-/* $Id: mandocdb.c,v 1.51 2012/06/08 12:05:27 kristaps Exp $ */
+/* $Id: mandocdb.c,v 1.116 2014/03/19 21:51:20 schwarze Exp $ */
/*
* Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
- * Copyright (c) 2011, 2012 Ingo Schwarze <schwarze@openbsd.org>
+ * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
*
* Permission to use, copy, modify, and distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
#include "config.h"
#endif
-#include <sys/param.h>
#include <sys/stat.h>
+#include <sys/wait.h>
#include <assert.h>
#include <ctype.h>
#include <fcntl.h>
#include <fts.h>
#include <getopt.h>
+#include <limits.h>
#include <stddef.h>
+#include <stdio.h>
#include <stdint.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
+#ifdef HAVE_OHASH
#include <ohash.h>
+#else
+#include "compat_ohash.h"
+#endif
#include <sqlite3.h>
#include "mdoc.h"
#include "man.h"
#include "mandoc.h"
-#include "mandocdb.h"
#include "manpath.h"
-
-/* Post a warning to stderr. */
-#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)
+#include "mansearch.h"
+
+extern int mansearch_keymax;
+extern const char *const mansearch_keynames[];
+
+#define SQL_EXEC(_v) \
+ if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
+ fprintf(stderr, "%s\n", sqlite3_errmsg(db))
+#define SQL_BIND_TEXT(_s, _i, _v) \
+ if (SQLITE_OK != sqlite3_bind_text \
+ ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
+ fprintf(stderr, "%s\n", sqlite3_errmsg(db))
+#define SQL_BIND_INT(_s, _i, _v) \
+ if (SQLITE_OK != sqlite3_bind_int \
+ ((_s), (_i)++, (_v))) \
+ fprintf(stderr, "%s\n", sqlite3_errmsg(db))
+#define SQL_BIND_INT64(_s, _i, _v) \
+ if (SQLITE_OK != sqlite3_bind_int64 \
+ ((_s), (_i)++, (_v))) \
+ fprintf(stderr, "%s\n", sqlite3_errmsg(db))
+#define SQL_STEP(_s) \
+ if (SQLITE_DONE != sqlite3_step((_s))) \
+ fprintf(stderr, "%s\n", sqlite3_errmsg(db))
enum op {
OP_DEFAULT = 0, /* new dbs from dir list or default config */
};
enum form {
- FORM_SRC, /* format is -man or -mdoc */
- FORM_CAT, /* format is cat */
- FORM_NONE /* format is unknown */
+ FORM_NONE, /* format is unknown */
+ FORM_SRC, /* format is -man or -mdoc */
+ FORM_CAT /* format is cat */
};
struct str {
- char *utf8; /* key in UTF-8 form */
- const struct of *of; /* if set, the owning parse */
- struct str *next; /* next in owning parse sequence */
+ char *rendered; /* key in UTF-8 or ASCII form */
+ const struct mpage *mpage; /* if set, the owning parse */
uint64_t mask; /* bitmask in sequence */
- char key[]; /* the string itself */
+ char key[]; /* may contain escape sequences */
};
-struct id {
- ino_t ino;
- dev_t dev;
+struct inodev {
+ ino_t st_ino;
+ dev_t st_dev;
};
-struct of {
- struct id id; /* used for hashing routine */
- struct of *next; /* next in ofs */
- 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) */
+struct mpage {
+ struct inodev inodev; /* used for hashing routine */
+ enum form form; /* format from file content */
+ char *sec; /* section from file content */
+ char *arch; /* architecture from file content */
+ char *title; /* title from file content */
+ char *desc; /* description from file content */
+ struct mlink *mlinks; /* singly linked list */
+};
+
+struct mlink {
+ char file[PATH_MAX]; /* filename rel. to manpath */
+ enum form dform; /* format from directory */
+ enum form fform; /* format from file name suffix */
+ char *dsec; /* section from directory */
+ char *arch; /* architecture from directory */
+ char *name; /* name from file name (not empty) */
+ char *fsec; /* section from file name suffix */
+ struct mlink *next; /* singly linked list */
};
enum stmt {
- STMT_DELETE = 0, /* delete manpage */
- STMT_INSERT_DOC, /* insert manpage */
- STMT_INSERT_KEY, /* insert parsed key */
+ STMT_DELETE_PAGE = 0, /* delete mpage */
+ STMT_INSERT_PAGE, /* insert mpage */
+ STMT_INSERT_LINK, /* insert mlink */
+ STMT_INSERT_KEY, /* insert parsed key */
STMT__MAX
};
-typedef int (*mdoc_fp)(struct of *, const struct mdoc_node *);
+typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_node *);
struct mdoc_handler {
mdoc_fp fp; /* optional handler */
uint64_t mask; /* set unless handler returns 0 */
- 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 dbclose(int);
+static void dbadd(const struct mpage *, struct mchars *);
+static int dbopen(int);
+static void dbprune(void);
+static void filescan(const char *);
static void *hash_alloc(size_t, void *);
static void hash_free(void *, size_t, void *);
static void *hash_halloc(size_t, void *);
-static void 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 *,
+static void mlink_add(struct mlink *, const struct stat *);
+static int mlink_check(struct mpage *, struct mlink *);
+static void mlink_free(struct mlink *);
+static void mlinks_undupe(struct mpage *);
+static void mpages_free(void);
+static void mpages_merge(struct mchars *, struct mparse *);
+static void parse_cat(struct mpage *);
+static void parse_man(struct mpage *, const struct man_node *);
+static void parse_mdoc(struct mpage *, const struct mdoc_node *);
+static int parse_mdoc_body(struct mpage *, const struct mdoc_node *);
+static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
+static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
+static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
+static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
+static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
+static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
+static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
+static void putkey(const struct mpage *, char *, uint64_t);
+static void putkeys(const struct mpage *,
+ const char *, size_t, uint64_t);
+static void putmdockey(const struct mpage *,
const struct mdoc_node *, uint64_t);
-static char *stradd(const char *);
-static char *straddbuf(const char *, size_t);
-static int treescan(const char *);
+static void render_key(struct mchars *, struct str *);
+static void say(const char *, const char *, ...);
+static int set_basedir(const char *);
+static int treescan(void);
static size_t utf8(unsigned int, char [7]);
-static void utf8key(struct mchars *, struct str *);
-static void wordaddbuf(const struct of *,
- const char *, size_t, uint64_t);
+static char tempfilename[32];
static char *progname;
-static int use_all; /* use all found files */
static int nodb; /* no database changes */
+static int mparse_options; /* abort the parse early */
