| version 1.32, 2011/12/10 16:53:39 |
version 1.51, 2012/06/08 12:05:27 |
|
|
| /* $Id$ */ |
/* $Id$ */ |
| /* |
/* |
| * 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 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 |
|
|
| #endif |
#endif |
| |
|
| #include <sys/param.h> |
#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 <stdio.h> |
#include <stddef.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 <ohash.h> |
| # include <endian.h> |
#include <sqlite3.h> |
| # include <db_185.h> |
|
| #elif defined(__APPLE__) |
|
| # include <libkern/OSByteOrder.h> |
|
| # include <db.h> |
|
| #else |
|
| # include <db.h> |
|
| #endif |
|
| |
|
| #include "man.h" |
|
| #include "mdoc.h" |
#include "mdoc.h" |
| |
#include "man.h" |
| #include "mandoc.h" |
#include "mandoc.h" |
| #include "mandocdb.h" |
#include "mandocdb.h" |
| #include "manpath.h" |
#include "manpath.h" |
| |
|
| #define MANDOC_BUFSZ BUFSIZ |
/* Post a warning to stderr. */ |
| #define MANDOC_SLOP 1024 |
#define WARNING(_f, _b, _fmt, _args...) \ |
| |
do if (warnings) { \ |
| |
fprintf(stderr, "%s: ", (_b)); \ |
| |
fprintf(stderr, (_fmt), ##_args); \ |
| |
if ('\0' != *(_f)) \ |
| |
fprintf(stderr, ": %s", (_f)); \ |
| |
fprintf(stderr, "\n"); \ |
| |
} while (/* CONSTCOND */ 0) |
| |
/* Post a "verbose" message to stderr. */ |
| |
#define DEBUG(_f, _b, _fmt, _args...) \ |
| |
do if (verb) { \ |
| |
fprintf(stderr, "%s: ", (_b)); \ |
| |
fprintf(stderr, (_fmt), ##_args); \ |
| |
fprintf(stderr, ": %s\n", (_f)); \ |
| |
} while (/* CONSTCOND */ 0) |
| |
|
| #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_SRC, /* format is -man or -mdoc */ |
| |
FORM_CAT, /* format is cat */ |
| |
FORM_NONE /* format is unknown */ |
| |
}; |
| |
|
| struct of { |
struct str { |
| char *fname; /* heap-allocated */ |
char *utf8; /* key in UTF-8 form */ |
| char *sec; |
const struct of *of; /* if set, the owning parse */ |
| char *arch; |
struct str *next; /* next in owning parse sequence */ |
| char *title; |
uint64_t mask; /* bitmask in sequence */ |
| int src_form; |
char key[]; /* the string itself */ |
| struct of *next; /* NULL for last one */ |
|
| struct of *first; /* first in list */ |
|
| }; |
}; |
| |
|
| /* Buffer for storing growable data. */ |
struct id { |
| |
ino_t ino; |
| |
dev_t dev; |
| |
}; |
| |
|
| struct buf { |
struct of { |
| char *cp; |
struct id id; /* used for hashing routine */ |
| size_t len; /* current length */ |
struct of *next; /* next in ofs */ |
| size_t size; /* total buffer size */ |
enum form dform; /* path-cued form */ |
| |
enum form sform; /* suffix-cued form */ |
| |
char file[MAXPATHLEN]; /* filename rel. to manpath */ |
| |
const char *desc; /* parsed description */ |
| |
const char *sec; /* suffix-cued section (or empty) */ |
| |
const char *dsec; /* path-cued section (or empty) */ |
| |
const char *arch; /* path-cued arch. (or empty) */ |
| |
const char *name; /* name (from filename) (not empty) */ |
| }; |
}; |
| |
|
| /* Operation we're going to perform. */ |
enum stmt { |
| |
STMT_DELETE = 0, /* delete manpage */ |
| enum op { |
STMT_INSERT_DOC, /* insert manpage */ |
| OP_NEW = 0, /* new database */ |
STMT_INSERT_KEY, /* insert parsed key */ |
| OP_UPDATE, /* delete/add entries in existing database */ |
STMT__MAX |
| OP_DELETE /* delete entries from existing database */ |
|
| }; |
}; |
| |
|
| #define MAN_ARGS DB *hash, \ |
typedef int (*mdoc_fp)(struct of *, const struct mdoc_node *); |
| struct buf *buf, \ |
|
| struct buf *dbuf, \ |
|
| const struct man_node *n |
|
| #define MDOC_ARGS DB *hash, \ |
|
| struct buf *buf, \ |
|
| struct buf *dbuf, \ |
|
| const struct mdoc_node *n, \ |
|
| const struct mdoc_meta *m |
|
| |
|
| static void buf_appendmdoc(struct buf *, |
|
| 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. */ |
int flags; /* for use by pmdoc_node */ |
| |
#define MDOCF_CHILD 0x01 /* automatically index child nodes */ |
| }; |
}; |
| |
|
| |
static void dbclose(const char *, int); |
| |
static void dbindex(struct mchars *, int, |
| |
const struct of *, const char *); |
| |
static int dbopen(const char *, int); |
| |
static void dbprune(const char *); |
| |
static void fileadd(struct of *); |
| |
static int filecheck(const char *); |
| |
static void filescan(const char *, const char *); |
| |
static struct str *hashget(const char *, size_t); |
| |
static void *hash_alloc(size_t, void *); |
| |
static void hash_free(void *, size_t, void *); |
| |
static void *hash_halloc(size_t, void *); |
| |
static void inoadd(const struct stat *, struct of *); |
| |
static int inocheck(const struct stat *); |
| |
static void ofadd(const char *, int, const char *, |
| |
const char *, const char *, const char *, |
| |
const char *, const struct stat *); |
| |
static void offree(void); |
| |
static int ofmerge(struct mchars *, struct mparse *, const char *); |
| |
static void parse_catpage(struct of *, const char *); |
| |
static int parse_man(struct of *, |
| |
const struct man_node *); |
| |
static void parse_mdoc(struct of *, const struct mdoc_node *); |
| |
static int parse_mdoc_body(struct of *, const struct mdoc_node *); |
| |
static int parse_mdoc_head(struct of *, const struct mdoc_node *); |
| |
static int parse_mdoc_Fd(struct of *, const struct mdoc_node *); |
| |
static int parse_mdoc_Fn(struct of *, const struct mdoc_node *); |
| |
static int parse_mdoc_In(struct of *, const struct mdoc_node *); |
| |
static int parse_mdoc_Nd(struct of *, const struct mdoc_node *); |
| |
static int parse_mdoc_Nm(struct of *, const struct mdoc_node *); |
| |
static int parse_mdoc_Sh(struct of *, const struct mdoc_node *); |
| |
static int parse_mdoc_St(struct of *, const struct mdoc_node *); |
| |
static int parse_mdoc_Xr(struct of *, const struct mdoc_node *); |
| |
static int path_reset(const char *, int, const char *); |
| |
static void putkey(const struct of *, |
| |
const char *, uint64_t); |
| |
static void putkeys(const struct of *, |
| |
const char *, int, uint64_t); |
| |
static void putmdockey(const struct of *, |
| |
const struct mdoc_node *, uint64_t); |
| |
static char *stradd(const char *); |
| |
static char *straddbuf(const char *, size_t); |
| |
static int treescan(const char *); |
| |
static size_t utf8(unsigned int, char [7]); |
| |
static void utf8key(struct mchars *, struct str *); |
| |
static void wordaddbuf(const struct of *, |
| |
const char *, size_t, uint64_t); |
| |
|
| |
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 enum op op; /* operational mode */ |
| |
static struct ohash inos; /* table of inodes/devices */ |
| |
static struct ohash filenames; /* table of filenames */ |
| |
static struct ohash strings; /* table of all strings */ |
| |
static struct of *ofs = NULL; /* vector of files to parse */ |
| |
static struct str *words = NULL; /* word list in current parse */ |
| |
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, 0 }, /* Ap */ |
| { NULL, 0, 0 }, /* Dd */ |
{ NULL, 0, 0 }, /* Dd */ |
| { NULL, 0, 0 }, /* Dt */ |
{ NULL, 0, 0 }, /* Dt */ |
| { NULL, 0, 0 }, /* Os */ |
{ NULL, 0, 0 }, /* Os */ |
| { pmdoc_Sh, TYPE_Sh, MDOCF_CHILD }, /* Sh */ |
{ parse_mdoc_Sh, TYPE_Sh, MDOCF_CHILD }, /* Sh */ |
| { pmdoc_head, TYPE_Ss, MDOCF_CHILD }, /* Ss */ |
{ parse_mdoc_head, TYPE_Ss, MDOCF_CHILD }, /* Ss */ |
| { NULL, 0, 0 }, /* Pp */ |
{ NULL, 0, 0 }, /* Pp */ |
| { NULL, 0, 0 }, /* D1 */ |
{ NULL, 0, 0 }, /* D1 */ |
| { NULL, 0, 0 }, /* Dl */ |
{ NULL, 0, 0 }, /* Dl */ |
| Line 162 static const struct mdoc_handler mdocs[MDOC_MAX] = { |
|
| Line 203 static const struct mdoc_handler mdocs[MDOC_MAX] = { |
|
| { NULL, TYPE_Ev, MDOCF_CHILD }, /* Ev */ |
{ NULL, TYPE_Ev, MDOCF_CHILD }, /* Ev */ |
| { NULL, 0, 0 }, /* Ex */ |
{ NULL, 0, 0 }, /* Ex */ |
| { NULL, TYPE_Fa, MDOCF_CHILD }, /* Fa */ |
{ NULL, TYPE_Fa, MDOCF_CHILD }, /* Fa */ |
| { pmdoc_Fd, TYPE_In, 0 }, /* Fd */ |
{ parse_mdoc_Fd, TYPE_In, 0 }, /* Fd */ |
| { NULL, TYPE_Fl, MDOCF_CHILD }, /* Fl */ |
{ NULL, TYPE_Fl, MDOCF_CHILD }, /* Fl */ |
| { pmdoc_Fn, 0, 0 }, /* Fn */ |
{ parse_mdoc_Fn, 0, 0 }, /* Fn */ |
| { NULL, TYPE_Ft, MDOCF_CHILD }, /* Ft */ |
{ NULL, TYPE_Ft, MDOCF_CHILD }, /* Ft */ |
| { NULL, TYPE_Ic, MDOCF_CHILD }, /* Ic */ |
{ NULL, TYPE_Ic, MDOCF_CHILD }, /* Ic */ |
| { pmdoc_In, TYPE_In, 0 }, /* In */ |
{ parse_mdoc_In, TYPE_In, MDOCF_CHILD }, /* In */ |
| { NULL, TYPE_Li, MDOCF_CHILD }, /* Li */ |
{ NULL, TYPE_Li, MDOCF_CHILD }, /* Li */ |
