version 1.34, 2011/12/12 02:00:49 |
version 1.100, 2014/01/05 04:13:52 |
|
|
/* $Id$ */ |
/* $Id$ */ |
/* |
/* |
* 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, 2012, 2013, 2014 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 |
|
|
#include "config.h" |
#include "config.h" |
#endif |
#endif |
|
|
#include <sys/param.h> |
|
#include <sys/types.h> |
|
#include <sys/stat.h> |
#include <sys/stat.h> |
|
|
#include <assert.h> |
#include <assert.h> |
#include <dirent.h> |
#include <ctype.h> |
|
#include <errno.h> |
#include <fcntl.h> |
#include <fcntl.h> |
|
#include <fts.h> |
#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__) |
#ifdef HAVE_OHASH |
# include <endian.h> |
#include <ohash.h> |
# include <db_185.h> |
|
#elif defined(__APPLE__) |
|
# include <libkern/OSByteOrder.h> |
|
# include <db.h> |
|
#else |
#else |
# include <db.h> |
#include "compat_ohash.h" |
#endif |
#endif |
|
#include <sqlite3.h> |
|
|
#include "man.h" |
|
#include "mdoc.h" |
#include "mdoc.h" |
|
#include "man.h" |
#include "mandoc.h" |
#include "mandoc.h" |
#include "mandocdb.h" |
|
#include "manpath.h" |
#include "manpath.h" |
|
#include "mansearch.h" |
|
|
#define MANDOC_BUFSZ BUFSIZ |
#define SQL_EXEC(_v) \ |
#define MANDOC_SLOP 1024 |
if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \ |
|
fprintf(stderr, "%s\n", sqlite3_errmsg(db)) |
|
#define SQL_BIND_TEXT(_s, _i, _v) \ |
|
if (SQLITE_OK != sqlite3_bind_text \ |
|
((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \ |
|
fprintf(stderr, "%s\n", sqlite3_errmsg(db)) |
|
#define SQL_BIND_INT(_s, _i, _v) \ |
|
if (SQLITE_OK != sqlite3_bind_int \ |
|
((_s), (_i)++, (_v))) \ |
|
fprintf(stderr, "%s\n", sqlite3_errmsg(db)) |
|
#define SQL_BIND_INT64(_s, _i, _v) \ |
|
if (SQLITE_OK != sqlite3_bind_int64 \ |
|
((_s), (_i)++, (_v))) \ |
|
fprintf(stderr, "%s\n", sqlite3_errmsg(db)) |
|
#define SQL_STEP(_s) \ |
|
if (SQLITE_DONE != sqlite3_step((_s))) \ |
|
fprintf(stderr, "%s\n", sqlite3_errmsg(db)) |
|
|
#define MANDOC_SRC 0x1 |
enum op { |
#define MANDOC_FORM 0x2 |
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 */ |
|
}; |
|
|
/* Tiny list for files. No need to bring in QUEUE. */ |
enum form { |
|
FORM_NONE, /* format is unknown */ |
|
FORM_SRC, /* format is -man or -mdoc */ |
|
FORM_CAT /* format is cat */ |
|
}; |
|
|
struct of { |
struct str { |
char *fname; /* heap-allocated */ |
char *rendered; /* key in UTF-8 or ASCII form */ |
char *sec; |
const struct mpage *mpage; /* if set, the owning parse */ |
char *arch; |
uint64_t mask; /* bitmask in sequence */ |
char *title; |
char key[]; /* may contain escape sequences */ |
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 */ |
enum form form; /* format from file content */ |
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 */ |
}; |
}; |
|
|
/* Operation we're going to perform. */ |
struct mlink { |
|
char file[PATH_MAX]; /* filename rel. to manpath */ |
|
enum form dform; /* format from directory */ |
|
enum form fform; /* format from file name suffix */ |
|
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 */ |
|
}; |
|
|
enum op { |
struct title { |
OP_NEW = 0, /* new database */ |
char *title; /* name(sec/arch) given inside the file */ |
OP_UPDATE, /* delete/add entries in existing database */ |
char *file; /* file name in case of mismatch */ |
OP_DELETE /* delete entries from existing database */ |
|
}; |
}; |
|
|
#define MAN_ARGS DB *hash, \ |
enum stmt { |
struct buf *buf, \ |
STMT_DELETE_PAGE = 0, /* delete mpage */ |
struct buf *dbuf, \ |
STMT_INSERT_PAGE, /* insert mpage */ |
const struct man_node *n |
STMT_INSERT_LINK, /* insert mlink */ |
#define MDOC_ARGS DB *hash, \ |
STMT_INSERT_KEY, /* insert parsed key */ |
struct buf *buf, \ |
STMT__MAX |
struct buf *dbuf, \ |
}; |
const struct mdoc_node *n, \ |
|
const struct mdoc_meta *m |
|
|
|
static void buf_appendmdoc(struct buf *, |
typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_node *); |
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 dbindex(const struct mpage *, struct mchars *); |
|
static int dbopen(int); |
|
static void dbprune(void); |
|
static void filescan(const char *); |
|
static void *hash_alloc(size_t, void *); |
|
static void hash_free(void *, size_t, void *); |
|
static void *hash_halloc(size_t, void *); |
|
static void mlink_add(struct mlink *, const struct stat *); |
|
static int 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 mchars *, struct mparse *, int); |
|
static void parse_cat(struct mpage *); |
|
static void parse_man(struct mpage *, const struct man_node *); |
|
static void parse_mdoc(struct mpage *, const struct mdoc_node *); |
|
static int parse_mdoc_body(struct mpage *, const struct mdoc_node *); |
|
static int parse_mdoc_head(struct mpage *, const struct mdoc_node *); |
|
static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *); |
|
static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *); |
|
static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *); |
|
static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *); |
|
static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *); |
|
static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *); |
|
static void putkey(const struct mpage *, |
|
const char *, uint64_t); |
|
static void putkeys(const struct mpage *, |
|
const char *, size_t, uint64_t); |
|
static void putmdockey(const struct mpage *, |
|
const struct mdoc_node *, uint64_t); |
|
static void render_key(struct mchars *, struct str *); |
|
static void say(const char *, const char *, ...); |
|
static int set_basedir(const char *); |
|
static int treescan(void); |
|
static size_t utf8(unsigned int, char [7]); |
|
|
|
static char *progname; |
|
static int use_all; /* use all found files */ |
|
static int nodb; /* no database changes */ |
|
static int verb; /* 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 strings; /* table of all strings */ |
|
static sqlite3 *db = NULL; /* current database */ |
|
static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */ |
|
|
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, TYPE_Nd }, /* Nd */ |
{ pmdoc_Nm, TYPE_Nm, MDOCF_CHILD }, /* Nm */ |
{ parse_mdoc_Nm, TYPE_Nm }, /* 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_head, 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 */ |
}; |
}; |
|
|
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[]) |
main(int argc, char *argv[]) |
{ |
{ |
struct mparse *mp; /* parse sequence */ |
int ch, i; |
struct manpaths dirs; |
size_t j, sz; |
enum op op; /* current operation */ |
const char *path_arg; |
const char *dir; |
struct mchars *mc; |
char *conf_file; |
struct manpaths dirs; |
char *cp; |
struct mparse *mp; |
char pbuf[PATH_MAX], |
struct ohash_info mpages_info, mlinks_info; |
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; |
|
|
|
|
memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *)); |
|
memset(&dirs, 0, sizeof(struct manpaths)); |
|
|
|
mpages_info.alloc = mlinks_info.alloc = hash_alloc; |
|
mpages_info.halloc = mlinks_info.halloc = hash_halloc; |
|
mpages_info.hfree = mlinks_info.hfree = hash_free; |
|
|
|
mpages_info.key_offset = offsetof(struct mpage, inodev); |
|
mlinks_info.key_offset = offsetof(struct mlink, file); |
|
|
progname = strrchr(argv[0], '/'); |
progname = strrchr(argv[0], '/'); |
if (progname == NULL) |
if (progname == NULL) |
progname = argv[0]; |
progname = argv[0]; |
else |
else |
++progname; |
++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)) { \ |
|
fprintf(stderr, "-%c: Conflicting option\n", (_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; |
|
conf_file = NULL; |
|
|
|
while (-1 != (ch = getopt(argc, argv, "aC:d:u:v"))) |
while (-1 != (ch = getopt(argc, argv, "aC:d:nT:tu:vW"))) |
switch (ch) { |
switch (ch) { |
case ('a'): |
case ('a'): |
use_all = 1; |
use_all = 1; |
break; |
break; |
case ('C'): |
case ('C'): |
conf_file = optarg; |
CHECKOP(op, ch); |
|
path_arg = optarg; |
|
op = OP_CONFFILE; |
