(const struct of *ofile, DB *db, const char *dbf, |
parse_cat(struct mpage *mpage, int fd) |
| DB *idx, const char *idxf, int verb, |
|
| recno_t *maxrec, recno_t **recs, size_t *recsz) |
|
| { |
{ |
| const struct of *of; |
FILE *stream; |
| const char *fn; |
char *line, *p, *title; |
| struct db_val *vbuf; |
size_t linesz, plen, titlesz; |
| unsigned seq, sseq; |
ssize_t len; |
| DBT key, val; |
int offs; |
| size_t reccur; |
|
| int ch; |
|
| |
|
| reccur = 0; |
stream = (-1 == fd) ? |
| seq = R_FIRST; |
fopen(mpage->mlinks->file, "r") : |
| while (0 == (ch = (*idx->seq)(idx, &key, &val, seq))) { |
fdopen(fd, "r"); |
| seq = R_NEXT; |
if (NULL == stream) { |
| *maxrec = *(recno_t *)key.data; |
if (-1 != fd) |
| if (0 == val.size) { |
close(fd); |
| if (reccur >= *recsz) { |
if (warnings) |
| *recsz += MANDOC_SLOP; |
say(mpage->mlinks->file, "&fopen"); |
| *recs = mandoc_realloc(*recs, |
return; |
| *recsz * sizeof(recno_t)); |
} |
| } |
|
| (*recs)[(int)reccur] = *maxrec; |
|
| reccur++; |
|
| continue; |
|
| } |
|
| |
|
| fn = (char *)val.data; |
line = NULL; |
| for (of = ofile; of; of = of->next) |
linesz = 0; |
| if (0 == strcmp(fn, of->fname)) |
|
| break; |
|
| |
|
| if (NULL == of) |
/* Skip to first blank line. */ |
| |
|
| |
while (getline(&line, &linesz, stream) != -1) |
| |
if (*line == '\n') |
| |
break; |
| |
|
| |
/* |
| |
* Assume the first line that is not indented |
| |
* is the first section header. Skip to it. |
| |
*/ |
| |
|
| |
while (getline(&line, &linesz, stream) != -1) |
| |
if (*line != '\n' && *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 ((len = getline(&line, &linesz, stream)) != -1) { |
| |
if (*line != ' ') |
| |
break; |
| |
offs = 0; |
| |
while (isspace((unsigned char)line[offs])) |
| |
offs++; |
| |
if (line[offs] == '\0') |
| continue; |
continue; |
| |
title = mandoc_realloc(title, titlesz + len - offs); |
| |
memcpy(title + titlesz, line + offs, len - offs); |
| |
titlesz += len - offs; |
| |
title[titlesz - 1] = ' '; |
| |
} |
| |
free(line); |
| |
|
| sseq = R_FIRST; |
/* |
| while (0 == (ch = (*db->seq)(db, &key, &val, sseq))) { |
* If no page content can be found, or the input line |
| sseq = R_NEXT; |
* is already the next section header, or there is no |
| assert(sizeof(struct db_val) == val.size); |
* trailing newline, reuse the page title as the page |
| vbuf = val.data; |
* description. |
| if (*maxrec != vbuf->rec) |
*/ |
| continue; |
|
| if (verb) |
|
| printf("%s: Deleted keyword: %s\n", |
|
| fn, (char *)key.data); |
|
| ch = (*db->del)(db, &key, R_CURSOR); |
|
| if (ch < 0) |
|
| break; |
|
| } |
|
| if (ch < 0) { |
|
| perror(dbf); |
|
| exit((int)MANDOCLEVEL_SYSERR); |
|
| } |
|
| |
|
| if (verb) |
if (NULL == title || '\0' == *title) { |
| printf("%s: Deleted index\n", fn); |
if (warnings) |
| |
say(mpage->mlinks->file, |
| |
"Cannot find NAME section"); |
| |
fclose(stream); |
| |
free(title); |
| |
return; |
| |
} |
| |
|
| val.size = 0; |
title[titlesz - 1] = '\0'; |
| ch = (*idx->put)(idx, &key, &val, R_CURSOR); |
|
| if (ch < 0) { |
|
| perror(idxf); |
|
| exit((int)MANDOCLEVEL_SYSERR); |
|
| } |
|
| |
|
| if (reccur >= *recsz) { |
/* |
| *recsz += MANDOC_SLOP; |
* Skip to the first dash. |
| *recs = mandoc_realloc |
* Use the remaining line as the description (no more than 70 |
| (*recs, *recsz * sizeof(recno_t)); |
* bytes). |
| } |
*/ |
| |
|
| (*recs)[(int)reccur] = *maxrec; |
if (NULL != (p = strstr(title, "- "))) { |
| reccur++; |
for (p += 2; ' ' == *p || '\b' == *p; p++) |
| |
/* Skip to next word. */ ; |
| |
} else { |
| |
if (warnings) |
| |
say(mpage->mlinks->file, |
| |
"No dash in title line"); |
| |
p = title; |
| } |
} |
| (*maxrec)++; |
|
| } |
|
| |
|
| /* |
plen = strlen(p); |
| * Grow the buffer (if necessary) and copy in a binary string. |
|
| */ |
|
| static void |
|
| buf_appendb(struct buf *buf, const void *cp, size_t sz) |
|
| { |
|
| |
|
| /* Overshoot by MANDOC_BUFSZ. */ |
/* Strip backspace-encoding from line. */ |
| |
|
| while (buf->len + sz >= buf->size) { |
while (NULL != (line = memchr(p, '\b', plen))) { |
| buf->size = buf->len + sz + MANDOC_BUFSZ; |
len = line - p; |
| buf->cp = mandoc_realloc(buf->cp, buf->size); |
if (0 == len) { |
| |
memmove(line, line + 1, plen--); |
| |
continue; |
| |
} |
| |
memmove(line - 1, line + 1, plen - len); |
| |
plen -= 2; |
| } |
} |
| |
|
| memcpy(buf->cp + (int)buf->len, cp, sz); |
mpage->desc = mandoc_strdup(p); |
| buf->len += sz; |
fclose(stream); |
| |
free(title); |
| } |
} |
| |
|
| /* |
/* |
| * Append a nil-terminated string to the buffer. |
* Put a type/word pair into the word database for this particular file. |
| * This can be invoked multiple times. |
|
| * The buffer string will be nil-terminated. |
|
| * If invoked multiple times, a space is put between strings. |
|
| */ |
*/ |
| static void |
static void |
| buf_append(struct buf *buf, const char *cp) |
