version 1.5, 2011/06/21 13:54:05 |
version 1.13, 2011/07/01 10:17:24 |
|
|
#include <sys/param.h> |
#include <sys/param.h> |
|
|
#include <assert.h> |
#include <assert.h> |
#ifdef __linux__ |
|
# include <db_185.h> |
|
#else |
|
# include <db.h> |
|
#endif |
|
#include <fcntl.h> |
#include <fcntl.h> |
#include <getopt.h> |
#include <getopt.h> |
#include <stdio.h> |
#include <stdio.h> |
|
|
#include <stdlib.h> |
#include <stdlib.h> |
#include <string.h> |
#include <string.h> |
|
|
|
#ifdef __linux__ |
|
# include <db_185.h> |
|
#else |
|
# include <db.h> |
|
#endif |
|
|
#include "man.h" |
#include "man.h" |
#include "mdoc.h" |
#include "mdoc.h" |
#include "mandoc.h" |
#include "mandoc.h" |
|
|
|
|
/* Bit-fields. See makewhatis.1. */ |
/* Bit-fields. See makewhatis.1. */ |
|
|
#define TYPE_NAME 0x01 |
#define TYPE_NAME 0x01 |
#define TYPE_FUNCTION 0x02 |
#define TYPE_FUNCTION 0x02 |
#define TYPE_UTILITY 0x04 |
#define TYPE_UTILITY 0x04 |
#define TYPE_INCLUDES 0x08 |
#define TYPE_INCLUDES 0x08 |
#define TYPE_VARIABLE 0x10 |
#define TYPE_VARIABLE 0x10 |
#define TYPE_STANDARD 0x20 |
#define TYPE_STANDARD 0x20 |
#define TYPE_AUTHOR 0x40 |
#define TYPE_AUTHOR 0x40 |
#define TYPE_CONFIG 0x80 |
#define TYPE_CONFIG 0x80 |
#define TYPE__MAX TYPE_CONFIG |
#define TYPE_DESC 0x100 |
|
#define TYPE_XREF 0x200 |
|
#define TYPE_PATH 0x400 |
|
|
/* Buffer for storing growable data. */ |
/* Buffer for storing growable data. */ |
|
|
|
|
const struct mdoc_node *n, \ |
const struct mdoc_node *n, \ |
const struct mdoc_meta *m |
const struct mdoc_meta *m |
|
|
|
static void buf_appendmdoc(struct buf *, |
|
const struct mdoc_node *); |
static void buf_append(struct buf *, const char *); |
static void buf_append(struct buf *, const char *); |
static void buf_appendb(struct buf *, |
static void buf_appendb(struct buf *, |
const void *, size_t); |
const void *, size_t); |
Line 87 static void pmdoc_Fn(MDOC_ARGS); |
|
Line 92 static void pmdoc_Fn(MDOC_ARGS); |
|
static void pmdoc_Fo(MDOC_ARGS); |
static void pmdoc_Fo(MDOC_ARGS); |
static void pmdoc_Nd(MDOC_ARGS); |
static void pmdoc_Nd(MDOC_ARGS); |
static void pmdoc_Nm(MDOC_ARGS); |
static void pmdoc_Nm(MDOC_ARGS); |
|
static void pmdoc_Pa(MDOC_ARGS); |
static void pmdoc_St(MDOC_ARGS); |
static void pmdoc_St(MDOC_ARGS); |
static void pmdoc_Vt(MDOC_ARGS); |
static void pmdoc_Vt(MDOC_ARGS); |
|
static void pmdoc_Xr(MDOC_ARGS); |
static void usage(void); |
static void usage(void); |
|
|
typedef void (*pmdoc_nf)(MDOC_ARGS); |
typedef void (*pmdoc_nf)(MDOC_ARGS); |
Line 129 static const pmdoc_nf mdocs[MDOC_MAX] = { |
|
Line 136 static const pmdoc_nf mdocs[MDOC_MAX] = { |