break; |
break; |
case ('d'): |
case ('d'): |
dir = optarg; |
CHECKOP(op, ch); |
|
path_arg = optarg; |
op = OP_UPDATE; |
op = OP_UPDATE; |
break; |
break; |
|
case ('n'): |
|
nodb = 1; |
|
break; |
|
case ('T'): |
|
if (strcmp(optarg, "utf8")) { |
|
fprintf(stderr, "-T%s: Unsupported " |
|
"output format\n", 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'): |
case ('u'): |
dir = optarg; |
CHECKOP(op, ch); |
|
path_arg = optarg; |
op = OP_DELETE; |
op = OP_DELETE; |
break; |
break; |
case ('v'): |
case ('v'): |
verb++; |
verb++; |
break; |
break; |
|
case ('W'): |
|
warnings = 1; |
|
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; |
fprintf(stderr, "-C: Too many arguments\n"); |
|
goto usage; |
|
} |
|
|
mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL); |
exitcode = (int)MANDOCLEVEL_OK; |
|
mp = mparse_alloc(MPARSE_AUTO, |
|
MANDOCLEVEL_FATAL, NULL, NULL, NULL); |
|
mc = mchars_alloc(); |
|
|
memset(&buf, 0, sizeof(struct buf)); |
ohash_init(&mpages, 6, &mpages_info); |
memset(&dbuf, 0, sizeof(struct buf)); |
ohash_init(&mlinks, 6, &mlinks_info); |
|
|
buf.size = dbuf.size = MANDOC_BUFSZ; |
if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) { |
|
/* |
|
* Force processing all files. |
|
*/ |
|
use_all = 1; |
|
|
buf.cp = mandoc_malloc(buf.size); |
/* |
dbuf.cp = mandoc_malloc(dbuf.size); |
* All of these deal with a specific directory. |
|
* Jump into that directory then collect files specified |
|
* on the command-line. |
|
*/ |
|
if (0 == set_basedir(path_arg)) |
|
goto out; |
|
for (i = 0; i < argc; i++) |
|
filescan(argv[i]); |
|
if (0 == dbopen(1)) |
|
goto out; |
|
if (OP_TEST != op) |
|
dbprune(); |
|
if (OP_DELETE != op) |
|
mpages_merge(mc, mp, 0); |
|
dbclose(1); |
|
} else { |
|
/* |
|
* If we have arguments, use them as our manpaths. |
|
* If we don't, grok from manpath(1) or however else |
|
* manpath_parse() wants to do it. |
|
*/ |
|
if (argc > 0) { |
|
dirs.paths = mandoc_calloc |
|
(argc, sizeof(char *)); |
|
dirs.sz = (size_t)argc; |
|
for (i = 0; i < argc; i++) |
|
dirs.paths[i] = mandoc_strdup(argv[i]); |
|
} else |
|
manpath_parse(&dirs, path_arg, NULL, NULL); |
|
|
flags = OP_NEW == op ? O_CREAT|O_TRUNC|O_RDWR : O_CREAT|O_RDWR; |
/* |
|
* 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 < dirs.sz; j++) { |
|
sz = strlen(dirs.paths[j]); |
|
if (sz && '/' == dirs.paths[j][sz - 1]) |
|
dirs.paths[j][--sz] = '\0'; |
|
if (0 == sz) |
|
continue; |
|
|
if (OP_UPDATE == op || OP_DELETE == op) { |
if (j) { |
ibuf[0] = fbuf[0] = '\0'; |
ohash_init(&mpages, 6, &mpages_info); |
|
ohash_init(&mlinks, 6, &mlinks_info); |
|
} |
|
|
strlcat(fbuf, dir, MAXPATHLEN); |
if (0 == set_basedir(dirs.paths[j])) |
strlcat(fbuf, "/", MAXPATHLEN); |
goto out; |
sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN); |
if (0 == treescan()) |
|
goto out; |
|
if (0 == set_basedir(dirs.paths[j])) |
|
goto out; |
|
if (0 == dbopen(0)) |
|
goto out; |
|
|
strlcat(ibuf, dir, MAXPATHLEN); |
mpages_merge(mc, mp, warnings && !use_all); |
strlcat(ibuf, "/", MAXPATHLEN); |
dbclose(0); |
sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN); |
|
|
|
if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) { |
if (j + 1 < dirs.sz) { |
fprintf(stderr, "%s: Path too long\n", dir); |
mpages_free(); |
exit((int)MANDOCLEVEL_BADARG); |
ohash_delete(&mpages); |
|
ohash_delete(&mlinks); |
|
} |
} |
} |
|
} |
|
out: |
|
set_basedir(NULL); |
|
manpath_free(&dirs); |
|
mchars_free(mc); |
|
mparse_free(mp); |
|
mpages_free(); |
|
ohash_delete(&mpages); |
|
ohash_delete(&mlinks); |
|
return(exitcode); |
|
usage: |
|
fprintf(stderr, "usage: %s [-anvW] [-C file] [-Tutf8]\n" |
|
" %s [-anvW] [-Tutf8] dir ...\n" |
|
" %s [-nvW] [-Tutf8] -d dir [file ...]\n" |
|
" %s [-nvW] -u dir [file ...]\n" |
|
" %s -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) |
|
{ |
|
FTS *f; |
|
FTSENT *ff; |
|
struct mlink *mlink; |
|
int dform; |
|
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); |
/* |
if (NULL == of) |
* Walk through all components under the directory, using the |
goto out; |
* logical descent of files. |
|
*/ |
|
f = fts_open((char * const *)argv, FTS_LOGICAL, NULL); |
|
if (NULL == f) { |
|
exitcode = (int)MANDOCLEVEL_SYSERR; |
|
say("", NULL); |
|
return(0); |
|
} |
|
|
of = of->first; |
dsec = arch = NULL; |
|
dform = FORM_NONE; |
|
|
index_prune(of, db, fbuf, idx, ibuf, |
while (NULL != (ff = fts_read(f))) { |
&maxrec, &recs, &recsz, &reccur); |
path = ff->fts_path + 2; |
|
|
/* |
/* |
* Go to the root of the respective manual tree |
* If we're a regular file, add an mlink by using the |
* such that .so links work. In case of failure, |
* stored directory data and handling the filename. |
* just prod on, even though .so links won't work. |
|
*/ |
*/ |
|
if (FTS_F == ff->fts_info) { |
|
if (0 == strcmp(path, MANDOC_DB)) |
|
continue; |
|
if ( ! use_all && ff->fts_level < 2) { |
|
if (warnings) |
|
say(path, "Extraneous file"); |
|
continue; |
|
} else if (NULL == (fsec = |
|
strrchr(ff->fts_name, '.'))) { |
|
if ( ! use_all) { |
|
if (warnings) |
|
say(path, |
|
"No filename suffix"); |
|
continue; |
|
} |
|
} else if (0 == strcmp(++fsec, "html")) { |
|
if (warnings) |
|
say(path, "Skip html"); |
|
continue; |
|
} else if (0 == strcmp(fsec, "gz")) { |
|
if (warnings) |
|
say(path, "Skip gz"); |
|
continue; |
|
} else if (0 == strcmp(fsec, "ps")) { |
|
if (warnings) |
|
say(path, "Skip ps"); |
|
continue; |
|
} else if (0 == strcmp(fsec, "pdf")) { |
|
if (warnings) |
|
say(path, "Skip pdf"); |
|
continue; |
|
} else if ( ! use_all && |
|
((FORM_SRC == dform && strcmp(fsec, dsec)) || |
|
(FORM_CAT == dform && strcmp(fsec, "0")))) { |
|
if (warnings) |
|
say(path, "Wrong filename suffix"); |
|
continue; |
|
} else |
|
fsec[-1] = '\0'; |
|
|
if (OP_UPDATE == op) { |
mlink = mandoc_calloc(1, sizeof(struct mlink)); |
chdir(dir); |
strlcpy(mlink->file, path, sizeof(mlink->file)); |
index_merge(of, mp, &dbuf, &buf, hash, |
mlink->dform = dform; |
db, fbuf, idx, ibuf, |
mlink->dsec = dsec; |
maxrec, recs, reccur); |
mlink->arch = arch; |
|
mlink->name = ff->fts_name; |
|
mlink->fsec = fsec; |
|
mlink_add(mlink, ff->fts_statp); |
|
continue; |
|
} else if (FTS_D != ff->fts_info && |
|
FTS_DP != ff->fts_info) { |
|
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 (FTS_DP == ff->fts_info) |
|
break; |
|
|
/* |
if (0 == strncmp(cp, "man", 3)) { |
* Configure the directories we're going to scan. |
dform = FORM_SRC; |
* If we have command-line arguments, use them. |
dsec = cp + 3; |
* If not, we use man(1)'s method (see mandocdb.8). |
} else if (0 == strncmp(cp, "cat", 3)) { |
*/ |
dform = FORM_CAT; |
|
dsec = cp + 3; |
if (argc > 0) { |
} else { |
dirs.paths = mandoc_calloc(argc, sizeof(char *)); |
dform = FORM_NONE; |
dirs.sz = argc; |
dsec = NULL; |
for (i = 0; i < argc; i++) { |
|
if (NULL == (cp = realpath(argv[i], pbuf))) { |
|
perror(argv[i]); |
|
goto out; |
|
} |
} |
dirs.paths[i] = mandoc_strdup(cp); |
|
} |
|
} else |
|
manpath_parse(&dirs, conf_file, NULL, NULL); |
|
|
|
for (i = 0; i < dirs.sz; i++) { |
if (NULL != dsec || use_all) |
ibuf[0] = fbuf[0] = '\0'; |
break; |
|
|
strlcat(fbuf, dirs.paths[i], MAXPATHLEN); |
if (warnings) |
strlcat(fbuf, "/", MAXPATHLEN); |
say(path, "Unknown directory part"); |
sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN); |
fts_set(f, ff, FTS_SKIP); |
|
break; |
|
case (2): |
|
/* |
|
* Possibly our architecture. |
|
* If we're descending, keep tabs on it. |
|
*/ |
|
if (FTS_DP != ff->fts_info && NULL != dsec) |
|
arch = ff->fts_name; |
|
else |
|
arch = NULL; |
|
break; |
|
default: |
|
if (FTS_DP == ff->fts_info || use_all) |
|
break; |
|
if (warnings) |
|
say(path, "Extraneous directory part"); |
|
fts_set(f, ff, FTS_SKIP); |
|
break; |
|
} |
|
} |
|
|
strlcat(ibuf, dirs.paths[i], MAXPATHLEN); |
fts_close(f); |
strlcat(ibuf, "/", MAXPATHLEN); |
return(1); |
sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN); |
} |
|
|