+static int use_all; /* use all found files */
static int verb; /* print what we're doing */
static int warnings; /* warn about crap */
+static int write_utf8; /* write UTF-8 output; else ASCII */
+static int exitcode; /* to be returned by main */
static enum op op; /* operational mode */
-static struct ohash inos; /* table of inodes/devices */
-static struct ohash filenames; /* table of filenames */
+static char basedir[PATH_MAX]; /* current base directory */
+static struct ohash mpages; /* table of distinct manual pages */
+static struct ohash mlinks; /* table of directory entries */
static struct ohash strings; /* table of all strings */
-static 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] = {
- { NULL, 0, 0 }, /* Ap */
- { NULL, 0, 0 }, /* Dd */
- { NULL, 0, 0 }, /* Dt */
- { NULL, 0, 0 }, /* Os */
- { parse_mdoc_Sh, TYPE_Sh, MDOCF_CHILD }, /* Sh */
- { parse_mdoc_head, TYPE_Ss, MDOCF_CHILD }, /* Ss */
- { NULL, 0, 0 }, /* Pp */
- { NULL, 0, 0 }, /* D1 */
- { NULL, 0, 0 }, /* Dl */
- { NULL, 0, 0 }, /* Bd */
- { NULL, 0, 0 }, /* Ed */
- { NULL, 0, 0 }, /* Bl */
- { NULL, 0, 0 }, /* El */
- { NULL, 0, 0 }, /* It */
- { NULL, 0, 0 }, /* Ad */
- { NULL, TYPE_An, MDOCF_CHILD }, /* An */
- { NULL, TYPE_Ar, MDOCF_CHILD }, /* Ar */
- { NULL, TYPE_Cd, MDOCF_CHILD }, /* Cd */
- { NULL, TYPE_Cm, MDOCF_CHILD }, /* Cm */
- { NULL, TYPE_Dv, MDOCF_CHILD }, /* Dv */
- { NULL, TYPE_Er, MDOCF_CHILD }, /* Er */
- { NULL, TYPE_Ev, MDOCF_CHILD }, /* Ev */
- { NULL, 0, 0 }, /* Ex */
- { NULL, TYPE_Fa, MDOCF_CHILD }, /* Fa */
- { parse_mdoc_Fd, TYPE_In, 0 }, /* Fd */
- { NULL, TYPE_Fl, MDOCF_CHILD }, /* Fl */
- { parse_mdoc_Fn, 0, 0 }, /* Fn */
- { NULL, TYPE_Ft, MDOCF_CHILD }, /* Ft */
- { NULL, TYPE_Ic, MDOCF_CHILD }, /* Ic */
- { parse_mdoc_In, TYPE_In, MDOCF_CHILD }, /* In */
- { NULL, TYPE_Li, MDOCF_CHILD }, /* Li */
- { parse_mdoc_Nd, TYPE_Nd, MDOCF_CHILD }, /* Nd */
- { parse_mdoc_Nm, TYPE_Nm, MDOCF_CHILD }, /* Nm */
- { NULL, 0, 0 }, /* Op */
- { NULL, 0, 0 }, /* Ot */
- { NULL, TYPE_Pa, MDOCF_CHILD }, /* Pa */
- { NULL, 0, 0 }, /* Rv */
- { parse_mdoc_St, TYPE_St, 0 }, /* St */
- { NULL, TYPE_Va, MDOCF_CHILD }, /* Va */
- { parse_mdoc_body, TYPE_Va, MDOCF_CHILD }, /* Vt */
- { parse_mdoc_Xr, TYPE_Xr, 0 }, /* Xr */
- { NULL, 0, 0 }, /* %A */
- { NULL, 0, 0 }, /* %B */
- { NULL, 0, 0 }, /* %D */
- { NULL, 0, 0 }, /* %I */
- { NULL, 0, 0 }, /* %J */
- { NULL, 0, 0 }, /* %N */
- { NULL, 0, 0 }, /* %O */
- { NULL, 0, 0 }, /* %P */
- { NULL, 0, 0 }, /* %R */
- { NULL, 0, 0 }, /* %T */
- { NULL, 0, 0 }, /* %V */
- { NULL, 0, 0 }, /* Ac */
- { NULL, 0, 0 }, /* Ao */
- { NULL, 0, 0 }, /* Aq */
- { NULL, TYPE_At, MDOCF_CHILD }, /* At */
- { NULL, 0, 0 }, /* Bc */
- { NULL, 0, 0 }, /* Bf */
- { NULL, 0, 0 }, /* Bo */
- { NULL, 0, 0 }, /* Bq */
- { NULL, TYPE_Bsx, MDOCF_CHILD }, /* Bsx */
- { NULL, TYPE_Bx, MDOCF_CHILD }, /* Bx */
- { NULL, 0, 0 }, /* Db */
- { NULL, 0, 0 }, /* Dc */
- { NULL, 0, 0 }, /* Do */
- { NULL, 0, 0 }, /* Dq */
- { NULL, 0, 0 }, /* Ec */
- { NULL, 0, 0 }, /* Ef */
- { NULL, TYPE_Em, MDOCF_CHILD }, /* Em */
- { NULL, 0, 0 }, /* Eo */
- { NULL, TYPE_Fx, MDOCF_CHILD }, /* Fx */
- { NULL, TYPE_Ms, MDOCF_CHILD }, /* Ms */
- { NULL, 0, 0 }, /* No */
- { NULL, 0, 0 }, /* Ns */
- { NULL, TYPE_Nx, MDOCF_CHILD }, /* Nx */
- { NULL, TYPE_Ox, MDOCF_CHILD }, /* Ox */
- { NULL, 0, 0 }, /* Pc */
- { NULL, 0, 0 }, /* Pf */
- { NULL, 0, 0 }, /* Po */
- { NULL, 0, 0 }, /* Pq */
- { NULL, 0, 0 }, /* Qc */
- { NULL, 0, 0 }, /* Ql */
- { NULL, 0, 0 }, /* Qo */
- { NULL, 0, 0 }, /* Qq */
- { NULL, 0, 0 }, /* Re */
- { NULL, 0, 0 }, /* Rs */
- { NULL, 0, 0 }, /* Sc */
- { NULL, 0, 0 }, /* So */
- { NULL, 0, 0 }, /* Sq */
- { NULL, 0, 0 }, /* Sm */
- { NULL, 0, 0 }, /* Sx */
- { NULL, TYPE_Sy, MDOCF_CHILD }, /* Sy */
- { NULL, TYPE_Tn, MDOCF_CHILD }, /* Tn */
- { NULL, 0, 0 }, /* Ux */
- { NULL, 0, 0 }, /* Xc */
- { NULL, 0, 0 }, /* Xo */
- { parse_mdoc_head, TYPE_Fn, 0 }, /* Fo */
- { NULL, 0, 0 }, /* Fc */
- { NULL, 0, 0 }, /* Oo */
- { NULL, 0, 0 }, /* Oc */
- { NULL, 0, 0 }, /* Bk */
- { NULL, 0, 0 }, /* Ek */
- { NULL, 0, 0 }, /* Bt */
- { NULL, 0, 0 }, /* Hf */
- { NULL, 0, 0 }, /* Fr */
- { NULL, 0, 0 }, /* Ud */
- { NULL, TYPE_Lb, MDOCF_CHILD }, /* Lb */
- { NULL, 0, 0 }, /* Lp */
- { NULL, TYPE_Lk, MDOCF_CHILD }, /* Lk */
- { NULL, TYPE_Mt, MDOCF_CHILD }, /* Mt */
- { NULL, 0, 0 }, /* Brq */
- { NULL, 0, 0 }, /* Bro */
- { NULL, 0, 0 }, /* Brc */
- { NULL, 0, 0 }, /* %C */
- { NULL, 0, 0 }, /* Es */
- { NULL, 0, 0 }, /* En */
- { NULL, TYPE_Dx, MDOCF_CHILD }, /* Dx */
- { NULL, 0, 0 }, /* %Q */
- { NULL, 0, 0 }, /* br */
- { NULL, 0, 0 }, /* sp */
- { NULL, 0, 0 }, /* %U */
- { NULL, 0, 0 }, /* Ta */
+ { NULL, 0 }, /* Ap */
+ { NULL, 0 }, /* Dd */
+ { NULL, 0 }, /* Dt */
+ { NULL, 0 }, /* Os */
+ { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
+ { parse_mdoc_head, TYPE_Ss }, /* Ss */
+ { NULL, 0 }, /* Pp */
+ { NULL, 0 }, /* D1 */
+ { NULL, 0 }, /* Dl */
+ { NULL, 0 }, /* Bd */
+ { NULL, 0 }, /* Ed */
+ { NULL, 0 }, /* Bl */
+ { NULL, 0 }, /* El */
+ { NULL, 0 }, /* It */
+ { NULL, 0 }, /* Ad */
+ { NULL, TYPE_An }, /* An */
+ { NULL, TYPE_Ar }, /* Ar */
+ { NULL, TYPE_Cd }, /* Cd */
+ { NULL, TYPE_Cm }, /* Cm */
+ { NULL, TYPE_Dv }, /* Dv */
+ { NULL, TYPE_Er }, /* Er */
+ { NULL, TYPE_Ev }, /* Ev */
+ { NULL, 0 }, /* Ex */
+ { NULL, TYPE_Fa }, /* Fa */
+ { parse_mdoc_Fd, 0 }, /* Fd */
+ { NULL, TYPE_Fl }, /* Fl */
+ { parse_mdoc_Fn, 0 }, /* Fn */
+ { NULL, TYPE_Ft }, /* Ft */
+ { NULL, TYPE_Ic }, /* Ic */
+ { NULL, TYPE_In }, /* In */
+ { NULL, TYPE_Li }, /* Li */
+ { parse_mdoc_Nd, TYPE_Nd }, /* Nd */
+ { parse_mdoc_Nm, TYPE_Nm }, /* Nm */
+ { NULL, 0 }, /* Op */
+ { NULL, 0 }, /* Ot */
+ { NULL, TYPE_Pa }, /* Pa */
+ { NULL, 0 }, /* Rv */
+ { NULL, TYPE_St }, /* St */
+ { NULL, TYPE_Va }, /* Va */
+ { parse_mdoc_body, TYPE_Va }, /* Vt */
+ { parse_mdoc_Xr, 0 }, /* Xr */
+ { NULL, 0 }, /* %A */
+ { NULL, 0 }, /* %B */
+ { NULL, 0 }, /* %D */
+ { NULL, 0 }, /* %I */
+ { NULL, 0 }, /* %J */
+ { NULL, 0 }, /* %N */
+ { NULL, 0 }, /* %O */
+ { NULL, 0 }, /* %P */
+ { NULL, 0 }, /* %R */
+ { NULL, 0 }, /* %T */
+ { NULL, 0 }, /* %V */
+ { NULL, 0 }, /* Ac */
+ { NULL, 0 }, /* Ao */
+ { NULL, 0 }, /* Aq */
+ { NULL, TYPE_At }, /* At */
+ { NULL, 0 }, /* Bc */
+ { NULL, 0 }, /* Bf */
+ { NULL, 0 }, /* Bo */
+ { NULL, 0 }, /* Bq */
+ { NULL, TYPE_Bsx }, /* Bsx */