|
pmdoc_Nm, /* Nm */ |
pmdoc_Nm, /* Nm */ |
NULL, /* Op */ |
NULL, /* Op */ |
NULL, /* Ot */ |
NULL, /* Ot */ |
NULL, /* Pa */ |
pmdoc_Pa, /* Pa */ |
NULL, /* Rv */ |
NULL, /* Rv */ |
pmdoc_St, /* St */ |
pmdoc_St, /* St */ |
pmdoc_Vt, /* Va */ |
pmdoc_Vt, /* Va */ |
pmdoc_Vt, /* Vt */ |
pmdoc_Vt, /* Vt */ |
NULL, /* Xr */ |
pmdoc_Xr, /* Xr */ |
NULL, /* %A */ |
NULL, /* %A */ |
NULL, /* %B */ |
NULL, /* %B */ |
NULL, /* %D */ |
NULL, /* %D */ |
Line 236 main(int argc, char *argv[]) |
|
Line 243 main(int argc, char *argv[]) |
|
fbuf[MAXPATHLEN], /* btree fname */ |
fbuf[MAXPATHLEN], /* btree fname */ |
fbbuf[MAXPATHLEN], /* btree backup fname */ |
fbbuf[MAXPATHLEN], /* btree backup fname */ |
vbuf[8]; /* stringified record number */ |
vbuf[8]; /* stringified record number */ |
int ch, seq; |
int ch, seq, verb; |
DB *idx, /* index database */ |
DB *idx, /* index database */ |
*db, /* keyword database */ |
*db, /* keyword database */ |
*hash; /* temporary keyword hashtable */ |
*hash; /* temporary keyword hashtable */ |
DBT key, val; |
DBT key, val; |
size_t sv, rsz; |
size_t sv; |
BTREEINFO info; /* btree configuration */ |
BTREEINFO info; /* btree configuration */ |
recno_t rec; /* current record number */ |
recno_t rec; /* current record number */ |
struct buf buf, /* keyword buffer */ |
struct buf buf, /* keyword buffer */ |
Line 256 main(int argc, char *argv[]) |
|
Line 263 main(int argc, char *argv[]) |
|
++progname; |
++progname; |
|
|
dir = ""; |
dir = ""; |
|
verb = 0; |
|
|
while (-1 != (ch = getopt(argc, argv, "d:"))) |
while (-1 != (ch = getopt(argc, argv, "d:v"))) |
switch (ch) { |
switch (ch) { |
case ('d'): |
case ('d'): |
dir = optarg; |
dir = optarg; |
break; |
break; |
|
case ('v'): |
|
verb++; |
|
break; |
default: |
default: |
usage(); |
usage(); |
return((int)MANDOCLEVEL_BADARG); |
return((int)MANDOCLEVEL_BADARG); |
Line 298 main(int argc, char *argv[]) |
|
Line 309 main(int argc, char *argv[]) |
|
'\0' != fbbuf[MAXPATHLEN - 2] || |
'\0' != fbbuf[MAXPATHLEN - 2] || |
'\0' != ibuf[MAXPATHLEN - 2] || |
'\0' != ibuf[MAXPATHLEN - 2] || |
'\0' != ibbuf[MAXPATHLEN - 2]) { |
'\0' != ibbuf[MAXPATHLEN - 2]) { |
fprintf(stderr, "%s: Path too long\n", progname); |
fprintf(stderr, "%s: Path too long\n", dir); |
exit((int)MANDOCLEVEL_SYSERR); |
exit((int)MANDOCLEVEL_SYSERR); |
} |
} |
|
|
Line 306 main(int argc, char *argv[]) |
|
Line 317 main(int argc, char *argv[]) |
|
* For the keyword database, open a BTREE database that allows |
* For the keyword database, open a BTREE database that allows |
* duplicates. |
* duplicates. |
* For the index database, use a standard RECNO database type. |