if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) { |
/* |
fprintf(stderr, "%s: Path too long\n", |
* Add a file to the mlinks table. |
dirs.paths[i]); |
* Do not verify that it's a "valid" looking manpage (we'll do that |
exit((int)MANDOCLEVEL_BADARG); |
* 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; |
|
|
if (db) |
assert(use_all); |
(*db->close)(db); |
|
if (idx) |
|
(*idx->close)(idx); |
|
|
|
db = dbopen(fbuf, flags, 0644, DB_BTREE, &info); |
if (0 == strncmp(file, "./", 2)) |
idx = dbopen(ibuf, flags, 0644, DB_RECNO, NULL); |
file += 2; |
|
|
if (NULL == db) { |
if (NULL == realpath(file, buf)) { |
perror(fbuf); |
exitcode = (int)MANDOCLEVEL_BADARG; |
exit((int)MANDOCLEVEL_SYSERR); |
say(file, NULL); |
} else if (NULL == idx) { |
return; |
perror(ibuf); |
} else if (OP_TEST != op && strstr(buf, basedir) != buf) { |
exit((int)MANDOCLEVEL_SYSERR); |
exitcode = (int)MANDOCLEVEL_BADARG; |
|
say("", "%s: outside base directory", buf); |
|
return; |
|
} else if (-1 == stat(buf, &st)) { |
|
exitcode = (int)MANDOCLEVEL_BADARG; |
|
say(file, NULL); |
|
return; |
|
} else if ( ! (S_IFREG & st.st_mode)) { |
|
exitcode = (int)MANDOCLEVEL_BADARG; |
|
say(file, "Not a regular file"); |
|
return; |
|
} |
|
start = buf + strlen(basedir); |
|
mlink = mandoc_calloc(1, sizeof(struct mlink)); |
|
strlcpy(mlink->file, start, sizeof(mlink->file)); |
|
|
|
/* |
|
* First try to guess our directory structure. |
|
* If we find a separator, try to look for man* or cat*. |
|
* If we find one of these and what's underneath is a directory, |
|
* assume it's an architecture. |
|
*/ |
|
if (NULL != (p = strchr(start, '/'))) { |
|
*p++ = '\0'; |
|
if (0 == strncmp(start, "man", 3)) { |
|
mlink->dform = FORM_SRC; |
|
mlink->dsec = start + 3; |
|
} else if (0 == strncmp(start, "cat", 3)) { |
|
mlink->dform = FORM_CAT; |
|
mlink->dsec = start + 3; |
} |
} |
|
|
if (verb > 2) { |
start = p; |
printf("%s: Truncated\n", fbuf); |
if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) { |
printf("%s: Truncated\n", ibuf); |
*p++ = '\0'; |
|
mlink->arch = start; |
|
start = p; |
} |
} |
|
} |
|
|
ofile_free(of); |
/* |
of = 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 ( ! ofile_dirbuild(dirs.paths[i], NULL, NULL, |
if ('.' == *p) { |
0, &of)) |
*p++ = '\0'; |
exit((int)MANDOCLEVEL_SYSERR); |
mlink->fsec = p; |
|
} |
|
|
if (NULL == of) |
/* |
continue; |
* 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); |
|
} |
|
|
of = of->first; |
static void |
|
mlink_add(struct mlink *mlink, const struct stat *st) |
|
{ |
|
struct inodev inodev; |
|
struct mpage *mpage; |
|
unsigned int slot; |
|
|
/* |
assert(NULL != mlink->file); |
* Go to the root of the respective manual tree |
|
* such that .so links work. In case of failure, |
|
* just prod on, even though .so links won't work. |
|
*/ |
|
|
|
chdir(dirs.paths[i]); |
mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : ""); |
index_merge(of, mp, &dbuf, &buf, hash, db, fbuf, |
mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : ""); |
idx, ibuf, maxrec, recs, reccur); |
mlink->name = mandoc_strdup(mlink->name ? mlink->name : ""); |
} |
mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : ""); |
|
|
out: |
if ('0' == *mlink->fsec) { |
if (db) |
free(mlink->fsec); |
(*db->close)(db); |
mlink->fsec = mandoc_strdup(mlink->dsec); |
if (idx) |
mlink->fform = FORM_CAT; |
(*idx->close)(idx); |
} else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec) |
if (hash) |
mlink->fform = FORM_SRC; |
(*hash->close)(hash); |
else |
if (mp) |
mlink->fform = FORM_NONE; |
mparse_free(mp); |
|
|
|
manpath_free(&dirs); |
slot = ohash_qlookup(&mlinks, mlink->file); |
ofile_free(of); |
assert(NULL == ohash_find(&mlinks, slot)); |
free(buf.cp); |
ohash_insert(&mlinks, slot, mlink); |
free(dbuf.cp); |
|
free(recs); |
|
|
|
return(MANDOCLEVEL_OK); |
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; |
} |
} |
|
|
void |
static void |
index_merge(const struct of *of, struct mparse *mp, |
mlink_free(struct mlink *mlink) |
struct buf *dbuf, struct buf *buf, DB *hash, |
|
DB *db, const char *dbf, DB *idx, const char *idxf, |
|
recno_t maxrec, const recno_t *recs, size_t reccur) |
|
{ |
{ |
recno_t rec; |
|
int ch; |
|
DBT key, val; |
|
struct mdoc *mdoc; |
|
struct man *man; |
|
const char *fn, *msec, *mtitle, *arch; |
|
size_t sv; |
|
unsigned seq; |
|
struct db_val vbuf; |
|
|
|
for (rec = 0; of; of = of->next) { |
free(mlink->dsec); |
fn = of->fname; |
free(mlink->arch); |
|
free(mlink->name); |
|
free(mlink->fsec); |
|
free(mlink); |
|
} |
|
|
/* |
static void |
* Try interpreting the file as mdoc(7) or man(7) |
mpages_free(void) |
* source code, unless it is already known to be |
{ |
* formatted. Fall back to formatted mode. |
struct mpage *mpage; |
*/ |
struct mlink *mlink; |
|
unsigned int slot; |
|
|
mparse_reset(mp); |
mpage = ohash_first(&mpages, &slot); |
mdoc = NULL; |
while (NULL != mpage) { |
man = NULL; |
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 ((MANDOC_SRC & of->src_form || |
/* |
! (MANDOC_FORM & of->src_form)) && |
* For each mlink to the mpage, check whether the path looks like |
MANDOCLEVEL_FATAL > mparse_readfd(mp, -1, fn)) |
* it is formatted, and if it does, check whether a source manual |
mparse_result(mp, &mdoc, &man); |
* exists by the same name, ignoring the suffix. |
|
* If both conditions hold, drop the mlink. |
|
*/ |
|
static void |
|
mlinks_undupe(struct mpage *mpage) |
|
{ |
|
char buf[PATH_MAX]; |
|
struct mlink **prev; |
|
struct mlink *mlink; |
|
char *bufp; |
|
|
if (NULL != mdoc) { |
mpage->form = FORM_CAT; |
msec = mdoc_meta(mdoc)->msec; |
prev = &mpage->mlinks; |
arch = mdoc_meta(mdoc)->arch; |
while (NULL != (mlink = *prev)) { |
mtitle = mdoc_meta(mdoc)->title; |
if (FORM_CAT != mlink->dform) { |
} else if (NULL != man) { |
mpage->form = FORM_NONE; |
msec = man_meta(man)->msec; |
goto nextlink; |
arch = NULL; |
|
mtitle = man_meta(man)->title; |
|
} else { |
|
msec = of->sec; |
|
arch = of->arch; |
|
mtitle = of->title; |
|
} |
} |
|
if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) { |
|
if (warnings) |
|
say(mlink->file, "Filename too long"); |
|
goto nextlink; |
|
} |
|
bufp = strstr(buf, "cat"); |
|
assert(NULL != bufp); |
|
memcpy(bufp, "man", 3); |
|
if (NULL != (bufp = strrchr(buf, '.'))) |
|
*++bufp = '\0'; |
|
strlcat(buf, mlink->dsec, PATH_MAX); |
|
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; |
|
} |
|
} |
|
|
/* |
static int |
* By default, skip a file if the manual section |
mlink_check(struct mpage *mpage, struct mlink *mlink) |
* and architecture given in the file disagree |
{ |
* with the directory where the file is located. |
int match; |
*/ |
|
|
|
if (0 == use_all) { |
match = 1; |
assert(of->sec); |
|
assert(msec); |
|
if (strcasecmp(msec, of->sec)) |
|
continue; |
|
|
|
if (NULL == arch) { |
/* |
if (NULL != of->arch) |
* Check whether the manual section given in a file |
continue; |
* agrees with the directory where the file is located. |
} else if (NULL == of->arch || |
* Some manuals have suffixes like (3p) on their |
strcasecmp(arch, of->arch)) |
* section number either inside the file or in the |
continue; |
* directory name, some are linked into more than one |
} |
* section, like encrypt(1) = makekey(8). |
|
*/ |
|
|
if (NULL == arch) |
if (FORM_SRC == mpage->form && |
arch = ""; |
strcasecmp(mpage->sec, mlink->dsec)) { |
|
match = 0; |
|
say(mlink->file, "Section \"%s\" manual in %s directory", |
|
mpage->sec, mlink->dsec); |
|
} |
|
|
/* |
/* |
* By default, skip a file if the title given |
* Manual page directories exist for each kernel |
* in the file disagrees with the file name. |
* architecture as returned by machine(1). |
* If both agree, use the file name as the title, |
* However, many manuals only depend on the |
* because the one in the file usually is all caps. |
* 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. |
|
*/ |
|
|
assert(of->title); |
if (strcasecmp(mpage->arch, mlink->arch)) { |