putkey(const struct mpage *mpage, char *value, uint64_t type) |
| { |
{ |
| size_t sz; |
putkeys(mpage, value, strlen(value), type); |
| |
|
| if (0 == (sz = strlen(cp))) |
|
| return; |
|
| |
|
| if (buf->len) |
|
| buf->cp[(int)buf->len - 1] = ' '; |
|
| |
|
| buf_appendb(buf, cp, sz + 1); |
|
| } |
} |
| |
|
| /* |
/* |
| * Recursively add all text from a given node. |
* Grok all nodes at or below a certain mdoc node into putkey(). |
| * This is optimised for general mdoc nodes in this context, which do |
|
| * not consist of subexpressions and having a recursive call for n->next |
|
| * would be wasteful. |
|
| * The "f" variable should be 0 unless called from pmdoc_Nd for the |
|
| * description buffer, which does not start at the beginning of the |
|
| * buffer. |
|
| */ |
*/ |
| static void |
static void |
| buf_appendmdoc(struct buf *buf, const struct mdoc_node *n, int f) |
putmdockey(const struct mpage *mpage, |
| |
const struct roff_node *n, uint64_t m, int taboo) |
| { |
{ |
| |
|
| for ( ; n; n = n->next) { |
for ( ; NULL != n; n = n->next) { |
| if (n->child) |
if (n->flags & taboo) |
| buf_appendmdoc(buf, n->child, f); |
continue; |
| |
if (NULL != n->child) |
| if (MDOC_TEXT == n->type && f) { |
putmdockey(mpage, n->child, m, taboo); |
| f = 0; |
if (n->type == ROFFT_TEXT) |
| buf_appendb(buf, n->string, |
putkey(mpage, n->string, m); |
| strlen(n->string) + 1); |
|
| } else if (MDOC_TEXT == n->type) |
|
| buf_append(buf, n->string); |
|
| |
|
| } |
} |
| } |
} |
| |
|
| /* ARGSUSED */ |
|
| static void |
static void |
| pmdoc_An(MDOC_ARGS) |
parse_man(struct mpage *mpage, const struct roff_meta *meta, |
| |
const struct roff_node *n) |
| { |
{ |
| |
const struct roff_node *head, *body; |
| if (SEC_AUTHORS != n->sec) |
char *start, *title; |
| |
char byte; |
| |
size_t sz; |
| |
|
| |
if (n == NULL) |
| return; |
return; |
| |
|
| buf_appendmdoc(buf, n->child, 0); |
/* |
| hash_put(hash, buf, TYPE_An); |
* We're only searching for one thing: the first text child in |
| |
* the BODY of a NAME section. Since we don't keep track of |
| |
* sections in -man, run some hoops to find out whether we're in |
| |
* the correct section or not. |
| |
*/ |
| |
|
| |
if (n->type == ROFFT_BODY && n->tok == MAN_SH) { |
| |
body = n; |
| |
if ((head = body->parent->head) != NULL && |
| |
(head = head->child) != NULL && |
| |
head->next == NULL && |
| |
head->type == ROFFT_TEXT && |
| |
strcmp(head->string, "NAME") == 0 && |
| |
body->child != NULL) { |
| |
|
| |
/* |
| |
* 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. |
| |
*/ |
| |
|
| |
title = NULL; |
| |
deroff(&title, body); |
| |
if (NULL == title) |
| |
return; |
| |
|
| |
/* |
| |
* Go through a special heuristic dance here. |
| |
* Conventionally, one or more manual names are |
| |
* comma-specified prior to a whitespace, then a |
| |
* dash, then a description. Try to puzzle out |
| |
* the name parts here. |
| |
*/ |
| |
|
| |
start = title; |
| |
for ( ;; ) { |
| |
sz = strcspn(start, " ,"); |
| |
if ('\0' == start[sz]) |
| |
break; |
| |
|
| |
byte = start[sz]; |
| |
start[sz] = '\0'; |
| |
|
| |
/* |
| |
* Assume a stray trailing comma in the |
| |
* name list if a name begins with a dash. |
| |
*/ |
| |
|
| |
if ('-' == start[0] || |
| |
('\\' == start[0] && '-' == start[1])) |
| |
break; |
| |
|
| |
putkey(mpage, start, NAME_TITLE); |
| |
if ( ! (mpage->name_head_done || |
| |
strcasecmp(start, meta->title))) { |
| |
putkey(mpage, start, NAME_HEAD); |
| |
mpage->name_head_done = 1; |
| |
} |
| |
|
| |
if (' ' == byte) { |
| |
start += sz + 1; |
| |
break; |
| |
} |
| |
|
| |
assert(',' == byte); |
| |
start += sz + 1; |
| |
while (' ' == *start) |
| |
start++; |
| |
} |
| |
|
| |
if (start == title) { |
| |
putkey(mpage, start, NAME_TITLE); |
| |
if ( ! (mpage->name_head_done || |
| |
strcasecmp(start, meta->title))) { |
| |
putkey(mpage, start, NAME_HEAD); |
| |
mpage->name_head_done = 1; |
| |
} |
| |
free(title); |
| |
return; |
| |
} |
| |
|
| |
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)) |
| |
start += 4; |
| |
else if (0 == strncmp(start, "\\(em", 4)) |
| |
start += 4; |
| |
|
| |
while (' ' == *start) |
| |
start++; |
| |
|
| |
mpage->desc = mandoc_strdup(start); |
| |
free(title); |
| |
return; |
| |
} |
| |
} |
| |
|
| |
for (n = n->child; n; n = n->next) { |
| |
if (NULL != mpage->desc) |
| |
break; |
| |
parse_man(mpage, meta, n); |
| |
} |
| } |
} |
| |
|
| static void |
static void |
| hash_reset(DB **db) |
parse_mdoc(struct mpage *mpage, const struct roff_meta *meta, |
| |
const struct roff_node *n) |
| { |
{ |
| DB *hash; |
|
| |
|
| if (NULL != (hash = *db)) |
assert(NULL != n); |
| (*hash->close)(hash); |
for (n = n->child; NULL != n; n = n->next) { |
| |
if (n->flags & mdocs[n->tok].taboo) |
| *db = dbopen(NULL, O_CREAT|O_RDWR, 0644, DB_HASH, NULL); |
continue; |
| if (NULL == *db) { |
switch (n->type) { |