+ { NULL, TYPE_Bx }, /* Bx */
+ { NULL, 0 }, /* Db */
+ { NULL, 0 }, /* Dc */
+ { NULL, 0 }, /* Do */
+ { NULL, 0 }, /* Dq */
+ { NULL, 0 }, /* Ec */
+ { NULL, 0 }, /* Ef */
+ { NULL, TYPE_Em }, /* Em */
+ { NULL, 0 }, /* Eo */
+ { NULL, TYPE_Fx }, /* Fx */
+ { NULL, TYPE_Ms }, /* Ms */
+ { NULL, 0 }, /* No */
+ { NULL, 0 }, /* Ns */
+ { NULL, TYPE_Nx }, /* Nx */
+ { NULL, TYPE_Ox }, /* Ox */
+ { NULL, 0 }, /* Pc */
+ { NULL, 0 }, /* Pf */
+ { NULL, 0 }, /* Po */
+ { NULL, 0 }, /* Pq */
+ { NULL, 0 }, /* Qc */
+ { NULL, 0 }, /* Ql */
+ { NULL, 0 }, /* Qo */
+ { NULL, 0 }, /* Qq */
+ { NULL, 0 }, /* Re */
+ { NULL, 0 }, /* Rs */
+ { NULL, 0 }, /* Sc */
+ { NULL, 0 }, /* So */
+ { NULL, 0 }, /* Sq */
+ { NULL, 0 }, /* Sm */
+ { NULL, 0 }, /* Sx */
+ { NULL, TYPE_Sy }, /* Sy */
+ { NULL, TYPE_Tn }, /* Tn */
+ { NULL, 0 }, /* Ux */
+ { NULL, 0 }, /* Xc */
+ { NULL, 0 }, /* Xo */
+ { parse_mdoc_head, 0 }, /* Fo */
+ { NULL, 0 }, /* Fc */
+ { NULL, 0 }, /* Oo */
+ { NULL, 0 }, /* Oc */
+ { NULL, 0 }, /* Bk */
+ { NULL, 0 }, /* Ek */
+ { NULL, 0 }, /* Bt */
+ { NULL, 0 }, /* Hf */
+ { NULL, 0 }, /* Fr */
+ { NULL, 0 }, /* Ud */
+ { NULL, TYPE_Lb }, /* Lb */
+ { NULL, 0 }, /* Lp */
+ { NULL, TYPE_Lk }, /* Lk */
+ { NULL, TYPE_Mt }, /* Mt */
+ { NULL, 0 }, /* Brq */
+ { NULL, 0 }, /* Bro */
+ { NULL, 0 }, /* Brc */
+ { NULL, 0 }, /* %C */
+ { NULL, 0 }, /* Es */
+ { NULL, 0 }, /* En */
+ { NULL, TYPE_Dx }, /* Dx */
+ { NULL, 0 }, /* %Q */
+ { NULL, 0 }, /* br */
+ { NULL, 0 }, /* sp */
+ { NULL, 0 }, /* %U */
+ { NULL, 0 }, /* Ta */
};
int
main(int argc, char *argv[])
{
- char cwd[MAXPATHLEN];
- int ch, rc, fd, i;
+ int ch, i;
size_t j, sz;
- const char *dir;
- struct str *s;
+ const char *path_arg;
struct mchars *mc;
struct manpaths dirs;
struct mparse *mp;
- struct ohash_info ino_info, filename_info, str_info;
+ struct ohash_info mpages_info, mlinks_info;
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;
+ mpages_info.alloc = mlinks_info.alloc = hash_alloc;
+ mpages_info.halloc = mlinks_info.halloc = hash_halloc;
+ mpages_info.hfree = mlinks_info.hfree = hash_free;
- ino_info.key_offset = offsetof(struct of, id);
- filename_info.key_offset = offsetof(struct of, file);
- str_info.key_offset = offsetof(struct str, key);
+ mpages_info.key_offset = offsetof(struct mpage, inodev);
+ mlinks_info.key_offset = offsetof(struct mlink, file);
progname = strrchr(argv[0], '/');
if (progname == NULL)
else
++progname;
- /*
- * 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);
- }
-
/*
* We accept a few different invocations.
* The CHECKOP macro makes sure that invocation styles don't
goto usage; \
} while (/*CONSTCOND*/0)
- dir = NULL;
+ path_arg = NULL;
op = OP_DEFAULT;
+ mparse_options = MPARSE_SO;
- while (-1 != (ch = getopt(argc, argv, "aC:d:ntu:vW")))
+ while (-1 != (ch = getopt(argc, argv, "aC:d:nQT:tu:vW")))
switch (ch) {
case ('a'):
use_all = 1;
break;
case ('C'):
CHECKOP(op, ch);
- dir = optarg;
+ path_arg = optarg;
op = OP_CONFFILE;
break;
case ('d'):
CHECKOP(op, ch);
- dir = optarg;
+ path_arg = optarg;
op = OP_UPDATE;
break;
case ('n'):
nodb = 1;
break;
+ case ('Q'):
+ mparse_options |= MPARSE_QUICK;
+ break;
+ case ('T'):
+ if (strcmp(optarg, "utf8")) {
+ fprintf(stderr, "-T%s: Unsupported "
+ "output format\n", optarg);
+ goto usage;
+ }
+ write_utf8 = 1;
+ break;
case ('t'):
CHECKOP(op, ch);
dup2(STDOUT_FILENO, STDERR_FILENO);
break;
case ('u'):
CHECKOP(op, ch);
- dir = optarg;
+ path_arg = optarg;
op = OP_DELETE;
break;
case ('v'):
goto usage;
}
- rc = 1;
- mp = mparse_alloc(MPARSE_AUTO,
- MANDOCLEVEL_FATAL, NULL, NULL, NULL);
+ exitcode = (int)MANDOCLEVEL_OK;
+ mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL, NULL);
mc = mchars_alloc();
- ohash_init(&strings, 6, &str_info);
- ohash_init(&inos, 6, &ino_info);
- ohash_init(&filenames, 6, &filename_info);
+ ohash_init(&mpages, 6, &mpages_info);
+ ohash_init(&mlinks, 6, &mlinks_info);
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))
+ if (0 == set_basedir(path_arg))
goto out;
for (i = 0; i < argc; i++)
- filescan(argv[i], dir);
- if (0 == dbopen(dir, 1))
+ filescan(argv[i]);
+ if (0 == dbopen(1))
goto out;
if (OP_TEST != op)
- dbprune(dir);
+ dbprune();
if (OP_DELETE != op)
- rc = ofmerge(mc, mp, dir);
- dbclose(dir, 1);
+ mpages_merge(mc, mp);
+ dbclose(1);
} else {
/*
* If we have arguments, use them as our manpaths.
for (i = 0; i < argc; i++)
dirs.paths[i] = mandoc_strdup(argv[i]);
} else
- manpath_parse(&dirs, dir, NULL, NULL);
+ manpath_parse(&dirs, path_arg, NULL, NULL);
/*
- * First scan the tree rooted at a base directory.
- * Then whak its database (if one exists), parse, and
- * build up the database.
+ * First scan the tree rooted at a base directory, then
+ * build a new database and finally move it into place.
* Ignore zero-length directories and strip trailing
* slashes.
*/
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]))
+
+ if (j) {
+ ohash_init(&mpages, 6, &mpages_info);
+ ohash_init(&mlinks, 6, &mlinks_info);
+ }
+
+ if (0 == set_basedir(dirs.paths[j]))
goto out;
- if (0 == path_reset(cwd, fd, dirs.paths[j]))
+ if (0 == treescan())
goto out;
- if (0 == dbopen(dirs.paths[j], 0))
+ if (0 == set_basedir(dirs.paths[j]))
goto out;
- if (0 == ofmerge(mc, mp, dirs.paths[j]))
+ if (0 == dbopen(0))
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);
+
+ mpages_merge(mc, mp);
+ dbclose(0);
+
+ if (j + 1 < dirs.sz) {
+ mpages_free();
+ ohash_delete(&mpages);
+ ohash_delete(&mlinks);
+ }
}
}
out:
- close(fd);
+ set_basedir(NULL);
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);
+ mpages_free();
+ ohash_delete(&mpages);
+ ohash_delete(&mlinks);
+ return(exitcode);
usage:
- fprintf(stderr, "usage: %s [-anvW] [-C file]\n"
- " %s [-anvW] dir ...\n"
- " %s [-nvW] -d dir [file ...]\n"
+ fprintf(stderr, "usage: %s [-anQvW] [-C file] [-Tutf8]\n"
+ " %s [-anQvW] [-Tutf8] dir ...\n"
+ " %s [-nQvW] [-Tutf8] -d dir [file ...]\n"
" %s [-nvW] -u dir [file ...]\n"
- " %s -t file ...\n",
+ " %s [-Q] -t file ...\n",
progname, progname, progname,
progname, progname);
- return(EXIT_FAILURE);
+ return((int)MANDOCLEVEL_BADARG);
}
/*
- * Scan a directory tree rooted at "base" for manpages.
+ * Scan a directory tree rooted at "basedir" for manpages.
* We use fts(), scanning directory parts along the way for clues to our
* section and architecture.
*
* TODO: accomodate for multi-language directories.