| { pmdoc_Nd, TYPE_Nd, MDOCF_CHILD }, /* Nd */ |
{ parse_mdoc_Nd, TYPE_Nd, MDOCF_CHILD }, /* Nd */ |
| { pmdoc_Nm, TYPE_Nm, MDOCF_CHILD }, /* Nm */ |
{ parse_mdoc_Nm, TYPE_Nm, MDOCF_CHILD }, /* Nm */ |
| { NULL, 0, 0 }, /* Op */ |
{ NULL, 0, 0 }, /* Op */ |
| { NULL, 0, 0 }, /* Ot */ |
{ NULL, 0, 0 }, /* Ot */ |
| { NULL, TYPE_Pa, MDOCF_CHILD }, /* Pa */ |
{ NULL, TYPE_Pa, MDOCF_CHILD }, /* Pa */ |
| { NULL, 0, 0 }, /* Rv */ |
{ NULL, 0, 0 }, /* Rv */ |
| { pmdoc_St, TYPE_St, 0 }, /* St */ |
{ parse_mdoc_St, TYPE_St, 0 }, /* St */ |
| { NULL, TYPE_Va, MDOCF_CHILD }, /* Va */ |
{ NULL, TYPE_Va, MDOCF_CHILD }, /* Va */ |
| { pmdoc_body, TYPE_Va, MDOCF_CHILD }, /* Vt */ |
{ parse_mdoc_body, TYPE_Va, MDOCF_CHILD }, /* Vt */ |
| { pmdoc_Xr, TYPE_Xr, 0 }, /* Xr */ |
{ parse_mdoc_Xr, TYPE_Xr, 0 }, /* Xr */ |
| { NULL, 0, 0 }, /* %A */ |
{ NULL, 0, 0 }, /* %A */ |
| { NULL, 0, 0 }, /* %B */ |
{ NULL, 0, 0 }, /* %B */ |
| { NULL, 0, 0 }, /* %D */ |
{ NULL, 0, 0 }, /* %D */ |
| Line 234 static const struct mdoc_handler mdocs[MDOC_MAX] = { |
|
| Line 275 static const struct mdoc_handler mdocs[MDOC_MAX] = { |
|
| { NULL, 0, 0 }, /* Ux */ |
{ NULL, 0, 0 }, /* Ux */ |
| { NULL, 0, 0 }, /* Xc */ |
{ NULL, 0, 0 }, /* Xc */ |
| { NULL, 0, 0 }, /* Xo */ |
{ NULL, 0, 0 }, /* Xo */ |
| { pmdoc_head, TYPE_Fn, 0 }, /* Fo */ |
{ parse_mdoc_head, TYPE_Fn, 0 }, /* Fo */ |
| { NULL, 0, 0 }, /* Fc */ |
{ NULL, 0, 0 }, /* Fc */ |
| { NULL, 0, 0 }, /* Oo */ |
{ NULL, 0, 0 }, /* Oo */ |
| { NULL, 0, 0 }, /* Oc */ |
{ NULL, 0, 0 }, /* Oc */ |
| Line 262 static const struct mdoc_handler mdocs[MDOC_MAX] = { |
|
| Line 303 static const struct mdoc_handler mdocs[MDOC_MAX] = { |
|
| { NULL, 0, 0 }, /* Ta */ |
{ NULL, 0, 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 */ |
char cwd[MAXPATHLEN]; |
| struct manpaths dirs; |
int ch, rc, fd, i; |
| enum op op; /* current operation */ |
size_t j, sz; |
| const char *dir; |
const char *dir; |
| char *cp; |
struct str *s; |
| char pbuf[PATH_MAX], |
struct mchars *mc; |
| ibuf[MAXPATHLEN], /* index fname */ |
struct manpaths dirs; |
| fbuf[MAXPATHLEN]; /* btree fname */ |
struct mparse *mp; |
| int ch, i, flags; |
struct ohash_info ino_info, filename_info, str_info; |
| 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)); |
| |
|
| |
ino_info.halloc = filename_info.halloc = |
| |
str_info.halloc = hash_halloc; |
| |
ino_info.hfree = filename_info.hfree = |
| |
str_info.hfree = hash_free; |
| |
ino_info.alloc = filename_info.alloc = |
| |
str_info.alloc = hash_alloc; |
| |
|
| |
ino_info.key_offset = offsetof(struct of, id); |
| |
filename_info.key_offset = offsetof(struct of, file); |
| |
str_info.key_offset = offsetof(struct str, key); |
| |
|
| 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)); |
/* |
| |
* Remember where we started by keeping a fd open to the origin |
| |
* 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 == getcwd(cwd, MAXPATHLEN)) { |
| |
perror(NULL); |
| |
return(EXIT_FAILURE); |
| |
} else if (-1 == (fd = open(cwd, O_RDONLY, 0))) { |
| |
perror(cwd); |
| |
return(EXIT_FAILURE); |
| |
} |
| |
|
| verb = 0; |
/* |
| use_all = 0; |
* We accept a few different invocations. |
| of = NULL; |
* The CHECKOP macro makes sure that invocation styles don't |
| db = idx = NULL; |
* clobber each other. |
| mp = NULL; |
*/ |
| hash = NULL; |
#define CHECKOP(_op, _ch) do \ |
| recs = NULL; |
if (OP_DEFAULT != (_op)) { \ |
| recsz = reccur = 0; |
fprintf(stderr, "-%c: Conflicting option\n", (_ch)); \ |
| maxrec = 0; |
goto usage; \ |
| op = OP_NEW; |
} while (/*CONSTCOND*/0) |
| |
|
| dir = NULL; |
dir = NULL; |
| |
op = OP_DEFAULT; |
| |
|
| while (-1 != (ch = getopt(argc, argv, "ad:u:v"))) |
while (-1 != (ch = getopt(argc, argv, "aC:d:ntu:vW"))) |
| switch (ch) { |
switch (ch) { |
| case ('a'): |
case ('a'): |
| use_all = 1; |
use_all = 1; |
| break; |
break; |
| |
case ('C'): |
| |
CHECKOP(op, ch); |
| |
dir = optarg; |
| |
op = OP_CONFFILE; |
| |
break; |
| case ('d'): |
case ('d'): |
| |
CHECKOP(op, ch); |
| dir = optarg; |
dir = optarg; |
| op = OP_UPDATE; |
op = OP_UPDATE; |
| break; |
break; |
| |
case ('n'): |
| |
nodb = 1; |
| |
break; |
| |
case ('t'): |
| |
CHECKOP(op, ch); |
| |
dup2(STDOUT_FILENO, STDERR_FILENO); |
| |
op = OP_TEST; |
| |
nodb = warnings = 1; |
| |
break; |
| case ('u'): |
case ('u'): |
| |
CHECKOP(op, ch); |
| dir = optarg; |
dir = 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); |
rc = 1; |
| |
mp = mparse_alloc(MPARSE_AUTO, |
| |
MANDOCLEVEL_FATAL, NULL, NULL, NULL); |
| |
mc = mchars_alloc(); |
| |
|
| memset(&buf, 0, sizeof(struct buf)); |
ohash_init(&strings, 6, &str_info); |
| memset(&dbuf, 0, sizeof(struct buf)); |
ohash_init(&inos, 6, &ino_info); |
| |
ohash_init(&filenames, 6, &filename_info); |
| |
|
| buf.size = dbuf.size = MANDOC_BUFSZ; |
if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) { |
| |
/* |
| |
* Force processing all files. |
| |
*/ |
| |
use_all = 1; |
| |
if (NULL == dir) |
| |
dir = cwd; |
| |
/* |
| |
* All of these deal with a specific directory. |
| |
* Jump into that directory then collect files specified |
| |
* on the command-line. |
| |
*/ |
| |
if (0 == path_reset(cwd, fd, dir)) |
| |
goto out; |
| |
for (i = 0; i < argc; i++) |
| |
filescan(argv[i], dir); |
| |
if (0 == dbopen(dir, 1)) |
| |
goto out; |
| |
if (OP_TEST != op) |
| |
dbprune(dir); |
| |
if (OP_DELETE != op) |
| |
rc = ofmerge(mc, mp, dir); |
| |
dbclose(dir, 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, dir, NULL, NULL); |
| |
|
| buf.cp = mandoc_malloc(buf.size); |
/* |
| dbuf.cp = mandoc_malloc(dbuf.size); |
* First scan the tree rooted at a base directory. |
| |
* Then whak its database (if one exists), parse, and |
| |
* build up the database. |
| |
* 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 (0 == path_reset(cwd, fd, dirs.paths[j])) |
| |
goto out; |
| |
if (0 == treescan(dirs.paths[j])) |
| |
goto out; |
| |
if (0 == path_reset(cwd, fd, dirs.paths[j])) |
| |
goto out; |
| |
if (0 == dbopen(dirs.paths[j], 0)) |
| |
goto out; |
| |
if (0 == ofmerge(mc, mp, dirs.paths[j])) |
| |
goto out; |
| |
dbclose(dirs.paths[j], 0); |
| |
offree(); |
| |
ohash_delete(&inos); |
| |
ohash_init(&inos, 6, &ino_info); |
| |
ohash_delete(&filenames); |
| |
ohash_init(&filenames, 6, &filename_info); |
| |
} |
| |
} |
| |
out: |
| |
close(fd); |
| |
manpath_free(&dirs); |
| |
mchars_free(mc); |
| |
mparse_free(mp); |
| |
for (s = ohash_first(&strings, &ch); |
| |
NULL != s; s = ohash_next(&strings, &ch)) { |
| |
if (s->utf8 != s->key) |
| |
free(s->utf8); |
| |
free(s); |
| |
} |
| |
ohash_delete(&strings); |
| |
ohash_delete(&inos); |
| |
ohash_delete(&filenames); |
| |
offree(); |
| |
return(rc ? EXIT_SUCCESS : EXIT_FAILURE); |
| |
usage: |
| |
fprintf(stderr, "usage: %s [-anvW] [-C file]\n" |
| |
" %s [-anvW] dir ...\n" |
| |
" %s [-nvW] -d dir [file ...]\n" |
| |
" %s [-nvW] -u dir [file ...]\n" |
| |
" %s -t file ...\n", |
| |
progname, progname, progname, |
| |
progname, progname); |
| |
|
| flags = OP_NEW == op ? O_CREAT|O_TRUNC|O_RDWR : O_CREAT|O_RDWR; |
return(EXIT_FAILURE); |
| |
} |
| |
|
| if (OP_UPDATE == op || OP_DELETE == op) { |
/* |
| ibuf[0] = fbuf[0] = '\0'; |
* Scan a directory tree rooted at "base" for manpages. |
| |
* We use fts(), scanning directory parts along the way for clues to our |
| |
* section and architecture. |
| |
* |
| |
* If use_all has been specified, grok all files. |
| |
* If not, sanitise paths to the following: |
| |
* |
| |
* [./]man*[/<arch>]/<name>.<section> |
| |
* or |
| |
* [./]cat<section>[/<arch>]/<name>.0 |
| |
* |
| |
* TODO: accomodate for multi-language directories. |
| |
*/ |
| |
static int |
| |
treescan(const char *base) |
| |
{ |
| |
FTS *f; |
| |
FTSENT *ff; |
| |
int dform; |
| |
char *sec; |
| |
const char *dsec, *arch, *cp, *name, *path; |
| |
const char *argv[2]; |
| |
|
| strlcat(fbuf, dir, MAXPATHLEN); |
argv[0] = "."; |
| strlcat(fbuf, "/", MAXPATHLEN); |
argv[1] = (char *)NULL; |
| sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN); |
|
| |
|
| strlcat(ibuf, dir, MAXPATHLEN); |
/* |
| strlcat(ibuf, "/", MAXPATHLEN); |
* Walk through all components under the directory, using the |
| sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN); |
* logical descent of files. |
| |
*/ |
| |
f = fts_open((char * const *)argv, FTS_LOGICAL, NULL); |
| |
if (NULL == f) { |
| |
perror(base); |
| |
return(0); |
| |
} |
| |
|