| perror("hash"); |
case ROFFT_ELEM: |
| exit((int)MANDOCLEVEL_SYSERR); |
case ROFFT_BLOCK: |
| |
case ROFFT_HEAD: |
| |
case ROFFT_BODY: |
| |
case ROFFT_TAIL: |
| |
if (NULL != mdocs[n->tok].fp) |
| |
if (0 == (*mdocs[n->tok].fp)(mpage, meta, n)) |
| |
break; |
| |
if (mdocs[n->tok].mask) |
| |
putmdockey(mpage, n->child, |
| |
mdocs[n->tok].mask, mdocs[n->tok].taboo); |
| |
break; |
| |
default: |
| |
assert(n->type != ROFFT_ROOT); |
| |
continue; |
| |
} |
| |
if (NULL != n->child) |
| |
parse_mdoc(mpage, meta, n); |
| } |
} |
| } |
} |
| |
|
| /* ARGSUSED */ |
static int |
| static void |
parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta, |
| pmdoc_Fd(MDOC_ARGS) |
const struct roff_node *n) |
| { |
{ |
| const char *start, *end; |
char *start, *end; |
| size_t sz; |
size_t sz; |
| |
|
| if (SEC_SYNOPSIS != n->sec) |
|
| return; |
|
| if (NULL == (n = n->child) || MDOC_TEXT != n->type) |
|
| return; |
|
| |
|
| |
if (SEC_SYNOPSIS != n->sec || |
| |
NULL == (n = n->child) || |
| |
n->type != ROFFT_TEXT) |
| |
return 0; |
| |
|
| /* |
/* |
| * Only consider those `Fd' macro fields that begin with an |
* Only consider those `Fd' macro fields that begin with an |
| * "inclusion" token (versus, e.g., #define). |
* "inclusion" token (versus, e.g., #define). |
| */ |
*/ |
| |
|
| if (strcmp("#include", n->string)) |
if (strcmp("#include", n->string)) |
| return; |
return 0; |
| |
|
| if (NULL == (n = n->next) || MDOC_TEXT != n->type) |
if ((n = n->next) == NULL || n->type != ROFFT_TEXT) |
| return; |
return 0; |
| |
|
| /* |
/* |
| * Strip away the enclosing angle brackets and make sure we're |
* Strip away the enclosing angle brackets and make sure we're |
| Line 781 pmdoc_Fd(MDOC_ARGS) |
|
| Line 1562 pmdoc_Fd(MDOC_ARGS) |
|
| start++; |
start++; |
| |
|
| if (0 == (sz = strlen(start))) |
if (0 == (sz = strlen(start))) |
| return; |
return 0; |
| |
|
| end = &start[(int)sz - 1]; |
end = &start[(int)sz - 1]; |
| if ('>' == *end || '"' == *end) |
if ('>' == *end || '"' == *end) |
| end--; |
end--; |
| |
|
| assert(end >= start); |
if (end > start) |
| |
putkeys(mpage, start, end - start + 1, TYPE_In); |
| buf_appendb(buf, start, (size_t)(end - start + 1)); |
return 0; |
| buf_appendb(buf, "", 1); |
|
| |
|
| hash_put(hash, buf, TYPE_In); |
|
| } |
} |
| |
|
| /* ARGSUSED */ |
|
| static void |
static void |
| pmdoc_Cd(MDOC_ARGS) |
parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n) |
| { |
{ |
| |
char *cp; |
| if (SEC_SYNOPSIS != n->sec) |
size_t sz; |
| |
|
| |
if (n->type != ROFFT_TEXT) |
| return; |
return; |
| |
|
| buf_appendmdoc(buf, n->child, 0); |
/* Skip function pointer punctuation. */ |
| hash_put(hash, buf, TYPE_Cd); |
|
| } |
|
| |
|
| /* ARGSUSED */ |
cp = n->string; |
| static void |
while (*cp == '(' || *cp == '*') |
| pmdoc_In(MDOC_ARGS) |
cp++; |
| { |
sz = strcspn(cp, "()"); |
| |
|
| if (SEC_SYNOPSIS != n->sec) |
|
| return; |
|
| if (NULL == n->child || MDOC_TEXT != n->child->type) |
|
| return; |
|
| |
|
| buf_append(buf, n->child->string); |
putkeys(mpage, cp, sz, TYPE_Fn); |
| hash_put(hash, buf, TYPE_In); |
if (n->sec == SEC_SYNOPSIS) |
| |
putkeys(mpage, cp, sz, NAME_SYN); |
| } |
} |
| |
|
| /* ARGSUSED */ |
static int |
| static void |
parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta, |
| pmdoc_Fn(MDOC_ARGS) |
const struct roff_node *n) |
| { |
{ |
| const char *cp; |
|
| |
|
| if (SEC_SYNOPSIS != n->sec) |
|
| return; |
|
| if (NULL == n->child || MDOC_TEXT != n->child->type) |
|
| return; |
|
| |
|
| /* .Fn "struct type *arg" "foo" */ |
if (n->child == NULL) |
| |
return 0; |
| |
|
| cp = strrchr(n->child->string, ' '); |
parse_mdoc_fname(mpage, n->child); |
| if (NULL == cp) |
|
| cp = n->child->string; |
|
| |
|
| /* Strip away pointer symbol. */ |
for (n = n->child->next; n != NULL; n = n->next) |
| |
if (n->type == ROFFT_TEXT) |
| |
putkey(mpage, n->string, TYPE_Fa); |
| |
|
| while ('*' == *cp) |
return 0; |
| cp++; |
|
| |
|
| buf_append(buf, cp); |
|
| hash_put(hash, buf, TYPE_Fn); |
|
| } |
} |
| |
|
| /* ARGSUSED */ |
static int |
| static void |
parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta, |
| pmdoc_St(MDOC_ARGS) |
const struct roff_node *n) |
| { |
{ |
| |
|
| if (SEC_STANDARDS != n->sec) |
|
| return; |
|
| if (NULL == n->child || MDOC_TEXT != n->child->type) |
|
| return; |
|
| |
|
| buf_append(buf, n->child->string); |
if (n->type != ROFFT_HEAD) |
| hash_put(hash, buf, TYPE_St); |
return 1; |
| } |
|
| |
|
| /* ARGSUSED */ |
if (n->child != NULL) |
| static void |
parse_mdoc_fname(mpage, n->child); |
| pmdoc_Xr(MDOC_ARGS) |
|
| { |
|
| |
|
| if (NULL == (n = n->child)) |
return 0; |
| return; |
|
| |
|
| buf_appendb(buf, n->string, strlen(n->string)); |
|
| |
|
| if (NULL != (n = n->next)) { |
|
| buf_appendb(buf, ".", 1); |
|
| buf_appendb(buf, n->string, strlen(n->string) + 1); |
|
| } else |
|
| buf_appendb(buf, ".", 2); |
