version 1.31, 2011/12/09 11:16:34 |
version 1.203, 2015/10/13 22:59:54 |
|
|
/* $Id$ */ |
/* $Id$ */ |
/* |
/* |
* Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv> |
* Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv> |
* Copyright (c) 2011 Ingo Schwarze <schwarze@openbsd.org> |
* Copyright (c) 2011-2015 Ingo Schwarze <schwarze@openbsd.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/types.h> |
#include <sys/stat.h> |
#include <sys/stat.h> |
|
#include <sys/wait.h> |
|
|
#include <assert.h> |
#include <assert.h> |
#include <dirent.h> |
#include <ctype.h> |
|
#include <err.h> |
|
#include <errno.h> |
#include <fcntl.h> |
#include <fcntl.h> |
|
#if HAVE_FTS |
|
#include <fts.h> |
|
#else |
|
#include "compat_fts.h" |
|
#endif |
#include <getopt.h> |
#include <getopt.h> |
|
#include <limits.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> |
#include <unistd.h> |
|
|
#if defined(__linux__) |
#include <sqlite3.h> |
# include <endian.h> |
|
# include <db_185.h> |
|
#elif defined(__APPLE__) |
|
# include <libkern/OSByteOrder.h> |
|
# include <db.h> |
|
#else |
|
# include <db.h> |
|
#endif |
|
|
|
#include "man.h" |
#include "mandoc_aux.h" |
#include "mdoc.h" |
#include "mandoc_ohash.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" |
|
|
#define MANDOC_BUFSZ BUFSIZ |
extern int mansearch_keymax; |
#define MANDOC_SLOP 1024 |
extern const char *const mansearch_keynames[]; |
|
|
#define MANDOC_SRC 0x1 |
#define SQL_EXEC(_v) \ |
#define MANDOC_FORM 0x2 |
if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \ |
|
say("", "%s: %s", (_v), sqlite3_errmsg(db)) |
|
#define SQL_BIND_TEXT(_s, _i, _v) \ |
|
if (SQLITE_OK != sqlite3_bind_text \ |
|
((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \ |
|
say(mlink->file, "%s", sqlite3_errmsg(db)) |
|
#define SQL_BIND_INT(_s, _i, _v) \ |
|
if (SQLITE_OK != sqlite3_bind_int \ |
|
((_s), (_i)++, (_v))) \ |
|
say(mlink->file, "%s", sqlite3_errmsg(db)) |
|
#define SQL_BIND_INT64(_s, _i, _v) \ |
|
if (SQLITE_OK != sqlite3_bind_int64 \ |
|
((_s), (_i)++, (_v))) \ |
|
say(mlink->file, "%s", sqlite3_errmsg(db)) |
|
#define SQL_STEP(_s) \ |
|
if (SQLITE_DONE != sqlite3_step((_s))) \ |
|
say(mlink->file, "%s", sqlite3_errmsg(db)) |
|
|
/* 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 */ |
char *sec; |
uint64_t mask; /* bitmask in sequence */ |
char *arch; |
char key[]; /* rendered text */ |
char *title; |
|
int src_form; |
|
struct of *next; /* NULL for last one */ |
|
struct of *first; /* first in list */ |
|
}; |
}; |
|
|
/* 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 */ |
int64_t pageid; /* pageid in mpages SQL table */ |
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 mlink *mlinks; /* singly linked list */ |
|
int form; /* format from file content */ |
|
int name_head_done; |
}; |
}; |
|
|
/* 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 dform; /* format from directory */ |
|
int fform; /* format from file name suffix */ |
|
int gzip; /* filename has a .gz suffix */ |
|
}; |
|
|
enum op { |
enum stmt { |
OP_NEW = 0, /* new database */ |
STMT_DELETE_PAGE = 0, /* delete mpage */ |
OP_UPDATE, /* delete/add entries in existing database */ |
STMT_INSERT_PAGE, /* insert mpage */ |
OP_DELETE /* delete entries from existing database */ |
STMT_INSERT_LINK, /* insert mlink */ |
|
STMT_INSERT_NAME, /* insert name */ |
|
STMT_SELECT_NAME, /* retrieve existing name flags */ |
|
STMT_INSERT_KEY, /* insert parsed key */ |
|
STMT__MAX |
}; |
}; |
|
|
#define MAN_ARGS DB *hash, \ |
typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *, |
struct buf *buf, \ |
const struct roff_node *); |
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 *, |
|
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 *, |
|
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 *, |
|
size_t *); |
|
static void ofile_argbuild(int, char *[], struct of **); |
|
static int ofile_dirbuild(const char *, const char *, |
|
const char *, int, struct of **); |
|
static void ofile_free(struct of *); |
|
static void pformatted(DB *, struct buf *, struct buf *, |
|
const struct of *); |
|
static int pman_node(MAN_ARGS); |
|
static void pmdoc_node(MDOC_ARGS); |
|
static int pmdoc_head(MDOC_ARGS); |
|
static int pmdoc_body(MDOC_ARGS); |
|
static int pmdoc_Fd(MDOC_ARGS); |
|
static int pmdoc_In(MDOC_ARGS); |
|
static int pmdoc_Fn(MDOC_ARGS); |
|
static int pmdoc_Nd(MDOC_ARGS); |
|
static int pmdoc_Nm(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. */ |
|
|
|
struct mdoc_handler { |
struct mdoc_handler { |
int (*fp)(MDOC_ARGS); /* Optional handler. */ |
mdoc_fp fp; /* optional handler */ |
uint64_t mask; /* Set unless handler returns 0. */ |
uint64_t mask; /* set unless handler returns 0 */ |
int flags; /* For use by pmdoc_node. */ |
|
}; |
}; |
|
|
|
static void dbclose(int); |
|
static void dbadd(struct mpage *); |
|
static void dbadd_mlink(const struct mlink *mlink); |
|
static void dbadd_mlink_name(const struct mlink *mlink); |
|
static int dbopen(int); |
|
static void dbprune(void); |
|
static void filescan(const char *); |
|
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 mparse *); |
|
static void names_check(void); |
|
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_body(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_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); |
|
static int render_string(char **, size_t *); |
|
static void say(const char *, const char *, ...); |
|
static int set_basedir(const char *, int); |
|
static int treescan(void); |
|
static size_t utf8(unsigned int, char [7]); |
|
|
|
extern char *__progname; |
|
|
|
static char tempfilename[32]; |
|
static int nodb; /* no database changes */ |
|
static int mparse_options; /* abort the parse early */ |
|
static int use_all; /* use all found files */ |
|
static int debug; /* print what we're doing */ |
|
static int warnings; /* warn about crap */ |
|
static int write_utf8; /* write UTF-8 output; else ASCII */ |
|
static int exitcode; /* to be returned by main */ |
|
static enum op op; /* operational mode */ |
|
static char basedir[PATH_MAX]; /* current base directory */ |
|
static struct ohash mpages; /* table of distinct manual pages */ |
|
static struct ohash mlinks; /* table of directory entries */ |
|
static struct ohash names; /* table of all names */ |
|
static struct ohash strings; /* table of all strings */ |
|
static sqlite3 *db = NULL; /* current database */ |
|
static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */ |
|
static uint64_t name_mask; |
|
|
static const struct mdoc_handler mdocs[MDOC_MAX] = { |
static const struct mdoc_handler mdocs[MDOC_MAX] = { |
{ NULL, 0, 0 }, /* Ap */ |
{ NULL, 0 }, /* Ap */ |
{ NULL, 0, 0 }, /* Dd */ |
{ NULL, 0 }, /* Dd */ |
{ NULL, 0, 0 }, /* Dt */ |
{ NULL, 0 }, /* Dt */ |
{ NULL, 0, 0 }, /* Os */ |
{ NULL, 0 }, /* Os */ |
{ pmdoc_Sh, TYPE_Sh, MDOCF_CHILD }, /* Sh */ |
{ parse_mdoc_Sh, TYPE_Sh }, /* Sh */ |
{ pmdoc_head, TYPE_Ss, MDOCF_CHILD }, /* Ss */ |
{ parse_mdoc_head, TYPE_Ss }, /* Ss */ |
{ NULL, 0, 0 }, /* Pp */ |
{ NULL, 0 }, /* Pp */ |
{ NULL, 0, 0 }, /* D1 */ |
{ NULL, 0 }, /* D1 */ |
{ NULL, 0, 0 }, /* Dl */ |
{ NULL, 0 }, /* Dl */ |
{ NULL, 0, 0 }, /* Bd */ |
{ NULL, 0 }, /* Bd */ |
{ NULL, 0, 0 }, /* Ed */ |
{ NULL, 0 }, /* Ed */ |
{ NULL, 0, 0 }, /* Bl */ |
{ NULL, 0 }, /* Bl */ |
{ NULL, 0, 0 }, /* El */ |
{ NULL, 0 }, /* El */ |
{ NULL, 0, 0 }, /* It */ |
{ NULL, 0 }, /* It */ |
{ NULL, 0, 0 }, /* Ad */ |
{ NULL, 0 }, /* Ad */ |
{ NULL, TYPE_An, MDOCF_CHILD }, /* An */ |
{ NULL, TYPE_An }, /* An */ |
{ NULL, TYPE_Ar, MDOCF_CHILD }, /* Ar */ |
{ NULL, TYPE_Ar }, /* Ar */ |
{ NULL, TYPE_Cd, MDOCF_CHILD }, /* Cd */ |
{ NULL, TYPE_Cd }, /* Cd */ |
{ NULL, TYPE_Cm, MDOCF_CHILD }, /* Cm */ |
{ NULL, TYPE_Cm }, /* Cm */ |
{ NULL, TYPE_Dv, MDOCF_CHILD }, /* Dv */ |
{ NULL, TYPE_Dv }, /* Dv */ |
{ NULL, TYPE_Er, MDOCF_CHILD }, /* Er */ |
{ NULL, TYPE_Er }, /* Er */ |
{ NULL, TYPE_Ev, MDOCF_CHILD }, /* Ev */ |
{ NULL, TYPE_Ev }, /* Ev */ |
{ NULL, 0, 0 }, /* Ex */ |
{ NULL, 0 }, /* Ex */ |
{ NULL, TYPE_Fa, MDOCF_CHILD }, /* Fa */ |
{ NULL, TYPE_Fa }, /* Fa */ |
{ pmdoc_Fd, TYPE_In, 0 }, /* Fd */ |
{ parse_mdoc_Fd, 0 }, /* Fd */ |
{ NULL, TYPE_Fl, MDOCF_CHILD }, /* Fl */ |
{ NULL, TYPE_Fl }, /* Fl */ |
{ pmdoc_Fn, 0, 0 }, /* Fn */ |
{ parse_mdoc_Fn, 0 }, /* Fn */ |
{ NULL, TYPE_Ft, MDOCF_CHILD }, /* Ft */ |
{ NULL, TYPE_Ft }, /* Ft */ |
{ NULL, TYPE_Ic, MDOCF_CHILD }, /* Ic */ |
{ NULL, TYPE_Ic }, /* Ic */ |
{ pmdoc_In, TYPE_In, 0 }, /* In */ |
{ NULL, TYPE_In }, /* In */ |
{ NULL, TYPE_Li, MDOCF_CHILD }, /* Li */ |
{ NULL, TYPE_Li }, /* Li */ |
{ pmdoc_Nd, TYPE_Nd, MDOCF_CHILD }, /* Nd */ |
{ parse_mdoc_Nd, 0 }, /* Nd */ |
{ pmdoc_Nm, TYPE_Nm, MDOCF_CHILD }, /* Nm */ |
{ parse_mdoc_Nm, 0 }, /* Nm */ |
{ NULL, 0, 0 }, /* Op */ |
{ NULL, 0 }, /* Op */ |
{ NULL, 0, 0 }, /* Ot */ |
{ NULL, 0 }, /* Ot */ |
{ NULL, TYPE_Pa, MDOCF_CHILD }, /* Pa */ |
{ NULL, TYPE_Pa }, /* Pa */ |
{ NULL, 0, 0 }, /* Rv */ |
{ NULL, 0 }, /* Rv */ |
{ pmdoc_St, TYPE_St, 0 }, /* St */ |
{ NULL, TYPE_St }, /* St */ |
{ NULL, TYPE_Va, MDOCF_CHILD }, /* Va */ |
{ NULL, TYPE_Va }, /* Va */ |
{ pmdoc_body, TYPE_Va, MDOCF_CHILD }, /* Vt */ |
{ parse_mdoc_body, TYPE_Va }, /* Vt */ |