*/
static int
-treescan(const char *base)
+treescan(void)
{
FTS *f;
FTSENT *ff;
+ struct mlink *mlink;
int dform;
- char *sec;
- const char *dsec, *arch, *cp, *name, *path;
+ char *dsec, *arch, *fsec, *cp;
+ const char *path;
const char *argv[2];
argv[0] = ".";
*/
f = fts_open((char * const *)argv, FTS_LOGICAL, NULL);
if (NULL == f) {
- perror(base);
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say("", NULL);
return(0);
}
while (NULL != (ff = fts_read(f))) {
path = ff->fts_path + 2;
/*
- * If we're a regular file, add an "of" by using the
+ * If we're a regular file, add an mlink 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");
+ if (0 == strcmp(path, MANDOC_DB))
continue;
- } else if (inocheck(ff->fts_statp)) {
- WARNING(path, base, "Duplicate file");
- continue;
- }
-
- cp = ff->fts_name;
-
- if (0 == strcmp(cp, "mandocdb.db")) {
- WARNING(path, base, "Skip database");
+ if ( ! use_all && ff->fts_level < 2) {
+ if (warnings)
+ say(path, "Extraneous file");
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");
+ } else if (NULL == (fsec =
+ strrchr(ff->fts_name, '.'))) {
+ if ( ! use_all) {
+ if (warnings)
+ say(path,
+ "No filename suffix");
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);
+ } else if (0 == strcmp(++fsec, "html")) {
+ if (warnings)
+ say(path, "Skip html");
+ continue;
+ } else if (0 == strcmp(fsec, "gz")) {
+ if (warnings)
+ say(path, "Skip gz");
+ continue;
+ } else if (0 == strcmp(fsec, "ps")) {
+ if (warnings)
+ say(path, "Skip ps");
+ continue;
+ } else if (0 == strcmp(fsec, "pdf")) {
+ if (warnings)
+ say(path, "Skip pdf");
+ continue;
+ } else if ( ! use_all &&
+ ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
+ (FORM_CAT == dform && strcmp(fsec, "0")))) {
+ if (warnings)
+ say(path, "Wrong filename suffix");
+ continue;
+ } else
+ fsec[-1] = '\0';
+
+ mlink = mandoc_calloc(1, sizeof(struct mlink));
+ strlcpy(mlink->file, path, sizeof(mlink->file));
+ mlink->dform = dform;
+ mlink->dsec = dsec;
+ mlink->arch = arch;
+ mlink->name = ff->fts_name;
+ mlink->fsec = fsec;
+ mlink_add(mlink, ff->fts_statp);
continue;
- } else if (FTS_D != ff->fts_info &&
- FTS_DP != ff->fts_info)
+ } else if (FTS_D != ff->fts_info &&
+ FTS_DP != ff->fts_info) {
+ if (warnings)
+ say(path, "Not a regular file");
continue;
+ }
switch (ff->fts_level) {
case (0):
* 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);
+ dsec = cp + 3;
} else if (0 == strncmp(cp, "cat", 3)) {
dform = FORM_CAT;
- dsec = stradd(cp + 3);
+ dsec = cp + 3;
+ } else {
+ dform = FORM_NONE;
+ dsec = NULL;
}
if (NULL != dsec || use_all)
break;
- WARNING(path, base, "Unknown directory part");
+ if (warnings)
+ say(path, "Unknown directory part");
fts_set(f, ff, FTS_SKIP);
break;
case (2):
* Possibly our architecture.
* If we're descending, keep tabs on it.
*/
- arch = NULL;
if (FTS_DP != ff->fts_info && NULL != dsec)
- arch = stradd(ff->fts_name);
+ arch = ff->fts_name;
+ else
+ arch = NULL;
break;
default:
if (FTS_DP == ff->fts_info || use_all)
break;
- WARNING(path, base, "Extraneous directory part");
+ if (warnings)
+ say(path, "Extraneous directory part");
fts_set(f, ff, FTS_SKIP);
break;
}
}
/*
- * Add a file to the file vector.
+ * Add a file to the mlinks table.
* Do not verify that it's a "valid" looking manpage (we'll do that
* later).
*
* 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)
+filescan(const char *file)
{
- const char *sec, *arch, *name, *dsec;
- char *p, *start, *buf;
- int dform;
+ char buf[PATH_MAX];
struct stat st;
+ struct mlink *mlink;
+ char *p, *start;
assert(use_all);
if (0 == strncmp(file, "./", 2))
file += 2;
- if (-1 == stat(file, &st)) {
- WARNING(file, base, "%s", strerror(errno));
+ if (NULL == realpath(file, buf)) {
+ exitcode = (int)MANDOCLEVEL_BADARG;
+ say(file, NULL);
return;
- } else if ( ! (S_IFREG & st.st_mode)) {
- WARNING(file, base, "Not a regular file");
+ }
+
+ if (strstr(buf, basedir) == buf)
+ start = buf + strlen(basedir) + 1;
+ else if (OP_TEST == op)
+ start = buf;
+ else {
+ exitcode = (int)MANDOCLEVEL_BADARG;
+ say("", "%s: outside base directory", buf);
return;
- } else if (inocheck(&st)) {
- WARNING(file, base, "Duplicate file");
+ }
+
+ if (-1 == stat(buf, &st)) {
+ exitcode = (int)MANDOCLEVEL_BADARG;
+ say(file, NULL);
+ return;
+ } else if ( ! (S_IFREG & st.st_mode)) {
+ exitcode = (int)MANDOCLEVEL_BADARG;
+ say(file, "Not a regular file");
return;
}
- buf = mandoc_strdup(file);
- start = buf;
- sec = arch = name = dsec = NULL;
- dform = FORM_NONE;
+ mlink = mandoc_calloc(1, sizeof(struct mlink));
+ strlcpy(mlink->file, start, sizeof(mlink->file));
/*
* First try to guess our directory structure.
if (NULL != (p = strchr(start, '/'))) {
*p++ = '\0';
if (0 == strncmp(start, "man", 3)) {
- dform = FORM_SRC;
- dsec = start + 3;
+ mlink->dform = FORM_SRC;
+ mlink->dsec = start + 3;
} else if (0 == strncmp(start, "cat", 3)) {
- dform = FORM_CAT;
- dsec = start + 3;
+ mlink->dform = FORM_CAT;
+ mlink->dsec = start + 3;
}
start = p;
- if (NULL != dsec && NULL != (p = strchr(start, '/'))) {
+ if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
*p++ = '\0';
- arch = start;
+ mlink->arch = start;
start = p;
- }
+ }
}
/*
if ('.' == *p) {
*p++ = '\0';
- sec = p;
+ mlink->fsec = p;
}
/*
* Now try to parse the name.
* Use the filename portion of the path.
*/
- name = start;
+ mlink->name = start;
if (NULL != (p = strrchr(start, '/'))) {
- name = p + 1;
+ mlink->name = p + 1;
*p = '\0';
- }
-
- ofadd(base, dform, file, name, dsec, sec, arch, &st);
- free(buf);
+ }
+ mlink_add(mlink, &st);
}
-/*
- * See fileadd().
- */
-static int
-filecheck(const char *name)
+static void
+mlink_add(struct mlink *mlink, const struct stat *st)
{
- unsigned int index;
-
- index = ohash_qlookup(&filenames, name);
- return(NULL != ohash_find(&filenames, index));
+ struct inodev inodev;
+ struct mpage *mpage;
+ unsigned int slot;
+
+ assert(NULL != mlink->file);
+
+ mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
+ mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
+ mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
+ mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
+
+ if ('0' == *mlink->fsec) {
+ free(mlink->fsec);
+ mlink->fsec = mandoc_strdup(mlink->dsec);
+ mlink->fform = FORM_CAT;
+ } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
+ mlink->fform = FORM_SRC;
+ else
+ mlink->fform = FORM_NONE;
+
+ slot = ohash_qlookup(&mlinks, mlink->file);
+ assert(NULL == ohash_find(&mlinks, slot));
+ ohash_insert(&mlinks, slot, mlink);
+
+ inodev.st_ino = st->st_ino;
+ inodev.st_dev = st->st_dev;
+ slot = ohash_lookup_memory(&mpages, (char *)&inodev,
+ sizeof(struct inodev), inodev.st_ino);
+ mpage = ohash_find(&mpages, slot);
+ if (NULL == mpage) {
+ mpage = mandoc_calloc(1, sizeof(struct mpage));
+ mpage->inodev.st_ino = inodev.st_ino;
+ mpage->inodev.st_dev = inodev.st_dev;
+ ohash_insert(&mpages, slot, mpage);
+ } else
+ mlink->next = mpage->mlinks;
+ mpage->mlinks = mlink;
}
-/*
- * Use the standard hashing mechanism (K&R) to see if the given filename
- * already exists.
- */
static void
-fileadd(struct of *of)
+mlink_free(struct mlink *mlink)
{
- unsigned int index;
- index = ohash_qlookup(&filenames, of->file);
- assert(NULL == ohash_find(&filenames, index));
- ohash_insert(&filenames, index, of);
+ free(mlink->dsec);
+ free(mlink->arch);
+ free(mlink->name);
+ free(mlink->fsec);
+ free(mlink);
}
-/*
- * See inoadd().