|
| |
|
| hash_put(hash, buf, TYPE_Xr); |
|
| } |
} |
| |
|
| /* ARGSUSED */ |
static int |
| static void |
parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta, |
| pmdoc_Vt(MDOC_ARGS) |
const struct roff_node *n) |
| { |
{ |
| const char *start; |
char *cp; |
| size_t sz; |
|
| |
|
| if (SEC_SYNOPSIS != n->sec) |
|
| return; |
|
| if (MDOC_Vt == n->tok && MDOC_BODY != n->type) |
|
| return; |
|
| if (NULL == n->last || MDOC_TEXT != n->last->type) |
|
| return; |
|
| |
|
| /* |
if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY) |
| * Strip away leading pointer symbol '*' and trailing ';'. |
return 0; |
| */ |
|
| |
|
| start = n->last->string; |
if (n->child != NULL && |
| |
n->child->next == NULL && |
| |
n->child->type == ROFFT_TEXT) |
| |
return 1; |
| |
|
| while ('*' == *start) |
cp = NULL; |
| start++; |
deroff(&cp, n); |
| |
if (cp != NULL) { |
| |
putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va || |
| |
n->type == ROFFT_BODY ? TYPE_Va : 0)); |
| |
free(cp); |
| |
} |
| |
|
| if (0 == (sz = strlen(start))) |
return 0; |
| return; |
|
| |
|
| if (';' == start[(int)sz - 1]) |
|
| sz--; |
|
| |
|
| if (0 == sz) |
|
| return; |
|
| |
|
| buf_appendb(buf, start, sz); |
|
| buf_appendb(buf, "", 1); |
|
| hash_put(hash, buf, TYPE_Va); |
|
| } |
} |
| |
|
| /* ARGSUSED */ |
static int |
| static void |
parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta, |
| pmdoc_Fo(MDOC_ARGS) |
const struct roff_node *n) |
| { |
{ |
| |
char *cp; |
| if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type) |
|
| return; |
|
| if (NULL == n->child || MDOC_TEXT != n->child->type) |
|
| return; |
|
| |
|
| buf_append(buf, n->child->string); |
if (NULL == (n = n->child)) |
| hash_put(hash, buf, TYPE_Fn); |
return 0; |
| } |
|
| |
|
| |
if (NULL == n->next) { |
| |
putkey(mpage, n->string, TYPE_Xr); |
| |
return 0; |
| |
} |
| |
|
| /* ARGSUSED */ |
mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string); |
| static void |
putkey(mpage, cp, TYPE_Xr); |
| pmdoc_Nd(MDOC_ARGS) |
free(cp); |
| { |
return 0; |
| |
|
| if (MDOC_BODY != n->type) |
|
| return; |
|
| |
|
| buf_appendmdoc(dbuf, n->child, 1); |
|
| buf_appendmdoc(buf, n->child, 0); |
|
| |
|
| hash_put(hash, buf, TYPE_Nd); |
|
| } |
} |
| |
|
| /* ARGSUSED */ |
static int |
| static void |
parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta, |
| pmdoc_Er(MDOC_ARGS) |
const struct roff_node *n) |
| { |
{ |
| |
|
| if (SEC_ERRORS != n->sec) |
if (n->type == ROFFT_BODY) |
| return; |
deroff(&mpage->desc, n); |
| |
return 0; |
| buf_appendmdoc(buf, n->child, 0); |
|
| hash_put(hash, buf, TYPE_Er); |
|
| } |
} |
| |
|
| /* ARGSUSED */ |
static int |
| static void |
parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta, |
| pmdoc_Ev(MDOC_ARGS) |
const struct roff_node *n) |
| { |
{ |
| |
|
| if (SEC_ENVIRONMENT != n->sec) |
if (SEC_NAME == n->sec) |
| return; |
putmdockey(mpage, n->child, NAME_TITLE, 0); |
| |
else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) { |
| buf_appendmdoc(buf, n->child, 0); |
if (n->child == NULL) |
| hash_put(hash, buf, TYPE_Ev); |
putkey(mpage, meta->name, NAME_SYN); |
| |
else |
| |
putmdockey(mpage, n->child, NAME_SYN, 0); |
| |
} |
| |
if ( ! (mpage->name_head_done || |
| |
n->child == NULL || n->child->string == NULL || |
| |
strcasecmp(n->child->string, meta->title))) { |
| |
putkey(mpage, n->child->string, NAME_HEAD); |
| |
mpage->name_head_done = 1; |
| |
} |
| |
return 0; |
| } |
} |
| |
|
| /* ARGSUSED */ |
static int |
| static void |
parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta, |
| pmdoc_Pa(MDOC_ARGS) |
const struct roff_node *n) |
| { |
{ |
| |
|
| if (SEC_FILES != n->sec) |
return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD; |
| return; |
|
| |
|
| buf_appendmdoc(buf, n->child, 0); |
|
| hash_put(hash, buf, TYPE_Pa); |
|
| } |
} |
| |
|
| /* ARGSUSED */ |
static int |
| static void |
parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta, |
| pmdoc_Nm(MDOC_ARGS) |
const struct roff_node *n) |
| { |
{ |
| |
|
| if (SEC_NAME == n->sec) { |
|
| buf_appendmdoc(buf, n->child, 0); |
|
| hash_put(hash, buf, TYPE_Nm); |
|
| return; |
|
| } else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type) |
|
| return; |
|
| |
|
| if (NULL == n->child) |
return n->type == ROFFT_HEAD; |
| buf_append(buf, m->name); |
|
| |
|
| buf_appendmdoc(buf, n->child, 0); |
|
| hash_put(hash, buf, TYPE_Nm); |
|
| } |
} |
| |
|
| |
/* |
| |
* Add a string to the hash table for the current manual. |
| |
* Each string has a bitmask telling which macros it belongs to. |
| |
* When we finish the manual, we'll dump the table. |
| |
*/ |
| static void |
static void |
| hash_put(DB *db, const struct buf *buf, uint64_t mask) |
putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v) |
| { |
{ |
| DBT key, val; |
struct ohash *htab; |
| int rc; |
struct str *s; |
| |