{ pmdoc_Xr, TYPE_Xr, 0 }, /* Xr */ |
{ parse_mdoc_Xr, 0 }, /* Xr */ |
{ NULL, 0, 0 }, /* %A */ |
{ NULL, 0 }, /* %A */ |
{ NULL, 0, 0 }, /* %B */ |
{ NULL, 0 }, /* %B */ |
{ NULL, 0, 0 }, /* %D */ |
{ NULL, 0 }, /* %D */ |
{ NULL, 0, 0 }, /* %I */ |
{ NULL, 0 }, /* %I */ |
{ NULL, 0, 0 }, /* %J */ |
{ NULL, 0 }, /* %J */ |
{ NULL, 0, 0 }, /* %N */ |
{ NULL, 0 }, /* %N */ |
{ NULL, 0, 0 }, /* %O */ |
{ NULL, 0 }, /* %O */ |
{ NULL, 0, 0 }, /* %P */ |
{ NULL, 0 }, /* %P */ |
{ NULL, 0, 0 }, /* %R */ |
{ NULL, 0 }, /* %R */ |
{ NULL, 0, 0 }, /* %T */ |
{ NULL, 0 }, /* %T */ |
{ NULL, 0, 0 }, /* %V */ |
{ NULL, 0 }, /* %V */ |
{ NULL, 0, 0 }, /* Ac */ |
{ NULL, 0 }, /* Ac */ |
{ NULL, 0, 0 }, /* Ao */ |
{ NULL, 0 }, /* Ao */ |
{ NULL, 0, 0 }, /* Aq */ |
{ NULL, 0 }, /* Aq */ |
{ NULL, TYPE_At, MDOCF_CHILD }, /* At */ |
{ NULL, TYPE_At }, /* At */ |
{ NULL, 0, 0 }, /* Bc */ |
{ NULL, 0 }, /* Bc */ |
{ NULL, 0, 0 }, /* Bf */ |
{ NULL, 0 }, /* Bf */ |
{ NULL, 0, 0 }, /* Bo */ |
{ NULL, 0 }, /* Bo */ |
{ NULL, 0, 0 }, /* Bq */ |
{ NULL, 0 }, /* Bq */ |
{ NULL, TYPE_Bsx, MDOCF_CHILD }, /* Bsx */ |
{ NULL, TYPE_Bsx }, /* Bsx */ |
{ NULL, TYPE_Bx, MDOCF_CHILD }, /* Bx */ |
{ NULL, TYPE_Bx }, /* Bx */ |
{ NULL, 0, 0 }, /* Db */ |
{ NULL, 0 }, /* Db */ |
{ NULL, 0, 0 }, /* Dc */ |
{ NULL, 0 }, /* Dc */ |
{ NULL, 0, 0 }, /* Do */ |
{ NULL, 0 }, /* Do */ |
{ NULL, 0, 0 }, /* Dq */ |
{ NULL, 0 }, /* Dq */ |
{ NULL, 0, 0 }, /* Ec */ |
{ NULL, 0 }, /* Ec */ |
{ NULL, 0, 0 }, /* Ef */ |
{ NULL, 0 }, /* Ef */ |
{ NULL, TYPE_Em, MDOCF_CHILD }, /* Em */ |
{ NULL, TYPE_Em }, /* Em */ |
{ NULL, 0, 0 }, /* Eo */ |
{ NULL, 0 }, /* Eo */ |
{ NULL, TYPE_Fx, MDOCF_CHILD }, /* Fx */ |
{ NULL, TYPE_Fx }, /* Fx */ |
{ NULL, TYPE_Ms, MDOCF_CHILD }, /* Ms */ |
{ NULL, TYPE_Ms }, /* Ms */ |
{ NULL, 0, 0 }, /* No */ |
{ NULL, 0 }, /* No */ |
{ NULL, 0, 0 }, /* Ns */ |
{ NULL, 0 }, /* Ns */ |
{ NULL, TYPE_Nx, MDOCF_CHILD }, /* Nx */ |
{ NULL, TYPE_Nx }, /* Nx */ |
{ NULL, TYPE_Ox, MDOCF_CHILD }, /* Ox */ |
{ NULL, TYPE_Ox }, /* Ox */ |
{ NULL, 0, 0 }, /* Pc */ |
{ NULL, 0 }, /* Pc */ |
{ NULL, 0, 0 }, /* Pf */ |
{ NULL, 0 }, /* Pf */ |
{ NULL, 0, 0 }, /* Po */ |
{ NULL, 0 }, /* Po */ |
{ NULL, 0, 0 }, /* Pq */ |
{ NULL, 0 }, /* Pq */ |
{ NULL, 0, 0 }, /* Qc */ |
{ NULL, 0 }, /* Qc */ |
{ NULL, 0, 0 }, /* Ql */ |
{ NULL, 0 }, /* Ql */ |
{ NULL, 0, 0 }, /* Qo */ |
{ NULL, 0 }, /* Qo */ |
{ NULL, 0, 0 }, /* Qq */ |
{ NULL, 0 }, /* Qq */ |
{ NULL, 0, 0 }, /* Re */ |
{ NULL, 0 }, /* Re */ |
{ NULL, 0, 0 }, /* Rs */ |
{ NULL, 0 }, /* Rs */ |
{ NULL, 0, 0 }, /* Sc */ |
{ NULL, 0 }, /* Sc */ |
{ NULL, 0, 0 }, /* So */ |
{ NULL, 0 }, /* So */ |
{ NULL, 0, 0 }, /* Sq */ |
{ NULL, 0 }, /* Sq */ |
{ NULL, 0, 0 }, /* Sm */ |
{ NULL, 0 }, /* Sm */ |
{ NULL, 0, 0 }, /* Sx */ |
{ NULL, 0 }, /* Sx */ |
{ NULL, TYPE_Sy, MDOCF_CHILD }, /* Sy */ |
{ NULL, TYPE_Sy }, /* Sy */ |
{ NULL, TYPE_Tn, MDOCF_CHILD }, /* Tn */ |
{ NULL, TYPE_Tn }, /* Tn */ |
{ NULL, 0, 0 }, /* Ux */ |
{ NULL, 0 }, /* Ux */ |
{ NULL, 0, 0 }, /* Xc */ |
{ NULL, 0 }, /* Xc */ |
{ NULL, 0, 0 }, /* Xo */ |
{ NULL, 0 }, /* Xo */ |
{ pmdoc_head, TYPE_Fn, 0 }, /* Fo */ |
{ parse_mdoc_Fo, 0 }, /* Fo */ |
{ NULL, 0, 0 }, /* Fc */ |
{ NULL, 0 }, /* Fc */ |
{ NULL, 0, 0 }, /* Oo */ |
{ NULL, 0 }, /* Oo */ |
{ NULL, 0, 0 }, /* Oc */ |
{ NULL, 0 }, /* Oc */ |
{ NULL, 0, 0 }, /* Bk */ |
{ NULL, 0 }, /* Bk */ |
{ NULL, 0, 0 }, /* Ek */ |
{ NULL, 0 }, /* Ek */ |
{ NULL, 0, 0 }, /* Bt */ |
{ NULL, 0 }, /* Bt */ |
{ NULL, 0, 0 }, /* Hf */ |
{ NULL, 0 }, /* Hf */ |
{ NULL, 0, 0 }, /* Fr */ |
{ NULL, 0 }, /* Fr */ |
{ NULL, 0, 0 }, /* Ud */ |
{ NULL, 0 }, /* Ud */ |
{ NULL, TYPE_Lb, MDOCF_CHILD }, /* Lb */ |
{ NULL, TYPE_Lb }, /* Lb */ |
{ NULL, 0, 0 }, /* Lp */ |
{ NULL, 0 }, /* Lp */ |
{ NULL, TYPE_Lk, MDOCF_CHILD }, /* Lk */ |
{ NULL, TYPE_Lk }, /* Lk */ |
{ NULL, TYPE_Mt, MDOCF_CHILD }, /* Mt */ |
{ NULL, TYPE_Mt }, /* Mt */ |
{ NULL, 0, 0 }, /* Brq */ |
{ NULL, 0 }, /* Brq */ |
{ NULL, 0, 0 }, /* Bro */ |
{ NULL, 0 }, /* Bro */ |
{ NULL, 0, 0 }, /* Brc */ |
{ NULL, 0 }, /* Brc */ |
{ NULL, 0, 0 }, /* %C */ |
{ NULL, 0 }, /* %C */ |
{ NULL, 0, 0 }, /* Es */ |
{ NULL, 0 }, /* Es */ |
{ NULL, 0, 0 }, /* En */ |
{ NULL, 0 }, /* En */ |
{ NULL, TYPE_Dx, MDOCF_CHILD }, /* Dx */ |
{ NULL, TYPE_Dx }, /* Dx */ |
{ NULL, 0, 0 }, /* %Q */ |
{ NULL, 0 }, /* %Q */ |
{ NULL, 0, 0 }, /* br */ |
{ NULL, 0 }, /* br */ |
{ NULL, 0, 0 }, /* sp */ |
{ NULL, 0 }, /* sp */ |
{ NULL, 0, 0 }, /* %U */ |
{ NULL, 0 }, /* %U */ |
{ NULL, 0, 0 }, /* Ta */ |
{ NULL, 0 }, /* Ta */ |
|
{ NULL, 0 }, /* ll */ |
}; |
}; |
|
|
static const char *progname; |
|
static int use_all; /* Use all directories and files. */ |
|
static int verb; /* Output verbosity level. */ |
|
|
|
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 */ |
const char *path_arg; |
const char *dir; |
size_t j, sz; |
char *cp; |
int ch, i; |
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 */ |
|
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 */ |
|
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], '/'); |
memset(&conf, 0, sizeof(conf)); |
if (progname == NULL) |
memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *)); |
progname = argv[0]; |
|
else |
|
++progname; |
|
|
|
memset(&dirs, 0, sizeof(struct manpaths)); |
/* |
|
* 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) |
|
|
verb = 0; |
path_arg = NULL; |
use_all = 0; |
op = OP_DEFAULT; |
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, "ad:u:v"))) |
while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v"))) |
switch (ch) { |
switch (ch) { |
case ('a'): |
case 'a': |
use_all = 1; |
use_all = 1; |
break; |
break; |
case ('d'): |
case 'C': |
dir = optarg; |
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 (OP_CONFFILE == op && argc > 0) { |
info.flags = R_DUP; |
warnx("-C: Too many arguments"); |
|
goto usage; |
|
} |
|
|
mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL); |
exitcode = (int)MANDOCLEVEL_OK; |
|
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)); |
|
|
memset(&buf, 0, sizeof(struct buf)); |
if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) { |
memset(&dbuf, 0, sizeof(struct buf)); |
|
|
|
buf.size = dbuf.size = MANDOC_BUFSZ; |
/* |
|
* 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; |
|
|
buf.cp = mandoc_malloc(buf.size); |
if (dbopen(1)) { |
dbuf.cp = mandoc_malloc(dbuf.size); |
/* |
|
* The existing database is usable. Process |
|
* all files specified on the command-line. |
|
*/ |
|
use_all = 1; |
|
for (i = 0; i < argc; i++) |
|
filescan(argv[i]); |
|
if (OP_TEST != op) |
|
dbprune(); |
|
} else { |
|
/* |
|
* Database missing or corrupt. |
|
* Recreate from scratch. |
|
*/ |
|
exitcode = (int)MANDOCLEVEL_OK; |
|
op = OP_DEFAULT; |
|
if (0 == treescan()) |
|
goto out; |
|
if (0 == dbopen(0)) |
|
goto out; |
|
} |
|
if (OP_DELETE != op) |
|
mpages_merge(mp); |
|
dbclose(OP_DEFAULT == op ? 0 : 1); |
|
} else { |
|
/* |
|
* If we have arguments, use them as our manpaths. |
|
* If we don't, grok from manpath(1) or however else |
|
* manconf_parse() wants to do it. |
|
*/ |
|
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); |
|
|
flags = OP_NEW == op ? O_CREAT|O_TRUNC|O_RDWR : O_CREAT|O_RDWR; |
if (conf.manpath.sz == 0) { |
|
exitcode = (int)MANDOCLEVEL_BADARG; |
|
say("", "Empty manpath"); |
|
} |
|
|
if (OP_UPDATE == op || OP_DELETE == op) { |
/* |
ibuf[0] = fbuf[0] = '\0'; |
* First scan the tree rooted at a base directory, then |
|
* 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(fbuf, dir, MAXPATHLEN); |
if (j) { |
strlcat(fbuf, "/", MAXPATHLEN); |
mandoc_ohash_init(&mpages, 6, |
sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN); |
offsetof(struct mpage, inodev)); |
|
mandoc_ohash_init(&mlinks, 6, |
|
offsetof(struct mlink, file)); |
|
} |
|
|
strlcat(ibuf, dir, MAXPATHLEN); |
if ( ! set_basedir(conf.manpath.paths[j], argc > 0)) |
strlcat(ibuf, "/", MAXPATHLEN); |
continue; |
sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN); |
if (0 == treescan()) |
|
continue; |
|
if (0 == dbopen(0)) |
|
continue; |
|
|
if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) { |
mpages_merge(mp); |
fprintf(stderr, "%s: Path too long\n", dir); |
if (warnings && !nodb && |
exit((int)MANDOCLEVEL_BADARG); |
! (MPARSE_QUICK & mparse_options)) |
|
names_check(); |
|
dbclose(0); |
|
|
|
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: |
|
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); |
* Scan a directory tree rooted at "basedir" for manpages. |
exit((int)MANDOCLEVEL_SYSERR); |
* We use fts(), scanning directory parts along the way for clues to our |
} else if (NULL == idx) { |
* section and architecture. |
perror(ibuf); |
* |
exit((int)MANDOCLEVEL_SYSERR); |
* 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: accomodate for multi-language directories. |
|
*/ |
|
static int |
|
treescan(void) |
|
{ |
|
char buf[PATH_MAX]; |
|
FTS *f; |
|
FTSENT *ff; |
|
struct mlink *mlink; |
|
int dform, gzip; |
|
char *dsec, *arch, *fsec, *cp; |
|
const char *path; |
|
const char *argv[2]; |
|
|
if (verb > 2) { |
argv[0] = "."; |
printf("%s: Opened\n", fbuf); |
argv[1] = (char *)NULL; |
printf("%s: Opened\n", ibuf); |
|
} |
|
|
|
ofile_argbuild(argc, argv, &of); |
f = fts_open((char * const *)argv, |