* For the index database, use a standard RECNO database type. |
* For the temporary keyword hashtable, use the HASH database |
|
* type. |
|
*/ |
*/ |
|
|
hash = dbopen(NULL, MANDOC_FLAGS, 0644, DB_HASH, NULL); |
|
if (NULL == hash) { |
|
perror("hash"); |
|
exit((int)MANDOCLEVEL_SYSERR); |
|
} |
|
|
|
memset(&info, 0, sizeof(BTREEINFO)); |
memset(&info, 0, sizeof(BTREEINFO)); |
info.flags = R_DUP; |
info.flags = R_DUP; |
db = dbopen(fbbuf, MANDOC_FLAGS, 0644, DB_BTREE, &info); |
db = dbopen(fbbuf, MANDOC_FLAGS, 0644, DB_BTREE, &info); |
|
|
if (NULL == db) { |
if (NULL == db) { |
perror(fbbuf); |
perror(fbbuf); |
(*hash->close)(hash); |
|
exit((int)MANDOCLEVEL_SYSERR); |
exit((int)MANDOCLEVEL_SYSERR); |
} |
} |
|
|
Line 331 main(int argc, char *argv[]) |
|
Line 333 main(int argc, char *argv[]) |
|
if (NULL == db) { |
if (NULL == db) { |
perror(ibbuf); |
perror(ibbuf); |
(*db->close)(db); |
(*db->close)(db); |
(*hash->close)(hash); |
|
exit((int)MANDOCLEVEL_SYSERR); |
exit((int)MANDOCLEVEL_SYSERR); |
} |
} |
|
|
/* |
/* |
* Try parsing the manuals given on the command line. If we |
* Try parsing each manual given on the command line. |
* totally fail, then just keep on going. Take resulting trees |
* If we fail, then emit an error and keep on going. |
* and push them down into the database code. |
* Take resulting trees and push them down into the database code. |
* Use the auto-parser and don't report any errors. |
* Use the auto-parser and don't report any errors. |
*/ |
*/ |
|
|
mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL); |
mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL); |
|
|
rec = 1; |
rec = 1; |
rsz = 0; |
hash = NULL; |
|
|
memset(&buf, 0, sizeof(struct buf)); |
memset(&buf, 0, sizeof(struct buf)); |
memset(&dbuf, 0, sizeof(struct buf)); |
memset(&dbuf, 0, sizeof(struct buf)); |
Line 358 main(int argc, char *argv[]) |
|
Line 359 main(int argc, char *argv[]) |
|
while (NULL != (fn = *argv++)) { |
while (NULL != (fn = *argv++)) { |
mparse_reset(mp); |
mparse_reset(mp); |
|
|
|
/* Initialise the in-memory hash of keywords. */ |
|
|
|
if (hash) |
|
(*hash->close)(hash); |
|
|
|
hash = dbopen(NULL, MANDOC_FLAGS, 0644, DB_HASH, NULL); |
|
|
|
if (NULL == hash) { |
|
perror("hash"); |
|
exit((int)MANDOCLEVEL_SYSERR); |
|
} |
|
|
/* Parse and get (non-empty) AST. */ |
/* Parse and get (non-empty) AST. */ |
|
|
if (mparse_readfd(mp, -1, fn) >= MANDOCLEVEL_FATAL) { |
if (mparse_readfd(mp, -1, fn) >= MANDOCLEVEL_FATAL) { |
fprintf(stderr, "%s: Parse failure\n", fn); |
fprintf(stderr, "%s: Parse failure\n", fn); |
continue; |
continue; |