| if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) { |
dsec = arch = NULL; |
| fprintf(stderr, "%s: Path too long\n", dir); |
dform = FORM_NONE; |
| exit((int)MANDOCLEVEL_BADARG); |
|
| } |
|
| |
|
| db = dbopen(fbuf, flags, 0644, DB_BTREE, &info); |
while (NULL != (ff = fts_read(f))) { |
| idx = dbopen(ibuf, flags, 0644, DB_RECNO, NULL); |
path = ff->fts_path + 2; |
| |
/* |
| |
* If we're a regular file, add an "of" by using the |
| |
* stored directory data and handling the filename. |
| |
* Disallow duplicate (hard-linked) files. |
| |
*/ |
| |
if (FTS_F == ff->fts_info) { |
| |
if ( ! use_all && ff->fts_level < 2) { |
| |
WARNING(path, base, "Extraneous file"); |
| |
continue; |
| |
} else if (inocheck(ff->fts_statp)) { |
| |
WARNING(path, base, "Duplicate file"); |
| |
continue; |
| |
} |
| |
|
| if (NULL == db) { |
cp = ff->fts_name; |
| perror(fbuf); |
|
| exit((int)MANDOCLEVEL_SYSERR); |
|
| } else if (NULL == idx) { |
|
| perror(ibuf); |
|
| exit((int)MANDOCLEVEL_SYSERR); |
|
| } |
|
| |
|
| if (verb > 2) { |
if (0 == strcmp(cp, "mandocdb.db")) { |
| printf("%s: Opened\n", fbuf); |
WARNING(path, base, "Skip database"); |
| printf("%s: Opened\n", ibuf); |
continue; |
| |
} else if (NULL != (cp = strrchr(cp, '.'))) { |
| |
if (0 == strcmp(cp + 1, "html")) { |
| |
WARNING(path, base, "Skip html"); |
| |
continue; |
| |
} else if (0 == strcmp(cp + 1, "gz")) { |
| |
WARNING(path, base, "Skip gz"); |
| |
continue; |
| |
} else if (0 == strcmp(cp + 1, "ps")) { |
| |
WARNING(path, base, "Skip ps"); |
| |
continue; |
| |
} else if (0 == strcmp(cp + 1, "pdf")) { |
| |
WARNING(path, base, "Skip pdf"); |
| |
continue; |
| |
} |
| |
} |
| |
|
| |
if (NULL != (sec = strrchr(ff->fts_name, '.'))) { |
| |
*sec = '\0'; |
| |
sec = stradd(sec + 1); |
| |
} |
| |
name = stradd(ff->fts_name); |
| |
ofadd(base, dform, path, |
| |
name, dsec, sec, arch, ff->fts_statp); |
| |
continue; |
| |
} else if (FTS_D != ff->fts_info && |
| |
FTS_DP != ff->fts_info) |
| |
continue; |
| |
|
| |
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. |
| |
*/ |
| |
dsec = NULL; |
| |
dform = FORM_NONE; |
| |
cp = ff->fts_name; |
| |
if (FTS_DP == ff->fts_info) |
| |
break; |
| |
|
| |
if (0 == strncmp(cp, "man", 3)) { |
| |
dform = FORM_SRC; |
| |
dsec = stradd(cp + 3); |
| |
} else if (0 == strncmp(cp, "cat", 3)) { |
| |
dform = FORM_CAT; |
| |
dsec = stradd(cp + 3); |
| |
} |
| |
|
| |
if (NULL != dsec || use_all) |
| |
break; |
| |
|
| |
WARNING(path, base, "Unknown directory part"); |
| |
fts_set(f, ff, FTS_SKIP); |
| |
break; |
| |
case (2): |
| |
/* |
| |
* Possibly our architecture. |
| |
* If we're descending, keep tabs on it. |
| |
*/ |
| |
arch = NULL; |
| |
if (FTS_DP != ff->fts_info && NULL != dsec) |
| |
arch = stradd(ff->fts_name); |
| |
break; |
| |
default: |
| |
if (FTS_DP == ff->fts_info || use_all) |
| |
break; |
| |
WARNING(path, base, "Extraneous directory part"); |
| |
fts_set(f, ff, FTS_SKIP); |
| |
break; |
| } |
} |
| |
} |
| |
|
| ofile_argbuild(argc, argv, &of); |
fts_close(f); |
| if (NULL == of) |
return(1); |
| goto out; |
} |
| |
|
| of = of->first; |
/* |
| |
* Add a file to the file vector. |
| |
* 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 |
| |
* |
| |
* Stuff this information directly into the "of" vector. |
| |
* See treescan() for the fts(3) version of this. |
| |
*/ |
| |
static void |
| |
filescan(const char *file, const char *base) |
| |
{ |
| |
const char *sec, *arch, *name, *dsec; |
| |
char *p, *start, *buf; |
| |
int dform; |
| |
struct stat st; |
| |
|
| index_prune(of, db, fbuf, idx, ibuf, |
assert(use_all); |
| &maxrec, &recs, &recsz, &reccur); |
|
| |
|
| /* |
if (0 == strncmp(file, "./", 2)) |
| * Go to the root of the respective manual tree |
file += 2; |
| * such that .so links work. In case of failure, |
|
| * just prod on, even though .so links won't work. |
|
| */ |
|
| |
|
| if (OP_UPDATE == op) { |
if (-1 == stat(file, &st)) { |
| chdir(dir); |
WARNING(file, base, "%s", strerror(errno)); |
| index_merge(of, mp, &dbuf, &buf, hash, |
return; |
| db, fbuf, idx, ibuf, |
} else if ( ! (S_IFREG & st.st_mode)) { |
| maxrec, recs, reccur); |
WARNING(file, base, "Not a regular file"); |
| |
return; |
| |
} else if (inocheck(&st)) { |
| |
WARNING(file, base, "Duplicate file"); |
| |
return; |
| |
} |
| |
|
| |
buf = mandoc_strdup(file); |
| |
start = buf; |
| |
sec = arch = name = dsec = NULL; |
| |
dform = FORM_NONE; |
| |
|
| |
/* |
| |
* 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)) { |
| |
dform = FORM_SRC; |
| |
dsec = start + 3; |
| |
} else if (0 == strncmp(start, "cat", 3)) { |
| |
dform = FORM_CAT; |
| |
dsec = start + 3; |
| } |
} |
| |
|
| goto out; |
start = p; |
| |
if (NULL != dsec && NULL != (p = strchr(start, '/'))) { |
| |
*p++ = '\0'; |
| |
arch = start; |
| |
start = p; |
| |
} |
| } |
} |
| |
|
| /* |
/* |
| * Configure the directories we're going to scan. |
* Now check the file suffix. |
| * If we have command-line arguments, use them. |
* Suffix of `.0' indicates a catpage, `.1-9' is a manpage. |
| * If not, we use man(1)'s method (see mandocdb.8). |
|
| */ |
*/ |
| |
p = strrchr(start, '\0'); |
| |
while (p-- > start && '/' != *p && '.' != *p) |
| |
/* Loop. */ ; |
| |
|
| if (argc > 0) { |
if ('.' == *p) { |
| dirs.paths = mandoc_calloc(argc, sizeof(char *)); |
*p++ = '\0'; |
| dirs.sz = argc; |
sec = p; |
| 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, NULL, NULL); |
|
| |
|
| for (i = 0; i < dirs.sz; i++) { |
/* |
| ibuf[0] = fbuf[0] = '\0'; |
* Now try to parse the name. |
| |
* Use the filename portion of the path. |
| |
*/ |
| |
name = start; |
| |
if (NULL != (p = strrchr(start, '/'))) { |
| |
name = p + 1; |
| |
*p = '\0'; |
| |
} |
| |
|
| strlcat(fbuf, dirs.paths[i], MAXPATHLEN); |
ofadd(base, dform, file, name, dsec, sec, arch, &st); |
| strlcat(fbuf, "/", MAXPATHLEN); |
free(buf); |
| sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN); |
} |
| |
|
| strlcat(ibuf, dirs.paths[i], MAXPATHLEN); |
/* |
| strlcat(ibuf, "/", MAXPATHLEN); |
* See fileadd(). |
| sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN); |
*/ |
| |
static int |
| |
filecheck(const char *name) |
| |
{ |
| |
unsigned int index; |
| |
|
| if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) { |
index = ohash_qlookup(&filenames, name); |
| fprintf(stderr, "%s: Path too long\n", |
return(NULL != ohash_find(&filenames, index)); |
| dirs.paths[i]); |
} |
| exit((int)MANDOCLEVEL_BADARG); |
|
| } |
|
| |
|
| if (db) |
/* |
| (*db->close)(db); |
* Use the standard hashing mechanism (K&R) to see if the given filename |
| if (idx) |
* already exists. |
| (*idx->close)(idx); |
*/ |
| |
static void |
| |
fileadd(struct of *of) |
| |
{ |
| |
unsigned int index; |
| |
|
| db = dbopen(fbuf, flags, 0644, DB_BTREE, &info); |
index = ohash_qlookup(&filenames, of->file); |
| idx = dbopen(ibuf, flags, 0644, DB_RECNO, NULL); |
assert(NULL == ohash_find(&filenames, index)); |
| |
ohash_insert(&filenames, index, of); |
| |
} |
| |
|
| if (NULL == db) { |
/* |
| perror(fbuf); |
* See inoadd(). |
| exit((int)MANDOCLEVEL_SYSERR); |
*/ |
| } else if (NULL == idx) { |
static int |
| perror(ibuf); |
inocheck(const struct stat *st) |
| exit((int)MANDOCLEVEL_SYSERR); |
{ |
| } |
struct id id; |
| |
uint32_t hash; |
| |
unsigned int index; |
| |
|
| if (verb > 2) { |
memset(&id, 0, sizeof(id)); |
| printf("%s: Truncated\n", fbuf); |
id.ino = hash = st->st_ino; |
| printf("%s: Truncated\n", ibuf); |
id.dev = st->st_dev; |
| } |
index = ohash_lookup_memory |
| |
(&inos, (char *)&id, sizeof(id), hash); |
| |
|
| ofile_free(of); |
return(NULL != ohash_find(&inos, index)); |
| of = NULL; |
} |
| |
|
| if ( ! ofile_dirbuild(dirs.paths[i], NULL, NULL, |
/* |
| 0, &of)) |
* The hashing function used here is quite simple: simply take the inode |
| exit((int)MANDOCLEVEL_SYSERR); |
* and use uint32_t of its bits. |
| |
* Then when we do the lookup, use both the inode and device identifier. |
| |
*/ |
| |
static void |
| |
inoadd(const struct stat *st, struct of *of) |
| |
{ |
| |
uint32_t hash; |
| |
unsigned int index; |
| |
|
| if (NULL == of) |
of->id.ino = hash = st->st_ino; |
| continue; |
of->id.dev = st->st_dev; |
| |
index = ohash_lookup_memory |
| |
(&inos, (char *)&of->id, sizeof(of->id), hash); |
| |
|
| of = of->first; |
assert(NULL == ohash_find(&inos, index)); |
| |
ohash_insert(&inos, index, of); |
| |
} |
| |
|
| /* |
static void |
| * Go to the root of the respective manual tree |