assert(mtitle); |
match = 0; |
|
say(mlink->file, "Architecture \"%s\" manual in " |
|
"\"%s\" directory", mpage->arch, mlink->arch); |
|
} |
|
|
if (0 == strcasecmp(mtitle, of->title)) |
if (strcasecmp(mpage->title, mlink->name)) |
mtitle = of->title; |
match = 0; |
else if (0 == use_all) |
|
continue; |
|
|
|
/* |
return(match); |
* The index record value consists of a nil-terminated |
} |
* filename, a nil-terminated manual section, and a |
|
* nil-terminated description. Since the description |
|
* may not be set, we set a sentinel to see if we're |
|
* going to write a nil byte in its place. |
|
*/ |
|
|
|
dbuf->len = 0; |
/* |
buf_append(dbuf, mdoc ? "mdoc" : (man ? "man" : "cat")); |
* Run through the files in the global vector "mpages" |
buf_appendb(dbuf, fn, strlen(fn) + 1); |
* and add them to the database specified in "basedir". |
buf_appendb(dbuf, msec, strlen(msec) + 1); |
* |
buf_appendb(dbuf, mtitle, strlen(mtitle) + 1); |
* This handles the parsing scheme itself, using the cues of directory |
buf_appendb(dbuf, arch, strlen(arch) + 1); |
* and filename to determine whether the file is parsable or not. |
|
*/ |
|
static void |
|
mpages_merge(struct mchars *mc, struct mparse *mp, int check_reachable) |
|
{ |
|
struct ohash title_table; |
|
struct ohash_info title_info, str_info; |
|
struct mpage *mpage; |
|
struct mlink *mlink; |
|
struct mdoc *mdoc; |
|
struct man *man; |
|
struct title *title_entry; |
|
char *title_str; |
|
const char *cp; |
|
int match; |
|
unsigned int pslot, tslot; |
|
enum mandoclevel lvl; |
|
|
sv = dbuf->len; |
str_info.alloc = hash_alloc; |
|
str_info.halloc = hash_halloc; |
|
str_info.hfree = hash_free; |
|
str_info.key_offset = offsetof(struct str, key); |
|
|
/* |
if (check_reachable) { |
* Collect keyword/mask pairs. |
title_info.alloc = hash_alloc; |
* Each pair will become a new btree node. |
title_info.halloc = hash_halloc; |
*/ |
title_info.hfree = hash_free; |
|
title_info.key_offset = offsetof(struct title, title); |
|
ohash_init(&title_table, 6, &title_info); |
|
} |
|
|
hash_reset(&hash); |
mpage = ohash_first(&mpages, &pslot); |
if (mdoc) |
while (NULL != mpage) { |
pmdoc_node(hash, buf, dbuf, |
mlinks_undupe(mpage); |
mdoc_node(mdoc), mdoc_meta(mdoc)); |
if (NULL == mpage->mlinks) { |
else if (man) |
mpage = ohash_next(&mpages, &pslot); |
pman_node(hash, buf, dbuf, man_node(man)); |
continue; |
else |
} |
pformatted(hash, buf, dbuf, of); |
|
|
|
/* |
ohash_init(&strings, 6, &str_info); |
* Reclaim an empty index record, if available. |
mparse_reset(mp); |
* Use its record number for all new btree nodes. |
mdoc = NULL; |
*/ |
man = NULL; |
|
|
if (reccur > 0) { |
|
--reccur; |
|
rec = recs[(int)reccur]; |
|
} else if (maxrec > 0) { |
|
rec = maxrec; |
|
maxrec = 0; |
|
} else |
|
rec++; |
|
vbuf.rec = htobe32(rec); |
|
|
|
/* |
/* |
* Copy from the in-memory hashtable of pending |
* Try interpreting the file as mdoc(7) or man(7) |
* keyword/mask pairs into the database. |
* source code, unless it is already known to be |
|
* formatted. Fall back to formatted mode. |
*/ |
*/ |
|
if (FORM_CAT != mpage->mlinks->dform || |
|
FORM_CAT != mpage->mlinks->fform) { |
|
lvl = mparse_readfd(mp, -1, mpage->mlinks->file); |
|
if (lvl < MANDOCLEVEL_FATAL) |
|
mparse_result(mp, &mdoc, &man); |
|
} |
|
|
seq = R_FIRST; |
if (NULL != mdoc) { |
while (0 == (ch = (*hash->seq)(hash, &key, &val, seq))) { |
mpage->form = FORM_SRC; |
seq = R_NEXT; |
mpage->sec = |
vbuf.mask = htobe64(*(uint64_t *)val.data); |
mandoc_strdup(mdoc_meta(mdoc)->msec); |
val.size = sizeof(struct db_val); |
mpage->arch = mdoc_meta(mdoc)->arch; |
val.data = &vbuf; |
mpage->arch = mandoc_strdup( |
dbt_put(db, dbf, &key, &val); |
NULL == mpage->arch ? "" : mpage->arch); |
|
mpage->title = |
|
mandoc_strdup(mdoc_meta(mdoc)->title); |
|
} else if (NULL != man) { |
|
mpage->form = FORM_SRC; |
|
mpage->sec = |
|
mandoc_strdup(man_meta(man)->msec); |
|
mpage->arch = |
|
mandoc_strdup(mpage->mlinks->arch); |
|
mpage->title = |
|
mandoc_strdup(man_meta(man)->title); |
|
} else { |
|
mpage->form = FORM_CAT; |
|
mpage->sec = |
|
mandoc_strdup(mpage->mlinks->dsec); |
|
mpage->arch = |
|
mandoc_strdup(mpage->mlinks->arch); |
|
mpage->title = |
|
mandoc_strdup(mpage->mlinks->name); |
} |
} |
if (ch < 0) { |
putkey(mpage, mpage->sec, TYPE_sec); |
perror("hash"); |
putkey(mpage, '\0' == *mpage->arch ? |
exit((int)MANDOCLEVEL_SYSERR); |
"any" : mpage->arch, TYPE_arch); |
|
|
|
for (mlink = mpage->mlinks; mlink; mlink = mlink->next) { |
|
if ('\0' != *mlink->dsec) |
|
putkey(mpage, mlink->dsec, TYPE_sec); |
|
if ('\0' != *mlink->fsec) |
|
putkey(mpage, mlink->fsec, TYPE_sec); |
|
putkey(mpage, '\0' == *mlink->arch ? |
|
"any" : mlink->arch, TYPE_arch); |
|
putkey(mpage, mlink->name, TYPE_Nm); |
} |
} |
|
|
/* |
|
* Apply to the index. If we haven't had a description |
|
* set, put an empty one in now. |
|
*/ |
|
|
|
if (dbuf->len == sv) |
if (warnings && !use_all) { |
buf_appendb(dbuf, "", 1); |
match = 0; |
|
for (mlink = mpage->mlinks; mlink; |
|
mlink = mlink->next) |
|
if (mlink_check(mpage, mlink)) |
|
match = 1; |
|
} else |
|
match = 1; |
|
|
key.data = &rec; |
if (NULL != mdoc) { |
key.size = sizeof(recno_t); |
if (NULL != (cp = mdoc_meta(mdoc)->name)) |
|
putkey(mpage, cp, TYPE_Nm); |
|
assert(NULL == mpage->desc); |
|
parse_mdoc(mpage, mdoc_node(mdoc)); |
|
putkey(mpage, NULL != mpage->desc ? |
|
mpage->desc : mpage->mlinks->name, TYPE_Nd); |
|
} else if (NULL != man) |
|
parse_man(mpage, man_node(man)); |
|
else |
|
parse_cat(mpage); |
|
|
val.data = dbuf->cp; |
/* |
val.size = dbuf->len; |
* Build a title string for the file. If it matches |
|
* the location of the file, remember the title as |
|
* found; else, remember it as missing. |
|
*/ |
|
|
if (verb) |
if (check_reachable) { |
printf("%s: Added index\n", fn); |
if (-1 == asprintf(&title_str, "%s(%s%s%s)", |
|
mpage->title, mpage->sec, |
|
'\0' == *mpage->arch ? "" : "/", |
|
mpage->arch)) { |
|
perror(NULL); |
|
exit((int)MANDOCLEVEL_SYSERR); |
|
} |
|
tslot = ohash_qlookup(&title_table, title_str); |
|
title_entry = ohash_find(&title_table, tslot); |
|
if (NULL == title_entry) { |
|
title_entry = mandoc_malloc( |
|
sizeof(struct title)); |
|
title_entry->title = title_str; |
|
title_entry->file = mandoc_strdup( |
|
match ? "" : mpage->mlinks->file); |
|
ohash_insert(&title_table, tslot, |
|
title_entry); |
|
} else { |
|
if (match) |
|
*title_entry->file = '\0'; |
|
free(title_str); |
|
} |
|
} |
|
|
dbt_put(idx, idxf, &key, &val); |
dbindex(mpage, mc); |
|
ohash_delete(&strings); |
|
mpage = ohash_next(&mpages, &pslot); |
} |
} |
|
|
|
if (check_reachable) { |
|
title_entry = ohash_first(&title_table, &tslot); |
|
while (NULL != title_entry) { |
|
if ('\0' != *title_entry->file) |
|
say(title_entry->file, |
|
"Probably unreachable, title is %s", |
|
title_entry->title); |
|
free(title_entry->title); |
|
free(title_entry->file); |
|
free(title_entry); |
|
title_entry = ohash_next(&title_table, &tslot); |
|
} |
|
ohash_delete(&title_table); |
|
} |
} |
} |
|
|
/* |
|
* 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) |
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; |
if (NULL == (stream = fopen(mpage->mlinks->file, "r"))) { |
seq = R_FIRST; |
if (warnings) |
while (0 == (ch = (*idx->seq)(idx, &key, &val, seq))) { |
say(mpage->mlinks->file, NULL); |
seq = R_NEXT; |
return; |
*maxrec = *(recno_t *)key.data; |
} |
cp = val.data; |
|
|
|
/* 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))) |
* Read up until the next section into a buffer. |
break; |
* Strip the leading and trailing newline from each read line, |
|
* appending a trailing space. |
|
* Ignore empty (whitespace-only) lines. |
|
*/ |
|
|
/* |
titlesz = 0; |
* Search for the file in those we care about. |
title = NULL; |
* XXX: build this into a tree. Too slow. |
|
*/ |
|
|
|
for (of = ofile; of; of = of->next) |
while (NULL != (line = fgetln(stream, &len))) { |