} |
} |
|
|
mparse_result(mp, &mdoc, &man); |
mparse_result(mp, &mdoc, &man); |
|
|
if (NULL == mdoc && NULL == man) |
if (NULL == mdoc && NULL == man) |
continue; |
continue; |
|
|
/* Manual section: can be empty string. */ |
|
|
|
msec = NULL != mdoc ? |
msec = NULL != mdoc ? |
mdoc_meta(mdoc)->msec : |
mdoc_meta(mdoc)->msec : |
man_meta(man)->msec; |
man_meta(man)->msec; |
Line 378 main(int argc, char *argv[]) |
|
Line 391 main(int argc, char *argv[]) |
|
man_meta(man)->title; |
man_meta(man)->title; |
arch = NULL != mdoc ? mdoc_meta(mdoc)->arch : NULL; |
arch = NULL != mdoc ? mdoc_meta(mdoc)->arch : NULL; |
|
|
assert(msec); |
|
assert(mtitle); |
|
|
|
/* |
/* |
* The index record value consists of a nil-terminated |
* The index record value consists of a nil-terminated |
* filename, a nil-terminated manual section, and a |
* filename, a nil-terminated manual section, and a |
Line 416 main(int argc, char *argv[]) |
|
Line 426 main(int argc, char *argv[]) |
|
|
|
seq = R_FIRST; |
seq = R_FIRST; |
while (0 == (ch = (*hash->seq)(hash, &key, &val, seq))) { |
while (0 == (ch = (*hash->seq)(hash, &key, &val, seq))) { |
|
seq = R_NEXT; |
|
|
memcpy(vbuf, val.data, sizeof(uint32_t)); |
memcpy(vbuf, val.data, sizeof(uint32_t)); |
val.size = sizeof(vbuf); |
val.size = sizeof(vbuf); |
val.data = vbuf; |
val.data = vbuf; |
|
|
|
if (verb > 1) |
|
printf("%s: Keyword %s (%zu): 0x%x\n", |
|
fn, (char *)key.data, key.size, |
|
*(int *)val.data); |
|
|
dbt_put(db, fbbuf, &key, &val); |
dbt_put(db, fbbuf, &key, &val); |
|
|
if ((*hash->del)(hash, &key, 0) < 0) { |
|
perror("hash"); |
|
exit((int)MANDOCLEVEL_SYSERR); |
|
} |
|
seq = R_NEXT; |
|
} |
} |
|
|
if (ch < 0) { |
if (ch < 0) { |
Line 447 main(int argc, char *argv[]) |
|
Line 460 main(int argc, char *argv[]) |
|
val.data = dbuf.cp; |
val.data = dbuf.cp; |
val.size = dbuf.len; |
val.size = dbuf.len; |
|
|
|
if (verb > 0) |
|
printf("%s: Indexed\n", fn); |
|
|
dbt_put(idx, ibbuf, &key, &val); |
dbt_put(idx, ibbuf, &key, &val); |
rec++; |
rec++; |
} |
} |
|
|
(*db->close)(db); |
(*db->close)(db); |
(*idx->close)(idx); |
(*idx->close)(idx); |
(*hash->close)(hash); |
|
|
|
|
if (hash) |
|
(*hash->close)(hash); |
|
|
mparse_free(mp); |
mparse_free(mp); |
|
|
free(buf.cp); |
free(buf.cp); |
Line 508 buf_append(struct buf *buf, const char *cp) |
|
Line 526 buf_append(struct buf *buf, const char *cp) |
|
buf_appendb(buf, cp, sz + 1); |
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. |
|
*/ |
|
static void |
|
buf_appendmdoc(struct buf *buf, const struct mdoc_node *n) |
|
{ |
|
|
|
for ( ; n; n = n->next) { |
|
if (n->child) |
|