ofadd(const char *base, int dform, const char *file, |
| * such that .so links work. In case of failure, |
const char *name, const char *dsec, const char *sec, |
| * just prod on, even though .so links won't work. |
const char *arch, const struct stat *st) |
| */ |
{ |
| |
struct of *of; |
| |
int sform; |
| |
|
| chdir(dirs.paths[i]); |
assert(NULL != file); |
| index_merge(of, mp, &dbuf, &buf, hash, db, fbuf, |
|
| idx, ibuf, maxrec, recs, reccur); |
if (NULL == name) |
| |
name = ""; |
| |
if (NULL == sec) |
| |
sec = ""; |
| |
if (NULL == dsec) |
| |
dsec = ""; |
| |
if (NULL == arch) |
| |
arch = ""; |
| |
|
| |
sform = FORM_NONE; |
| |
if (NULL != sec && *sec <= '9' && *sec >= '1') |
| |
sform = FORM_SRC; |
| |
else if (NULL != sec && *sec == '0') { |
| |
sec = dsec; |
| |
sform = FORM_CAT; |
| } |
} |
| |
|
| out: |
of = mandoc_calloc(1, sizeof(struct of)); |
| if (db) |
strlcpy(of->file, file, MAXPATHLEN); |
| (*db->close)(db); |
of->name = name; |
| if (idx) |
of->sec = sec; |
| (*idx->close)(idx); |
of->dsec = dsec; |
| if (hash) |
of->arch = arch; |
| (*hash->close)(hash); |
of->sform = sform; |
| if (mp) |
of->dform = dform; |
| mparse_free(mp); |
of->next = ofs; |
| |
ofs = of; |
| |
|
| manpath_free(&dirs); |
/* |
| ofile_free(of); |
* Add to unique identifier hash. |
| free(buf.cp); |
* Then if it's a source manual and we're going to use source in |
| free(dbuf.cp); |
* favour of catpages, add it to that hash. |
| free(recs); |
*/ |
| |
inoadd(st, of); |
| |
fileadd(of); |
| |
} |
| |
|
| return(MANDOCLEVEL_OK); |
static void |
| |
offree(void) |
| |
{ |
| |
struct of *of; |
| |
|
| |
while (NULL != (of = ofs)) { |
| |
ofs = of->next; |
| |
free(of); |
| |
} |
| } |
} |
| |
|
| void |
/* |
| index_merge(const struct of *of, struct mparse *mp, |
* Run through the files in the global vector "ofs" and add them to the |
| struct buf *dbuf, struct buf *buf, DB *hash, |
* database specified in "base". |
| 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 int |
| |
ofmerge(struct mchars *mc, struct mparse *mp, const char *base) |
| { |
{ |
| recno_t rec; |
int form; |
| int ch; |
size_t sz; |
| DBT key, val; |
|
| struct mdoc *mdoc; |
struct mdoc *mdoc; |
| struct man *man; |
struct man *man; |
| const char *fn, *msec, *mtitle, *arch; |
char buf[MAXPATHLEN]; |
| size_t sv; |
char *bufp; |
| unsigned seq; |
const char *msec, *march, *mtitle, *cp; |
| struct db_val vbuf; |
struct of *of; |
| |
enum mandoclevel lvl; |
| |
|
| for (rec = 0; of; of = of->next) { |
for (of = ofs; NULL != of; of = of->next) { |
| fn = of->fname; |
|
| |
|
| /* |
/* |
| * Reclaim an empty index record, if available. |
* If we're a catpage (as defined by our path), then see |
| |
* if a manpage exists by the same name (ignoring the |
| |
* suffix). |
| |
* If it does, then we want to use it instead of our |
| |
* own. |
| */ |
*/ |
| |
if ( ! use_all && FORM_CAT == of->dform) { |
| |
sz = strlcpy(buf, of->file, MAXPATHLEN); |
| |
if (sz >= MAXPATHLEN) { |
| |
WARNING(of->file, base, |
| |
"Filename too long"); |
| |
continue; |
| |
} |
| |
bufp = strstr(buf, "cat"); |
| |
assert(NULL != bufp); |
| |
memcpy(bufp, "man", 3); |
| |
if (NULL != (bufp = strrchr(buf, '.'))) |
| |
*++bufp = '\0'; |
| |
strlcat(buf, of->dsec, MAXPATHLEN); |
| |
if (filecheck(buf)) { |
| |
WARNING(of->file, base, "Man " |
| |
"source exists: %s", buf); |
| |
continue; |
| |
} |
| |
} |
| |
|
| if (reccur > 0) { |
words = NULL; |
| --reccur; |
|
| rec = recs[(int)reccur]; |
|
| } else if (maxrec > 0) { |
|
| rec = maxrec; |
|
| maxrec = 0; |
|
| } else |
|
| rec++; |
|
| |
|
| mparse_reset(mp); |
mparse_reset(mp); |
| hash_reset(&hash); |
|
| mdoc = NULL; |
mdoc = NULL; |
| man = NULL; |
man = NULL; |
| |
form = 0; |
| |
msec = of->dsec; |
| |
march = of->arch; |
| |
mtitle = of->name; |
| |
|
| /* |
/* |
| * Try interpreting the file as mdoc(7) or man(7) |
* Try interpreting the file as mdoc(7) or man(7) |
| * source code, unless it is already known to be |
* source code, unless it is already known to be |
| * formatted. Fall back to formatted mode. |
* formatted. Fall back to formatted mode. |
| */ |
*/ |
| |
if (FORM_SRC == of->dform || FORM_SRC == of->sform) { |
| |
lvl = mparse_readfd(mp, -1, of->file); |
| |
if (lvl < MANDOCLEVEL_FATAL) |
| |
mparse_result(mp, &mdoc, &man); |
| |
} |
| |
|
| 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) { |
if (NULL != mdoc) { |
| |
form = 1; |
| msec = mdoc_meta(mdoc)->msec; |
msec = mdoc_meta(mdoc)->msec; |
| arch = mdoc_meta(mdoc)->arch; |
march = mdoc_meta(mdoc)->arch; |
| mtitle = mdoc_meta(mdoc)->title; |
mtitle = mdoc_meta(mdoc)->title; |
| } else if (NULL != man) { |
} else if (NULL != man) { |
| |
form = 1; |
| msec = man_meta(man)->msec; |
msec = man_meta(man)->msec; |
| arch = NULL; |
march = ""; |
| mtitle = man_meta(man)->title; |
mtitle = man_meta(man)->title; |
| } else { |
} |
| msec = of->sec; |
|
| arch = of->arch; |
|
| mtitle = of->title; |
|
| } |
|
| |
|
| |
if (NULL == msec) |
| |
msec = ""; |
| |
if (NULL == march) |
| |
march = ""; |
| |
if (NULL == mtitle) |
| |
mtitle = ""; |
| |
|
| /* |
/* |
| * By default, skip a file if the manual section |
* Check whether the manual section given in a file |
| * and architecture given in the file disagree |
* agrees with the directory where the file is located. |
| * with the directory where the file is located. |
* Some manuals have suffixes like (3p) on their |
| |
* section number either inside the file or in the |
| |
* directory name, some are linked into more than one |
| |
* section, like encrypt(1) = makekey(8). Do not skip |
| |
* manuals for such reasons. |
| */ |
*/ |
| |
if ( ! use_all && form && strcasecmp(msec, of->dsec)) |
| |
WARNING(of->file, base, "Section \"%s\" " |
| |
"manual in %s directory", |
| |
msec, of->dsec); |
| |
|
| if (0 == use_all) { |
/* |
| assert(of->sec); |
* Manual page directories exist for each kernel |
| assert(msec); |
* architecture as returned by machine(1). |
| if (strcasecmp(msec, of->sec)) |
* However, many manuals only depend on the |
| continue; |
* application architecture as returned by arch(1). |
| |
* For example, some (2/ARM) manuals are shared |
| |
* across the "armish" and "zaurus" kernel |
| |
* architectures. |
| |
* A few manuals are even shared across completely |
| |
* different architectures, for example fdformat(1) |
| |
* on amd64, i386, sparc, and sparc64. |
| |
* Thus, warn about architecture mismatches, |
| |
* but don't skip manuals for this reason. |
| |
*/ |
| |
if ( ! use_all && strcasecmp(march, of->arch)) |
| |
WARNING(of->file, base, "Architecture \"%s\" " |
| |
"manual in \"%s\" directory", |
| |
march, of->arch); |
| |
|
| if (NULL == arch) { |
putkey(of, of->name, TYPE_Nm); |
| if (NULL != of->arch) |
|
| continue; |
|
| } else if (NULL == of->arch || |
|
| strcasecmp(arch, of->arch)) |
|
| continue; |
|
| } |
|
| |
|
| if (NULL == arch) |
if (NULL != mdoc) { |
| arch = ""; |
if (NULL != (cp = mdoc_meta(mdoc)->name)) |
| |
putkey(of, cp, TYPE_Nm); |
| |
parse_mdoc(of, mdoc_node(mdoc)); |
| |
} else if (NULL != man) |
| |
parse_man(of, man_node(man)); |
| |
else |
| |
parse_catpage(of, base); |
| |
|
| /* |
dbindex(mc, form, of, base); |
| * 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); |
return(1); |
| assert(mtitle); |
} |
| |
|
| if (0 == strcasecmp(mtitle, of->title)) |
static void |
| mtitle = of->title; |
parse_catpage(struct of *of, const char *base) |
| else if (0 == use_all) |
{ |
| continue; |
FILE *stream; |
| |
char *line, *p, *title; |
| |
size_t len, plen, titlesz; |
| |
|
| /* |
if (NULL == (stream = fopen(of->file, "r"))) { |
| * The index record value consists of a nil-terminated |
WARNING(of->file, base, "%s", strerror(errno)); |
| * filename, a nil-terminated manual section, and a |
return; |
| * 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; |
/* Skip to first blank line. */ |
| buf_append(dbuf, mdoc ? "mdoc" : (man ? "man" : "cat")); |
|
| buf_appendb(dbuf, fn, strlen(fn) + 1); |
|
| buf_appendb(dbuf, msec, strlen(msec) + 1); |
|
| buf_appendb(dbuf, mtitle, strlen(mtitle) + 1); |
|
| buf_appendb(dbuf, arch, strlen(arch) + 1); |
|
| |
|
| sv = dbuf->len; |
while (NULL != (line = fgetln(stream, &len))) |
| |
if ('\n' == *line) |
| |
break; |
| |
|
| /* Fix the record number in the btree value. */ |
/* |
| |
* Assume the first line that is not indented |
| |