if (0 == strcmp(fn, of->fname)) |
if (' ' != *line || '\n' != line[len - 1]) |
break; |
break; |
|
while (len > 0 && isspace((unsigned char)*line)) { |
if (NULL == of) |
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; |
assert(NULL == mpage->desc); |
|
mpage->desc = mandoc_strdup(mpage->mlinks->name); |
|
putkey(mpage, mpage->mlinks->name, TYPE_Nd); |
|
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); |
} |
plen -= 2; |
|
|
(*recs)[(int)*reccur] = *maxrec; |
|
(*reccur)++; |
|
} |
} |
|
|
if (ch < 0) { |
assert(NULL == mpage->desc); |
perror(idxf); |
mpage->desc = mandoc_strdup(p); |
exit((int)MANDOCLEVEL_SYSERR); |
putkey(mpage, mpage->desc, TYPE_Nd); |
} else if (1 != ch) { |
fclose(stream); |
fprintf(stderr, "%s: Corrupt index\n", idxf); |
free(title); |
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, const char *value, uint64_t type) |
{ |
{ |
|
|
/* Overshoot by MANDOC_BUFSZ. */ |
assert(NULL != value); |
|
putkeys(mpage, value, strlen(value), type); |
while (buf->len + sz >= buf->size) { |
|
buf->size = buf->len + sz + MANDOC_BUFSZ; |
|
buf->cp = mandoc_realloc(buf->cp, buf->size); |
|
} |
|
|
|
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 mdoc_node *n, uint64_t m) |
{ |
{ |
size_t sz; |
|
|
|
if (0 == (sz = strlen(cp))) |
for ( ; NULL != n; n = n->next) { |
return; |
if (NULL != n->child) |
|
putmdockey(mpage, n->child, m); |
if (buf->len) |
if (MDOC_TEXT == n->type) |
buf->cp[(int)buf->len - 1] = ' '; |
putkey(mpage, n->string, m); |
|
} |
buf_appendb(buf, cp, sz + 1); |
|
} |
} |
|
|
/* |
|
* Recursively add all text from a given node. |
|
* This is optimised for general mdoc nodes in this context, which do |
|
* not consist of subexpressions and having a recursive call for n->next |
|
* would be wasteful. |
|
* The "f" variable should be 0 unless called from pmdoc_Nd for the |
|
* description buffer, which does not start at the beginning of the |
|
* buffer. |
|
*/ |
|
static void |
static void |
buf_appendmdoc(struct buf *buf, const struct mdoc_node *n, int f) |
parse_man(struct mpage *mpage, const struct man_node *n) |
{ |
{ |
|
const struct man_node *head, *body; |
|
char *start, *sv, *title; |
|
char byte; |
|
size_t sz, titlesz; |
|
|
for ( ; n; n = n->next) { |
if (NULL == n) |
if (n->child) |
return; |
buf_appendmdoc(buf, n->child, f); |
|
|
|
if (MDOC_TEXT == n->type && f) { |
/* |
f = 0; |
* We're only searching for one thing: the first text child in |
buf_appendb(buf, n->string, |
* the BODY of a NAME section. Since we don't keep track of |
strlen(n->string) + 1); |
* sections in -man, run some hoops to find out whether we're in |
} else if (MDOC_TEXT == n->type) |
* the correct section or not. |
buf_append(buf, n->string); |
*/ |
|
|
} |
if (MAN_BODY == n->type && MAN_SH == n->tok) { |
} |
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) { |
|
|
static void |
title = NULL; |
hash_reset(DB **db) |
titlesz = 0; |
{ |
|
DB *hash; |
|
|
|
if (NULL != (hash = *db)) |
/* |
(*hash->close)(hash); |
* Suck the entire NAME section into memory. |
|
* Yes, we might run away. |
|
* But too many manuals have big, spread-out |
|
* NAME sections over many lines. |
|
*/ |
|
|
*db = dbopen(NULL, O_CREAT|O_RDWR, 0644, DB_HASH, NULL); |
for ( ; NULL != body; body = body->next) { |
if (NULL == *db) { |
if (MAN_TEXT != body->type) |
perror("hash"); |
break; |
exit((int)MANDOCLEVEL_SYSERR); |
if (0 == (sz = strlen(body->string))) |
} |
continue; |
} |
title = mandoc_realloc |
|
(title, titlesz + sz + 1); |
|
memcpy(title + titlesz, body->string, sz); |
|
titlesz += sz + 1; |
|
title[titlesz - 1] = ' '; |
|
} |
|
if (NULL == title) |
|
return; |
|
|
/* ARGSUSED */ |
title = mandoc_realloc(title, titlesz + 1); |
static int |
title[titlesz] = '\0'; |
pmdoc_head(MDOC_ARGS) |
|
{ |
|
|
|
return(MDOC_HEAD == n->type); |
/* Skip leading space. */ |
|
|
|
sv = title; |
|
while (isspace((unsigned char)*sv)) |
|
sv++; |
|
|
|
if (0 == (sz = strlen(sv))) { |
|
free(title); |
|
return; |
|
} |
|
|
|
/* Erase trailing space. */ |
|
|
|
start = &sv[sz - 1]; |
|
while (start > sv && isspace((unsigned char)*start)) |
|
*start-- = '\0'; |
|
|
|
if (start == sv) { |
|
free(title); |
|
return; |
|
} |
|
|
|
start = sv; |
|
|
|
/* |
|
* Go through a special heuristic dance here. |
|
* Conventionally, one or more manual names are |
|
* comma-specified prior to a whitespace, then a |
|
* dash, then a description. Try to puzzle out |
|
* the name parts here. |
|
*/ |
|
|
|
for ( ;; ) { |
|
sz = strcspn(start, " ,"); |
|
if ('\0' == start[sz]) |
|
break; |
|
|
|
byte = start[sz]; |
|
start[sz] = '\0'; |
|
|
|
putkey(mpage, start, TYPE_Nm); |
|
|
|
if (' ' == byte) { |
|
start += sz + 1; |
|
break; |
|
} |
|
|
|
assert(',' == byte); |
|
start += sz + 1; |
|
while (' ' == *start) |
|
start++; |
|
} |
|
|
|
if (sv == start) { |
|
putkey(mpage, start, TYPE_Nm); |
|
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++; |
|
|
|
assert(NULL == mpage->desc); |
|
mpage->desc = mandoc_strdup(start); |
|
putkey(mpage, mpage->desc, TYPE_Nd); |
|
free(title); |
|
return; |
|
} |
|
} |
|
|
|
for (n = n->child; n; n = n->next) { |
|
if (NULL != mpage->desc) |
|
break; |
|
parse_man(mpage, n); |
|
} |
} |
} |
|
|
/* ARGSUSED */ |
static void |
static int |
parse_mdoc(struct mpage *mpage, const struct mdoc_node *n) |
pmdoc_body(MDOC_ARGS) |
|
{ |
{ |
|
|
return(MDOC_BODY == n->type); |
assert(NULL != n); |
|
for (n = n->child; NULL != n; n = n->next) { |
|
switch (n->type) { |
|
case (MDOC_ELEM): |
|
/* FALLTHROUGH */ |
|
case (MDOC_BLOCK): |
|
/* FALLTHROUGH */ |
|
case (MDOC_HEAD): |
|
/* FALLTHROUGH */ |
|
case (MDOC_BODY): |
|
/* FALLTHROUGH */ |
|
case (MDOC_TAIL): |
|
if (NULL != mdocs[n->tok].fp) |
|
if (0 == (*mdocs[n->tok].fp)(mpage, n)) |
|
break; |
|
if (mdocs[n->tok].mask) |
|
putmdockey(mpage, n->child, |
|
mdocs[n->tok].mask); |
|
break; |
|
default: |
|
assert(MDOC_ROOT != n->type); |
|
continue; |
|
} |
|
if (NULL != n->child) |
|
parse_mdoc(mpage, n); |
|
} |
} |
} |
|
|
/* ARGSUSED */ |
|
static int |
static int |
pmdoc_Fd(MDOC_ARGS) |
parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n) |
{ |
{ |
const char *start, *end; |
const char *start, *end; |
size_t sz; |
size_t sz; |
|
|
if (SEC_SYNOPSIS != n->sec) |
if (SEC_SYNOPSIS != n->sec || |
|
NULL == (n = n->child) || |
|
MDOC_TEXT != n->type) |
return(0); |
return(0); |
if (NULL == (n = n->child) || MDOC_TEXT != n->type) |
|
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); |
|
|
Line 931 pmdoc_Fd(MDOC_ARGS) |
|
Line 1467 pmdoc_Fd(MDOC_ARGS) |
|
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 int |
static int |
pmdoc_In(MDOC_ARGS) |
parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n) |
{ |
{ |
|
|
if (NULL == n->child || MDOC_TEXT != n->child->type) |
|
return(0); |
|
|
|
buf_append(buf, n->child->string); |
|
return(1); |
|
} |
|
|
|
/* ARGSUSED */ |
|
static int |
|
pmdoc_Fn(MDOC_ARGS) |
|
{ |
|
struct mdoc_node *nn; |
|
const char *cp; |
const char *cp; |
|
|
nn = n->child; |
if (NULL == (n = n->child) || MDOC_TEXT != n->type) |
|
|
if (NULL == nn || MDOC_TEXT != nn->type) |
|
return(0); |
return(0); |
|
|
/* .Fn "struct type *name" "char *arg" */ |
/* |
|
* Parse: .Fn "struct type *name" "char *arg". |
|
* First strip away pointer symbol. |
|
* Then store the function name, then type. |
|
* Finally, store the arguments. |
|
*/ |
|
|
cp = strrchr(nn->string, ' '); |
if (NULL == (cp = strrchr(n->string, ' '))) |
if (NULL == cp) |
cp = n->string; |
cp = nn->string; |
|
|
|
/* Strip away pointer symbol. */ |
|
|
|
while ('*' == *cp) |
while ('*' == *cp) |
cp++; |
cp++; |
|
|
/* Store the function name. */ |
putkey(mpage, cp, TYPE_Fn); |
|
|
buf_append(buf, cp); |
if (n->string < cp) |
hash_put(hash, buf, TYPE_Fn); |
putkeys(mpage, n->string, cp - n->string, TYPE_Ft); |
|
|
/* Store the function type. */ |
for (n = n->next; NULL != n; n = n->next) |
|
if (MDOC_TEXT == n->type) |
|
putkey(mpage, n->string, TYPE_Fa); |