buf_appendmdoc(buf, n->child); |
|
if (MDOC_TEXT == n->type) |
|
buf_append(buf, n->string); |
|
} |
|
} |
|
|
/* ARGSUSED */ |
/* ARGSUSED */ |
static void |
static void |
pmdoc_An(MDOC_ARGS) |
pmdoc_An(MDOC_ARGS) |
Line 516 pmdoc_An(MDOC_ARGS) |
|
Line 552 pmdoc_An(MDOC_ARGS) |
|
if (SEC_AUTHORS != n->sec) |
if (SEC_AUTHORS != n->sec) |
return; |
return; |
|
|
for (n = n->child; n; n = n->next) |
buf_appendmdoc(buf, n->child); |
if (MDOC_TEXT == n->type) |
|
buf_append(buf, n->string); |
|
|
|
hash_put(hash, buf, TYPE_AUTHOR); |
hash_put(hash, buf, TYPE_AUTHOR); |
} |
} |
|
|
Line 577 pmdoc_Cd(MDOC_ARGS) |
|
Line 610 pmdoc_Cd(MDOC_ARGS) |
|
if (SEC_SYNOPSIS != n->sec) |
if (SEC_SYNOPSIS != n->sec) |
return; |
return; |
|
|
for (n = n->child; n; n = n->next) |
buf_appendmdoc(buf, n->child); |
if (MDOC_TEXT == n->type) |
|
buf_append(buf, n->string); |
|
|
|
hash_put(hash, buf, TYPE_CONFIG); |
hash_put(hash, buf, TYPE_CONFIG); |
} |
} |
|
|
Line 640 pmdoc_St(MDOC_ARGS) |
|
Line 670 pmdoc_St(MDOC_ARGS) |
|
|
|
/* ARGSUSED */ |
/* ARGSUSED */ |
static void |
static void |
|
pmdoc_Xr(MDOC_ARGS) |
|
{ |
|
|
|
if (NULL == (n = n->child)) |
|
return; |
|
|
|
buf_appendb(buf, n->string, strlen(n->string)); |
|
|
|
if (NULL != (n = n->next)) { |
|
buf_appendb(buf, ".", 1); |
|
buf_appendb(buf, n->string, strlen(n->string) + 1); |
|
} else |
|
buf_appendb(buf, ".", 2); |
|
|
|
hash_put(hash, buf, TYPE_XREF); |
|
} |
|
|
|
/* ARGSUSED */ |
|
static void |
pmdoc_Vt(MDOC_ARGS) |
pmdoc_Vt(MDOC_ARGS) |
{ |
{ |
const char *start; |
const char *start; |
Line 694 pmdoc_Fo(MDOC_ARGS) |
|
Line 743 pmdoc_Fo(MDOC_ARGS) |
|
static void |
static void |
pmdoc_Nd(MDOC_ARGS) |
pmdoc_Nd(MDOC_ARGS) |
{ |
{ |
int first; |
size_t sz; |
|
|
|
if (MDOC_BODY != n->type) |
|
return; |
|
else if (NULL == (n = n->child)) |
|
return; |
|
|
|
/* FIXME: don't assume this. */ |
|
assert(MDOC_TEXT == n->type); |
|
|
|
sz = strlen(n->string) + 1; |
|
buf_appendb(dbuf, n->string, sz); |
|
buf_appendb(buf, n->string, sz); |
|
|
for (first = 1, n = n->child; n; n = n->next) { |
buf_appendmdoc(dbuf, n->next); |
if (MDOC_TEXT != n->type) |
buf_appendmdoc(buf, n->next); |
continue; |
|
if (first) |
hash_put(hash, buf, TYPE_DESC); |
buf_appendb(dbuf, n->string, strlen(n->string) + 1); |
|
else |
|
buf_append(dbuf, n->string); |
|
first = 0; |
|
} |
|
} |
} |
|
|
/* ARGSUSED */ |
/* ARGSUSED */ |
static void |
static void |
|
pmdoc_Pa(MDOC_ARGS) |
|
{ |
|
|
|
if (SEC_FILES != n->sec) |
|
return; |
|
|
|
buf_appendmdoc(buf, n->child); |
|
hash_put(hash, buf, TYPE_PATH); |
|
} |
|
|
|
/* ARGSUSED */ |
|
static void |
pmdoc_Nm(MDOC_ARGS) |