if (NULL == of) |
FTS_PHYSICAL | FTS_NOCHDIR, NULL); |
goto out; |
if (f == NULL) { |
|
exitcode = (int)MANDOCLEVEL_SYSERR; |
|
say("", "&fts_open"); |
|
return 0; |
|
} |
|
|
of = of->first; |
dsec = arch = NULL; |
|
dform = FORM_NONE; |
|
|
index_prune(of, db, fbuf, idx, ibuf, |
while ((ff = fts_read(f)) != NULL) { |
&maxrec, &recs, &recsz, &reccur); |
path = ff->fts_path + 2; |
|
switch (ff->fts_info) { |
|
|
/* |
/* |
* Go to the root of the respective manual tree |
* Symbolic links require various sanity checks, |
* such that .so links work. In case of failure, |
* then get handled just like regular files. |
* just prod on, even though .so links won't work. |
|
*/ |
*/ |
|
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 (OP_UPDATE == op) { |
/* |
chdir(dir); |
* If we're a regular file, add an mlink by using the |
index_merge(of, mp, &dbuf, &buf, hash, |
* stored directory data and handling the filename. |
db, fbuf, idx, ibuf, |
*/ |
maxrec, recs, reccur); |
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; |
} |
} |
|
|
goto out; |
switch (ff->fts_level) { |
|
case 0: |
|
/* 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; |
|
} |
} |
} |
|
|
|
fts_close(f); |
|
return 1; |
|
} |
|
|
|
/* |
|
* 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; |
|
|
|
assert(use_all); |
|
|
|
if (0 == strncmp(file, "./", 2)) |
|
file += 2; |
|
|
/* |
/* |
* Configure the directories we're going to scan. |
* We have to do lstat(2) before realpath(3) loses |
* If we have command-line arguments, use them. |
* the information whether this is a symbolic link. |
* If not, we use man(1)'s method (see mandocdb.8). |
* 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; |
|
} |
|
|
if (argc > 0) { |
/* |
dirs.paths = mandoc_calloc(argc, sizeof(char *)); |
* We have to resolve the file name to the real path |
dirs.sz = argc; |
* in any case for the base directory check. |
for (i = 0; i < argc; i++) { |
*/ |
if (NULL == (cp = realpath(argv[i], pbuf))) { |
if (NULL == realpath(file, buf)) { |
perror(argv[i]); |
exitcode = (int)MANDOCLEVEL_BADARG; |
goto out; |
say(file, "&realpath"); |
} |
return; |
dirs.paths[i] = mandoc_strdup(cp); |
} |
} |
|
} else |
|
manpath_parse(&dirs, NULL, NULL); |
|
|
|
for (i = 0; i < dirs.sz; i++) { |
if (OP_TEST == op) |
ibuf[0] = fbuf[0] = '\0'; |
start = buf; |
|
else if (strstr(buf, basedir) == buf) |
|
start = buf + strlen(basedir); |
|
#ifdef HOMEBREWDIR |
|
else if (strstr(buf, HOMEBREWDIR) == buf) |
|
start = buf; |
|
#endif |
|
else { |
|
exitcode = (int)MANDOCLEVEL_BADARG; |
|
say("", "%s: outside base directory", buf); |
|
return; |
|
} |
|
|
strlcat(fbuf, dirs.paths[i], MAXPATHLEN); |
/* |
strlcat(fbuf, "/", MAXPATHLEN); |
* Now we are sure the file is inside our tree. |
sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN); |
* 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(ibuf, dirs.paths[i], MAXPATHLEN); |
mlink = mandoc_calloc(1, sizeof(struct mlink)); |
strlcat(ibuf, "/", MAXPATHLEN); |
mlink->dform = FORM_NONE; |
sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN); |
if (strlcpy(mlink->file, start, sizeof(mlink->file)) >= |
|
sizeof(mlink->file)) { |
|
say(start, "Filename too long"); |
|
free(mlink); |
|
return; |
|
} |
|
|
if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) { |
/* |
fprintf(stderr, "%s: Path too long\n", |
* First try to guess our directory structure. |
dirs.paths[i]); |
* If we find a separator, try to look for man* or cat*. |
exit((int)MANDOCLEVEL_BADARG); |
* 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 (db) |
start = p; |
(*db->close)(db); |
if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) { |
if (idx) |
*p++ = '\0'; |
(*idx->close)(idx); |
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 == idx) { |
} |
perror(ibuf); |
|
exit((int)MANDOCLEVEL_SYSERR); |
/* |
|
* Now try to parse the name. |
|
* 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); |
|
} |
|
|
|
static void |
|
mlink_add(struct mlink *mlink, const struct stat *st) |
|
{ |
|
struct inodev inodev; |
|
struct mpage *mpage; |
|
unsigned int slot; |
|
|
|
assert(NULL != mlink->file); |
|
|
|
mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : ""); |
|
mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : ""); |
|
mlink->name = mandoc_strdup(mlink->name ? mlink->name : ""); |
|
mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : ""); |
|
|
|
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; |
|
|
|
slot = ohash_qlookup(&mlinks, mlink->file); |
|
assert(NULL == ohash_find(&mlinks, slot)); |
|
ohash_insert(&mlinks, slot, mlink); |
|
|
|
memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */ |
|
inodev.st_ino = st->st_ino; |
|
inodev.st_dev = st->st_dev; |
|
slot = ohash_lookup_memory(&mpages, (char *)&inodev, |
|
sizeof(struct inodev), inodev.st_ino); |
|
mpage = ohash_find(&mpages, slot); |
|
if (NULL == mpage) { |
|
mpage = mandoc_calloc(1, sizeof(struct mpage)); |
|
mpage->inodev.st_ino = inodev.st_ino; |
|
mpage->inodev.st_dev = inodev.st_dev; |
|
ohash_insert(&mpages, slot, mpage); |
|
} else |
|
mlink->next = mpage->mlinks; |
|
mpage->mlinks = mlink; |
|
mlink->mpage = mpage; |
|
} |
|
|
|
static void |
|
mlink_free(struct mlink *mlink) |
|
{ |
|
|
|
free(mlink->dsec); |
|
free(mlink->arch); |
|
free(mlink->name); |
|
free(mlink->fsec); |
|
free(mlink); |
|
} |
|
|
|
static void |
|
mpages_free(void) |
|
{ |
|
struct mpage *mpage; |
|
struct mlink *mlink; |
|
unsigned int slot; |
|
|
|
mpage = ohash_first(&mpages, &slot); |
|
while (NULL != mpage) { |
|
while (NULL != (mlink = mpage->mlinks)) { |
|
mpage->mlinks = mlink->next; |
|
mlink_free(mlink); |
} |
} |
|
free(mpage->sec); |
|
free(mpage->arch); |
|
free(mpage->title); |
|
free(mpage->desc); |
|
free(mpage); |
|
mpage = ohash_next(&mpages, &slot); |
|
} |
|
} |
|
|
if (verb > 2) { |
/* |
printf("%s: Truncated\n", fbuf); |
* For each mlink to the mpage, check whether the path looks like |
printf("%s: Truncated\n", ibuf); |
* 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; |
|
|
|
mpage->form = FORM_CAT; |
|
prev = &mpage->mlinks; |
|
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; |
|
} |
|
} |
|
|
ofile_free(of); |
static void |
of = NULL; |
mlink_check(struct mpage *mpage, struct mlink *mlink) |
|
{ |
|
struct str *str; |
|
unsigned int slot; |
|
|
if ( ! ofile_dirbuild(dirs.paths[i], NULL, NULL, |
/* |
0, &of)) |
* Check whether the manual section given in a file |
exit((int)MANDOCLEVEL_SYSERR); |
* 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). |
|
*/ |
|
|
if (NULL == of) |
if (FORM_SRC == mpage->form && |
continue; |
strcasecmp(mpage->sec, mlink->dsec)) |
|
say(mlink->file, "Section \"%s\" manual in %s directory", |
|
mpage->sec, mlink->dsec); |
|
|
of = of->first; |
/* |
|
* 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. |
|
*/ |
|
|
/* |
if (strcasecmp(mpage->arch, mlink->arch)) |
* Go to the root of the respective manual tree |
say(mlink->file, "Architecture \"%s\" manual in " |
* such that .so links work. In case of failure, |
"\"%s\" directory", mpage->arch, mlink->arch); |
* just prod on, even though .so links won't work. |
|
*/ |
|
|
|
chdir(dirs.paths[i]); |
/* |
index_merge(of, mp, &dbuf, &buf, hash, db, fbuf, |
* XXX |
idx, ibuf, maxrec, recs, reccur); |
* parse_cat() doesn't set NAME_TITLE yet. |
} |
*/ |
|
|
out: |
if (FORM_CAT == mpage->form) |
if (db) |
return; |
(*db->close)(db); |
|
if (idx) |
|
(*idx->close)(idx); |
|
if (hash) |
|
(*hash->close)(hash); |
|
if (mp) |
|
mparse_free(mp); |
|
|
|
manpath_free(&dirs); |
/* |
ofile_free(of); |
* Check whether this mlink |
free(buf.cp); |
* appears as a name in the NAME section. |
free(dbuf.cp); |
*/ |
free(recs); |
|
|
|
return(MANDOCLEVEL_OK); |
slot = ohash_qlookup(&names, mlink->name); |
|
str = ohash_find(&names, slot); |
|
assert(NULL != str); |
|
if ( ! (NAME_TITLE & str->mask)) |
|
say(mlink->file, "Name missing in NAME section"); |
} |
} |
|
|
void |
/* |
index_merge(const struct of *of, struct mparse *mp, |
* Run through the files in the global vector "mpages" |
struct buf *dbuf, struct buf *buf, DB *hash, |
* and add them to the database specified in "basedir". |
DB *db, const char *dbf, DB *idx, const char *idxf, |
* |
recno_t maxrec, const recno_t *recs, size_t reccur) |
* 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 mparse *mp) |
{ |
{ |
recno_t rec; |
char any[] = "any"; |
int ch; |
struct mpage *mpage, *mpage_dest; |
DBT key, val; |
struct mlink *mlink, *mlink_dest; |
struct mdoc *mdoc; |
struct roff_man *man; |
struct man *man; |
char *sodest; |
const char *fn, *msec, *mtitle, *arch; |
char *cp; |
size_t sv; |
int fd; |
unsigned seq; |
unsigned int pslot; |
struct db_val vbuf; |
|
|
|
for (rec = 0; of; of = of->next) { |
if ( ! nodb) |
fn = of->fname; |
SQL_EXEC("BEGIN TRANSACTION"); |
|
|
/* |
mpage = ohash_first(&mpages, &pslot); |
* Reclaim an empty index record, if available. |
while (mpage != NULL) { |
*/ |
mlinks_undupe(mpage); |
|
if ((mlink = mpage->mlinks) == NULL) { |
|
mpage = ohash_next(&mpages, &pslot); |
|
continue; |
|
} |
|
|
if (reccur > 0) { |
name_mask = NAME_MASK; |
--reccur; |
mandoc_ohash_init(&names, 4, offsetof(struct str, key)); |
rec = recs[(int)reccur]; |