|
|
if (nn->string < cp) { |
|
buf->len = 0; |
|
buf_appendb(buf, nn->string, cp - nn->string); |
|
buf_appendb(buf, "", 1); |
|
hash_put(hash, buf, TYPE_Ft); |
|
} |
|
|
|
/* Store the arguments. */ |
|
|
|
for (nn = nn->next; nn; nn = nn->next) { |
|
if (MDOC_TEXT != nn->type) |
|
continue; |
|
buf->len = 0; |
|
buf_append(buf, nn->string); |
|
hash_put(hash, buf, TYPE_Fa); |
|
} |
|
|
|
return(0); |
return(0); |
} |
} |
|
|
/* ARGSUSED */ |
|
static int |
static int |
pmdoc_St(MDOC_ARGS) |
parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n) |
{ |
{ |
|
char *cp; |
|
|
if (NULL == n->child || MDOC_TEXT != n->child->type) |
if (NULL == (n = n->child)) |
return(0); |
return(0); |
|
|
buf_append(buf, n->child->string); |
if (NULL == n->next) { |
return(1); |
putkey(mpage, n->string, TYPE_Xr); |
|
return(0); |
|
} |
|
|
|
if (-1 == asprintf(&cp, "%s(%s)", n->string, n->next->string)) { |
|
perror(NULL); |
|
exit((int)MANDOCLEVEL_SYSERR); |
|
} |
|
putkey(mpage, cp, TYPE_Xr); |
|
free(cp); |
|
return(0); |
} |
} |
|
|
/* ARGSUSED */ |
|
static int |
static int |
pmdoc_Xr(MDOC_ARGS) |
parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n) |
{ |
{ |
|
size_t sz; |
|
|
if (NULL == (n = n->child)) |
if (MDOC_BODY != n->type) |
return(0); |
return(0); |
|
|
buf_appendb(buf, n->string, strlen(n->string)); |
/* |
|
* Special-case the `Nd' because we need to put the description |
|
* into the document table. |
|
*/ |
|
|
if (NULL != (n = n->next)) { |
for (n = n->child; NULL != n; n = n->next) { |
buf_appendb(buf, ".", 1); |
if (MDOC_TEXT == n->type) { |
buf_appendb(buf, n->string, strlen(n->string) + 1); |
if (NULL != mpage->desc) { |
} else |
sz = strlen(mpage->desc) + |
buf_appendb(buf, ".", 2); |
strlen(n->string) + 2; |
|
mpage->desc = mandoc_realloc( |
|
mpage->desc, sz); |
|
strlcat(mpage->desc, " ", sz); |
|
strlcat(mpage->desc, n->string, sz); |
|
} else |
|
mpage->desc = mandoc_strdup(n->string); |
|
} |
|
if (NULL != n->child) |
|
parse_mdoc_Nd(mpage, n); |
|
} |
return(1); |
return(1); |
} |
} |
|
|
/* ARGSUSED */ |
|
static int |
static int |
pmdoc_Nd(MDOC_ARGS) |
parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n) |
{ |
{ |
|
|
if (MDOC_BODY != n->type) |
return(SEC_NAME == n->sec || |
return(0); |
(SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type)); |
|
|
buf_appendmdoc(dbuf, n->child, 1); |
|
return(1); |
|
} |
} |
|
|
/* ARGSUSED */ |
|
static int |
static int |
pmdoc_Nm(MDOC_ARGS) |
parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n) |
{ |
{ |
|
|
if (SEC_NAME == n->sec) |
return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type); |
return(1); |
} |
else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type) |
|
return(0); |
|
|
|
if (NULL == n->child) |
static int |
buf_append(buf, m->name); |
parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n) |
|
{ |
|
|
return(1); |
return(MDOC_HEAD == n->type); |
} |
} |
|
|
/* ARGSUSED */ |
|
static int |
static int |
pmdoc_Sh(MDOC_ARGS) |
parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n) |
{ |
{ |
|
|
return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type); |
return(MDOC_BODY == n->type); |
} |
} |
|
|
|
/* |
|
* 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, |
|
const char *cp, size_t sz, uint64_t v) |
{ |
{ |
DBT key, val; |
struct str *s; |
int rc; |
unsigned int slot; |
|
const char *end; |
|
|
if (buf->len < 2) |
if (0 == sz) |
return; |
return; |
|
|
key.data = buf->cp; |
end = cp + sz; |
key.size = buf->len; |
slot = ohash_qlookupi(&strings, cp, &end); |
|
s = ohash_find(&strings, slot); |
|
|
if ((rc = (*db->get)(db, &key, &val, 0)) < 0) { |
if (NULL != s && mpage == s->mpage) { |
perror("hash"); |
s->mask |= v; |
exit((int)MANDOCLEVEL_SYSERR); |
return; |
} else if (0 == rc) |
} else if (NULL == s) { |
mask |= *(uint64_t *)val.data; |
s = mandoc_calloc(sizeof(struct str) + sz + 1, 1); |
|
memcpy(s->key, cp, sz); |
val.data = &mask; |
ohash_insert(&strings, slot, s); |
val.size = sizeof(uint64_t); |
} |
|
s->mpage = mpage; |
if ((rc = (*db->put)(db, &key, &val, 0)) < 0) { |
s->mask = v; |
perror("hash"); |
|
exit((int)MANDOCLEVEL_SYSERR); |
|
} |
|
} |
} |
|
|
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 |
* Store the rendered version of a key, or alias the pointer |
* key. If the macro sets the database key, flush it to the database. |
* if the key contains no escape sequences. |
*/ |
*/ |
static void |
static void |
pmdoc_node(MDOC_ARGS) |
render_key(struct mchars *mc, struct str *key) |
{ |
{ |
|
size_t sz, bsz, pos; |
|
char utfbuf[7], res[5]; |
|
char *buf; |
|
const char *seq, *cpp, *val; |
|
int len, u; |
|
enum mandoc_esc esc; |
|
|
if (NULL == n) |
assert(NULL == key->rendered); |
|
|
|
res[0] = '\\'; |
|
res[1] = '\t'; |
|
res[2] = ASCII_NBRSP; |
|
res[3] = ASCII_HYPH; |
|
res[4] = '\0'; |
|
|
|
val = key->key; |
|
bsz = strlen(val); |
|
|
|
/* |
|
* Pre-check: if we have no stop-characters, then set the |
|
* pointer as ourselvse and get out of here. |
|
*/ |
|
if (strcspn(val, res) == bsz) { |
|
key->rendered = key->key; |
return; |
return; |
|
} |
|
|
switch (n->type) { |
/* Pre-allocate by the length of the input */ |
case (MDOC_HEAD): |
|
/* FALLTHROUGH */ |
|
case (MDOC_BODY): |
|
/* FALLTHROUGH */ |
|
case (MDOC_TAIL): |
|
/* FALLTHROUGH */ |
|
case (MDOC_BLOCK): |
|
/* FALLTHROUGH */ |
|
case (MDOC_ELEM): |
|
buf->len = 0; |
|
|
|
|
buf = mandoc_malloc(++bsz); |
|
pos = 0; |
|
|
|
while ('\0' != *val) { |
/* |
/* |
* Both NULL handlers and handlers returning true |
* Halt on the first escape sequence. |
* request using the data. Only skip the element |
* This also halts on the end of string, in which case |
* when the handler returns false. |
* we just copy, fallthrough, and exit the loop. |
*/ |
*/ |
|
if ((sz = strcspn(val, res)) > 0) { |
|
memcpy(&buf[pos], val, sz); |
|
pos += sz; |
|
val += sz; |
|
} |
|
|
if (NULL != mdocs[n->tok].fp && |
if (ASCII_HYPH == *val) { |
0 == (*mdocs[n->tok].fp)(hash, buf, dbuf, n, m)) |
buf[pos++] = '-'; |
|
val++; |
|
continue; |
|
} else if ('\t' == *val || ASCII_NBRSP == *val) { |
|
buf[pos++] = ' '; |
|
val++; |
|
continue; |
|
} else if ('\\' != *val) |
break; |
break; |
|
|
|
/* Read past the slash. */ |
|
|
|
val++; |
|
|
/* |
/* |
* For many macros, use the text from all children. |
* Parse the escape sequence and see if it's a |
* Set zero flags for macros not needing this. |
* predefined character or special character. |
* In that case, the handler must fill the buffer. |
|
*/ |
*/ |
|
|
if (MDOCF_CHILD & mdocs[n->tok].flags) |
esc = mandoc_escape |
buf_appendmdoc(buf, n->child, 0); |
((const char **)&val, &seq, &len); |
|
if (ESCAPE_ERROR == esc) |
|
break; |
|
if (ESCAPE_SPECIAL != esc) |
|
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); |
if (0 == (u = mchars_spec2cp(mc, seq, len))) |
break; |
continue; |
default: |
cpp = utfbuf; |
break; |
if (0 == (sz = utf8(u, utfbuf))) |
} |
continue; |
|
sz = strlen(cpp); |
pmdoc_node(hash, buf, dbuf, n->child, m); |
} else { |
pmdoc_node(hash, buf, dbuf, n->next, m); |
cpp = mchars_spec2str(mc, seq, len, &sz); |
} |
if (NULL == cpp) |
|
continue; |
static int |
if (ASCII_NBRSP == *cpp) { |
pman_node(MAN_ARGS) |
cpp = " "; |
{ |
sz = 1; |
const struct man_node *head, *body; |
|
const char *start, *sv; |
|
size_t sz; |
|
|
|
if (NULL == n) |
|
return(0); |
|
|
|
/* |
|
* We're only searching for one thing: the first text child in |
|
* the BODY of a NAME section. Since we don't keep track of |
|
* sections in -man, run some hoops to find out whether we're in |
|
* the correct section or not. |
|
*/ |
|
|
|
if (MAN_BODY == n->type && MAN_SH == n->tok) { |
|
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); |
|
start = sv = body->string; |
|
|
|
/* |
|
* Go through a special heuristic dance here. |
|
* This is why -man manuals are great! |
|
* (I'm being sarcastic: my eyes are bleeding.) |
|
* Conventionally, one or more manual names are |
|