const char *end; |
| |
unsigned int slot; |
| |
int i, mustfree; |
| |
|
| if (buf->len < 2) |
if (0 == sz) |
| return; |
return; |
| |
|
| key.data = buf->cp; |
mustfree = render_string(&cp, &sz); |
| key.size = buf->len; |
|
| |
|
| if ((rc = (*db->get)(db, &key, &val, 0)) < 0) { |
if (TYPE_Nm & v) { |
| perror("hash"); |
htab = &names; |
| exit((int)MANDOCLEVEL_SYSERR); |
v &= name_mask; |
| } else if (0 == rc) |
if (v & NAME_FIRST) |
| mask |= *(uint64_t *)val.data; |
name_mask &= ~NAME_FIRST; |
| |
if (debug > 1) |
| |
say(mpage->mlinks->file, |
| |
"Adding name %*s, bits=0x%llx", (int)sz, cp, |
| |
(unsigned long long)v); |
| |
} else { |
| |
htab = &strings; |
| |
if (debug > 1) |
| |
for (i = 0; i < KEY_MAX; i++) |
| |
if ((uint64_t)1 << i & v) |
| |
say(mpage->mlinks->file, |
| |
"Adding key %s=%*s", |
| |
mansearch_keynames[i], (int)sz, cp); |
| |
} |
| |
|
| val.data = &mask; |
end = cp + sz; |
| val.size = sizeof(uint64_t); |
slot = ohash_qlookupi(htab, cp, &end); |
| |
s = ohash_find(htab, slot); |
| |
|
| if ((rc = (*db->put)(db, &key, &val, 0)) < 0) { |
if (NULL != s && mpage == s->mpage) { |
| perror("hash"); |
s->mask |= v; |
| exit((int)MANDOCLEVEL_SYSERR); |
return; |
| } |
} else if (NULL == s) { |
| |
s = mandoc_calloc(1, sizeof(struct str) + sz + 1); |
| |
memcpy(s->key, cp, sz); |
| |
ohash_insert(htab, slot, s); |
| |
} |
| |
s->mpage = mpage; |
| |
s->mask = v; |
| |
|
| |
if (mustfree) |
| |
free(cp); |
| } |
} |
| |
|
| static void |
/* |
| dbt_put(DB *db, const char *dbn, DBT *key, DBT *val) |
* Take a Unicode codepoint and produce its UTF-8 encoding. |
| |
* This isn't the best way to do this, but it works. |
| |
* The magic numbers are from the UTF-8 packaging. |
| |
* They're not as scary as they seem: read the UTF-8 spec for details. |
| |
*/ |
| |
static size_t |
| |
utf8(unsigned int cp, char out[7]) |
| { |
{ |
| |
size_t rc; |
| |
|
| assert(key->size); |
rc = 0; |
| assert(val->size); |
if (cp <= 0x0000007F) { |
| |
rc = 1; |
| |
out[0] = (char)cp; |
| |
} else if (cp <= 0x000007FF) { |
| |
rc = 2; |
| |
out[0] = (cp >> 6 & 31) | 192; |
| |
out[1] = (cp & 63) | 128; |
| |
} else if (cp <= 0x0000FFFF) { |
| |
rc = 3; |
| |
out[0] = (cp >> 12 & 15) | 224; |
| |
out[1] = (cp >> 6 & 63) | 128; |
| |
out[2] = (cp & 63) | 128; |
| |
} else if (cp <= 0x001FFFFF) { |
| |
rc = 4; |
| |
out[0] = (cp >> 18 & 7) | 240; |
| |
out[1] = (cp >> 12 & 63) | 128; |
| |
out[2] = (cp >> 6 & 63) | 128; |
| |
out[3] = (cp & 63) | 128; |
| |
} else if (cp <= 0x03FFFFFF) { |
| |
rc = 5; |
| |
out[0] = (cp >> 24 & 3) | 248; |
| |
out[1] = (cp >> 18 & 63) | 128; |
| |
out[2] = (cp >> 12 & 63) | 128; |
| |
out[3] = (cp >> 6 & 63) | 128; |
| |
out[4] = (cp & 63) | 128; |
| |
} else if (cp <= 0x7FFFFFFF) { |
| |
rc = 6; |
| |
out[0] = (cp >> 30 & 1) | 252; |
| |
out[1] = (cp >> 24 & 63) | 128; |
| |
out[2] = (cp >> 18 & 63) | 128; |
| |
out[3] = (cp >> 12 & 63) | 128; |
| |
out[4] = (cp >> 6 & 63) | 128; |
| |
out[5] = (cp & 63) | 128; |
| |
} else |
| |
return 0; |
| |
|
| if (0 == (*db->put)(db, key, val, 0)) |
out[rc] = '\0'; |
| return; |
return rc; |
| |
|
| perror(dbn); |
|
| exit((int)MANDOCLEVEL_SYSERR); |
|
| /* NOTREACHED */ |
|
| } |
} |
| |
|
| /* |
/* |
| * Call out to per-macro handlers after clearing the persistent database |
* If the string contains escape sequences, |
| * key. If the macro sets the database key, flush it to the database. |
* replace it with an allocated rendering and return 1, |
| |
* such that the caller can free it after use. |
| |
* Otherwise, do nothing and return 0. |
| */ |
*/ |
| static void |
static int |
| pmdoc_node(MDOC_ARGS) |
render_string(char **public, size_t *psz) |
| { |
{ |
| |
const char *src, *scp, *addcp, *seq; |
| |
char *dst; |
| |
size_t ssz, dsz, addsz; |
| |
char utfbuf[7], res[6]; |
| |
int seqlen, unicode; |
| |
|
| if (NULL == n) |
res[0] = '\\'; |
| return; |
res[1] = '\t'; |
| |
res[2] = ASCII_NBRSP; |
| |
res[3] = ASCII_HYPH; |
| |
res[4] = ASCII_BREAK; |
| |
res[5] = '\0'; |
| |
|
| switch (n->type) { |
src = scp = *public; |
| case (MDOC_HEAD): |
ssz = *psz; |
| /* FALLTHROUGH */ |
dst = NULL; |
| case (MDOC_BODY): |
dsz = 0; |
| /* FALLTHROUGH */ |
|
| case (MDOC_TAIL): |
|
| /* FALLTHROUGH */ |
|
| case (MDOC_BLOCK): |
|
| /* FALLTHROUGH */ |
|
| case (MDOC_ELEM): |
|
| if (NULL == mdocs[n->tok]) |
|
| break; |
|
| |
|
| buf->len = 0; |
while (scp < src + *psz) { |
| (*mdocs[n->tok])(hash, buf, dbuf, n, m); |
|
| break; |
|
| default: |
|
| break; |
|
| } |
|
| |
|
| pmdoc_node(hash, buf, dbuf, n->child, m); |
/* Leave normal characters unchanged. */ |
| pmdoc_node(hash, buf, dbuf, n->next, m); |
|
| } |
|
| |
|
| static int |
if (strchr(res, *scp) == NULL) { |
| pman_node(MAN_ARGS) |
if (dst != NULL) |
| { |
dst[dsz++] = *scp; |
| const struct man_node *head, *body; |
scp++; |
| const char *start, *sv; |
continue; |
| size_t sz; |
} |