* is the first section header. Skip to it. |
| |
*/ |
| |
|
| if (mdoc) |
while (NULL != (line = fgetln(stream, &len))) |
| pmdoc_node(hash, buf, dbuf, |
if ('\n' != *line && ' ' != *line) |
| mdoc_node(mdoc), mdoc_meta(mdoc)); |
break; |
| else if (man) |
|
| pman_node(hash, buf, dbuf, man_node(man)); |
/* |
| else |
* Read up until the next section into a buffer. |
| pformatted(hash, buf, dbuf, of); |
* Strip the leading and trailing newline from each read line, |
| |
* appending a trailing space. |
| |
* Ignore empty (whitespace-only) lines. |
| |
*/ |
| |
|
| /* |
titlesz = 0; |
| * Copy from the in-memory hashtable of pending keywords |
title = NULL; |
| * into the database. |
|
| */ |
|
| |
|
| vbuf.rec = htobe32(rec); |
while (NULL != (line = fgetln(stream, &len))) { |
| seq = R_FIRST; |
if (' ' != *line || '\n' != line[len - 1]) |
| while (0 == (ch = (*hash->seq)(hash, &key, &val, seq))) { |
break; |
| seq = R_NEXT; |
while (len > 0 && isspace((unsigned char)*line)) { |
| vbuf.mask = htobe64(*(uint64_t *)val.data); |
line++; |
| val.size = sizeof(struct db_val); |
len--; |
| val.data = &vbuf; |
|
| dbt_put(db, dbf, &key, &val); |
|
| } |
} |
| if (ch < 0) { |
if (1 == len) |
| perror("hash"); |
continue; |
| exit((int)MANDOCLEVEL_SYSERR); |
title = mandoc_realloc(title, titlesz + len); |
| } |
memcpy(title + titlesz, line, len); |
| |
titlesz += len; |
| /* |
title[titlesz - 1] = ' '; |
| * Apply to the index. If we haven't had a description |
} |
| * set, put an empty one in now. |
|
| */ |
|
| |
|
| if (dbuf->len == sv) |
/* |
| buf_appendb(dbuf, "", 1); |
* 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. |
| |
*/ |
| |
|
| key.data = &rec; |
if (NULL == title || '\0' == *title) { |
| key.size = sizeof(recno_t); |
WARNING(of->file, base, "Cannot find NAME section"); |
| |
fclose(stream); |
| |
free(title); |
| |
return; |
| |
} |
| |
|
| val.data = dbuf->cp; |
title = mandoc_realloc(title, titlesz + 1); |
| val.size = dbuf->len; |
title[titlesz] = '\0'; |
| |
|
| if (verb) |
/* |
| printf("%s: Added index\n", fn); |
* Skip to the first dash. |
| |
* Use the remaining line as the description (no more than 70 |
| |
* bytes). |
| |
*/ |
| |
|
| dbt_put(idx, idxf, &key, &val); |
if (NULL != (p = strstr(title, "- "))) { |
| |
for (p += 2; ' ' == *p || '\b' == *p; p++) |
| |
/* Skip to next word. */ ; |
| |
} else { |
| |
WARNING(of->file, base, "No dash in title line"); |
| |
p = title; |
| } |
} |
| |
|
| |
plen = strlen(p); |
| |
|
| |
/* Strip backspace-encoding from line. */ |
| |
|
| |
while (NULL != (line = memchr(p, '\b', plen))) { |
| |
len = line - p; |
| |
if (0 == len) { |
| |
memmove(line, line + 1, plen--); |
| |
continue; |
| |
} |
| |
memmove(line - 1, line + 1, plen - len); |
| |
plen -= 2; |
| |
} |
| |
|
| |
of->desc = stradd(p); |
| |
putkey(of, p, TYPE_Nd); |
| |
fclose(stream); |
| |
free(title); |
| } |
} |
| |
|
| /* |
/* |
| * Scan through all entries in the index file `idx' and prune those |
* Put a type/word pair into the word database for this particular file. |
| * 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, |
putkey(const struct of *of, const char *value, uint64_t type) |
| DB *idx, const char *idxf, recno_t *maxrec, |
|
| recno_t **recs, size_t *recsz, size_t *reccur) |
|
| { |
{ |
| const struct of *of; |
|
| const char *fn, *cp; |
|
| struct db_val *vbuf; |
|
| unsigned seq, sseq; |
|
| DBT key, val; |
|
| int ch; |
|
| |
|
| *reccur = 0; |
assert(NULL != value); |
| seq = R_FIRST; |
wordaddbuf(of, value, strlen(value), type); |
| while (0 == (ch = (*idx->seq)(idx, &key, &val, seq))) { |
} |
| seq = R_NEXT; |
|
| *maxrec = *(recno_t *)key.data; |
|
| cp = val.data; |
|
| |
|
| /* Deleted records are zero-sized. Skip them. */ |
/* |
| |
* Like putkey() but for unterminated strings. |
| |
*/ |
| |
static void |
| |
putkeys(const struct of *of, const char *value, int sz, uint64_t type) |
| |
{ |
| |
|
| if (0 == val.size) |
wordaddbuf(of, value, sz, type); |
| goto cont; |
} |
| |
|
| /* |
/* |
| * Make sure we're sane. |
* Grok all nodes at or below a certain mdoc node into putkey(). |
| * Read past our mdoc/man/cat type to the next string, |
*/ |
| * then make sure it's bounded by a NUL. |
static void |
| * Failing any of these, we go into our error handler. |
putmdockey(const struct of *of, const struct mdoc_node *n, uint64_t m) |
| */ |
{ |
| |
|
| if (NULL == (fn = memchr(cp, '\0', val.size))) |
for ( ; NULL != n; n = n->next) { |
| break; |
if (NULL != n->child) |
| if (++fn - cp >= (int)val.size) |
putmdockey(of, n->child, m); |
| break; |
if (MDOC_TEXT == n->type) |
| if (NULL == memchr(fn, '\0', val.size - (fn - cp))) |
putkey(of, n->string, m); |
| break; |
} |
| |
} |
| |
|
| /* |
static int |
| * Search for the file in those we care about. |
parse_man(struct of *of, const struct man_node *n) |
| * XXX: build this into a tree. Too slow. |
{ |
| */ |
const struct man_node *head, *body; |
| |
char *start, *sv, *title; |
| |
char byte; |
| |
size_t sz, titlesz; |
| |
|
| for (of = ofile; of; of = of->next) |
if (NULL == n) |
| if (0 == strcmp(fn, of->fname)) |
return(0); |
| break; |
|
| |
|
| if (NULL == of) |
/* |
| continue; |
* 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) { |
| * Search through the keyword database, throwing out all |
body = n; |
| * references to our file. |
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) { |
| |
|
| sseq = R_FIRST; |
title = NULL; |
| while (0 == (ch = (*db->seq)(db, &key, &val, sseq))) { |
titlesz = 0; |
| sseq = R_NEXT; |
|
| if (sizeof(struct db_val) != val.size) |
|
| break; |
|
| |
|
| vbuf = val.data; |
/* |
| if (*maxrec != betoh32(vbuf->rec)) |
* Suck the entire NAME section into memory. |
| continue; |
* Yes, we might run away. |
| |
* But too many manuals have big, spread-out |
| |
* NAME sections over many lines. |
| |
*/ |
| |
|
| if ((ch = (*db->del)(db, &key, R_CURSOR)) < 0) |
for ( ; NULL != body; body = body->next) { |
| break; |
if (MAN_TEXT != body->type) |
| } |
break; |
| |
if (0 == (sz = strlen(body->string))) |
| |
continue; |
| |
title = mandoc_realloc |
| |
(title, titlesz + sz + 1); |
| |
memcpy(title + titlesz, body->string, sz); |
| |
titlesz += sz + 1; |
| |
title[titlesz - 1] = ' '; |
| |
} |
| |
if (NULL == title) |
| |
return(1); |
| |
|
| if (ch < 0) { |
title = mandoc_realloc(title, titlesz + 1); |
| perror(dbf); |
title[titlesz] = '\0'; |
| exit((int)MANDOCLEVEL_SYSERR); |
|
| } else if (1 != ch) { |
|
| fprintf(stderr, "%s: Corrupt database\n", dbf); |
|
| exit((int)MANDOCLEVEL_SYSERR); |
|
| } |
|
| |
|
| if (verb) |
/* Skip leading space. */ |
| printf("%s: Deleted index\n", fn); |
|
| |
|
| val.size = 0; |
sv = title; |
| ch = (*idx->put)(idx, &key, &val, R_CURSOR); |
while (isspace((unsigned char)*sv)) |
| |
sv++; |
| |
|
| if (ch < 0) |
if (0 == (sz = strlen(sv))) { |
| break; |
free(title); |
| cont: |
return(1); |
| if (*reccur >= *recsz) { |
} |
| *recsz += MANDOC_SLOP; |
|
| *recs = mandoc_realloc |
|
| (*recs, *recsz * sizeof(recno_t)); |
|
| } |
|
| |
|
| (*recs)[(int)*reccur] = *maxrec; |
/* Erase trailing space. */ |
| (*reccur)++; |
|
| } |
|
| |
|
| if (ch < 0) { |
start = &sv[sz - 1]; |
| perror(idxf); |
while (start > sv && isspace((unsigned char)*start)) |
| exit((int)MANDOCLEVEL_SYSERR); |
*start-- = '\0'; |
| } else if (1 != ch) { |
|
| fprintf(stderr, "%s: Corrupt index\n", idxf); |
|
| exit((int)MANDOCLEVEL_SYSERR); |
|
| } |
|
| |
|
| (*maxrec)++; |
if (start == sv) { |
| } |
free(title); |
| |
return(1); |
| |
} |
| |
|
| /* |
start = sv; |
| * Grow the buffer (if necessary) and copy in a binary string. |
|
| */ |
|
| static void |
|
| buf_appendb(struct buf *buf, const void *cp, size_t sz) |
|
| { |
|
| |
|
| /* Overshoot by MANDOC_BUFSZ. */ |
/* |
| |
* 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. |
| |
*/ |
| |
|
| while (buf->len + sz >= buf->size) { |
for ( ;; ) { |
| buf->size = buf->len + sz + MANDOC_BUFSZ; |
sz = strcspn(start, " ,"); |
| buf->cp = mandoc_realloc(buf->cp, buf->size); |
if ('\0' == start[sz]) |
| } |
break; |
| |
|
| memcpy(buf->cp + (int)buf->len, cp, sz); |
byte = start[sz]; |
| buf->len += sz; |
start[sz] = '\0'; |
| } |
|
| |
|
| /* |
putkey(of, start, TYPE_Nm); |
| * Append a nil-terminated string to the buffer. |
|
| * 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 |
|
| buf_append(struct buf *buf, const char *cp) |
|
| { |
|
| size_t sz; |
|
| |
|
| if (0 == (sz = strlen(cp))) |