- */
-static int
-inocheck(const struct stat *st)
+static void
+mpages_free(void)
{
- struct id id;
- uint32_t hash;
- unsigned int index;
-
- memset(&id, 0, sizeof(id));
- id.ino = hash = st->st_ino;
- id.dev = st->st_dev;
- index = ohash_lookup_memory
- (&inos, (char *)&id, sizeof(id), hash);
-
- return(NULL != ohash_find(&inos, index));
+ struct mpage *mpage;
+ struct mlink *mlink;
+ unsigned int slot;
+
+ mpage = ohash_first(&mpages, &slot);
+ while (NULL != mpage) {
+ while (NULL != (mlink = mpage->mlinks)) {
+ mpage->mlinks = mlink->next;
+ mlink_free(mlink);
+ }
+ free(mpage->sec);
+ free(mpage->arch);
+ free(mpage->title);
+ free(mpage->desc);
+ free(mpage);
+ mpage = ohash_next(&mpages, &slot);
+ }
}
/*
- * The hashing function used here is quite simple: simply take the inode
- * and use uint32_t of its bits.
- * Then when we do the lookup, use both the inode and device identifier.
+ * For each mlink to the mpage, check whether the path looks like
+ * it is formatted, and if it does, check whether a source manual
+ * exists by the same name, ignoring the suffix.
+ * If both conditions hold, drop the mlink.
*/
static void
-inoadd(const struct stat *st, struct of *of)
+mlinks_undupe(struct mpage *mpage)
{
- uint32_t hash;
- unsigned int index;
-
- of->id.ino = hash = st->st_ino;
- of->id.dev = st->st_dev;
- index = ohash_lookup_memory
- (&inos, (char *)&of->id, sizeof(of->id), hash);
-
- assert(NULL == ohash_find(&inos, index));
- ohash_insert(&inos, index, of);
+ char buf[PATH_MAX];
+ struct mlink **prev;
+ struct mlink *mlink;
+ char *bufp;
+
+ mpage->form = FORM_CAT;
+ prev = &mpage->mlinks;
+ while (NULL != (mlink = *prev)) {
+ if (FORM_CAT != mlink->dform) {
+ mpage->form = FORM_NONE;
+ goto nextlink;
+ }
+ if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) {
+ if (warnings)
+ say(mlink->file, "Filename too long");
+ goto nextlink;
+ }
+ bufp = strstr(buf, "cat");
+ assert(NULL != bufp);
+ memcpy(bufp, "man", 3);
+ if (NULL != (bufp = strrchr(buf, '.')))
+ *++bufp = '\0';
+ strlcat(buf, mlink->dsec, PATH_MAX);
+ if (NULL == ohash_find(&mlinks,
+ ohash_qlookup(&mlinks, buf)))
+ goto nextlink;
+ if (warnings)
+ say(mlink->file, "Man source exists: %s", buf);
+ if (use_all)
+ goto nextlink;
+ *prev = mlink->next;
+ mlink_free(mlink);
+ continue;
+nextlink:
+ prev = &(*prev)->next;
+ }
}
-static void
-ofadd(const char *base, int dform, const char *file,
- const char *name, const char *dsec, const char *sec,
- const char *arch, const struct stat *st)
+static int
+mlink_check(struct mpage *mpage, struct mlink *mlink)
{
- struct of *of;
- int sform;
-
- assert(NULL != file);
-
- 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;
- }
+ int match;
- of = mandoc_calloc(1, sizeof(struct of));
- strlcpy(of->file, file, MAXPATHLEN);
- of->name = name;
- of->sec = sec;
- of->dsec = dsec;
- of->arch = arch;
- of->sform = sform;
- of->dform = dform;
- of->next = ofs;
- ofs = of;
+ match = 1;
/*
- * Add to unique identifier hash.
- * Then if it's a source manual and we're going to use source in
- * favour of catpages, add it to that hash.
+ * Check whether the manual section given in a file
+ * agrees with the directory where the file is located.
+ * Some manuals have suffixes like (3p) on their
+ * section number either inside the file or in the
+ * directory name, some are linked into more than one
+ * section, like encrypt(1) = makekey(8).
*/
- inoadd(st, of);
- fileadd(of);
-}
-static void
-offree(void)
-{
- struct of *of;
+ if (FORM_SRC == mpage->form &&
+ strcasecmp(mpage->sec, mlink->dsec)) {
+ match = 0;
+ say(mlink->file, "Section \"%s\" manual in %s directory",
+ mpage->sec, mlink->dsec);
+ }
- while (NULL != (of = ofs)) {
- ofs = of->next;
- free(of);
+ /*
+ * Manual page directories exist for each kernel
+ * architecture as returned by machine(1).
+ * However, many manuals only depend on the
+ * application architecture as returned by arch(1).
+ * For example, some (2/ARM) manuals are shared
+ * across the "armish" and "zaurus" kernel
+ * architectures.
+ * A few manuals are even shared across completely
+ * different architectures, for example fdformat(1)
+ * on amd64, i386, sparc, and sparc64.
+ */
+
+ if (strcasecmp(mpage->arch, mlink->arch)) {
+ match = 0;
+ say(mlink->file, "Architecture \"%s\" manual in "
+ "\"%s\" directory", mpage->arch, mlink->arch);
}
+
+ if (strcasecmp(mpage->title, mlink->name))
+ match = 0;
+
+ return(match);
}
/*
- * Run through the files in the global vector "ofs" and add them to the
- * database specified in "base".
+ * Run through the files in the global vector "mpages"
+ * and add them to the database specified in "basedir".
*
* 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)
+static void
+mpages_merge(struct mchars *mc, struct mparse *mp)
{
- int form;
- size_t sz;
- struct mdoc *mdoc;
- struct man *man;
- char buf[MAXPATHLEN];
- char *bufp;
- const char *msec, *march, *mtitle, *cp;
- struct of *of;
- enum mandoclevel lvl;
-
- for (of = ofs; NULL != of; of = of->next) {
- /*
- * 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;
- }
+ char any[] = "any";
+ struct ohash_info str_info;
+ struct mpage *mpage;
+ struct mlink *mlink;
+ struct mdoc *mdoc;
+ struct man *man;
+ char *cp;
+ int match;
+ unsigned int pslot;
+ enum mandoclevel lvl;
+
+ str_info.alloc = hash_alloc;
+ str_info.halloc = hash_halloc;
+ str_info.hfree = hash_free;
+ str_info.key_offset = offsetof(struct str, key);
+
+ if (0 == nodb)
+ SQL_EXEC("BEGIN TRANSACTION");
+
+ mpage = ohash_first(&mpages, &pslot);
+ while (NULL != mpage) {
+ mlinks_undupe(mpage);
+ if (NULL == mpage->mlinks) {
+ mpage = ohash_next(&mpages, &pslot);
+ continue;
}
- words = NULL;
+ ohash_init(&strings, 6, &str_info);
mparse_reset(mp);
mdoc = NULL;
man = NULL;
- form = 0;
- msec = of->dsec;
- march = of->arch;
- mtitle = of->name;
/*
* Try interpreting the file as mdoc(7) or man(7)
* source code, unless it is already known to be
* formatted. Fall back to formatted mode.
*/
- if (FORM_SRC == of->dform || FORM_SRC == of->sform) {
- lvl = mparse_readfd(mp, -1, of->file);
+ if (FORM_CAT != mpage->mlinks->dform ||
+ FORM_CAT != mpage->mlinks->fform) {
+ lvl = mparse_readfd(mp, -1, mpage->mlinks->file);
if (lvl < MANDOCLEVEL_FATAL)
mparse_result(mp, &mdoc, &man);
- }
+ }
if (NULL != mdoc) {
- form = 1;
- msec = mdoc_meta(mdoc)->msec;
- march = mdoc_meta(mdoc)->arch;
- mtitle = mdoc_meta(mdoc)->title;
+ mpage->form = FORM_SRC;
+ mpage->sec =
+ mandoc_strdup(mdoc_meta(mdoc)->msec);
+ mpage->arch = mdoc_meta(mdoc)->arch;
+ mpage->arch = mandoc_strdup(
+ NULL == mpage->arch ? "" : mpage->arch);
+ mpage->title =
+ mandoc_strdup(mdoc_meta(mdoc)->title);
} else if (NULL != man) {
- form = 1;
- msec = man_meta(man)->msec;
- march = "";
- mtitle = man_meta(man)->title;
- }
-
- if (NULL == msec)
- msec = "";
- if (NULL == march)
- march = "";
- if (NULL == mtitle)
- mtitle = "";
-
- /*
- * Check whether the manual section given in a file
- * agrees with the directory where the file is located.
- * Some manuals have suffixes like (3p) on their
- * section number either inside the file or in the
- * directory name, some are linked into more than one
- * section, like encrypt(1) = makekey(8). 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);
-
- /*
- * Manual page directories exist for each kernel
- * architecture as returned by machine(1).
- * However, many manuals only depend on the
- * application architecture as returned by arch(1).
- * For example, some (2/ARM) manuals are shared
- * across the "armish" and "zaurus" kernel
- * architectures.
- * A few manuals are even shared across completely
- * different architectures, for example fdformat(1)
- * on amd64, i386, sparc, and sparc64.