| |
|
| if (NULL == n) |
/* |
| return(0); |
* Found something that requires replacing, |
| |
* make sure we have a destination buffer. |
| |
*/ |
| |
|
| /* |
if (dst == NULL) { |
| * We're only searching for one thing: the first text child in |
dst = mandoc_malloc(ssz + 1); |
| * the BODY of a NAME section. Since we don't keep track of |
dsz = scp - src; |
| * sections in -man, run some hoops to find out whether we're in |
memcpy(dst, src, dsz); |
| * the correct section or not. |
} |
| */ |
|
| |
|
| if (MAN_BODY == n->type && MAN_SH == n->tok) { |
/* Handle single-char special characters. */ |
| body = n; |
|
| assert(body->parent); |
|
| if (NULL != (head = body->parent->head) && |
|
| 1 == head->nchild && |
|
| NULL != (head = (head->child)) && |
|
| MAN_TEXT == head->type && |
|
| 0 == strcmp(head->string, "NAME") && |
|
| NULL != (body = body->child) && |
|
| MAN_TEXT == body->type) { |
|
| |
|
| assert(body->string); |
switch (*scp) { |
| start = sv = body->string; |
case '\\': |
| |
break; |
| |
case '\t': |
| |
case ASCII_NBRSP: |
| |
dst[dsz++] = ' '; |
| |
scp++; |
| |
continue; |
| |
case ASCII_HYPH: |
| |
dst[dsz++] = '-'; |
| |
/* FALLTHROUGH */ |
| |
case ASCII_BREAK: |
| |
scp++; |
| |
continue; |
| |
default: |
| |
abort(); |
| |
} |
| |
|
| /* |
/* |
| * Go through a special heuristic dance here. |
* Found an escape sequence. |
| * This is why -man manuals are great! |
* Read past the slash, then parse it. |
| * (I'm being sarcastic: my eyes are bleeding.) |
* Ignore everything except characters. |
| * Conventionally, one or more manual names are |
*/ |
| * comma-specified prior to a whitespace, then a |
|
| * dash, then a description. Try to puzzle out |
|
| * the name parts here. |
|
| */ |
|
| |
|
| for ( ;; ) { |
scp++; |
| sz = strcspn(start, " ,"); |
if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL) |
| if ('\0' == start[(int)sz]) |
continue; |
| break; |
|
| |
|
| buf->len = 0; |
/* |
| buf_appendb(buf, start, sz); |
* Render the special character |
| buf_appendb(buf, "", 1); |
* as either UTF-8 or ASCII. |
| |
*/ |
| |
|
| hash_put(hash, buf, TYPE_Nm); |
if (write_utf8) { |
| |
unicode = mchars_spec2cp(seq, seqlen); |
| if (' ' == start[(int)sz]) { |
if (unicode <= 0) |
| start += (int)sz + 1; |
continue; |
| break; |
addsz = utf8(unicode, utfbuf); |
| } |
if (addsz == 0) |
| |
continue; |
| assert(',' == start[(int)sz]); |
addcp = utfbuf; |
| start += (int)sz + 1; |
} else { |
| while (' ' == *start) |
addcp = mchars_spec2str(seq, seqlen, &addsz); |
| start++; |
if (addcp == NULL) |
| |
continue; |
| |
if (*addcp == ASCII_NBRSP) { |
| |
addcp = " "; |
| |
addsz = 1; |
| } |
} |
| |
} |
| |
|
| buf->len = 0; |
/* Copy the rendered glyph into the stream. */ |
| |
|
| if (sv == start) { |
ssz += addsz; |
| buf_append(buf, start); |
dst = mandoc_realloc(dst, ssz + 1); |
| return(1); |
memcpy(dst + dsz, addcp, addsz); |
| } |
dsz += addsz; |
| |
} |
| |
if (dst != NULL) { |
| |
*public = dst; |
| |
*psz = dsz; |
| |
} |
| |
|
| while (' ' == *start) |
/* Trim trailing whitespace and NUL-terminate. */ |
| start++; |
|
| |
|
| if (0 == strncmp(start, "-", 1)) |
while (*psz > 0 && (*public)[*psz - 1] == ' ') |
| start += 1; |
--*psz; |
| else if (0 == strncmp(start, "\\-", 2)) |
if (dst != NULL) { |
| start += 2; |
(*public)[*psz] = '\0'; |
| else if (0 == strncmp(start, "\\(en", 4)) |
return 1; |
| start += 4; |
} else |
| else if (0 == strncmp(start, "\\(em", 4)) |
return 0; |
| start += 4; |
} |
| |
|
| while (' ' == *start) |
static void |
| start++; |
dbadd_mlink(const struct mlink *mlink) |
| |
{ |
| |
dba_page_alias(mlink->mpage->dba, mlink->name, NAME_FILE); |
| |
dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->dsec); |
| |
dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->fsec); |
| |
dba_page_add(mlink->mpage->dba, DBP_ARCH, mlink->arch); |
| |
dba_page_add(mlink->mpage->dba, DBP_FILE, mlink->file); |
| |
} |
| |
|
| sz = strlen(start) + 1; |
/* |
| buf_appendb(dbuf, start, sz); |
* Flush the current page's terms (and their bits) into the database. |
| buf_appendb(buf, start, sz); |
* Also, handle escape sequences at the last possible moment. |
| |
*/ |
| |
static void |
| |
dbadd(struct dba *dba, struct mpage *mpage) |
| |
{ |
| |
struct mlink *mlink; |
| |
struct str *key; |
| |
char *cp; |
| |
uint64_t mask; |
| |
size_t i; |
| |
unsigned int slot; |
| |
int mustfree; |
| |
|
| hash_put(hash, buf, TYPE_Nd); |
mlink = mpage->mlinks; |
| |
|
| |
if (nodb) { |
| |
for (key = ohash_first(&names, &slot); NULL != key; |
| |
key = ohash_next(&names, &slot)) |
| |
free(key); |
| |
for (key = ohash_first(&strings, &slot); NULL != key; |
| |
key = ohash_next(&strings, &slot)) |
| |
free(key); |
| |
if (0 == debug) |
| |
return; |
| |
while (NULL != mlink) { |
| |
fputs(mlink->name, stdout); |
| |
if (NULL == mlink->next || |
| |
strcmp(mlink->dsec, mlink->next->dsec) || |