if (' ' == byte) { |
| return; |
start += sz + 1; |
| |
break; |
| |
} |
| |
|
| if (buf->len) |
assert(',' == byte); |
| buf->cp[(int)buf->len - 1] = ' '; |
start += sz + 1; |
| |
while (' ' == *start) |
| |
start++; |
| |
} |
| |
|
| buf_appendb(buf, cp, sz + 1); |
if (sv == start) { |
| } |
putkey(of, start, TYPE_Nm); |
| |
free(title); |
| |
return(1); |
| |
} |
| |
|
| /* |
while (isspace((unsigned char)*start)) |
| * Recursively add all text from a given node. |
start++; |
| * 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 |
|
| buf_appendmdoc(struct buf *buf, const struct mdoc_node *n, int f) |
|
| { |
|
| |
|
| for ( ; n; n = n->next) { |
if (0 == strncmp(start, "-", 1)) |
| if (n->child) |
start += 1; |
| buf_appendmdoc(buf, n->child, f); |
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; |
| |
|
| if (MDOC_TEXT == n->type && f) { |
while (' ' == *start) |
| f = 0; |
start++; |
| buf_appendb(buf, n->string, |
|
| strlen(n->string) + 1); |
|
| } else if (MDOC_TEXT == n->type) |
|
| buf_append(buf, n->string); |
|
| |
|
| |
assert(NULL == of->desc); |
| |
of->desc = stradd(start); |
| |
putkey(of, start, TYPE_Nd); |
| |
free(title); |
| |
return(1); |
| |
} |
| } |
} |
| |
|
| |
for (n = n->child; n; n = n->next) |
| |
if (parse_man(of, n)) |
| |
return(1); |
| |
|
| |
return(0); |
| } |
} |
| |
|
| static void |
static void |
| hash_reset(DB **db) |
parse_mdoc(struct of *of, const struct mdoc_node *n) |
| { |
{ |
| DB *hash; |
|
| |
|
| if (NULL != (hash = *db)) |
assert(NULL != n); |
| (*hash->close)(hash); |
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)(of, n)) |
| |
break; |
| |
|
| *db = dbopen(NULL, O_CREAT|O_RDWR, 0644, DB_HASH, NULL); |
if (MDOCF_CHILD & mdocs[n->tok].flags) |
| if (NULL == *db) { |
putmdockey(of, n->child, mdocs[n->tok].mask); |
| perror("hash"); |
break; |
| exit((int)MANDOCLEVEL_SYSERR); |
default: |
| |
assert(MDOC_ROOT != n->type); |
| |
continue; |
| |
} |
| |
if (NULL != n->child) |
| |
parse_mdoc(of, n); |
| } |
} |
| } |
} |
| |
|
| /* ARGSUSED */ |
|
| static int |
static int |
| pmdoc_head(MDOC_ARGS) |
parse_mdoc_Fd(struct of *of, const struct mdoc_node *n) |
| { |
{ |
| |
|
| return(MDOC_HEAD == n->type); |
|
| } |
|
| |
|
| /* ARGSUSED */ |
|
| static int |
|
| pmdoc_body(MDOC_ARGS) |
|
| { |
|
| |
|
| return(MDOC_BODY == n->type); |
|
| } |
|
| |
|
| /* ARGSUSED */ |
|
| static int |
|
| pmdoc_Fd(MDOC_ARGS) |
|
| { |
|
| 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 922 pmdoc_Fd(MDOC_ARGS) |
|
| Line 1369 pmdoc_Fd(MDOC_ARGS) |
|
| if ('>' == *end || '"' == *end) |
if ('>' == *end || '"' == *end) |
| end--; |
end--; |
| |
|
| assert(end >= start); |
if (end > start) |
| |
putkeys(of, start, end - start + 1, TYPE_In); |
| buf_appendb(buf, start, (size_t)(end - start + 1)); |
|
| buf_appendb(buf, "", 1); |
|
| return(1); |
return(1); |
| } |
} |
| |
|
| /* ARGSUSED */ |
|
| static int |
static int |
| pmdoc_In(MDOC_ARGS) |
parse_mdoc_In(struct of *of, const struct mdoc_node *n) |
| { |
{ |
| |
|
| if (NULL == n->child || MDOC_TEXT != n->child->type) |
if (NULL != n->child && MDOC_TEXT == n->child->type) |
| return(0); |
return(0); |
| |
|
| buf_append(buf, n->child->string); |
putkey(of, n->child->string, TYPE_In); |
| return(1); |
return(1); |
| } |
} |
| |
|
| /* ARGSUSED */ |
|
| static int |
static int |
| pmdoc_Fn(MDOC_ARGS) |
parse_mdoc_Fn(struct of *of, const struct mdoc_node *n) |
| { |
{ |
| 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(of, cp, TYPE_Fn); |
| |
|
| buf_append(buf, cp); |
if (n->string < cp) |
| hash_put(hash, buf, TYPE_Fn); |
putkeys(of, 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(of, 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_St(struct of *of, const struct mdoc_node *n) |
| { |
{ |
| |
|
| if (NULL == n->child || MDOC_TEXT != n->child->type) |
if (NULL == n->child || MDOC_TEXT != n->child->type) |
| return(0); |
return(0); |
| |
|
| buf_append(buf, n->child->string); |
putkey(of, n->child->string, TYPE_St); |
| return(1); |
return(1); |
| } |
} |
| |
|
| /* ARGSUSED */ |
|
| static int |
static int |
| pmdoc_Xr(MDOC_ARGS) |
parse_mdoc_Xr(struct of *of, const struct mdoc_node *n) |
| { |
{ |
| |
|
| if (NULL == (n = n->child)) |
if (NULL == (n = n->child)) |
| return(0); |
return(0); |
| |
|
| buf_appendb(buf, n->string, strlen(n->string)); |
putkey(of, n->string, TYPE_Xr); |
| |
|
| if (NULL != (n = n->next)) { |
|
| buf_appendb(buf, ".", 1); |
|
| buf_appendb(buf, n->string, strlen(n->string) + 1); |
|
| } else |
|
| buf_appendb(buf, ".", 2); |
|
| |
|
| return(1); |
return(1); |
| } |
} |
| |
|
| /* ARGSUSED */ |
|
| static int |
static int |
| pmdoc_Nd(MDOC_ARGS) |
parse_mdoc_Nd(struct of *of, const struct mdoc_node *n) |
| { |
{ |
| |
size_t sz; |
| |
char *sv, *desc; |
| |
|
| if (MDOC_BODY != n->type) |
if (MDOC_BODY != n->type) |
| return(0); |
return(0); |
| |
|
| buf_appendmdoc(dbuf, n->child, 1); |
/* |
| |
* Special-case the `Nd' because we need to put the description |
| |
* into the document table. |
| |
*/ |
| |
|
| |
desc = NULL; |
| |
for (n = n->child; NULL != n; n = n->next) { |
| |
if (MDOC_TEXT == n->type) { |
| |
sz = strlen(n->string) + 1; |
| |
if (NULL != (sv = desc)) |
| |
sz += strlen(desc) + 1; |
| |
desc = mandoc_realloc(desc, sz); |
| |
if (NULL != sv) |
| |
strlcat(desc, " ", sz); |
| |
else |
| |
*desc = '\0'; |
| |
strlcat(desc, n->string, sz); |
| |
} |
| |
if (NULL != n->child) |
| |
parse_mdoc_Nd(of, n); |
| |
} |
| |
|
| |
of->desc = NULL != desc ? stradd(desc) : NULL; |
| |
free(desc); |
| return(1); |
return(1); |
| } |
} |
| |
|
| /* ARGSUSED */ |
|
| static int |
static int |
| pmdoc_Nm(MDOC_ARGS) |
parse_mdoc_Nm(struct of *of, const struct mdoc_node *n) |
| { |
{ |
| |
|
| if (SEC_NAME == n->sec) |
if (SEC_NAME == n->sec) |
| Line 1044 pmdoc_Nm(MDOC_ARGS) |
|
| Line 1485 pmdoc_Nm(MDOC_ARGS) |
|
| else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type) |
else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type) |
| return(0); |
return(0); |
| |
|
| if (NULL == n->child) |
|
| buf_append(buf, m->name); |
|
| |
|
| return(1); |
return(1); |
| } |
} |
| |
|
| /* ARGSUSED */ |
|
| static int |
static int |
| pmdoc_Sh(MDOC_ARGS) |
parse_mdoc_Sh(struct of *of, const struct mdoc_node *n) |
| { |
{ |
| |
|
| return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type); |
return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type); |
| } |
} |
| |
|
| static void |
static int |
| hash_put(DB *db, const struct buf *buf, uint64_t mask) |
parse_mdoc_head(struct of *of, const struct mdoc_node *n) |
| { |
{ |
| DBT key, val; |
|
| int rc; |
|
| |
|
| if (buf->len < 2) |
return(MDOC_HEAD == n->type); |
| return; |
} |
| |
|
| key.data = buf->cp; |
static int |
| key.size = buf->len; |
parse_mdoc_body(struct of *of, const struct mdoc_node *n) |
| |
{ |
| |
|
| if ((rc = (*db->get)(db, &key, &val, 0)) < 0) { |
return(MDOC_BODY == n->type); |
| perror("hash"); |
} |
| exit((int)MANDOCLEVEL_SYSERR); |
|
| } else if (0 == rc) |
|
| mask |= *(uint64_t *)val.data; |
|
| |
|
| val.data = &mask; |
/* |
| val.size = sizeof(uint64_t); |
* See straddbuf(). |
| |
*/ |
| |
static char * |
| |
stradd(const char *cp) |
| |
{ |
| |
|
| if ((rc = (*db->put)(db, &key, &val, 0)) < 0) { |
return(straddbuf(cp, strlen(cp))); |
| perror("hash"); |
|
| exit((int)MANDOCLEVEL_SYSERR); |
|
| } |
|
| } |
} |
| |
|
| static void |
/* |
| dbt_put(DB *db, const char *dbn, DBT *key, DBT *val) |
* This looks up or adds a string to the string table. |
| |
* The string table is a table of all strings encountered during parse |
| |
* or file scan. |
| |
* In using it, we avoid having thousands of (e.g.) "cat1" string |
| |
* allocations for the "of" table. |
| |
* We also have a layer atop the string table for keeping track of words |
| |
* in a parse sequence (see wordaddbuf()). |
| |
*/ |
| |
static char * |
| |
straddbuf(const char *cp, size_t sz) |
| { |
{ |
| |
struct str *s; |
| |
unsigned int index; |
| |
const char *end; |
| |
|
| assert(key->size); |
if (NULL != (s = hashget(cp, sz))) |
| assert(val->size); |
return(s->key); |
| |
|
| if (0 == (*db->put)(db, key, val, 0)) |
s = mandoc_calloc(sizeof(struct str) + sz + 1, 1); |
| return; |
memcpy(s->key, cp, sz); |
| |
|
| perror(dbn); |
end = cp + sz; |
| exit((int)MANDOCLEVEL_SYSERR); |
index = ohash_qlookupi(&strings, cp, &end); |
| /* NOTREACHED */ |
assert(NULL == ohash_find(&strings, index)); |