mandoc_ohash_init(&strings, 6, offsetof(struct str, key)); |
} else if (maxrec > 0) { |
|
rec = maxrec; |
|
maxrec = 0; |
|
} else |
|
rec++; |
|
|
|
mparse_reset(mp); |
mparse_reset(mp); |
hash_reset(&hash); |
|
mdoc = NULL; |
|
man = NULL; |
man = NULL; |
|
sodest = NULL; |
|
|
/* |
mparse_open(mp, &fd, mlink->file); |
* Try interpreting the file as mdoc(7) or man(7) |
if (fd == -1) { |
* source code, unless it is already known to be |
say(mlink->file, "&open"); |
* formatted. Fall back to formatted mode. |
goto nextpage; |
*/ |
|
|
|
if ((MANDOC_SRC & of->src_form || |
|
! (MANDOC_FORM & of->src_form)) && |
|
MANDOCLEVEL_FATAL > mparse_readfd(mp, -1, fn)) |
|
mparse_result(mp, &mdoc, &man); |
|
|
|
if (NULL != mdoc) { |
|
msec = mdoc_meta(mdoc)->msec; |
|
arch = mdoc_meta(mdoc)->arch; |
|
mtitle = mdoc_meta(mdoc)->title; |
|
} else if (NULL != man) { |
|
msec = man_meta(man)->msec; |
|
arch = NULL; |
|
mtitle = man_meta(man)->title; |
|
} else { |
|
msec = of->sec; |
|
arch = of->arch; |
|
mtitle = of->title; |
|
} |
} |
|
|
/* |
/* |
* By default, skip a file if the manual section |
* Interpret the file as mdoc(7) or man(7) source |
* and architecture given in the file disagree |
* code, unless it is known to be formatted. |
* with the directory where the file is located. |
|
*/ |
*/ |
|
if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) { |
if (0 == use_all) { |
mparse_readfd(mp, fd, mlink->file); |
assert(of->sec); |
mparse_result(mp, &man, &sodest); |
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; |
|
} |
} |
|
|
if (NULL == arch) |
if (sodest != NULL) { |
arch = ""; |
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) { |
|
|
/* |
/* The .so target exists. */ |
* 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. |
|
*/ |
|
|
|
assert(of->title); |
mpage_dest = mlink_dest->mpage; |
assert(mtitle); |
while (1) { |
|
mlink->mpage = mpage_dest; |
|
|
if (0 == strcasecmp(mtitle, of->title)) |
/* |
mtitle = of->title; |
* If the target was already |
else if (0 == use_all) |
* processed, add the links |
continue; |
* to the database now. |
|
* Otherwise, this will |
|
* happen when we come |
|
* to the target. |
|
*/ |
|
|
/* |
if (mpage_dest->pageid) |
* The index record value consists of a nil-terminated |
dbadd_mlink_name(mlink); |
* 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. |
|
*/ |
|
|
|
dbuf->len = 0; |
if (mlink->next == NULL) |
buf_append(dbuf, mdoc ? "mdoc" : (man ? "man" : "cat")); |
break; |
buf_appendb(dbuf, fn, strlen(fn) + 1); |
mlink = mlink->next; |
buf_appendb(dbuf, msec, strlen(msec) + 1); |
} |
buf_appendb(dbuf, mtitle, strlen(mtitle) + 1); |
|
buf_appendb(dbuf, arch, strlen(arch) + 1); |
|
|
|
sv = dbuf->len; |
/* Move all links to the target. */ |
|
|
/* Fix the record number in the btree value. */ |
mlink->next = mlink_dest->next; |
|
mlink_dest->next = mpage->mlinks; |
|
mpage->mlinks = NULL; |
|
} |
|
goto nextpage; |
|
} else if (man != NULL && man->macroset == MACROSET_MDOC) { |
|
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) { |
|
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); |
|
} |
|
putkey(mpage, mpage->sec, TYPE_sec); |
|
if (*mpage->arch != '\0') |
|
putkey(mpage, mpage->arch, TYPE_arch); |
|
|
if (mdoc) |
for ( ; mlink != NULL; mlink = mlink->next) { |
pmdoc_node(hash, buf, dbuf, |
if ('\0' != *mlink->dsec) |
mdoc_node(mdoc), mdoc_meta(mdoc)); |
putkey(mpage, mlink->dsec, TYPE_sec); |
else if (man) |
if ('\0' != *mlink->fsec) |
pman_node(hash, buf, dbuf, man_node(man)); |
putkey(mpage, mlink->fsec, TYPE_sec); |
|
putkey(mpage, '\0' == *mlink->arch ? |
|
any : mlink->arch, TYPE_arch); |
|
putkey(mpage, mlink->name, NAME_FILE); |
|
} |
|
|
|
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 |
else |
pformatted(hash, buf, dbuf, of); |
parse_cat(mpage, fd); |
|
if (mpage->desc == NULL) |
|
mpage->desc = mandoc_strdup(mpage->mlinks->name); |
|
|
/* |
if (warnings && !use_all) |
* Copy from the in-memory hashtable of pending keywords |
for (mlink = mpage->mlinks; mlink; |
* into the database. |
mlink = mlink->next) |
*/ |
mlink_check(mpage, mlink); |
|
|
vbuf.rec = htobe32(rec); |
dbadd(mpage); |
seq = R_FIRST; |
mlink = mpage->mlinks; |
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); |
|
val.data = &vbuf; |
|
dbt_put(db, dbf, &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. |
|
*/ |
|
|
|
if (dbuf->len == sv) |
nextpage: |
buf_appendb(dbuf, "", 1); |
ohash_delete(&strings); |
|
ohash_delete(&names); |
|
mpage = ohash_next(&mpages, &pslot); |
|
} |
|
|
key.data = &rec; |
if (0 == nodb) |
key.size = sizeof(recno_t); |
SQL_EXEC("END TRANSACTION"); |
|
} |
|
|
val.data = dbuf->cp; |
static void |
val.size = dbuf->len; |
names_check(void) |
|
{ |
|
sqlite3_stmt *stmt; |
|
const char *name, *sec, *arch, *key; |
|
|
if (verb) |
sqlite3_prepare_v2(db, |
printf("%s: Added index\n", fn); |
"SELECT name, sec, arch, key FROM (" |
|
"SELECT name AS key, pageid FROM names " |
|
"WHERE bits & ? AND NOT EXISTS (" |
|
"SELECT pageid FROM mlinks " |
|
"WHERE mlinks.pageid == names.pageid " |
|
"AND mlinks.name == names.name" |
|
")" |
|
") JOIN (" |
|
"SELECT sec, arch, name, pageid FROM mlinks " |
|
"GROUP BY pageid" |
|
") USING (pageid);", |
|
-1, &stmt, NULL); |
|
|
dbt_put(idx, idxf, &key, &val); |
if (sqlite3_bind_int64(stmt, 1, NAME_TITLE) != SQLITE_OK) |
|
say("", "%s", sqlite3_errmsg(db)); |
|
|
|
while (sqlite3_step(stmt) == SQLITE_ROW) { |
|
name = (const char *)sqlite3_column_text(stmt, 0); |
|
sec = (const char *)sqlite3_column_text(stmt, 1); |
|
arch = (const char *)sqlite3_column_text(stmt, 2); |
|
key = (const char *)sqlite3_column_text(stmt, 3); |
|
say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec, |
|
'\0' == *arch ? "" : "/", |
|
'\0' == *arch ? "" : arch, key); |
} |
} |
|
sqlite3_finalize(stmt); |
} |
} |
|
|
/* |
|
* 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, recno_t *maxrec, |
|
recno_t **recs, size_t *recsz, size_t *reccur) |
|
{ |
{ |
const struct of *of; |
FILE *stream; |
const char *fn, *cp; |
char *line, *p, *title; |
struct db_val *vbuf; |
size_t len, plen, titlesz; |
unsigned seq, sseq; |
|
DBT key, val; |
|
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) |
cp = val.data; |
close(fd); |
|
if (warnings) |
|
say(mpage->mlinks->file, "&fopen"); |
|
return; |
|
} |
|
|
/* Deleted records are zero-sized. Skip them. */ |
/* Skip to first blank line. */ |
|
|
if (0 == val.size) |
while (NULL != (line = fgetln(stream, &len))) |
goto cont; |
if ('\n' == *line) |
|
break; |
|
|
/* |
/* |
* Make sure we're sane. |
* Assume the first line that is not indented |
* Read past our mdoc/man/cat type to the next string, |
* is the first section header. Skip to it. |
* then make sure it's bounded by a NUL. |
*/ |
* Failing any of these, we go into our error handler. |
|
*/ |
|
|
|
if (NULL == (fn = memchr(cp, '\0', val.size))) |
while (NULL != (line = fgetln(stream, &len))) |
|
if ('\n' != *line && ' ' != *line) |
break; |
break; |
if (++fn - cp >= (int)val.size) |
|
break; |
|
if (NULL == memchr(fn, '\0', val.size - (fn - cp))) |
|
break; |
|
|
|
/* |
/* |
* Search for the file in those we care about. |
* Read up until the next section into a buffer. |
* XXX: build this into a tree. Too slow. |
* Strip the leading and trailing newline from each read line, |
*/ |
* appending a trailing space. |
|
* Ignore empty (whitespace-only) lines. |
|
*/ |
|
|
for (of = ofile; of; of = of->next) |
titlesz = 0; |
if (0 == strcmp(fn, of->fname)) |
title = NULL; |
break; |
|
|
|
if (NULL == of) |
while (NULL != (line = fgetln(stream, &len))) { |
|
if (' ' != *line || '\n' != line[len - 1]) |
|
break; |
|
while (len > 0 && isspace((unsigned char)*line)) { |
|
line++; |
|
len--; |
|
} |
|
if (1 == len) |
continue; |
continue; |
|
title = mandoc_realloc(title, titlesz + len); |
|
memcpy(title + titlesz, line, len); |
|
titlesz += len; |
|
title[titlesz - 1] = ' '; |
|
} |
|
|
/* |
/* |
* Search through the keyword database, throwing out all |
* If no page content can be found, or the input line |
* references to our file. |
* is already the next section header, or there is no |
*/ |
* trailing newline, reuse the page title as the page |
|
* description. |
|
*/ |
|
|
sseq = R_FIRST; |
if (NULL == title || '\0' == *title) { |
while (0 == (ch = (*db->seq)(db, &key, &val, sseq))) { |
if (warnings) |
sseq = R_NEXT; |
say(mpage->mlinks->file, |
if (sizeof(struct db_val) != val.size) |
"Cannot find NAME section"); |
break; |
fclose(stream); |
|
free(title); |
|
return; |
|
} |
|
|
vbuf = val.data; |
title = mandoc_realloc(title, titlesz + 1); |
if (*maxrec != betoh32(vbuf->rec)) |
title[titlesz] = '\0'; |
continue; |
|
|
|
if ((ch = (*db->del)(db, &key, R_CURSOR)) < 0) |
/* |
break; |
* Skip to the first dash. |
} |
* Use the remaining line as the description (no more than 70 |
|
* bytes). |
|
*/ |
|
|
if (ch < 0) { |
if (NULL != (p = strstr(title, "- "))) { |
perror(dbf); |
for (p += 2; ' ' == *p || '\b' == *p; p++) |
exit((int)MANDOCLEVEL_SYSERR); |
/* Skip to next word. */ ; |
} else if (1 != ch) { |
} else { |
fprintf(stderr, "%s: Corrupt database\n", dbf); |
if (warnings) |
exit((int)MANDOCLEVEL_SYSERR); |
say(mpage->mlinks->file, |
} |
"No dash in title line"); |
|