* comma-specified prior to a whitespace, then a |
|
* dash, then a description. Try to puzzle out |
|
* the name parts here. |
|
*/ |
|
|
|
for ( ;; ) { |
|
sz = strcspn(start, " ,"); |
|
if ('\0' == start[(int)sz]) |
|
break; |
|
|
|
buf->len = 0; |
|
buf_appendb(buf, start, sz); |
|
buf_appendb(buf, "", 1); |
|
|
|
hash_put(hash, buf, TYPE_Nm); |
|
|
|
if (' ' == start[(int)sz]) { |
|
start += (int)sz + 1; |
|
break; |
|
} |
|
|
|
assert(',' == start[(int)sz]); |
|
start += (int)sz + 1; |
|
while (' ' == *start) |
|
start++; |
|
} |
} |
|
} |
|
|
buf->len = 0; |
/* Copy the rendered glyph into the stream. */ |
|
|
if (sv == start) { |
bsz += sz; |
buf_append(buf, start); |
buf = mandoc_realloc(buf, bsz); |
return(1); |
memcpy(&buf[pos], cpp, sz); |
} |
pos += sz; |
|
|
while (' ' == *start) |
|
start++; |
|
|
|
if (0 == strncmp(start, "-", 1)) |
|
start += 1; |
|
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++; |
|
|
|
sz = strlen(start) + 1; |
|
buf_appendb(dbuf, start, sz); |
|
buf_appendb(buf, start, sz); |
|
|
|
hash_put(hash, buf, TYPE_Nd); |
|
} |
|
} |
} |
|
|
for (n = n->child; n; n = n->next) |
buf[pos] = '\0'; |
if (pman_node(hash, buf, dbuf, n)) |
key->rendered = buf; |
return(1); |
|
|
|
return(0); |
|
} |
} |
|
|
/* |
/* |
* Parse a formatted manual page. |
* Flush the current page's terms (and their bits) into the database. |
* By necessity, this involves rather crude guesswork. |
* Wrap the entire set of additions in a transaction to make sqlite be a |
|
* little faster. |
|
* Also, handle escape sequences at the last possible moment. |
*/ |
*/ |
static void |
static void |
pformatted(DB *hash, struct buf *buf, struct buf *dbuf, |
dbindex(const struct mpage *mpage, struct mchars *mc) |
const struct of *of) |
|
{ |
{ |
FILE *stream; |
struct mlink *mlink; |
char *line, *p; |
struct str *key; |
size_t len, plen; |
const char *desc; |
|
int64_t recno; |
|
size_t i; |
|
unsigned int slot; |
|
|
if (NULL == (stream = fopen(of->fname, "r"))) { |
if (verb) |
perror(of->fname); |
say(mpage->mlinks->file, "Adding to index"); |
return; |
|
} |
|
|
|
/* |
if (nodb) |
* Always use the title derived from the filename up front, |
|
* 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; |
|
buf_append(buf, of->title); |
|
hash_put(hash, buf, TYPE_Nm); |
|
|
|
/* Skip to first blank line. */ |
|
|
|
while (NULL != (line = fgetln(stream, &len))) |
|
if ('\n' == *line) |
|
break; |
|
|
|
/* |
|
* Assume the first line that is not indented |
|
* is the first section header. Skip to it. |
|
*/ |
|
|
|
while (NULL != (line = fgetln(stream, &len))) |
|
if ('\n' != *line && ' ' != *line) |
|
break; |
|
|
|
/* |
|
* 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; |
|
|
|
desc = ""; |
|
if (NULL != mpage->desc && '\0' != *mpage->desc) { |
|
key = ohash_find(&strings, |
|
ohash_qlookup(&strings, mpage->desc)); |
|
assert(NULL != key); |
|
if (NULL == key->rendered) |
|
render_key(mc, key); |
|
desc = key->rendered; |
} |
} |
|
|
line[(int)--len] = '\0'; |
SQL_EXEC("BEGIN TRANSACTION"); |
|
|
/* |
i = 1; |
* Skip to the first dash. |
SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, desc); |
* Use the remaining line as the description (no more than 70 |
SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form); |
* bytes). |
SQL_STEP(stmts[STMT_INSERT_PAGE]); |
*/ |
recno = sqlite3_last_insert_rowid(db); |
|
sqlite3_reset(stmts[STMT_INSERT_PAGE]); |
|
|
if (NULL != (p = strstr(line, "- "))) { |
for (mlink = mpage->mlinks; mlink; mlink = mlink->next) { |
for (p += 2; ' ' == *p || '\b' == *p; p++) |
i = 1; |
/* Skip to next word. */ ; |
SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->file); |
} else |
SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec); |
p = line; |
SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch); |
|
SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name); |
if ((plen = strlen(p)) > 70) { |
SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, recno); |
plen = 70; |
SQL_STEP(stmts[STMT_INSERT_LINK]); |
p[plen] = '\0'; |
sqlite3_reset(stmts[STMT_INSERT_LINK]); |
} |
} |
|
|
/* Strip backspace-encoding from line. */ |
for (key = ohash_first(&strings, &slot); NULL != key; |
|
key = ohash_next(&strings, &slot)) { |
while (NULL != (line = memchr(p, '\b', plen))) { |
assert(key->mpage == mpage); |
len = line - p; |
if (NULL == key->rendered) |
if (0 == len) { |
render_key(mc, key); |
memmove(line, line + 1, plen--); |
i = 1; |
continue; |
SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask); |
} |
SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered); |
memmove(line - 1, line + 1, plen - len); |
SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, recno); |
plen -= 2; |
SQL_STEP(stmts[STMT_INSERT_KEY]); |
|
sqlite3_reset(stmts[STMT_INSERT_KEY]); |
|
if (key->rendered != key->key) |
|
free(key->rendered); |
|
free(key); |
} |
} |
|
|
buf_appendb(dbuf, p, plen + 1); |
SQL_EXEC("END TRANSACTION"); |
buf->len = 0; |
|
buf_appendb(buf, p, plen + 1); |
|
hash_put(hash, buf, TYPE_Nd); |
|
fclose(stream); |
|
} |
} |
|
|
static void |
static void |
ofile_argbuild(int argc, char *argv[], struct of **of) |
dbprune(void) |
{ |
{ |
char buf[MAXPATHLEN]; |
struct mpage *mpage; |
char *sec, *arch, *title, *p; |
struct mlink *mlink; |
int i, src_form; |
size_t i; |
struct of *nof; |
unsigned int slot; |
|
|
for (i = 0; i < argc; i++) { |
if (nodb) |
|
return; |
|
|
/* |
mpage = ohash_first(&mpages, &slot); |
* Try to infer the manual section, architecture and |
while (NULL != mpage) { |
* page title from the path, assuming it looks like |
mlink = mpage->mlinks; |
* man*[/<arch>]/<title>.<section> or |
i = 1; |
* cat<section>[/<arch>]/<title>.0 |
SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE], i, mlink->file); |
*/ |
SQL_STEP(stmts[STMT_DELETE_PAGE]); |
|
sqlite3_reset(stmts[STMT_DELETE_PAGE]); |
|
if (verb) |
|
say(mlink->file, "Deleted from index"); |
|
mpage = ohash_next(&mpages, &slot); |
|
} |
|
} |
|
|
if (strlcpy(buf, argv[i], sizeof(buf)) >= sizeof(buf)) { |
/* |
fprintf(stderr, "%s: Path too long\n", argv[i]); |
* Close an existing database and its prepared statements. |
continue; |
* If "real" is not set, rename the temporary file into the real one. |
} |
*/ |
sec = arch = title = NULL; |
static void |
src_form = 0; |
dbclose(int real) |
p = strrchr(buf, '\0'); |
{ |
while (p-- > buf) { |
size_t i; |
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; |
|
|
|
/* |
if (nodb) |
* Build the file structure. |
return; |
*/ |
|
|
|
nof = mandoc_calloc(1, sizeof(struct of)); |
for (i = 0; i < STMT__MAX; i++) { |
nof->fname = mandoc_strdup(argv[i]); |
sqlite3_finalize(stmts[i]); |
if (NULL != sec) |
stmts[i] = NULL; |
nof->sec = mandoc_strdup(sec); |
} |
if (NULL != arch) |
|
nof->arch = mandoc_strdup(arch); |
|
nof->title = mandoc_strdup(title); |
|
nof->src_form = src_form; |
|
|
|
/* |
sqlite3_close(db); |
* Add the structure to the list. |
db = NULL; |
*/ |
|
|
|
if (verb > 2) |
if (real) |
printf("%s: Scheduling\n", argv[i]); |
return; |
if (NULL == *of) { |
|
*of = nof; |
if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) { |
(*of)->first = nof; |
exitcode = (int)MANDOCLEVEL_SYSERR; |
} else { |
say(MANDOC_DB, NULL); |
nof->first = (*of)->first; |
|
(*of)->next = nof; |
|
*of = nof; |
|
} |
|
} |
} |
} |
} |
|
|
/* |
/* |
* Recursively build up a list of files to parse. |
* This is straightforward stuff. |
* We use this instead of ftw() and so on because I don't want global |
* Open a database connection to a "temporary" database, then open a set |
* variables hanging around. |
* of prepared statements we'll use over and over again. |
* This ignores the mandoc.db and mandoc.index files, but assumes that |
* If "real" is set, we use the existing database; if not, we truncate a |
* everything else is a manual. |
* temporary one. |
* Pass in a pointer to a NULL structure for the first invocation. |
* Must be matched by dbclose(). |