| |
ohash_insert(&strings, index, s); |
| |
return(s->key); |
| } |
} |
| |
|
| |
static struct str * |
| |
hashget(const char *cp, size_t sz) |
| |
{ |
| |
unsigned int index; |
| |
const char *end; |
| |
|
| |
end = cp + sz; |
| |
index = ohash_qlookupi(&strings, cp, &end); |
| |
return(ohash_find(&strings, index)); |
| |
} |
| |
|
| /* |
/* |
| * Call out to per-macro handlers after clearing the persistent database |
* Add a word to the current parse sequence. |
| * key. If the macro sets the database key, flush it to the database. |
* Within the hashtable of strings, we maintain a list of strings that |
| |
* are currently indexed. |
| |
* Each of these ("words") has a bitmask modified within the parse. |
| |
* When we finish a parse, we'll dump the list, then remove the head |
| |
* entry -- since the next parse will have a new "of", it can keep track |
| |
* of its entries without conflict. |
| */ |
*/ |
| static void |
static void |
| pmdoc_node(MDOC_ARGS) |
wordaddbuf(const struct of *of, |
| |
const char *cp, size_t sz, uint64_t v) |
| { |
{ |
| |
struct str *s; |
| |
unsigned int index; |
| |
const char *end; |
| |
|
| if (NULL == n) |
if (0 == sz) |
| return; |
return; |
| |
|
| switch (n->type) { |
s = hashget(cp, sz); |
| case (MDOC_HEAD): |
|
| /* FALLTHROUGH */ |
|
| case (MDOC_BODY): |
|
| /* FALLTHROUGH */ |
|
| case (MDOC_TAIL): |
|
| /* FALLTHROUGH */ |
|
| case (MDOC_BLOCK): |
|
| /* FALLTHROUGH */ |
|
| case (MDOC_ELEM): |
|
| buf->len = 0; |
|
| |
|
| /* |
if (NULL != s && of == s->of) { |
| * Both NULL handlers and handlers returning true |
s->mask |= v; |
| * request using the data. Only skip the element |
return; |
| * when the handler returns false. |
} else if (NULL == s) { |
| */ |
s = mandoc_calloc(sizeof(struct str) + sz + 1, 1); |
| |
memcpy(s->key, cp, sz); |
| |
end = cp + sz; |
| |
index = ohash_qlookupi(&strings, cp, &end); |
| |
assert(NULL == ohash_find(&strings, index)); |
| |
ohash_insert(&strings, index, s); |
| |
} |
| |
|
| if (NULL != mdocs[n->tok].fp && |
s->next = words; |
| 0 == (*mdocs[n->tok].fp)(hash, buf, dbuf, n, m)) |
s->of = of; |
| break; |
s->mask = v; |
| |
words = s; |
| |
} |
| |
|
| /* |
/* |
| * For many macros, use the text from all children. |
* Take a Unicode codepoint and produce its UTF-8 encoding. |
| * Set zero flags for macros not needing this. |
* This isn't the best way to do this, but it works. |
| * In that case, the handler must fill the buffer. |
* 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; |
| |
|
| if (MDOCF_CHILD & mdocs[n->tok].flags) |
rc = 0; |
| buf_appendmdoc(buf, n->child, 0); |
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); |
| |
|
| /* |
out[rc] = '\0'; |
| * Cover the most common case: |
return(rc); |
| * Automatically stage one string per element. |
|
| * Set a zero mask for macros not needing this. |
|
| * Additional staging can be done in the handler. |
|
| */ |
|
| |
|
| if (mdocs[n->tok].mask) |
|
| hash_put(hash, buf, mdocs[n->tok].mask); |
|
| break; |
|
| default: |
|
| break; |
|
| } |
|
| |
|
| pmdoc_node(hash, buf, dbuf, n->child, m); |
|
| pmdoc_node(hash, buf, dbuf, n->next, m); |
|
| } |
} |
| |
|
| static int |
/* |
| pman_node(MAN_ARGS) |
* Store the UTF-8 version of a key, or alias the pointer if the key has |
| |
* no UTF-8 transcription marks in it. |
| |
*/ |
| |
static void |
| |
utf8key(struct mchars *mc, struct str *key) |
| { |
{ |
| const struct man_node *head, *body; |
size_t sz, bsz, pos; |
| const char *start, *sv; |
char utfbuf[7], res[5]; |
| size_t sz; |
char *buf; |
| |
const char *seq, *cpp, *val; |
| |
int len, u; |
| |
enum mandoc_esc esc; |
| |
|
| if (NULL == n) |
assert(NULL == key->utf8); |
| return(0); |
|
| |
|
| |
res[0] = '\\'; |
| |
res[1] = '\t'; |
| |
res[2] = ASCII_NBRSP; |
| |
res[3] = ASCII_HYPH; |
| |
res[4] = '\0'; |
| |
|
| |
val = key->key; |
| |
bsz = strlen(val); |
| |
|
| /* |
/* |
| * We're only searching for one thing: the first text child in |
* Pre-check: if we have no stop-characters, then set the |
| * the BODY of a NAME section. Since we don't keep track of |
* pointer as ourselvse and get out of here. |
| * sections in -man, run some hoops to find out whether we're in |
|
| * the correct section or not. |
|
| */ |
*/ |
| |
if (strcspn(val, res) == bsz) { |
| |
key->utf8 = key->key; |
| |
return; |
| |
} |
| |
|
| if (MAN_BODY == n->type && MAN_SH == n->tok) { |
/* Pre-allocate by the length of the input */ |
| 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); |
buf = mandoc_malloc(++bsz); |
| start = sv = body->string; |
pos = 0; |
| |
|
| /* |
while ('\0' != *val) { |
| * Go through a special heuristic dance here. |
/* |
| * This is why -man manuals are great! |
* Halt on the first escape sequence. |
| * (I'm being sarcastic: my eyes are bleeding.) |
* This also halts on the end of string, in which case |
| * Conventionally, one or more manual names are |
* we just copy, fallthrough, and exit the loop. |
| * comma-specified prior to a whitespace, then a |
*/ |
| * dash, then a description. Try to puzzle out |
if ((sz = strcspn(val, res)) > 0) { |
| * the name parts here. |
memcpy(&buf[pos], val, sz); |
| */ |
pos += sz; |
| |
val += sz; |
| |
} |
| |
|
| for ( ;; ) { |
if (ASCII_HYPH == *val) { |
| sz = strcspn(start, " ,"); |
buf[pos++] = '-'; |
| if ('\0' == start[(int)sz]) |
val++; |
| break; |
continue; |
| |
} else if ('\t' == *val || ASCII_NBRSP == *val) { |
| |
buf[pos++] = ' '; |
| |
val++; |
| |
continue; |
| |
} else if ('\\' != *val) |
| |
break; |
| |
|
| buf->len = 0; |
/* Read past the slash. */ |
| buf_appendb(buf, start, sz); |
|
| buf_appendb(buf, "", 1); |
|
| |
|
| hash_put(hash, buf, TYPE_Nm); |
val++; |
| |
u = 0; |
| |
|
| if (' ' == start[(int)sz]) { |
/* |
| start += (int)sz + 1; |
* Parse the escape sequence and see if it's a |
| break; |
* predefined character or special character. |
| } |
*/ |
| |
esc = mandoc_escape |
| |
((const char **)&val, &seq, &len); |
| |
if (ESCAPE_ERROR == esc) |
| |
break; |
| |
|
| assert(',' == start[(int)sz]); |
if (ESCAPE_SPECIAL != esc) |
| start += (int)sz + 1; |
continue; |
| while (' ' == *start) |
if (0 == (u = mchars_spec2cp(mc, seq, len))) |
| start++; |
continue; |
| } |
|
| |
|
| buf->len = 0; |
/* |
| |
* If we have a Unicode codepoint, try to convert that |
| |
* to a UTF-8 byte string. |
| |
*/ |
| |
cpp = utfbuf; |
| |
if (0 == (sz = utf8(u, utfbuf))) |
| |
continue; |
| |
|
| if (sv == start) { |
/* Copy the rendered glyph into the stream. */ |
| buf_append(buf, start); |
|
| return(1); |
|
| } |
|
| |
|
| while (' ' == *start) |
sz = strlen(cpp); |
| start++; |
bsz += sz; |
| |
|
| if (0 == strncmp(start, "-", 1)) |
buf = mandoc_realloc(buf, bsz); |
| 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) |
memcpy(&buf[pos], cpp, sz); |
| start++; |
pos += sz; |
| |
|
| 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->utf8 = 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, UTF-8-encode the description at the last possible moment. |
| */ |
*/ |
| static void |
static void |
| pformatted(DB *hash, struct buf *buf, struct buf *dbuf, |
dbindex(struct mchars *mc, int form, |
| const struct of *of) |
const struct of *of, const char *base) |
| { |
{ |
| FILE *stream; |
struct str *key; |
| char *line, *p; |
const char *desc; |
| size_t len, plen; |
int64_t recno; |
| |
|
| if (NULL == (stream = fopen(of->fname, "r"))) { |
DEBUG(of->file, base, "Adding to index"); |
| perror(of->fname); |
|
| |
if (nodb) |
| return; |
return; |
| |
|
| |
desc = ""; |
| |
if (NULL != of->desc) { |
| |
key = hashget(of->desc, strlen(of->desc)); |
| |
assert(NULL != key); |
| |
if (NULL == key->utf8) |
| |
utf8key(mc, key); |
| |
desc = key->utf8; |
| } |
} |
| |
|
| /* |
sqlite3_exec(db, "BEGIN TRANSACTION", NULL, NULL, NULL); |
| * 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; |
sqlite3_bind_text |
| buf_append(buf, of->title); |
(stmts[STMT_INSERT_DOC], 1, |
| hash_put(hash, buf, TYPE_Nm); |
of->file, -1, SQLITE_STATIC); |
| |
sqlite3_bind_text |
| |
(stmts[STMT_INSERT_DOC], 2, |
| |
of->sec, -1, SQLITE_STATIC); |
| |
sqlite3_bind_text |
| |
(stmts[STMT_INSERT_DOC], 3, |
| |
of->arch, -1, SQLITE_STATIC); |
| |
sqlite3_bind_text |
| |
(stmts[STMT_INSERT_DOC], 4, |
| |
desc, -1, SQLITE_STATIC); |
| |
sqlite3_bind_int |
| |
(stmts[STMT_INSERT_DOC], 5, form); |
| |
sqlite3_step(stmts[STMT_INSERT_DOC]); |
| |