pmdoc_Nm(MDOC_ARGS) |
{ |
{ |
|
|
if (SEC_NAME == n->sec) { |
if (SEC_NAME == n->sec) { |
for (n = n->child; n; n = n->next) |
buf_appendmdoc(buf, n->child); |
if (MDOC_TEXT == n->type) |
|
buf_append(buf, n->string); |
|
hash_put(hash, buf, TYPE_NAME); |
hash_put(hash, buf, TYPE_NAME); |
return; |
return; |
} else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type) |
} else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type) |
Line 724 pmdoc_Nm(MDOC_ARGS) |
|
Line 790 pmdoc_Nm(MDOC_ARGS) |
|
if (NULL == n->child) |
if (NULL == n->child) |
buf_append(buf, m->name); |
buf_append(buf, m->name); |
|
|
for (n = n->child; n; n = n->next) |
buf_appendmdoc(buf, n->child); |
if (MDOC_TEXT == n->type) |
|
buf_append(buf, n->string); |
|
|
|
hash_put(hash, buf, TYPE_UTILITY); |
hash_put(hash, buf, TYPE_UTILITY); |
} |
} |
|
|
Line 737 hash_put(DB *db, const struct buf *buf, int mask) |
|
Line 800 hash_put(DB *db, const struct buf *buf, int mask) |
|
DBT key, val; |
DBT key, val; |
int rc; |
int rc; |
|
|
key.data = buf->cp; |
if (buf->len < 2) |
|
|
if ((key.size = buf->len) < 2) |
|
return; |
return; |
|
|
|
key.data = buf->cp; |
|
key.size = buf->len; |
|
|
if ((rc = (*db->get)(db, &key, &val, 0)) < 0) { |
if ((rc = (*db->get)(db, &key, &val, 0)) < 0) { |
perror("hash"); |
perror("hash"); |
exit((int)MANDOCLEVEL_SYSERR); |
exit((int)MANDOCLEVEL_SYSERR); |
Line 751 hash_put(DB *db, const struct buf *buf, int mask) |
|
Line 815 hash_put(DB *db, const struct buf *buf, int mask) |
|
val.data = &mask; |
val.data = &mask; |
val.size = sizeof(int); |
val.size = sizeof(int); |
|
|
/*fprintf(stderr, "Hashing: [%s] (0x%x)\n", |
|
(char *)key.data, mask);*/ |
|
|
|
if ((rc = (*db->put)(db, &key, &val, 0)) < 0) { |
if ((rc = (*db->put)(db, &key, &val, 0)) < 0) { |
perror("hash"); |
perror("hash"); |
exit((int)MANDOCLEVEL_SYSERR); |
exit((int)MANDOCLEVEL_SYSERR); |
|
|
dbt_put(DB *db, const char *dbn, DBT *key, DBT *val) |
dbt_put(DB *db, const char *dbn, DBT *key, DBT *val) |
{ |
{ |
|
|
if (0 == key->size) |
|
return; |
|
|
|
assert(key->size); |
assert(key->size); |
assert(val->size); |
assert(val->size); |
|
|
Line 897 pman_node(MAN_ARGS) |
|
Line 955 pman_node(MAN_ARGS) |
|
while (' ' == *start) |
while (' ' == *start) |
start++; |
start++; |
|
|
buf_appendb(dbuf, start, strlen(start) + 1); |
sz = strlen(start) + 1; |
|
buf_appendb(dbuf, start, sz); |
|
buf_appendb(buf, start, sz); |
} |
} |
} |
} |
|
|
|
|
usage(void) |
usage(void) |
{ |
{ |
|
|
fprintf(stderr, "usage: %s [-d path] [file...]\n", progname); |
fprintf(stderr, "usage: %s [-v] [-d path] [file...]\n", |
|
progname); |
} |
} |