*/ |
*/ |
static int |
static int |
ofile_dirbuild(const char *dir, const char* psec, const char *parch, |
dbopen(int real) |
int p_src_form, struct of **of) |
|
{ |
{ |
char buf[MAXPATHLEN]; |
const char *file, *sql; |
struct stat sb; |
int rc, ofl; |
size_t sz; |
|
DIR *d; |
|
const char *fn, *sec, *arch; |
|
char *p, *q, *suffix; |
|
struct of *nof; |
|
struct dirent *dp; |
|
int src_form; |
|
|
|
if (NULL == (d = opendir(dir))) { |
if (nodb) |
perror(dir); |
return(1); |
|
|
|
ofl = SQLITE_OPEN_READWRITE; |
|
if (0 == real) { |
|
file = MANDOC_DB "~"; |
|
if (-1 == remove(file) && ENOENT != errno) { |
|
exitcode = (int)MANDOCLEVEL_SYSERR; |
|
say(file, NULL); |
|
return(0); |
|
} |
|
ofl |= SQLITE_OPEN_EXCLUSIVE; |
|
} else |
|
file = MANDOC_DB; |
|
|
|
rc = sqlite3_open_v2(file, &db, ofl, NULL); |
|
if (SQLITE_OK == rc) |
|
goto prepare_statements; |
|
if (SQLITE_CANTOPEN != rc) { |
|
exitcode = (int)MANDOCLEVEL_SYSERR; |
|
say(file, NULL); |
return(0); |
return(0); |
} |
} |
|
|
while (NULL != (dp = readdir(d))) { |
sqlite3_close(db); |
fn = dp->d_name; |
db = NULL; |
|
|
if ('.' == *fn) |
if (SQLITE_OK != (rc = sqlite3_open(file, &db))) { |
continue; |
exitcode = (int)MANDOCLEVEL_SYSERR; |
|
say(file, NULL); |
|
return(0); |
|
} |
|
|
src_form = p_src_form; |
sql = "CREATE TABLE \"mpages\" (\n" |
|
" \"desc\" TEXT NOT NULL,\n" |
|
" \"form\" INTEGER NOT NULL,\n" |
|
" \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n" |
|
");\n" |
|
"\n" |
|
"CREATE TABLE \"mlinks\" (\n" |
|
" \"file\" TEXT NOT NULL,\n" |
|
" \"sec\" TEXT NOT NULL,\n" |
|
" \"arch\" TEXT NOT NULL,\n" |
|
" \"name\" TEXT NOT NULL,\n" |
|
" \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) " |
|
"ON DELETE CASCADE,\n" |
|
" \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n" |
|
");\n" |
|
"\n" |
|
"CREATE TABLE \"keys\" (\n" |
|
" \"bits\" INTEGER NOT NULL,\n" |
|
" \"key\" TEXT NOT NULL,\n" |
|
" \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) " |
|
"ON DELETE CASCADE,\n" |
|
" \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n" |
|
");\n" |
|
"\n" |
|
"CREATE INDEX \"key_index\" ON keys (key);\n"; |
|
|
if (DT_DIR == dp->d_type) { |
if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) { |
sec = psec; |
exitcode = (int)MANDOCLEVEL_SYSERR; |
arch = parch; |
say(file, "%s", sqlite3_errmsg(db)); |
|
return(0); |
|
} |
|
|
/* |
prepare_statements: |
* By default, only use directories called: |
SQL_EXEC("PRAGMA foreign_keys = ON"); |
* man<section>/[<arch>/] or |
sql = "DELETE FROM mpages where file=?"; |
* cat<section>/[<arch>/] |
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 " |
|
"(file,sec,arch,name,pageid) VALUES (?,?,?,?,?)"; |
|
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL); |
|
sql = "INSERT INTO keys " |
|
"(bits,key,pageid) VALUES (?,?,?)"; |
|
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL); |
|
|
if (NULL == sec) { |
#ifndef __APPLE__ |
if(0 == strncmp("man", fn, 3)) { |
/* |
src_form |= MANDOC_SRC; |
* When opening a new database, we can turn off |
sec = fn + 3; |
* synchronous mode for much better performance. |
} else if (0 == strncmp("cat", fn, 3)) { |
*/ |
src_form |= MANDOC_FORM; |
|
sec = fn + 3; |
|
} else if (use_all) |
|
sec = fn; |
|
else |
|
continue; |
|
} else if (NULL == arch && (use_all || |
|
NULL == strchr(fn, '.'))) |
|
arch = fn; |
|
else if (0 == use_all) |
|
continue; |
|
|
|
buf[0] = '\0'; |
if (real) |
strlcat(buf, dir, MAXPATHLEN); |
SQL_EXEC("PRAGMA synchronous = OFF"); |
strlcat(buf, "/", MAXPATHLEN); |
#endif |
sz = strlcat(buf, fn, MAXPATHLEN); |
|
|
|
if (MAXPATHLEN <= sz) { |
return(1); |
fprintf(stderr, "%s: Path too long\n", dir); |
} |
return(0); |
|
} |
|
|
|
if (verb > 2) |
|
printf("%s: Scanning\n", buf); |
|
|
|
if ( ! ofile_dirbuild(buf, sec, arch, |
static void * |
src_form, of)) |
hash_halloc(size_t sz, void *arg) |
return(0); |
{ |
} |
|
if (DT_REG != dp->d_type || |
|
(NULL == psec && !use_all) || |
|
!strcmp(MANDOC_DB, fn) || |
|
!strcmp(MANDOC_IDX, fn)) |
|
continue; |
|
|
|
/* |
return(mandoc_calloc(sz, 1)); |
* By default, skip files where the file name suffix |
} |
* does not agree with the section directory |
|
* they are located in. |
|
*/ |
|
|
|
suffix = strrchr(fn, '.'); |
static void * |
if (0 == use_all) { |
hash_alloc(size_t sz, void *arg) |
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; |
|
} |
|
|
|
|
return(mandoc_malloc(sz)); |
|
} |
|
|
/* |
static void |
* Skip formatted manuals if a source version is |
hash_free(void *p, size_t sz, void *arg) |
* available. Ignore the age: it is very unlikely |
{ |
* that people install newer formatted base manuals |
|
* when they used to have source manuals before, |
|
* and in ports, old manuals get removed on update. |
|
*/ |
|
if (0 == use_all && MANDOC_FORM & src_form && |
|
NULL != psec) { |
|
buf[0] = '\0'; |
|
strlcat(buf, dir, MAXPATHLEN); |
|
p = strrchr(buf, '/'); |
|
if (NULL != parch && NULL != p) |
|
for (p--; p > buf; p--) |
|
if ('/' == *p) |
|
break; |
|
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; |
|
} |
|
} |
|
|
|
buf[0] = '\0'; |
free(p); |
strlcat(buf, dir, MAXPATHLEN); |
} |
strlcat(buf, "/", MAXPATHLEN); |
|
sz = strlcat(buf, fn, MAXPATHLEN); |
|
if (sz >= MAXPATHLEN) { |
|
fprintf(stderr, "%s: Path too long\n", dir); |
|
continue; |
|
} |
|
|
|
nof = mandoc_calloc(1, sizeof(struct of)); |
static int |
nof->fname = mandoc_strdup(buf); |
set_basedir(const char *targetdir) |
if (NULL != psec) |
{ |
nof->sec = mandoc_strdup(psec); |
static char startdir[PATH_MAX]; |
if (NULL != parch) |
static int fd; |
nof->arch = mandoc_strdup(parch); |
|
nof->src_form = src_form; |
|
|
|
/* |
/* |
* Remember the file name without the extension, |
* Remember where we started by keeping a fd open to the origin |
* to be used as the page title in the database. |
* path component: throughout this utility, we chdir() a lot to |
*/ |
* handle relative paths, and by doing this, we can return to |
|
* the starting point. |
if (NULL != suffix) |
*/ |
*suffix = '\0'; |
if ('\0' == *startdir) { |
nof->title = mandoc_strdup(fn); |
if (NULL == getcwd(startdir, PATH_MAX)) { |
|
exitcode = (int)MANDOCLEVEL_SYSERR; |
/* |
if (NULL != targetdir) |
* Add the structure to the list. |
say(".", NULL); |
*/ |
return(0); |
|
|
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; |
|
} |
} |
|
if (-1 == (fd = open(startdir, O_RDONLY, 0))) { |
|
exitcode = (int)MANDOCLEVEL_SYSERR; |
|
say(startdir, NULL); |
|
return(0); |
|
} |
|
if (NULL == targetdir) |
|
targetdir = startdir; |
|
} else { |
|
if (-1 == fd) |
|
return(0); |
|
if (-1 == fchdir(fd)) { |
|
close(fd); |
|
basedir[0] = '\0'; |
|
exitcode = (int)MANDOCLEVEL_SYSERR; |
|
say(startdir, NULL); |
|
return(0); |
|
} |
|
if (NULL == targetdir) { |
|
close(fd); |
|
return(1); |
|
} |
} |
} |
|
if (NULL == realpath(targetdir, basedir)) { |
closedir(d); |
basedir[0] = '\0'; |
|
exitcode = (int)MANDOCLEVEL_BADARG; |
|
say(targetdir, NULL); |
|
return(0); |
|
} else if (-1 == chdir(basedir)) { |
|
exitcode = (int)MANDOCLEVEL_BADARG; |
|
say("", NULL); |
|
return(0); |
|
} |
return(1); |
return(1); |
} |
} |
|
|
static void |
static void |
ofile_free(struct of *of) |
say(const char *file, const char *format, ...) |
{ |
{ |
struct of *nof; |
va_list ap; |
|
|
while (of) { |
if ('\0' != *basedir) |
nof = of->next; |
fprintf(stderr, "%s", basedir); |
free(of->fname); |
if ('\0' != *basedir && '\0' != *file) |
free(of->sec); |
fputs("//", stderr); |
free(of->arch); |
if ('\0' != *file) |
free(of->title); |
fprintf(stderr, "%s", file); |
free(of); |
fputs(": ", stderr); |
of = nof; |
|
|
if (NULL == format) { |
|
perror(NULL); |
|
return; |
} |
} |
} |
|
|
|
static void |
va_start(ap, format); |
usage(void) |
vfprintf(stderr, format, ap); |
{ |
va_end(ap); |
|
|
fprintf(stderr, "usage: %s [-v] " |
fputc('\n', stderr); |
"[-C file] |" |
|
" dir ... |" |
|
" -d dir [file ...] |" |
|
" -u dir [file ...]\n", progname); |
|
} |
} |