recno = sqlite3_last_insert_rowid(db); |
| |
sqlite3_reset(stmts[STMT_INSERT_DOC]); |
| |
|
| /* Skip to first blank line. */ |
for (key = words; NULL != key; key = key->next) { |
| |
assert(key->of == of); |
| |
if (NULL == key->utf8) |
| |
utf8key(mc, key); |
| |
sqlite3_bind_int64 |
| |
(stmts[STMT_INSERT_KEY], 1, key->mask); |
| |
sqlite3_bind_text |
| |
(stmts[STMT_INSERT_KEY], 2, |
| |
key->utf8, -1, SQLITE_STATIC); |
| |
sqlite3_bind_int64 |
| |
(stmts[STMT_INSERT_KEY], 3, recno); |
| |
sqlite3_step(stmts[STMT_INSERT_KEY]); |
| |
sqlite3_reset(stmts[STMT_INSERT_KEY]); |
| |
} |
| |
|
| while (NULL != (line = fgetln(stream, &len))) |
sqlite3_exec(db, "COMMIT TRANSACTION", NULL, NULL, NULL); |
| 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))) |
static void |
| if ('\n' != *line && ' ' != *line) |
dbprune(const char *base) |
| break; |
{ |
| |
struct of *of; |
| |
|
| /* |
if (nodb) |
| * 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'; |
for (of = ofs; NULL != of; of = of->next) { |
| |
sqlite3_bind_text |
| /* |
(stmts[STMT_DELETE], 1, |
| * Skip to the first dash. |
of->file, -1, SQLITE_STATIC); |
| * Use the remaining line as the description (no more than 70 |
sqlite3_step(stmts[STMT_DELETE]); |
| * bytes). |
sqlite3_reset(stmts[STMT_DELETE]); |
| */ |
DEBUG(of->file, base, "Deleted from index"); |
| |
|
| if (NULL != (p = strstr(line, "- "))) { |
|
| for (p += 2; ' ' == *p || '\b' == *p; p++) |
|
| /* Skip to next word. */ ; |
|
| } else |
|
| p = line; |
|
| |
|
| if ((plen = strlen(p)) > 70) { |
|
| plen = 70; |
|
| p[plen] = '\0'; |
|
| } |
} |
| |
|
| /* Strip backspace-encoding from line. */ |
|
| |
|
| while (NULL != (line = memchr(p, '\b', plen))) { |
|
| len = line - p; |
|
| if (0 == len) { |
|
| memmove(line, line + 1, plen--); |
|
| continue; |
|
| } |
|
| memmove(line - 1, line + 1, plen - len); |
|
| plen -= 2; |
|
| } |
|
| |
|
| buf_appendb(dbuf, p, plen + 1); |
|
| buf->len = 0; |
|
| buf_appendb(buf, p, plen + 1); |
|
| hash_put(hash, buf, TYPE_Nd); |
|
| fclose(stream); |
|
| } |
} |
| |
|
| |
/* |
| |
* Close an existing database and its prepared statements. |
| |
* If "real" is not set, rename the temporary file into the real one. |
| |
*/ |
| static void |
static void |
| ofile_argbuild(int argc, char *argv[], struct of **of) |
dbclose(const char *base, int real) |
| { |
{ |
| char buf[MAXPATHLEN]; |
size_t i; |
| char *sec, *arch, *title, *p; |
char file[MAXPATHLEN]; |
| int i, src_form; |
|
| struct of *nof; |
|
| |
|
| for (i = 0; i < argc; i++) { |
if (nodb) |
| |
return; |
| |
|
| /* |
for (i = 0; i < STMT__MAX; i++) { |
| * Try to infer the manual section, architecture and |
sqlite3_finalize(stmts[i]); |
| * page title from the path, assuming it looks like |
stmts[i] = NULL; |
| * man*[/<arch>]/<title>.<section> or |
} |
| * cat<section>[/<arch>]/<title>.0 |
|
| */ |
|
| |
|
| if (strlcpy(buf, argv[i], sizeof(buf)) >= sizeof(buf)) { |
sqlite3_close(db); |
| fprintf(stderr, "%s: Path too long\n", argv[i]); |
db = NULL; |
| continue; |
|
| } |
|
| sec = arch = title = NULL; |
|
| 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; |
|
| |
|
| /* |
if (real) |
| * Build the file structure. |
return; |
| */ |
|
| |
|
| nof = mandoc_calloc(1, sizeof(struct of)); |
strlcpy(file, MANDOC_DB, MAXPATHLEN); |
| nof->fname = mandoc_strdup(argv[i]); |
strlcat(file, "~", MAXPATHLEN); |
| if (NULL != sec) |
if (-1 == rename(file, MANDOC_DB)) |
| nof->sec = mandoc_strdup(sec); |
perror(MANDOC_DB); |
| if (NULL != arch) |
|
| nof->arch = mandoc_strdup(arch); |
|
| nof->title = mandoc_strdup(title); |
|
| nof->src_form = src_form; |
|
| |
|
| /* |
|
| * Add the structure to the list. |
|
| */ |
|
| |
|
| if (verb > 2) |
|
| printf("%s: Scheduling\n", argv[i]); |
|
| if (NULL == *of) { |
|
| *of = nof; |
|
| (*of)->first = nof; |
|
| } else { |
|
| 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(const char *base, int real) |
| int p_src_form, struct of **of) |
|
| { |
{ |
| char buf[MAXPATHLEN]; |
char file[MAXPATHLEN]; |
| struct stat sb; |
const char *sql; |
| |
int rc, ofl; |
| size_t sz; |
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); |
| |
|
| |
sz = strlcpy(file, MANDOC_DB, MAXPATHLEN); |
| |
if ( ! real) |
| |
sz = strlcat(file, "~", MAXPATHLEN); |
| |
|
| |
if (sz >= MAXPATHLEN) { |
| |
fprintf(stderr, "%s: Path too long\n", file); |
| return(0); |
return(0); |
| } |
} |
| |
|
| while (NULL != (dp = readdir(d))) { |
if ( ! real) |
| fn = dp->d_name; |
remove(file); |
| |
|
| if ('.' == *fn) |
ofl = SQLITE_OPEN_PRIVATECACHE | SQLITE_OPEN_READWRITE | |
| continue; |
(0 == real ? SQLITE_OPEN_EXCLUSIVE : 0); |
| |
|
| src_form = p_src_form; |
rc = sqlite3_open_v2(file, &db, ofl, NULL); |
| |
if (SQLITE_OK == rc) |
| |
return(1); |
| |
if (SQLITE_CANTOPEN != rc) { |
| |
perror(file); |
| |
return(0); |
| |
} |
| |
|
| if (DT_DIR == dp->d_type) { |
sqlite3_close(db); |
| sec = psec; |
db = NULL; |
| arch = parch; |
|
| |
|
| /* |
if (SQLITE_OK != (rc = sqlite3_open(file, &db))) { |
| * By default, only use directories called: |
perror(file); |
| * man<section>/[<arch>/] or |
return(0); |
| * cat<section>/[<arch>/] |
} |
| */ |
|
| |
|
| if (NULL == sec) { |
sql = "CREATE TABLE \"docs\" (\n" |
| if(0 == strncmp("man", fn, 3)) { |
" \"file\" TEXT NOT NULL,\n" |
| src_form |= MANDOC_SRC; |
" \"sec\" TEXT NOT NULL,\n" |
| sec = fn + 3; |
" \"arch\" TEXT NOT NULL,\n" |
| } else if (0 == strncmp("cat", fn, 3)) { |
" \"desc\" TEXT NOT NULL,\n" |
| src_form |= MANDOC_FORM; |
" \"form\" INTEGER NOT NULL,\n" |
| sec = fn + 3; |
" \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n" |
| } else if (use_all) |
");\n" |
| sec = fn; |
"\n" |
| else |
"CREATE TABLE \"keys\" (\n" |
| continue; |
" \"bits\" INTEGER NOT NULL,\n" |
| } else if (NULL == arch && (use_all || |
" \"key\" TEXT NOT NULL,\n" |
| NULL == strchr(fn, '.'))) |
" \"docid\" INTEGER NOT NULL REFERENCES docs(id) " |
| arch = fn; |
"ON DELETE CASCADE,\n" |
| else if (0 == use_all) |
" \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n" |
| continue; |
");\n" |
| |
"\n" |
| |
"CREATE INDEX \"key_index\" ON keys (key);\n"; |
| |
|
| buf[0] = '\0'; |
if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) { |
| strlcat(buf, dir, MAXPATHLEN); |
perror(sqlite3_errmsg(db)); |
| strlcat(buf, "/", MAXPATHLEN); |
return(0); |
| sz = strlcat(buf, fn, MAXPATHLEN); |
} |
| |
|
| if (MAXPATHLEN <= sz) { |
sql = "DELETE FROM docs where file=?"; |
| fprintf(stderr, "%s: Path too long\n", dir); |
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE], NULL); |
| return(0); |
sql = "INSERT INTO docs " |
| } |
"(file,sec,arch,desc,form) VALUES (?,?,?,?,?)"; |
| |
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_DOC], NULL); |
| if (verb > 2) |
sql = "INSERT INTO keys " |
| printf("%s: Scanning\n", buf); |
"(bits,key,docid) VALUES (?,?,?)"; |
| |
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL); |
| |
return(1); |
| |
} |
| |
|
| 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)); |
| /* |
|
| * Skip formatted manuals if a source version is |
|
| * 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'; |
|
| 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)); |
|
| nof->fname = mandoc_strdup(buf); |
|
| if (NULL != psec) |
|
| nof->sec = mandoc_strdup(psec); |
|
| if (NULL != parch) |
|
| nof->arch = mandoc_strdup(parch); |
|
| nof->src_form = src_form; |
|
| |
|
| /* |
|
| * 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; |
|
| } |
|
| } |
|
| |
|
| closedir(d); |
|
| return(1); |
|
| } |
} |
| |
|
| static void |
static void |
| ofile_free(struct of *of) |
hash_free(void *p, size_t sz, void *arg) |
| { |
{ |
| struct of *nof; |
|
| |
|
| while (of) { |
free(p); |
| nof = of->next; |
|
| free(of->fname); |
|
| free(of->sec); |
|
| free(of->arch); |
|
| free(of->title); |
|
| free(of); |
|
| of = nof; |
|
| } |
|
| } |
} |
| |
|
| static void |
static int |
| usage(void) |
path_reset(const char *cwd, int fd, const char *base) |
| { |
{ |
| |
|
| fprintf(stderr, "usage: %s [-v] " |
if (-1 == fchdir(fd)) { |
| "[-d dir [files...] |" |
perror(cwd); |
| " -u dir [files...] |" |
return(0); |
| " dir...]\n", progname); |
} else if (-1 == chdir(base)) { |
| |
perror(base); |
| |
return(0); |
| |
} |
| |
return(1); |
| } |
} |
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to mandocdb.c CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [cvsweb.bsd.lv] / mandoc