| |
strcmp(mlink->fsec, mlink->next->fsec) || |
| |
strcmp(mlink->arch, mlink->next->arch)) { |
| |
putchar('('); |
| |
if ('\0' == *mlink->dsec) |
| |
fputs(mlink->fsec, stdout); |
| |
else |
| |
fputs(mlink->dsec, stdout); |
| |
if ('\0' != *mlink->arch) |
| |
printf("/%s", mlink->arch); |
| |
putchar(')'); |
| |
} |
| |
mlink = mlink->next; |
| |
if (NULL != mlink) |
| |
fputs(", ", stdout); |
| } |
} |
| |
printf(" - %s\n", mpage->desc); |
| |
return; |
| } |
} |
| |
|
| for (n = n->child; n; n = n->next) |
if (debug) |
| if (pman_node(hash, buf, dbuf, n)) |
say(mlink->file, "Adding to database"); |
| return(1); |
|
| |
|
| return(0); |
cp = mpage->desc; |
| |
i = strlen(cp); |
| |
mustfree = render_string(&cp, &i); |
| |
mpage->dba = dba_page_new(dba->pages, |
| |
*mpage->arch == '\0' ? mlink->arch : mpage->arch, |
| |
cp, mlink->file, mpage->form); |
| |
if (mustfree) |
| |
free(cp); |
| |
dba_page_add(mpage->dba, DBP_SECT, mpage->sec); |
| |
|
| |
while (mlink != NULL) { |
| |
dbadd_mlink(mlink); |
| |
mlink = mlink->next; |
| |
} |
| |
|
| |
for (key = ohash_first(&names, &slot); NULL != key; |
| |
key = ohash_next(&names, &slot)) { |
| |
assert(key->mpage == mpage); |
| |
dba_page_alias(mpage->dba, key->key, key->mask); |
| |
free(key); |
| |
} |
| |
for (key = ohash_first(&strings, &slot); NULL != key; |
| |
key = ohash_next(&strings, &slot)) { |
| |
assert(key->mpage == mpage); |
| |
i = 0; |
| |
for (mask = TYPE_Xr; mask <= TYPE_Lb; mask *= 2) { |
| |
if (key->mask & mask) |
| |
dba_macro_add(dba->macros, i, |
| |
key->key, mpage->dba); |
| |
i++; |
| |
} |
| |
free(key); |
| |
} |
| } |
} |
| |
|
| static void |
static void |
| ofile_argbuild(char *argv[], int argc, int verb, struct of **of) |
dbprune(struct dba *dba) |
| { |
{ |
| int i; |
struct dba_array *page, *files; |
| struct of *nof; |
char *file; |
| |
|
| for (i = 0; i < argc; i++) { |
dba_array_FOREACH(dba->pages, page) { |
| nof = mandoc_calloc(1, sizeof(struct of)); |
files = dba_array_get(page, DBP_FILE); |
| nof->fname = strdup(argv[i]); |
dba_array_FOREACH(files, file) { |
| if (verb > 2) |
if (*file < ' ') |
| printf("%s: Scheduling\n", argv[i]); |
file++; |
| if (NULL == *of) { |
if (ohash_find(&mlinks, ohash_qlookup(&mlinks, |
| *of = nof; |
file)) != NULL) { |
| (*of)->first = nof; |
if (debug) |
| } else { |
say(file, "Deleting from database"); |
| nof->first = (*of)->first; |
dba_array_del(dba->pages); |
| (*of)->next = nof; |
break; |
| *of = nof; |
} |
| } |
} |
| } |
} |
| } |
} |
| |
|
| /* |
/* |
| * Recursively build up a list of files to parse. |
* Write the database from memory to disk. |
| * 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 |
|
| * 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, int verb, struct of **of) |
dbwrite(struct dba *dba) |
| { |
{ |
| char buf[MAXPATHLEN]; |
char tfn[32]; |
| size_t sz; |
int status; |
| DIR *d; |
pid_t child; |
| const char *fn; |
|
| struct of *nof; |
|
| struct dirent *dp; |
|
| |
|
| if (NULL == (d = opendir(dir))) { |
if (dba_write(MANDOC_DB "~", dba) != -1) { |
| perror(dir); |
if (rename(MANDOC_DB "~", MANDOC_DB) == -1) { |
| return(0); |
exitcode = (int)MANDOCLEVEL_SYSERR; |
| |
say(MANDOC_DB, "&rename"); |
| |
unlink(MANDOC_DB "~"); |
| |
} |
| |
return; |
| } |
} |
| |
|
| while (NULL != (dp = readdir(d))) { |
(void)strlcpy(tfn, "/tmp/mandocdb.XXXXXXXX", sizeof(tfn)); |
| fn = dp->d_name; |
if (mkdtemp(tfn) == NULL) { |
| if (DT_DIR == dp->d_type) { |
exitcode = (int)MANDOCLEVEL_SYSERR; |
| if (0 == strcmp(".", fn)) |
say("", "&%s", tfn); |
| continue; |
return; |
| if (0 == strcmp("..", fn)) |
} |
| continue; |
|
| |
|
| buf[0] = '\0'; |
(void)strlcat(tfn, "/" MANDOC_DB, sizeof(tfn)); |
| strlcat(buf, dir, MAXPATHLEN); |
if (dba_write(tfn, dba) == -1) { |
| strlcat(buf, "/", MAXPATHLEN); |
exitcode = (int)MANDOCLEVEL_SYSERR; |
| sz = strlcat(buf, fn, MAXPATHLEN); |
say(tfn, "&dba_write"); |
| |
goto out; |
| |
} |
| |
|
| if (sz < MAXPATHLEN) { |
switch (child = fork()) { |
| if ( ! ofile_dirbuild(buf, verb, of)) |
case -1: |
| return(0); |
exitcode = (int)MANDOCLEVEL_SYSERR; |
| continue; |
say("", "&fork cmp"); |
| } else if (sz < MAXPATHLEN) |
return; |
| continue; |
case 0: |
| |
execlp("cmp", "cmp", "-s", tfn, MANDOC_DB, (char *)NULL); |
| |
say("", "&exec cmp"); |
| |
exit(0); |
| |
default: |
| |
break; |
| |
} |
| |
if (waitpid(child, &status, 0) == -1) { |
| |
exitcode = (int)MANDOCLEVEL_SYSERR; |
| |
say("", "&wait cmp"); |
| |
} else if (WIFSIGNALED(status)) { |
| |
exitcode = (int)MANDOCLEVEL_SYSERR; |
| |
say("", "cmp died from signal %d", WTERMSIG(status)); |
| |
} else if (WEXITSTATUS(status)) { |
| |
exitcode = (int)MANDOCLEVEL_SYSERR; |
| |
say(MANDOC_DB, |
| |
"Data changed, but cannot replace database"); |
| |
} |
| |
|
| fprintf(stderr, "%s: Path too long\n", dir); |
out: |
| return(0); |
*strrchr(tfn, '/') = '\0'; |