p = title; |
|
} |
|
|
if (verb) |
plen = strlen(p); |
printf("%s: Deleted index\n", fn); |
|
|
|
val.size = 0; |
/* Strip backspace-encoding from line. */ |
ch = (*idx->put)(idx, &key, &val, R_CURSOR); |
|
|
|
if (ch < 0) |
while (NULL != (line = memchr(p, '\b', plen))) { |
break; |
len = line - p; |
cont: |
if (0 == len) { |
if (*reccur >= *recsz) { |
memmove(line, line + 1, plen--); |
*recsz += MANDOC_SLOP; |
continue; |
*recs = mandoc_realloc |
|
(*recs, *recsz * sizeof(recno_t)); |
|
} |
} |
|
memmove(line - 1, line + 1, plen - len); |
(*recs)[(int)*reccur] = *maxrec; |
plen -= 2; |
(*reccur)++; |
|
} |
} |
|
|
if (ch < 0) { |
mpage->desc = mandoc_strdup(p); |
perror(idxf); |
fclose(stream); |
exit((int)MANDOCLEVEL_SYSERR); |
free(title); |
} else if (1 != ch) { |
|
fprintf(stderr, "%s: Corrupt index\n", idxf); |
|
exit((int)MANDOCLEVEL_SYSERR); |
|
} |
|
|
|
(*maxrec)++; |
|
} |
} |
|
|
/* |
/* |
* Grow the buffer (if necessary) and copy in a binary string. |
* Put a type/word pair into the word database for this particular file. |
*/ |
*/ |
static void |
static void |
buf_appendb(struct buf *buf, const void *cp, size_t sz) |
putkey(const struct mpage *mpage, char *value, uint64_t type) |
{ |
{ |
|
char *cp; |
|
|
/* Overshoot by MANDOC_BUFSZ. */ |
assert(NULL != value); |
|
if (TYPE_arch == type) |
while (buf->len + sz >= buf->size) { |
for (cp = value; *cp; cp++) |
buf->size = buf->len + sz + MANDOC_BUFSZ; |
if (isupper((unsigned char)*cp)) |
buf->cp = mandoc_realloc(buf->cp, buf->size); |
*cp = _tolower((unsigned char)*cp); |
} |
putkeys(mpage, value, strlen(value), type); |
|
|
memcpy(buf->cp + (int)buf->len, cp, sz); |
|
buf->len += sz; |
|
} |
} |
|
|
/* |
/* |
* Append a nil-terminated string to the buffer. |
* Grok all nodes at or below a certain mdoc node into putkey(). |
* 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) |
putmdockey(const struct mpage *mpage, |
|
const struct roff_node *n, uint64_t m) |
{ |
{ |
|
|
|
for ( ; NULL != n; n = n->next) { |
|
if (NULL != n->child) |
|
putmdockey(mpage, n->child, m); |
|
if (n->type == ROFFT_TEXT) |
|
putkey(mpage, n->string, m); |
|
} |
|
} |
|
|
|
static void |
|
parse_man(struct mpage *mpage, const struct roff_meta *meta, |
|
const struct roff_node *n) |
|
{ |
|
const struct roff_node *head, *body; |
|
char *start, *title; |
|
char byte; |
size_t sz; |
size_t sz; |
|
|
if (0 == (sz = strlen(cp))) |
if (NULL == n) |
return; |
return; |
|
|
if (buf->len) |
/* |
buf->cp[(int)buf->len - 1] = ' '; |
* 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. |
|
*/ |
|
|
buf_appendb(buf, cp, sz + 1); |
if (n->type == ROFFT_BODY && n->tok == MAN_SH) { |
} |
body = n; |
|
assert(body->parent); |
|
if (NULL != (head = body->parent->head) && |
|
1 == head->nchild && |
|
NULL != (head = (head->child)) && |
|
head->type == ROFFT_TEXT && |
|
0 == strcmp(head->string, "NAME") && |
|
NULL != body->child) { |
|
|
/* |
/* |
* Recursively add all text from a given node. |
* Suck the entire NAME section into memory. |
* This is optimised for general mdoc nodes in this context, which do |
* Yes, we might run away. |
* not consist of subexpressions and having a recursive call for n->next |
* But too many manuals have big, spread-out |
* would be wasteful. |
* NAME sections over many lines. |
* 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 |
|
buf_appendmdoc(struct buf *buf, const struct mdoc_node *n, int f) |
|
{ |
|
|
|
for ( ; n; n = n->next) { |
title = NULL; |
if (n->child) |
deroff(&title, body); |
buf_appendmdoc(buf, n->child, f); |
if (NULL == title) |
|
return; |
|
|
if (MDOC_TEXT == n->type && f) { |
/* |
f = 0; |
* Go through a special heuristic dance here. |
buf_appendb(buf, n->string, |
* Conventionally, one or more manual names are |
strlen(n->string) + 1); |
* comma-specified prior to a whitespace, then a |
} else if (MDOC_TEXT == n->type) |
* dash, then a description. Try to puzzle out |
buf_append(buf, n->string); |
* the name parts here. |
|
*/ |
|
|
} |
start = title; |
} |
for ( ;; ) { |
|
sz = strcspn(start, " ,"); |
|
if ('\0' == start[sz]) |
|
break; |
|
|
static void |
byte = start[sz]; |
hash_reset(DB **db) |
start[sz] = '\0'; |
{ |
|
DB *hash; |
|
|
|
if (NULL != (hash = *db)) |
/* |
(*hash->close)(hash); |
* Assume a stray trailing comma in the |
|
* name list if a name begins with a dash. |
|
*/ |
|
|
*db = dbopen(NULL, O_CREAT|O_RDWR, 0644, DB_HASH, NULL); |
if ('-' == start[0] || |
if (NULL == *db) { |
('\\' == start[0] && '-' == start[1])) |
perror("hash"); |
break; |
exit((int)MANDOCLEVEL_SYSERR); |
|
} |
|
} |
|
|
|
/* ARGSUSED */ |
putkey(mpage, start, NAME_TITLE); |
static int |
if ( ! (mpage->name_head_done || |
pmdoc_head(MDOC_ARGS) |
strcasecmp(start, meta->title))) { |
{ |
putkey(mpage, start, NAME_HEAD); |
|
mpage->name_head_done = 1; |
|
} |
|
|
return(MDOC_HEAD == n->type); |
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); |
|
} |
} |
} |
|
|
/* ARGSUSED */ |
static void |
static int |
parse_mdoc(struct mpage *mpage, const struct roff_meta *meta, |
pmdoc_body(MDOC_ARGS) |
const struct roff_node *n) |
{ |
{ |
|
|
return(MDOC_BODY == n->type); |
assert(NULL != n); |
|
for (n = n->child; NULL != n; n = n->next) { |
|
switch (n->type) { |
|
case ROFFT_ELEM: |
|
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); |
|
break; |
|
default: |
|
assert(n->type != ROFFT_ROOT); |
|
continue; |
|
} |
|
if (NULL != n->child) |
|
parse_mdoc(mpage, meta, n); |
|
} |
} |
} |
|
|
/* ARGSUSED */ |
|
static int |
static int |
pmdoc_Fd(MDOC_ARGS) |
parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta, |
|
const struct roff_node *n) |
{ |
{ |
const char *start, *end; |
char *start, *end; |
size_t sz; |
size_t sz; |
|
|
if (SEC_SYNOPSIS != n->sec) |
if (SEC_SYNOPSIS != n->sec || |
return(0); |
NULL == (n = n->child) || |
if (NULL == (n = n->child) || MDOC_TEXT != n->type) |
n->type != ROFFT_TEXT) |
return(0); |
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(0); |
return 0; |
|
|
if (NULL == (n = n->next) || MDOC_TEXT != n->type) |
if ((n = n->next) == NULL || n->type != ROFFT_TEXT) |
return(0); |
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 916 pmdoc_Fd(MDOC_ARGS) |
|
Line 1594 pmdoc_Fd(MDOC_ARGS) |
|
start++; |
start++; |
|
|
if (0 == (sz = strlen(start))) |
if (0 == (sz = strlen(start))) |
return(0); |
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); |
|
return(1); |
|
} |
} |
|
|
/* ARGSUSED */ |
static void |
static int |
parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n) |
pmdoc_In(MDOC_ARGS) |
|
{ |
{ |
|
char *cp; |
|
size_t sz; |
|
|
if (NULL == n->child || MDOC_TEXT != n->child->type) |
if (n->type != ROFFT_TEXT) |
return(0); |
return; |
|
|
buf_append(buf, n->child->string); |
/* Skip function pointer punctuation. */ |
return(1); |
|
|
cp = n->string; |
|
while (*cp == '(' || *cp == '*') |
|
cp++; |
|
sz = strcspn(cp, "()"); |
|
|
|
putkeys(mpage, cp, sz, TYPE_Fn); |
|
if (n->sec == SEC_SYNOPSIS) |
|
putkeys(mpage, cp, sz, NAME_SYN); |
} |
} |
|
|
/* ARGSUSED */ |
|
static int |
static int |
pmdoc_Fn(MDOC_ARGS) |
parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta, |
|
const struct roff_node *n) |
{ |
{ |
struct mdoc_node *nn; |
|
const char *cp; |
|
|
|
nn = n->child; |
if (n->child == NULL) |
|
return 0; |
|
|
if (NULL == nn || MDOC_TEXT != nn->type) |
parse_mdoc_fname(mpage, n->child); |
return(0); |
|
|
|
/* .Fn "struct type *name" "char *arg" */ |
for (n = n->child->next; n != NULL; n = n->next) |
|
if (n->type == ROFFT_TEXT) |
|
putkey(mpage, n->string, TYPE_Fa); |
|
|
cp = strrchr(nn->string, ' '); |
return 0; |
if (NULL == cp) |
} |
cp = nn->string; |
|
|
|
/* Strip away pointer symbol. */ |
static int |
|
parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta, |
|
const struct roff_node *n) |
|
{ |
|
|
while ('*' == *cp) |
if (n->type != ROFFT_HEAD) |
cp++; |
return 1; |
|
|
/* Store the function name. */ |
if (n->child != NULL) |
|
parse_mdoc_fname(mpage, n->child); |
|
|
buf_append(buf, cp); |
return 0; |
hash_put(hash, buf, TYPE_Fn); |
} |
|
|
/* Store the function type. */ |
static int |
|
parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta, |
|
const struct roff_node *n) |
|
{ |
|
char *cp; |
|
|
if (nn->string < cp) { |
if (NULL == (n = n->child)) |
buf->len = 0; |
return 0; |
buf_appendb(buf, nn->string, cp - nn->string); |
|
buf_appendb(buf, "", 1); |
|
hash_put(hash, buf, TYPE_Ft); |
|
} |
|
|
|
/* Store the arguments. */ |
if (NULL == n->next) { |
|
putkey(mpage, n->string, TYPE_Xr); |
for (nn = nn->next; nn; nn = nn->next) { |
return 0; |
if (MDOC_TEXT != nn->type) |
|
continue; |
|
buf->len = 0; |
|
buf_append(buf, nn->string); |
|
hash_put(hash, buf, TYPE_Fa); |
|
} |
} |
|
|
return(0); |
mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string); |
|
putkey(mpage, cp, TYPE_Xr); |
|
free(cp); |
|
return 0; |
} |
} |
|
|
/* ARGSUSED */ |
|
static int |
static int |
pmdoc_St(MDOC_ARGS) |
parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta, |
|
const struct roff_node *n) |
{ |
{ |
|
|
if (NULL == n->child || MDOC_TEXT != n->child->type) |
if (n->type == ROFFT_BODY) |
return(0); |
deroff(&mpage->desc, n); |
|
return 0; |
buf_append(buf, n->child->string); |
|
return(1); |
|
} |
} |
|
|