- * 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);
+ mpage->form = FORM_SRC;
+ mpage->sec =
+ mandoc_strdup(man_meta(man)->msec);
+ mpage->arch =
+ mandoc_strdup(mpage->mlinks->arch);
+ mpage->title =
+ mandoc_strdup(man_meta(man)->title);
+ } else {
+ mpage->form = FORM_CAT;
+ mpage->sec =
+ mandoc_strdup(mpage->mlinks->dsec);
+ mpage->arch =
+ mandoc_strdup(mpage->mlinks->arch);
+ mpage->title =
+ mandoc_strdup(mpage->mlinks->name);
+ }
+ putkey(mpage, mpage->sec, TYPE_sec);
+ putkey(mpage, '\0' == *mpage->arch ?
+ any : mpage->arch, TYPE_arch);
+
+ for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
+ if ('\0' != *mlink->dsec)
+ putkey(mpage, mlink->dsec, TYPE_sec);
+ if ('\0' != *mlink->fsec)
+ putkey(mpage, mlink->fsec, TYPE_sec);
+ putkey(mpage, '\0' == *mlink->arch ?
+ any : mlink->arch, TYPE_arch);
+ putkey(mpage, mlink->name, TYPE_Nm);
+ }
- putkey(of, of->name, TYPE_Nm);
+ if (warnings && !use_all) {
+ match = 0;
+ for (mlink = mpage->mlinks; mlink;
+ mlink = mlink->next)
+ if (mlink_check(mpage, mlink))
+ match = 1;
+ } else
+ match = 1;
if (NULL != mdoc) {
if (NULL != (cp = mdoc_meta(mdoc)->name))
- putkey(of, cp, TYPE_Nm);
- parse_mdoc(of, mdoc_node(mdoc));
+ putkey(mpage, cp, TYPE_Nm);
+ assert(NULL == mpage->desc);
+ parse_mdoc(mpage, mdoc_node(mdoc));
+ putkey(mpage, NULL != mpage->desc ?
+ mpage->desc : mpage->mlinks->name, TYPE_Nd);
} else if (NULL != man)
- parse_man(of, man_node(man));
+ parse_man(mpage, man_node(man));
else
- parse_catpage(of, base);
+ parse_cat(mpage);
- dbindex(mc, form, of, base);
+ dbadd(mpage, mc);
+ ohash_delete(&strings);
+ mpage = ohash_next(&mpages, &pslot);
}
- return(1);
+ if (0 == nodb)
+ SQL_EXEC("END TRANSACTION");
}
static void
-parse_catpage(struct of *of, const char *base)
+parse_cat(struct mpage *mpage)
{
FILE *stream;
char *line, *p, *title;
size_t len, plen, titlesz;
- if (NULL == (stream = fopen(of->file, "r"))) {
- WARNING(of->file, base, "%s", strerror(errno));
+ if (NULL == (stream = fopen(mpage->mlinks->file, "r"))) {
+ if (warnings)
+ say(mpage->mlinks->file, NULL);
return;
}
*/
if (NULL == title || '\0' == *title) {
- WARNING(of->file, base, "Cannot find NAME section");
+ if (warnings)
+ say(mpage->mlinks->file,
+ "Cannot find NAME section");
+ assert(NULL == mpage->desc);
+ mpage->desc = mandoc_strdup(mpage->mlinks->name);
+ putkey(mpage, mpage->mlinks->name, TYPE_Nd);
fclose(stream);
free(title);
return;
for (p += 2; ' ' == *p || '\b' == *p; p++)
/* Skip to next word. */ ;
} else {
- WARNING(of->file, base, "No dash in title line");
+ if (warnings)
+ say(mpage->mlinks->file,
+ "No dash in title line");
p = title;
}
plen -= 2;
}
- of->desc = stradd(p);
- putkey(of, p, TYPE_Nd);
+ assert(NULL == mpage->desc);
+ mpage->desc = mandoc_strdup(p);
+ putkey(mpage, mpage->desc, TYPE_Nd);
fclose(stream);
free(title);
}
* Put a type/word pair into the word database for this particular file.
*/
static void
-putkey(const struct of *of, const char *value, uint64_t type)
+putkey(const struct mpage *mpage, char *value, uint64_t type)
{
+ char *cp;
assert(NULL != value);
- wordaddbuf(of, value, strlen(value), type);
-}
-
-/*
- * Like putkey() but for unterminated strings.
- */
-static void
-putkeys(const struct of *of, const char *value, int sz, uint64_t type)
-{
-
- wordaddbuf(of, value, sz, type);
+ if (TYPE_arch == type)
+ for (cp = value; *cp; cp++)
+ if (isupper((unsigned char)*cp))
+ *cp = _tolower((unsigned char)*cp);
+ putkeys(mpage, value, strlen(value), type);
}
/*
* Grok all nodes at or below a certain mdoc node into putkey().
*/
static void
-putmdockey(const struct of *of, const struct mdoc_node *n, uint64_t m)
+putmdockey(const struct mpage *mpage,
+ const struct mdoc_node *n, uint64_t m)
{
for ( ; NULL != n; n = n->next) {
if (NULL != n->child)
- putmdockey(of, n->child, m);
+ putmdockey(mpage, n->child, m);
if (MDOC_TEXT == n->type)
- putkey(of, n->string, m);
+ putkey(mpage, n->string, m);
}
}
-static int
-parse_man(struct of *of, const struct man_node *n)
+static void
+parse_man(struct mpage *mpage, const struct man_node *n)
{
const struct man_node *head, *body;
char *start, *sv, *title;
size_t sz, titlesz;
if (NULL == n)
- return(0);
+ return;
/*
* We're only searching for one thing: the first text child in
title[titlesz - 1] = ' ';
}
if (NULL == title)
- return(1);
+ return;
title = mandoc_realloc(title, titlesz + 1);
title[titlesz] = '\0';
if (0 == (sz = strlen(sv))) {
free(title);
- return(1);
+ return;
}
/* Erase trailing space. */
if (start == sv) {
free(title);
- return(1);
+ return;
}
start = sv;
byte = start[sz];
start[sz] = '\0';
- putkey(of, start, TYPE_Nm);
+ /*
+ * Assume a stray trailing comma in the
+ * name list if a name begins with a dash.
+ */
+
+ if ('-' == start[0] ||
+ ('\\' == start[0] && '-' == start[1]))
+ break;
+
+ putkey(mpage, start, TYPE_Nm);
if (' ' == byte) {
start += sz + 1;
}
if (sv == start) {
- putkey(of, start, TYPE_Nm);
+ putkey(mpage, start, TYPE_Nm);
free(title);
- return(1);
+ return;
}
while (isspace((unsigned char)*start))
while (' ' == *start)
start++;
- assert(NULL == of->desc);
- of->desc = stradd(start);
- putkey(of, start, TYPE_Nd);
+ assert(NULL == mpage->desc);
+ mpage->desc = mandoc_strdup(start);
+ putkey(mpage, mpage->desc, TYPE_Nd);
free(title);
- return(1);
+ return;
}
}
- for (n = n->child; n; n = n->next)
- if (parse_man(of, n))
- return(1);
-
- return(0);
+ for (n = n->child; n; n = n->next) {
+ if (NULL != mpage->desc)
+ break;
+ parse_man(mpage, n);
+ }
}
static void
-parse_mdoc(struct of *of, const struct mdoc_node *n)
+parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
{
assert(NULL != n);
/* FALLTHROUGH */
case (MDOC_TAIL):
if (NULL != mdocs[n->tok].fp)
- if (0 == (*mdocs[n->tok].fp)(of, n))
+ if (0 == (*mdocs[n->tok].fp)(mpage, n))
break;
-
- if (MDOCF_CHILD & mdocs[n->tok].flags)
- putmdockey(of, n->child, mdocs[n->tok].mask);
+ if (mdocs[n->tok].mask)
+ putmdockey(mpage, n->child,
+ mdocs[n->tok].mask);
break;
default:
assert(MDOC_ROOT != n->type);
continue;
}
if (NULL != n->child)
- parse_mdoc(of, n);
+ parse_mdoc(mpage, n);
}
}
static int
-parse_mdoc_Fd(struct of *of, const struct mdoc_node *n)
+parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
{
const char *start, *end;
size_t sz;
end--;
if (end > start)
- putkeys(of, start, end - start + 1, TYPE_In);
- return(1);
-}
-
-static int
-parse_mdoc_In(struct of *of, const struct mdoc_node *n)
-{
-
- if (NULL != n->child && MDOC_TEXT == n->child->type)
- return(0);
-
- putkey(of, n->child->string, TYPE_In);
- return(1);
+ putkeys(mpage, start, end - start + 1, TYPE_In);
+ return(0);
}
static int
-parse_mdoc_Fn(struct of *of, const struct mdoc_node *n)
+parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
{
- const char *cp;
+ char *cp;
if (NULL == (n = n->child) || MDOC_TEXT != n->type)
return(0);
while ('*' == *cp)
cp++;
- putkey(of, cp, TYPE_Fn);
+ putkey(mpage, cp, TYPE_Fn);
if (n->string < cp)
- putkeys(of, n->string, cp - n->string, TYPE_Ft);
+ putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
for (n = n->next; NULL != n; n = n->next)
if (MDOC_TEXT == n->type)
- putkey(of, n->string, TYPE_Fa);
+ putkey(mpage, n->string, TYPE_Fa);
return(0);
}
static int
-parse_mdoc_St(struct of *of, const struct mdoc_node *n)
+parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
{
+ char *cp;
- if (NULL == n->child || MDOC_TEXT != n->child->type)
+ if (NULL == (n = n->child))
return(0);
- putkey(of, n->child->string, TYPE_St);
- return(1);
-}
-
-static int
-parse_mdoc_Xr(struct of *of, const struct mdoc_node *n)
-{
-
- if (NULL == (n = n->child))
+ if (NULL == n->next) {
+ putkey(mpage, n->string, TYPE_Xr);
return(0);
+ }
- putkey(of, n->string, TYPE_Xr);
- return(1);
+ if (-1 == asprintf(&cp, "%s(%s)", n->string, n->next->string)) {
+ perror(NULL);
+ exit((int)MANDOCLEVEL_SYSERR);
+ }
+ putkey(mpage, cp, TYPE_Xr);
+ free(cp);
+ return(0);
}
static int
-parse_mdoc_Nd(struct of *of, const struct mdoc_node *n)
+parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
{
size_t sz;
- char *sv, *desc;
if (MDOC_BODY != n->type)
return(0);
* 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 != mpage->desc) {
+ sz = strlen(mpage->desc) +
+ strlen(n->string) + 2;
+ mpage->desc = mandoc_realloc(
+ mpage->desc, sz);
+ strlcat(mpage->desc, " ", sz);
+ strlcat(mpage->desc, n->string, sz);
+ } else
+ mpage->desc = mandoc_strdup(n->string);
}
if (NULL != n->child)
- parse_mdoc_Nd(of, n);
+ parse_mdoc_Nd(mpage, n);
}
-
- of->desc = NULL != desc ? stradd(desc) : NULL;
- free(desc);
return(1);
}
static int
-parse_mdoc_Nm(struct of *of, const struct mdoc_node *n)
+parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
{
- if (SEC_NAME == n->sec)
- return(1);
- else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
- return(0);
-
- return(1);
+ return(SEC_NAME == n->sec ||
+ (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type));
}
static int
-parse_mdoc_Sh(struct of *of, const struct mdoc_node *n)
+parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
{
return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
}
static int
-parse_mdoc_head(struct of *of, const struct mdoc_node *n)
+parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
{
return(MDOC_HEAD == n->type);
}
static int
-parse_mdoc_body(struct of *of, const struct mdoc_node *n)
+parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
{
return(MDOC_BODY == n->type);
}
/*
- * See straddbuf().