| } |
switch (child = fork()) { |
| if (DT_REG != dp->d_type) |
case -1: |
| continue; |
exitcode = (int)MANDOCLEVEL_SYSERR; |
| |
say("", "&fork rm"); |
| |
return; |
| |
case 0: |
| |
execlp("rm", "rm", "-rf", tfn, (char *)NULL); |
| |
say("", "&exec rm"); |
| |
exit((int)MANDOCLEVEL_SYSERR); |
| |
default: |
| |
break; |
| |
} |
| |
if (waitpid(child, &status, 0) == -1) { |
| |
exitcode = (int)MANDOCLEVEL_SYSERR; |
| |
say("", "&wait rm"); |
| |
} else if (WIFSIGNALED(status) || WEXITSTATUS(status)) { |
| |
exitcode = (int)MANDOCLEVEL_SYSERR; |
| |
say("", "%s: Cannot remove temporary directory", tfn); |
| |
} |
| |
} |
| |
|
| if (0 == strcmp(MANDOC_DB, fn) || |
static int |
| 0 == strcmp(MANDOC_IDX, fn)) |
set_basedir(const char *targetdir, int report_baddir) |
| continue; |
{ |
| |
static char startdir[PATH_MAX]; |
| |
static int getcwd_status; /* 1 = ok, 2 = failure */ |
| |
static int chdir_status; /* 1 = changed directory */ |
| |
char *cp; |
| |
|
| buf[0] = '\0'; |
/* |
| strlcat(buf, dir, MAXPATHLEN); |
* Remember the original working directory, if possible. |
| strlcat(buf, "/", MAXPATHLEN); |
* This will be needed if the second or a later directory |
| sz = strlcat(buf, fn, MAXPATHLEN); |
* on the command line is given as a relative path. |
| if (sz >= MAXPATHLEN) { |
* Do not error out if the current directory is not |
| fprintf(stderr, "%s: Path too long\n", dir); |
* searchable: Maybe it won't be needed after all. |
| return(0); |
*/ |
| } |
if (0 == getcwd_status) { |
| |
if (NULL == getcwd(startdir, sizeof(startdir))) { |
| |
getcwd_status = 2; |
| |
(void)strlcpy(startdir, strerror(errno), |
| |
sizeof(startdir)); |
| |
} else |
| |
getcwd_status = 1; |
| |
} |
| |
|
| nof = mandoc_calloc(1, sizeof(struct of)); |
/* |
| nof->fname = mandoc_strdup(buf); |
* We are leaving the old base directory. |
| |
* Do not use it any longer, not even for messages. |
| |
*/ |
| |
*basedir = '\0'; |
| |
|
| if (verb > 2) |
/* |
| printf("%s: Scheduling\n", buf); |
* If and only if the directory was changed earlier and |
| |
* the next directory to process is given as a relative path, |
| if (NULL == *of) { |
* first go back, or bail out if that is impossible. |
| *of = nof; |
*/ |
| (*of)->first = nof; |
if (chdir_status && '/' != *targetdir) { |
| } else { |
if (2 == getcwd_status) { |
| nof->first = (*of)->first; |
exitcode = (int)MANDOCLEVEL_SYSERR; |
| (*of)->next = nof; |
say("", "getcwd: %s", startdir); |
| *of = nof; |
return 0; |
| } |
} |
| |
if (-1 == chdir(startdir)) { |
| |
exitcode = (int)MANDOCLEVEL_SYSERR; |
| |
say("", "&chdir %s", startdir); |
| |
return 0; |
| |
} |
| } |
} |
| |
|
| closedir(d); |
/* |
| return(1); |
* Always resolve basedir to the canonicalized absolute |
| } |
* pathname and append a trailing slash, such that |
| |
* we can reliably check whether files are inside. |
| static void |
*/ |
| ofile_free(struct of *of) |
if (NULL == realpath(targetdir, basedir)) { |
| { |
if (report_baddir || errno != ENOENT) { |
| struct of *nof; |
exitcode = (int)MANDOCLEVEL_BADARG; |
| |
say("", "&%s: realpath", targetdir); |
| while (of) { |
} |
| nof = of->next; |
return 0; |
| free(of->fname); |
} else if (-1 == chdir(basedir)) { |
| free(of); |
if (report_baddir || errno != ENOENT) { |
| of = nof; |
exitcode = (int)MANDOCLEVEL_BADARG; |
| |
say("", "&chdir"); |
| |
} |
| |
return 0; |
| } |
} |
| |
chdir_status = 1; |
| |
cp = strchr(basedir, '\0'); |
| |
if ('/' != cp[-1]) { |
| |
if (cp - basedir >= PATH_MAX - 1) { |
| |
exitcode = (int)MANDOCLEVEL_SYSERR; |
| |
say("", "Filename too long"); |
| |
return 0; |
| |
} |
| |
*cp++ = '/'; |
| |
*cp = '\0'; |
| |
} |
| |
return 1; |
| } |
} |
| |
|
| static void |
static void |
| usage(void) |
say(const char *file, const char *format, ...) |
| { |
{ |
| |
va_list ap; |
| |
int use_errno; |
| |
|
| fprintf(stderr, "usage: %s [-v] " |
if ('\0' != *basedir) |
| "[-d dir [files...] |" |
fprintf(stderr, "%s", basedir); |
| " -u dir [files...] |" |
if ('\0' != *basedir && '\0' != *file) |
| " dir...]\n", progname); |
fputc('/', stderr); |
| |
if ('\0' != *file) |
| |
fprintf(stderr, "%s", file); |
| |
|
| |
use_errno = 1; |
| |
if (NULL != format) { |
| |
switch (*format) { |
| |
case '&': |
| |
format++; |
| |
break; |
| |
case '\0': |
| |
format = NULL; |
| |
break; |
| |
default: |
| |
use_errno = 0; |
| |
break; |
| |
} |
| |
} |
| |
if (NULL != format) { |
| |
if ('\0' != *basedir || '\0' != *file) |
| |
fputs(": ", stderr); |
| |
va_start(ap, format); |
| |
vfprintf(stderr, format, ap); |
| |
va_end(ap); |
| |
} |
| |
if (use_errno) { |
| |
if ('\0' != *basedir || '\0' != *file || NULL != format) |
| |
fputs(": ", stderr); |
| |
perror(NULL); |
| |
} else |
| |
fputc('\n', stderr); |
| } |
} |
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to mandocdb.c CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [cvsweb.bsd.lv] / mandoc