/* ARGSUSED */ |
|
static int |
static int |
pmdoc_Xr(MDOC_ARGS) |
parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta, |
|
const struct roff_node *n) |
{ |
{ |
|
|
if (NULL == (n = n->child)) |
if (SEC_NAME == n->sec) |
return(0); |
putmdockey(mpage, n->child, NAME_TITLE); |
|
else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) { |
buf_appendb(buf, n->string, strlen(n->string)); |
if (n->child == NULL) |
|
putkey(mpage, meta->name, NAME_SYN); |
if (NULL != (n = n->next)) { |
else |
buf_appendb(buf, ".", 1); |
putmdockey(mpage, n->child, NAME_SYN); |
buf_appendb(buf, n->string, strlen(n->string) + 1); |
} |
} else |
if ( ! (mpage->name_head_done || |
buf_appendb(buf, ".", 2); |
n->child == NULL || n->child->string == NULL || |
|
strcasecmp(n->child->string, meta->title))) { |
return(1); |
putkey(mpage, n->child->string, ROFFT_HEAD); |
|
mpage->name_head_done = 1; |
|
} |
|
return 0; |
} |
} |
|
|
/* ARGSUSED */ |
|
static int |
static int |
pmdoc_Nd(MDOC_ARGS) |
parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta, |
|
const struct roff_node *n) |
{ |
{ |
|
|
if (MDOC_BODY != n->type) |
return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD; |
return(0); |
|
|
|
buf_appendmdoc(dbuf, n->child, 1); |
|
return(1); |
|
} |
} |
|
|
/* ARGSUSED */ |
|
static int |
static int |
pmdoc_Nm(MDOC_ARGS) |
parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta, |
|
const struct roff_node *n) |
{ |
{ |
|
|
if (SEC_NAME == n->sec) |
return n->type == ROFFT_HEAD; |
return(1); |
|
else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type) |
|
return(0); |
|
|
|
if (NULL == n->child) |
|
buf_append(buf, m->name); |
|
|
|
return(1); |
|
} |
} |
|
|
/* ARGSUSED */ |
|
static int |
static int |
pmdoc_Sh(MDOC_ARGS) |
parse_mdoc_body(struct mpage *mpage, const struct roff_meta *meta, |
|
const struct roff_node *n) |
{ |
{ |
|
|
return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type); |
return n->type == ROFFT_BODY; |
} |
} |
|
|
|
/* |
|
* 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=%d", sz, cp, v); |
|
} else { |
|
htab = &strings; |
|
if (debug > 1) |
|
for (i = 0; i < mansearch_keymax; i++) |
|
if ((uint64_t)1 << i & v) |
|
say(mpage->mlinks->file, |
|
"Adding key %s=%*s", |
|
mansearch_keynames[i], 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): |
|
buf->len = 0; |
|
|
|
|
while (scp < src + *psz) { |
|
|
|
/* Leave normal characters unchanged. */ |
|
|
|
if (strchr(res, *scp) == NULL) { |
|
if (dst != NULL) |
|
dst[dsz++] = *scp; |
|
scp++; |
|
continue; |
|
} |
|
|
/* |
/* |
* Both NULL handlers and handlers returning true |
* Found something that requires replacing, |
* request using the data. Only skip the element |
* make sure we have a destination buffer. |
* when the handler returns false. |
|
*/ |
*/ |
|
|
if (NULL != mdocs[n->tok].fp && |
if (dst == NULL) { |
0 == (*mdocs[n->tok].fp)(hash, buf, dbuf, n, m)) |
dst = mandoc_malloc(ssz + 1); |
|
dsz = scp - src; |
|
memcpy(dst, src, dsz); |
|
} |
|
|
|
/* Handle single-char special characters. */ |
|
|
|
switch (*scp) { |
|
case '\\': |
break; |
break; |
|
case '\t': |
|
case ASCII_NBRSP: |
|
dst[dsz++] = ' '; |
|
scp++; |
|
continue; |
|
case ASCII_HYPH: |
|
dst[dsz++] = '-'; |
|
/* FALLTHROUGH */ |
|
case ASCII_BREAK: |
|
scp++; |
|
continue; |
|
default: |
|
abort(); |
|
} |
|
|
/* |
/* |
* For many macros, use the text from all children. |
* Found an escape sequence. |
* Set zero flags for macros not needing this. |
* Read past the slash, then parse it. |
* In that case, the handler must fill the buffer. |
* Ignore everything except characters. |
*/ |
*/ |
|
|
if (MDOCF_CHILD & mdocs[n->tok].flags) |
scp++; |
buf_appendmdoc(buf, n->child, 0); |
if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL) |
|
continue; |
|
|
/* |
/* |
* Cover the most common case: |
* Render the special character |
* Automatically stage one string per element. |
* as either UTF-8 or ASCII. |
* Set a zero mask for macros not needing this. |
|
* Additional staging can be done in the handler. |
|
*/ |
*/ |
|
|
if (mdocs[n->tok].mask) |
if (write_utf8) { |
hash_put(hash, buf, mdocs[n->tok].mask); |
unicode = mchars_spec2cp(seq, seqlen); |
break; |
if (unicode <= 0) |
default: |
continue; |
break; |
addsz = utf8(unicode, utfbuf); |
|
if (addsz == 0) |
|
continue; |
|
addcp = utfbuf; |
|
} else { |
|
addcp = mchars_spec2str(seq, seqlen, &addsz); |
|
if (addcp == NULL) |
|
continue; |
|
if (*addcp == ASCII_NBRSP) { |
|
addcp = " "; |
|
addsz = 1; |
|
} |
|
} |
|
|
|
/* Copy the rendered glyph into the stream. */ |
|
|
|
ssz += addsz; |
|
dst = mandoc_realloc(dst, ssz + 1); |
|
memcpy(dst + dsz, addcp, addsz); |
|
dsz += addsz; |
} |
} |
|
if (dst != NULL) { |
|
*public = dst; |
|
*psz = dsz; |
|
} |
|
|
pmdoc_node(hash, buf, dbuf, n->child, m); |
/* Trim trailing whitespace and NUL-terminate. */ |
pmdoc_node(hash, buf, dbuf, n->next, m); |
|
|
while (*psz > 0 && (*public)[*psz - 1] == ' ') |
|
--*psz; |
|
if (dst != NULL) { |
|
(*public)[*psz] = '\0'; |
|
return 1; |
|
} else |
|
return 0; |
} |
} |
|
|
static int |
static void |
pman_node(MAN_ARGS) |
dbadd_mlink(const struct mlink *mlink) |
{ |
{ |
const struct man_node *head, *body; |
size_t i; |
const char *start, *sv; |
|
size_t sz; |
|
|
|
if (NULL == n) |
i = 1; |
return(0); |
SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec); |
|
SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch); |
|
SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name); |
|
SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid); |
|
SQL_STEP(stmts[STMT_INSERT_LINK]); |
|
sqlite3_reset(stmts[STMT_INSERT_LINK]); |
|
} |
|
|
/* |
static void |
* We're only searching for one thing: the first text child in |
dbadd_mlink_name(const struct mlink *mlink) |
* 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 |
uint64_t bits; |
* the correct section or not. |
size_t i; |
*/ |
|
|
|
if (MAN_BODY == n->type && MAN_SH == n->tok) { |
dbadd_mlink(mlink); |
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); |
i = 1; |
start = sv = body->string; |
SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid); |
|
bits = NAME_FILE & NAME_MASK; |
|
if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) { |
|
bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0); |
|
sqlite3_reset(stmts[STMT_SELECT_NAME]); |
|
} |
|
|
/* |
i = 1; |
* Go through a special heuristic dance here. |
SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits); |
* This is why -man manuals are great! |
SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name); |
* (I'm being sarcastic: my eyes are bleeding.) |
SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid); |
* Conventionally, one or more manual names are |
SQL_STEP(stmts[STMT_INSERT_NAME]); |
* comma-specified prior to a whitespace, then a |
sqlite3_reset(stmts[STMT_INSERT_NAME]); |
* dash, then a description. Try to puzzle out |
} |
* the name parts here. |
|
*/ |
|
|
|
for ( ;; ) { |
/* |
sz = strcspn(start, " ,"); |
* Flush the current page's terms (and their bits) into the database. |
if ('\0' == start[(int)sz]) |
* Wrap the entire set of additions in a transaction to make sqlite be a |
break; |
* little faster. |
|
* Also, handle escape sequences at the last possible moment. |
|
*/ |
|
static void |
|
dbadd(struct mpage *mpage) |
|
{ |
|
struct mlink *mlink; |
|
struct str *key; |
|
char *cp; |
|
size_t i; |
|
unsigned int slot; |
|
int mustfree; |
|
|
buf->len = 0; |
mlink = mpage->mlinks; |
buf_appendb(buf, start, sz); |
|
buf_appendb(buf, "", 1); |
|
|
|
hash_put(hash, buf, TYPE_Nm); |
if (nodb) { |
|
for (key = ohash_first(&names, &slot); NULL != key; |
if (' ' == start[(int)sz]) { |
key = ohash_next(&names, &slot)) |
start += (int)sz + 1; |
free(key); |
break; |
for (key = ohash_first(&strings, &slot); NULL != key; |
} |
key = ohash_next(&strings, &slot)) |
|
free(key); |
assert(',' == start[(int)sz]); |
if (0 == debug) |
start += (int)sz + 1; |
return; |
while (' ' == *start) |
while (NULL != mlink) { |
start++; |
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; |
|
} |
|
|
buf->len = 0; |
if (debug) |
|
say(mlink->file, "Adding to database"); |
|
|
if (sv == start) { |
cp = mpage->desc; |
buf_append(buf, start); |
i = strlen(cp); |
return(1); |
mustfree = render_string(&cp, &i); |
} |
i = 1; |
|
SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp); |
|
SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form); |
|
SQL_STEP(stmts[STMT_INSERT_PAGE]); |
|
mpage->pageid = sqlite3_last_insert_rowid(db); |
|
sqlite3_reset(stmts[STMT_INSERT_PAGE]); |
|
if (mustfree) |
|
free(cp); |
|
|
while (' ' == *start) |
while (NULL != mlink) { |
start++; |
dbadd_mlink(mlink); |
|
mlink = mlink->next; |
|
} |
|
mlink = mpage->mlinks; |
|
|
if (0 == strncmp(start, "-", 1)) |
for (key = ohash_first(&names, &slot); NULL != key; |
start += 1; |
key = ohash_next(&names, &slot)) { |
else if (0 == strncmp(start, "\\-", 2)) |
assert(key->mpage == mpage); |
start += 2; |
i = 1; |
else if (0 == strncmp(start, "\\(en", 4)) |
SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask); |
start += 4; |
SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key); |
else if (0 == strncmp(start, "\\(em", 4)) |
SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid); |
start += 4; |
SQL_STEP(stmts[STMT_INSERT_NAME]); |
|
sqlite3_reset(stmts[STMT_INSERT_NAME]); |
|
free(key); |
|
} |
|
for (key = ohash_first(&strings, &slot); NULL != key; |
|
key = ohash_next(&strings, &slot)) { |
|
assert(key->mpage == mpage); |
|
i = 1; |
|
SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask); |
|
SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key); |
|
SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid); |
|
SQL_STEP(stmts[STMT_INSERT_KEY]); |
|
sqlite3_reset(stmts[STMT_INSERT_KEY]); |
|
free(key); |
|
} |
|
} |
|
|
while (' ' == *start) |
static void |
start++; |
dbprune(void) |
|
{ |
|
struct mpage *mpage; |
|
struct mlink *mlink; |
|
size_t i; |
|
unsigned int slot; |
|
|
sz = strlen(start) + 1; |
if (0 == nodb) |
buf_appendb(dbuf, start, sz); |
SQL_EXEC("BEGIN TRANSACTION"); |
buf_appendb(buf, start, sz); |
|
|
|
hash_put(hash, buf, TYPE_Nd); |
for (mpage = ohash_first(&mpages, &slot); NULL != mpage; |
|
mpage = ohash_next(&mpages, &slot)) { |
|
mlink = mpage->mlinks; |
|
if (debug) |
|
say(mlink->file, "Deleting from database"); |
|
if (nodb) |
|
continue; |
|
for ( ; NULL != mlink; mlink = mlink->next) { |
|
i = 1; |
|
SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE], |
|
i, mlink->dsec); |
|
SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE], |
|
i, mlink->arch); |
|
SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE], |
|
i, mlink->name); |
|
SQL_STEP(stmts[STMT_DELETE_PAGE]); |
|
sqlite3_reset(stmts[STMT_DELETE_PAGE]); |
} |
} |
} |
} |
|
|
for (n = n->child; n; n = n->next) |
if (0 == nodb) |
if (pman_node(hash, buf, dbuf, n)) |
SQL_EXEC("END TRANSACTION"); |
return(1); |
|
|
|
return(0); |
|
} |
} |
|
|
/* |
/* |
* Parse a formatted manual page. |
* Close an existing database and its prepared statements. |
* By necessity, this involves rather crude guesswork. |
* If "real" is not set, rename the temporary file into the real one. |
*/ |
*/ |
static void |
static void |
pformatted(DB *hash, struct buf *buf, struct buf *dbuf, |
dbclose(int real) |
const struct of *of) |
|
{ |
{ |
FILE *stream; |
size_t i; |
char *line, *p; |
int status; |
size_t len, plen; |
pid_t child; |
|
|
if (NULL == (stream = fopen(of->fname, "r"))) { |
if (nodb) |
perror(of->fname); |
|
return; |
return; |
|
|
|
for (i = 0; i < STMT__MAX; i++) { |
|
sqlite3_finalize(stmts[i]); |
|
stmts[i] = NULL; |
} |
} |
|
|
/* |
sqlite3_close(db); |
* Always use the title derived from the filename up front, |
db = NULL; |
* do not even try to find it in the file. This also makes |
|
* sure we don't end up with an orphan index record, even if |
|
* the file content turns out to be completely unintelligible. |
|
*/ |
|
|
|
buf->len = 0; |
if (real) |
buf_append(buf, of->title); |
return; |
hash_put(hash, buf, TYPE_Nm); |
|
|
|
/* Skip to first blank line. */ |
if ('\0' == *tempfilename) { |
|
if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) { |
while (NULL != (line = fgetln(stream, &len))) |
exitcode = (int)MANDOCLEVEL_SYSERR; |
if ('\n' == *line) |
say(MANDOC_DB, "&rename"); |
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; |
|
|
|
/* |
|
* 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. |
|
*/ |
|
|
|
line = fgetln(stream, &len); |
|
if (NULL == line || ' ' != *line || '\n' != line[(int)len - 1]) { |
|
buf_appendb(dbuf, buf->cp, buf->size); |
|
hash_put(hash, buf, TYPE_Nd); |
|
fclose(stream); |
|
return; |
return; |
} |
} |
|
|
line[(int)--len] = '\0'; |
switch (child = fork()) { |
|
case -1: |
/* |
exitcode = (int)MANDOCLEVEL_SYSERR; |
* Skip to the first dash. |
say("", "&fork cmp"); |
* Use the remaining line as the description (no more than 70 |
return; |
* bytes). |
case 0: |
*/ |
execlp("cmp", "cmp", "-s", |
|
tempfilename, MANDOC_DB, (char *)NULL); |
if (NULL != (p = strstr(line, "- "))) { |
say("", "&exec cmp"); |
for (p += 2; ' ' == *p || '\b' == *p; p++) |
exit(0); |
/* Skip to next word. */ ; |
default: |
} else |
break; |
p = line; |
|
|
|
if ((plen = strlen(p)) > 70) { |
|
plen = 70; |
|
p[plen] = '\0'; |
|
} |
} |
|
if (-1 == waitpid(child, &status, 0)) { |
/* Strip backspace-encoding from line. */ |
exitcode = (int)MANDOCLEVEL_SYSERR; |
|
say("", "&wait cmp"); |
while (NULL != (line = memchr(p, '\b', plen))) { |
} else if (WIFSIGNALED(status)) { |
len = line - p; |
exitcode = (int)MANDOCLEVEL_SYSERR; |
if (0 == len) { |
say("", "cmp died from signal %d", WTERMSIG(status)); |
memmove(line, line + 1, plen--); |
} else if (WEXITSTATUS(status)) { |
continue; |
exitcode = (int)MANDOCLEVEL_SYSERR; |
} |
say(MANDOC_DB, |
memmove(line - 1, line + 1, plen - len); |
"Data changed, but cannot replace database"); |
plen -= 2; |
|
} |
} |
|
|
buf_appendb(dbuf, p, plen + 1); |
*strrchr(tempfilename, '/') = '\0'; |
buf->len = 0; |
switch (child = fork()) { |
buf_appendb(buf, p, plen + 1); |
case -1: |
hash_put(hash, buf, TYPE_Nd); |
exitcode = (int)MANDOCLEVEL_SYSERR; |
fclose(stream); |
say("", "&fork rm"); |
|
return; |
|
case 0: |
|
execlp("rm", "rm", "-rf", tempfilename, (char *)NULL); |
|
say("", "&exec rm"); |
|
exit((int)MANDOCLEVEL_SYSERR); |
|
default: |
|
break; |
|
} |
|
if (-1 == waitpid(child, &status, 0)) { |
|
exitcode = (int)MANDOCLEVEL_SYSERR; |
|
say("", "&wait rm"); |
|
} else if (WIFSIGNALED(status) || WEXITSTATUS(status)) { |
|
exitcode = (int)MANDOCLEVEL_SYSERR; |
|
say("", "%s: Cannot remove temporary directory", |
|
tempfilename); |
|
} |
} |
} |
|
|
static void |
/* |
ofile_argbuild(int argc, char *argv[], struct of **of) |
* This is straightforward stuff. |
|
* Open a database connection to a "temporary" database, then open a set |
|
* of prepared statements we'll use over and over again. |
|
* If "real" is set, we use the existing database; if not, we truncate a |
|
* temporary one. |
|
* Must be matched by dbclose(). |
|
*/ |
|
static int |
|
dbopen(int real) |
{ |
{ |
char buf[MAXPATHLEN]; |
const char *sql; |
char *sec, *arch, *title, *p; |
int rc, ofl; |
int i, src_form; |
|
struct of *nof; |
|
|
|
for (i = 0; i < argc; i++) { |
if (nodb) |
|
return 1; |
|
|
/* |
*tempfilename = '\0'; |
* Try to infer the manual section, architecture and |
ofl = SQLITE_OPEN_READWRITE; |
* page title from the path, assuming it looks like |
|
* man*[/<arch>]/<title>.<section> or |
|
* cat<section>[/<arch>]/<title>.0 |
|
*/ |
|
|
|
if (strlcpy(buf, argv[i], sizeof(buf)) >= sizeof(buf)) { |
if (real) { |
fprintf(stderr, "%s: Path too long\n", argv[i]); |
rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL); |
continue; |
if (SQLITE_OK != rc) { |
|
exitcode = (int)MANDOCLEVEL_SYSERR; |
|
if (SQLITE_CANTOPEN != rc) |
|
say(MANDOC_DB, "%s", sqlite3_errstr(rc)); |
|
return 0; |
} |
} |
sec = arch = title = NULL; |
goto prepare_statements; |
src_form = 0; |
} |
p = strrchr(buf, '\0'); |
|
while (p-- > buf) { |
|
if (NULL == sec && '.' == *p) { |
|
sec = p + 1; |
|
*p = '\0'; |
|
if ('0' == *sec) |
|
src_form |= MANDOC_FORM; |
|
else if ('1' <= *sec && '9' >= *sec) |
|
src_form |= MANDOC_SRC; |
|
continue; |
|
} |
|
if ('/' != *p) |
|
continue; |
|
if (NULL == title) { |
|
title = p + 1; |
|
*p = '\0'; |
|
continue; |
|
} |
|
if (0 == strncmp("man", p + 1, 3)) |
|
src_form |= MANDOC_SRC; |
|
else if (0 == strncmp("cat", p + 1, 3)) |
|
src_form |= MANDOC_FORM; |
|
else |
|
arch = p + 1; |
|
break; |
|
} |
|
if (NULL == title) |
|
title = buf; |
|
|
|
/* |
ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE; |
* Build the file structure. |
|
*/ |
|
|
|
nof = mandoc_calloc(1, sizeof(struct of)); |
remove(MANDOC_DB "~"); |
nof->fname = mandoc_strdup(argv[i]); |
rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL); |
if (NULL != sec) |
if (SQLITE_OK == rc) |
nof->sec = mandoc_strdup(sec); |
goto create_tables; |
if (NULL != arch) |
if (MPARSE_QUICK & mparse_options) { |
nof->arch = mandoc_strdup(arch); |
exitcode = (int)MANDOCLEVEL_SYSERR; |
nof->title = mandoc_strdup(title); |
say(MANDOC_DB "~", "%s", sqlite3_errstr(rc)); |
nof->src_form = src_form; |
return 0; |
|
} |
|
|
/* |
(void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX", |
* Add the structure to the list. |
sizeof(tempfilename)); |
*/ |
if (NULL == mkdtemp(tempfilename)) { |
|
exitcode = (int)MANDOCLEVEL_SYSERR; |
if (verb > 2) |
say("", "&%s", tempfilename); |
printf("%s: Scheduling\n", argv[i]); |
return 0; |
if (NULL == *of) { |
|
*of = nof; |
|
(*of)->first = nof; |
|
} else { |
|
nof->first = (*of)->first; |
|
(*of)->next = nof; |
|
*of = nof; |
|
} |
|
} |
} |
} |
(void)strlcat(tempfilename, "/" MANDOC_DB, |
|
sizeof(tempfilename)); |
|
rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL); |
|
if (SQLITE_OK != rc) { |
|
exitcode = (int)MANDOCLEVEL_SYSERR; |
|
say("", "%s: %s", tempfilename, sqlite3_errstr(rc)); |
|
return 0; |
|
} |
|
|
/* |
create_tables: |
* Recursively build up a list of files to parse. |
sql = "CREATE TABLE \"mpages\" (\n" |
* We use this instead of ftw() and so on because I don't want global |
" \"desc\" TEXT NOT NULL,\n" |
* variables hanging around. |
" \"form\" INTEGER NOT NULL,\n" |
* This ignores the mandoc.db and mandoc.index files, but assumes that |
" \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n" |
* everything else is a manual. |
");\n" |