- */
-static char *
-stradd(const char *cp)
-{
-
- return(straddbuf(cp, strlen(cp)));
-}
-
-/*
- * 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;
-
- if (NULL != (s = hashget(cp, sz)))
- return(s->key);
-
- 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);
- 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));
-}
-
-/*
- * Add a word to the current parse sequence.
- * 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.
+ * Add a string to the hash table for the current manual.
+ * Each string has a bitmask telling which macros it belongs to.
+ * When we finish the manual, we'll dump the table.
*/
static void
-wordaddbuf(const struct of *of,
- const char *cp, size_t sz, uint64_t v)
+putkeys(const struct mpage *mpage,
+ const char *cp, size_t sz, uint64_t v)
{
struct str *s;
- unsigned int index;
const char *end;
+ uint64_t mask;
+ unsigned int slot;
+ int i;
if (0 == sz)
return;
- s = hashget(cp, sz);
+ if (verb > 1) {
+ for (i = 0, mask = 1;
+ i < mansearch_keymax;
+ i++, mask <<= 1)
+ if (mask & v)
+ break;
+ say(mpage->mlinks->file, "Adding key %s=%*s",
+ mansearch_keynames[i], sz, cp);
+ }
- if (NULL != s && of == s->of) {
+ end = cp + sz;
+ slot = ohash_qlookupi(&strings, cp, &end);
+ s = ohash_find(&strings, slot);
+
+ if (NULL != s && mpage == s->mpage) {
s->mask |= v;
return;
} 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);
+ ohash_insert(&strings, slot, s);
}
-
- s->next = words;
- s->of = of;
+ s->mpage = mpage;
s->mask = v;
- words = s;
}
/*
}
/*
- * Store the UTF-8 version of a key, or alias the pointer if the key has
- * no UTF-8 transcription marks in it.
+ * Store the rendered version of a key, or alias the pointer
+ * if the key contains no escape sequences.
*/
static void
-utf8key(struct mchars *mc, struct str *key)
+render_key(struct mchars *mc, struct str *key)
{
size_t sz, bsz, pos;
- char utfbuf[7], res[5];
+ char utfbuf[7], res[6];
char *buf;
const char *seq, *cpp, *val;
int len, u;
enum mandoc_esc esc;
- assert(NULL == key->utf8);
+ assert(NULL == key->rendered);
res[0] = '\\';
res[1] = '\t';
res[2] = ASCII_NBRSP;
res[3] = ASCII_HYPH;
- res[4] = '\0';
+ res[4] = ASCII_BREAK;
+ res[5] = '\0';
val = key->key;
bsz = strlen(val);
* pointer as ourselvse and get out of here.
*/
if (strcspn(val, res) == bsz) {
- key->utf8 = key->key;
+ key->rendered = key->key;
return;
}
val += sz;
}
- if (ASCII_HYPH == *val) {
+ switch (*val) {
+ case (ASCII_HYPH):
buf[pos++] = '-';
val++;
continue;
- } else if ('\t' == *val || ASCII_NBRSP == *val) {
+ case ('\t'):
+ /* FALLTHROUGH */
+ case (ASCII_NBRSP):
buf[pos++] = ' ';
val++;
+ /* FALLTHROUGH */
+ case (ASCII_BREAK):
continue;
- } else if ('\\' != *val)
+ default:
+ break;
+ }
+ if ('\\' != *val)
break;
/* Read past the slash. */
val++;
- u = 0;
/*
* Parse the escape sequence and see if it's a
* predefined character or special character.
*/
+
esc = mandoc_escape
((const char **)&val, &seq, &len);
if (ESCAPE_ERROR == esc)
break;
-
if (ESCAPE_SPECIAL != esc)
continue;
- if (0 == (u = mchars_spec2cp(mc, seq, len)))
- continue;
/*
- * If we have a Unicode codepoint, try to convert that
- * to a UTF-8 byte string.
+ * Render the special character
+ * as either UTF-8 or ASCII.
*/
- cpp = utfbuf;
- if (0 == (sz = utf8(u, utfbuf)))
- continue;
+
+ if (write_utf8) {
+ if (0 == (u = mchars_spec2cp(mc, seq, len)))
+ continue;
+ cpp = utfbuf;
+ if (0 == (sz = utf8(u, utfbuf)))
+ continue;
+ sz = strlen(cpp);
+ } else {
+ cpp = mchars_spec2str(mc, seq, len, &sz);
+ if (NULL == cpp)
+ continue;
+ if (ASCII_NBRSP == *cpp) {
+ cpp = " ";
+ sz = 1;
+ }
+ }
/* Copy the rendered glyph into the stream. */
- sz = strlen(cpp);
bsz += sz;
-
buf = mandoc_realloc(buf, bsz);
-
memcpy(&buf[pos], cpp, sz);
pos += sz;
}
buf[pos] = '\0';
- key->utf8 = buf;
+ key->rendered = buf;
}
/*
* Flush the current page's terms (and their bits) into the database.
* 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.
+ * Also, handle escape sequences at the last possible moment.
*/
static void
-dbindex(struct mchars *mc, int form,
- const struct of *of, const char *base)
+dbadd(const struct mpage *mpage, struct mchars *mc)
{
+ struct mlink *mlink;
struct str *key;
- const char *desc;
int64_t recno;
+ size_t i;
+ unsigned int slot;
- DEBUG(of->file, base, "Adding to index");
+ if (verb)
+ say(mpage->mlinks->file, "Adding to database");
if (nodb)
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;
+ i = 1;
+ SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
+ SQL_STEP(stmts[STMT_INSERT_PAGE]);
+ recno = sqlite3_last_insert_rowid(db);
+ sqlite3_reset(stmts[STMT_INSERT_PAGE]);
+
+ for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
+ i = 1;
+ SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
+ SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
+ SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
+ SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, recno);
+ SQL_STEP(stmts[STMT_INSERT_LINK]);
+ sqlite3_reset(stmts[STMT_INSERT_LINK]);
}
- sqlite3_exec(db, "BEGIN TRANSACTION", NULL, NULL, NULL);
-
- sqlite3_bind_text
- (stmts[STMT_INSERT_DOC], 1,
- 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]);
-
- 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]);
+ for (key = ohash_first(&strings, &slot); NULL != key;
+ key = ohash_next(&strings, &slot)) {
+ assert(key->mpage == mpage);
+ if (NULL == key->rendered)
+ render_key(mc, key);
+ i = 1;
+ SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
+ SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
+ SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, recno);
+ SQL_STEP(stmts[STMT_INSERT_KEY]);
sqlite3_reset(stmts[STMT_INSERT_KEY]);
+ if (key->rendered != key->key)
+ free(key->rendered);
+ free(key);
}
-
- sqlite3_exec(db, "COMMIT TRANSACTION", NULL, NULL, NULL);
-
}
static void
-dbprune(const char *base)
+dbprune(void)
{
- struct of *of;
+ struct mpage *mpage;
+ struct mlink *mlink;
+ size_t i;
+ unsigned int slot;
- if (nodb)
- return;
+ if (0 == nodb)
+ SQL_EXEC("BEGIN TRANSACTION");
- for (of = ofs; NULL != of; of = of->next) {
- sqlite3_bind_text
- (stmts[STMT_DELETE], 1,
- of->file, -1, SQLITE_STATIC);
- sqlite3_step(stmts[STMT_DELETE]);
- sqlite3_reset(stmts[STMT_DELETE]);
- DEBUG(of->file, base, "Deleted from index");
+ for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
+ mpage = ohash_next(&mpages, &slot)) {
+ mlink = mpage->mlinks;
+ if (verb)
+ say(mlink->file, "Deleting from database");
+ if (nodb)
+ continue;
+ for ( ; NULL != mlink; mlink = mlink->next) {
+ i = 1;
+ SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
+ i, mlink->dsec);
+ SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
+ i, mlink->arch);
+ SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
+ i, mlink->name);
+ SQL_STEP(stmts[STMT_DELETE_PAGE]);
+ sqlite3_reset(stmts[STMT_DELETE_PAGE]);
+ }
}
+
+ if (0 == nodb)
+ SQL_EXEC("END TRANSACTION");
}
/*
* If "real" is not set, rename the temporary file into the real one.