* Pass in a pointer to a NULL structure for the first invocation. |
"\n" |
*/ |
"CREATE TABLE \"mlinks\" (\n" |
static int |
" \"sec\" TEXT NOT NULL,\n" |
ofile_dirbuild(const char *dir, const char* psec, const char *parch, |
" \"arch\" TEXT NOT NULL,\n" |
int p_src_form, struct of **of) |
" \"name\" TEXT NOT NULL,\n" |
{ |
" \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) " |
char buf[MAXPATHLEN]; |
"ON DELETE CASCADE\n" |
struct stat sb; |
");\n" |
size_t sz; |
"CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n" |
DIR *d; |
"\n" |
const char *fn, *sec, *arch; |
"CREATE TABLE \"names\" (\n" |
char *p, *q, *suffix; |
" \"bits\" INTEGER NOT NULL,\n" |
struct of *nof; |
" \"name\" TEXT NOT NULL,\n" |
struct dirent *dp; |
" \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) " |
int src_form; |
"ON DELETE CASCADE,\n" |
|
" UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n" |
|
");\n" |
|
"\n" |
|
"CREATE TABLE \"keys\" (\n" |
|
" \"bits\" INTEGER NOT NULL,\n" |
|
" \"key\" TEXT NOT NULL,\n" |
|
" \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) " |
|
"ON DELETE CASCADE\n" |
|
");\n" |
|
"CREATE INDEX keys_pageid_idx ON keys (pageid);\n"; |
|
|
if (NULL == (d = opendir(dir))) { |
if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) { |
perror(dir); |
exitcode = (int)MANDOCLEVEL_SYSERR; |
return(0); |
say(MANDOC_DB, "%s", sqlite3_errmsg(db)); |
|
sqlite3_close(db); |
|
return 0; |
} |
} |
|
|
while (NULL != (dp = readdir(d))) { |
prepare_statements: |
fn = dp->d_name; |
if (SQLITE_OK != sqlite3_exec(db, |
|
"PRAGMA foreign_keys = ON", NULL, NULL, NULL)) { |
|
exitcode = (int)MANDOCLEVEL_SYSERR; |
|
say(MANDOC_DB, "PRAGMA foreign_keys: %s", |
|
sqlite3_errmsg(db)); |
|
sqlite3_close(db); |
|
return 0; |
|
} |
|
|
if ('.' == *fn) |
sql = "DELETE FROM mpages WHERE pageid IN " |
continue; |
"(SELECT pageid FROM mlinks WHERE " |
|
"sec=? AND arch=? AND name=?)"; |
|
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL); |
|
sql = "INSERT INTO mpages " |
|
"(desc,form) VALUES (?,?)"; |
|
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL); |
|
sql = "INSERT INTO mlinks " |
|
"(sec,arch,name,pageid) VALUES (?,?,?,?)"; |
|
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL); |
|
sql = "SELECT bits FROM names where pageid = ?"; |
|
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL); |
|
sql = "INSERT INTO names " |
|
"(bits,name,pageid) VALUES (?,?,?)"; |
|
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL); |
|
sql = "INSERT INTO keys " |
|
"(bits,key,pageid) VALUES (?,?,?)"; |
|
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL); |
|
|
src_form = p_src_form; |
#ifndef __APPLE__ |
|
/* |
|
* When opening a new database, we can turn off |
|
* synchronous mode for much better performance. |
|
*/ |
|
|
if (DT_DIR == dp->d_type) { |
if (real && SQLITE_OK != sqlite3_exec(db, |
sec = psec; |
"PRAGMA synchronous = OFF", NULL, NULL, NULL)) { |
arch = parch; |
exitcode = (int)MANDOCLEVEL_SYSERR; |
|
say(MANDOC_DB, "PRAGMA synchronous: %s", |
|
sqlite3_errmsg(db)); |
|
sqlite3_close(db); |
|
return 0; |
|
} |
|
#endif |
|
|
/* |
return 1; |
* By default, only use directories called: |
} |
* man<section>/[<arch>/] or |
|
* cat<section>/[<arch>/] |
|
*/ |
|
|
|
if (NULL == sec) { |
static int |
if(0 == strncmp("man", fn, 3)) { |
set_basedir(const char *targetdir, int report_baddir) |
src_form |= MANDOC_SRC; |
{ |
sec = fn + 3; |
static char startdir[PATH_MAX]; |
} else if (0 == strncmp("cat", fn, 3)) { |
static int getcwd_status; /* 1 = ok, 2 = failure */ |
src_form |= MANDOC_FORM; |
static int chdir_status; /* 1 = changed directory */ |
sec = fn + 3; |
char *cp; |
} else if (use_all) |
|
sec = fn; |
|
else |
|
continue; |
|
} else if (NULL == arch && (use_all || |
|
NULL == strchr(fn, '.'))) |
|
arch = fn; |
|
else if (0 == use_all) |
|
continue; |
|
|
|
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. |
|
* Do not error out if the current directory is not |
|
* searchable: Maybe it won't be needed after all. |
|
*/ |
|
if (0 == getcwd_status) { |
|
if (NULL == getcwd(startdir, sizeof(startdir))) { |
|
getcwd_status = 2; |
|
(void)strlcpy(startdir, strerror(errno), |
|
sizeof(startdir)); |
|
} else |
|
getcwd_status = 1; |
|
} |
|
|
if (MAXPATHLEN <= sz) { |
/* |
fprintf(stderr, "%s: Path too long\n", dir); |
* We are leaving the old base directory. |
return(0); |
* Do not use it any longer, not even for messages. |
} |
*/ |
|
*basedir = '\0'; |
if (verb > 2) |
|
printf("%s: Scanning\n", buf); |
|
|
|
if ( ! ofile_dirbuild(buf, sec, arch, |
/* |
src_form, of)) |
* If and only if the directory was changed earlier and |
return(0); |
* the next directory to process is given as a relative path, |
|
* first go back, or bail out if that is impossible. |
|
*/ |
|
if (chdir_status && '/' != *targetdir) { |
|
if (2 == getcwd_status) { |
|
exitcode = (int)MANDOCLEVEL_SYSERR; |
|
say("", "getcwd: %s", startdir); |
|
return 0; |
} |
} |
if (DT_REG != dp->d_type || |
if (-1 == chdir(startdir)) { |
(NULL == psec && !use_all) || |
exitcode = (int)MANDOCLEVEL_SYSERR; |
!strcmp(MANDOC_DB, fn) || |
say("", "&chdir %s", startdir); |
!strcmp(MANDOC_IDX, fn)) |
return 0; |
continue; |
|
|
|
/* |
|
* By default, skip files where the file name suffix |
|
* does not agree with the section directory |
|
* they are located in. |
|
*/ |
|
|
|
suffix = strrchr(fn, '.'); |
|
if (0 == use_all) { |
|
if (NULL == suffix) |
|
continue; |
|
if ((MANDOC_SRC & src_form && |
|
strcmp(suffix + 1, psec)) || |
|
(MANDOC_FORM & src_form && |
|
strcmp(suffix + 1, "0"))) |
|
continue; |
|
} |
} |
if (NULL != suffix) { |
} |
if ('0' == suffix[1]) |
|
src_form |= MANDOC_FORM; |
|
else if ('1' <= suffix[1] && '9' >= suffix[1]) |
|
src_form |= MANDOC_SRC; |
|
} |
|
|
|
|
/* |
/* |
* Always resolve basedir to the canonicalized absolute |
* Skip formatted manuals if a source version is |
* pathname and append a trailing slash, such that |
* available. Ignore the age: it is very unlikely |
* we can reliably check whether files are inside. |
* that people install newer formatted base manuals |
*/ |
* when they used to have source manuals before, |
if (NULL == realpath(targetdir, basedir)) { |
* and in ports, old manuals get removed on update. |
if (report_baddir || errno != ENOENT) { |
*/ |
exitcode = (int)MANDOCLEVEL_BADARG; |
if (0 == use_all && MANDOC_FORM & src_form && |
say("", "&%s: realpath", targetdir); |
NULL != psec) { |
|
buf[0] = '\0'; |
|
strlcat(buf, dir, MAXPATHLEN); |
|
p = strrchr(buf, '/'); |
|
if (NULL == p) |
|
p = buf; |
|
else |
|
p++; |
|
if (0 == strncmp("cat", p, 3)) |
|
memcpy(p, "man", 3); |
|
strlcat(buf, "/", MAXPATHLEN); |
|
sz = strlcat(buf, fn, MAXPATHLEN); |
|
if (sz >= MAXPATHLEN) { |
|
fprintf(stderr, "%s: Path too long\n", buf); |
|
continue; |
|
} |
|
q = strrchr(buf, '.'); |
|
if (NULL != q && p < q++) { |
|
*q = '\0'; |
|
sz = strlcat(buf, psec, MAXPATHLEN); |
|
if (sz >= MAXPATHLEN) { |
|
fprintf(stderr, |
|
"%s: Path too long\n", buf); |
|
continue; |
|
} |
|
if (0 == stat(buf, &sb)) |
|
continue; |
|
} |
|
} |
} |
|
return 0; |
buf[0] = '\0'; |
} else if (-1 == chdir(basedir)) { |
strlcat(buf, dir, MAXPATHLEN); |
if (report_baddir || errno != ENOENT) { |
strlcat(buf, "/", MAXPATHLEN); |
exitcode = (int)MANDOCLEVEL_BADARG; |
sz = strlcat(buf, fn, MAXPATHLEN); |
say("", "&chdir"); |
if (sz >= MAXPATHLEN) { |
|
fprintf(stderr, "%s: Path too long\n", dir); |
|
continue; |
|
} |
} |
|
return 0; |
nof = mandoc_calloc(1, sizeof(struct of)); |
} |
nof->fname = mandoc_strdup(buf); |
chdir_status = 1; |
if (NULL != psec) |
cp = strchr(basedir, '\0'); |
nof->sec = mandoc_strdup(psec); |
if ('/' != cp[-1]) { |
if (NULL != parch) |
if (cp - basedir >= PATH_MAX - 1) { |
nof->arch = mandoc_strdup(parch); |
exitcode = (int)MANDOCLEVEL_SYSERR; |
nof->src_form = src_form; |
say("", "Filename too long"); |
|
return 0; |
/* |
|
* Remember the file name without the extension, |
|
* to be used as the page title in the database. |
|
*/ |
|
|
|
if (NULL != suffix) |
|
*suffix = '\0'; |
|
nof->title = mandoc_strdup(fn); |
|
|
|
/* |
|
* Add the structure to the list. |
|
*/ |
|
|
|
if (verb > 2) |
|
printf("%s: Scheduling\n", buf); |
|
if (NULL == *of) { |
|
*of = nof; |
|
(*of)->first = nof; |
|
} else { |
|
nof->first = (*of)->first; |
|
(*of)->next = nof; |
|
*of = nof; |
|
} |
} |
|
*cp++ = '/'; |
|
*cp = '\0'; |
} |
} |
|
return 1; |
closedir(d); |
|
return(1); |
|
} |
} |
|
|
static void |
static void |
ofile_free(struct of *of) |
say(const char *file, const char *format, ...) |
{ |
{ |
struct of *nof; |
va_list ap; |
|
int use_errno; |
|
|
while (of) { |
if ('\0' != *basedir) |
nof = of->next; |
fprintf(stderr, "%s", basedir); |
free(of->fname); |
if ('\0' != *basedir && '\0' != *file) |
free(of->sec); |
fputc('/', stderr); |
free(of->arch); |
if ('\0' != *file) |
free(of->title); |
fprintf(stderr, "%s", file); |
free(of); |
|
of = nof; |
|
} |
|
} |
|
|
|
static void |
use_errno = 1; |
usage(void) |
if (NULL != format) { |
{ |
switch (*format) { |
|
case '&': |
fprintf(stderr, "usage: %s [-v] " |
format++; |
"[-d dir [files...] |" |
break; |
" -u dir [files...] |" |
case '\0': |
" dir...]\n", progname); |
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); |
} |
} |