*/
static void
-dbclose(const char *base, int real)
+dbclose(int real)
{
size_t i;
- char file[MAXPATHLEN];
+ int status;
+ pid_t child;
if (nodb)
return;
if (real)
return;
- strlcpy(file, MANDOC_DB, MAXPATHLEN);
- strlcat(file, "~", MAXPATHLEN);
- if (-1 == rename(file, MANDOC_DB))
- perror(MANDOC_DB);
+ if ('\0' == *tempfilename) {
+ if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say(MANDOC_DB, "%s", strerror(errno));
+ }
+ return;
+ }
+
+ switch (child = fork()) {
+ case (-1):
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say("fork cmp", "%s", strerror(errno));
+ return;
+ case (0):
+ execlp("cmp", "cmp", "-s",
+ tempfilename, MANDOC_DB, NULL);
+ say("exec cmp", "%s", strerror(errno));
+ exit(0);
+ default:
+ break;
+ }
+ if (-1 == waitpid(child, &status, 0)) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say("wait cmp", "%s", strerror(errno));
+ } else if (WIFSIGNALED(status)) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say("cmp", "Died from a signal");
+ } else if (WEXITSTATUS(status)) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say(MANDOC_DB,
+ "Data changed, but cannot replace database");
+ }
+
+ *strrchr(tempfilename, '/') = '\0';
+ switch (child = fork()) {
+ case (-1):
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say("fork rm", "%s", strerror(errno));
+ return;
+ case (0):
+ execlp("rm", "rm", "-rf", tempfilename, NULL);
+ say("exec rm", "%s", strerror(errno));
+ exit((int)MANDOCLEVEL_SYSERR);
+ default:
+ break;
+ }
+ if (-1 == waitpid(child, &status, 0)) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say("wait rm", "%s", strerror(errno));
+ } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say(tempfilename,
+ "Cannot remove temporary directory");
+ }
}
/*
* Must be matched by dbclose().
*/
static int
-dbopen(const char *base, int real)
+dbopen(int real)
{
- char file[MAXPATHLEN];
const char *sql;
int rc, ofl;
- size_t sz;
if (nodb)
return(1);
- sz = strlcpy(file, MANDOC_DB, MAXPATHLEN);
- if ( ! real)
- sz = strlcat(file, "~", MAXPATHLEN);
+ *tempfilename = '\0';
+ ofl = SQLITE_OPEN_READWRITE;
- if (sz >= MAXPATHLEN) {
- fprintf(stderr, "%s: Path too long\n", file);
- return(0);
+ if (real) {
+ rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
+ if (SQLITE_OK != rc) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say(MANDOC_DB, "%s", sqlite3_errmsg(db));
+ return(0);
+ }
+ goto prepare_statements;
}
- if ( ! real)
- remove(file);
+ ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
- ofl = SQLITE_OPEN_PRIVATECACHE | SQLITE_OPEN_READWRITE |
- (0 == real ? SQLITE_OPEN_EXCLUSIVE : 0);
-
- rc = sqlite3_open_v2(file, &db, ofl, NULL);
+ remove(MANDOC_DB "~");
+ rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
if (SQLITE_OK == rc)
- return(1);
- if (SQLITE_CANTOPEN != rc) {
- perror(file);
+ goto create_tables;
+ if (MPARSE_QUICK & mparse_options) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say(MANDOC_DB "~", "%s", sqlite3_errmsg(db));
return(0);
}
- sqlite3_close(db);
- db = NULL;
-
- if (SQLITE_OK != (rc = sqlite3_open(file, &db))) {
- perror(file);
+ if (strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
+ sizeof(tempfilename)) >= sizeof(tempfilename)) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say("/tmp/mandocdb.XXXXXX", "Filename too long");
+ return(0);
+ }
+ if (NULL == mkdtemp(tempfilename)) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say(tempfilename, "%s", strerror(errno));
+ return(0);
+ }
+ if (strlcat(tempfilename, "/" MANDOC_DB,
+ sizeof(tempfilename)) >= sizeof(tempfilename)) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say(tempfilename, "Filename too long");
+ return(0);
+ }
+ rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
+ if (SQLITE_OK != rc) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say(tempfilename, "%s", sqlite3_errmsg(db));
return(0);
}
- sql = "CREATE TABLE \"docs\" (\n"
- " \"file\" TEXT NOT NULL,\n"
- " \"sec\" TEXT NOT NULL,\n"
- " \"arch\" TEXT NOT NULL,\n"
- " \"desc\" TEXT NOT NULL,\n"
+create_tables:
+ sql = "CREATE TABLE \"mpages\" (\n"
" \"form\" INTEGER NOT NULL,\n"
" \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
");\n"
"\n"
+ "CREATE TABLE \"mlinks\" (\n"
+ " \"sec\" TEXT NOT NULL,\n"
+ " \"arch\" TEXT NOT NULL,\n"
+ " \"name\" TEXT NOT NULL,\n"
+ " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
+ "ON DELETE CASCADE\n"
+ ");\n"
+ "\n"
"CREATE TABLE \"keys\" (\n"
" \"bits\" INTEGER NOT NULL,\n"
" \"key\" TEXT NOT NULL,\n"
- " \"docid\" INTEGER NOT NULL REFERENCES docs(id) "
- "ON DELETE CASCADE,\n"
- " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
- ");\n"
- "\n"
- "CREATE INDEX \"key_index\" ON keys (key);\n";
+ " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
+ "ON DELETE CASCADE\n"
+ ");\n";
if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
- perror(sqlite3_errmsg(db));
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say(MANDOC_DB, "%s", sqlite3_errmsg(db));
return(0);
}
- sql = "DELETE FROM docs where file=?";
- sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE], NULL);
- sql = "INSERT INTO docs "
- "(file,sec,arch,desc,form) VALUES (?,?,?,?,?)";
- sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_DOC], NULL);
+prepare_statements:
+ SQL_EXEC("PRAGMA foreign_keys = ON");
+ sql = "DELETE FROM mpages WHERE id IN "
+ "(SELECT pageid FROM mlinks WHERE "
+ "sec=? AND arch=? AND name=?)";
+ sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
+ sql = "INSERT INTO mpages "
+ "(form) VALUES (?)";
+ sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
+ sql = "INSERT INTO mlinks "
+ "(sec,arch,name,pageid) VALUES (?,?,?,?)";
+ sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
sql = "INSERT INTO keys "
- "(bits,key,docid) VALUES (?,?,?)";
+ "(bits,key,pageid) VALUES (?,?,?)";
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
+
+#ifndef __APPLE__
+ /*
+ * When opening a new database, we can turn off
+ * synchronous mode for much better performance.
+ */
+
+ if (real)
+ SQL_EXEC("PRAGMA synchronous = OFF");
+#endif
+
return(1);
}
}
static int
-path_reset(const char *cwd, int fd, const char *base)
+set_basedir(const char *targetdir)
{
+ static char startdir[PATH_MAX];
+ static int fd;
- if (-1 == fchdir(fd)) {
- perror(cwd);
+ /*
+ * 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 ('\0' == *startdir) {
+ if (NULL == getcwd(startdir, PATH_MAX)) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ if (NULL != targetdir)
+ say(".", NULL);
+ return(0);
+ }
+ if (-1 == (fd = open(startdir, O_RDONLY, 0))) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say(startdir, NULL);
+ return(0);
+ }
+ if (NULL == targetdir)
+ targetdir = startdir;
+ } else {
+ if (-1 == fd)
+ return(0);
+ if (-1 == fchdir(fd)) {
+ close(fd);
+ basedir[0] = '\0';
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say(startdir, NULL);
+ return(0);
+ }
+ if (NULL == targetdir) {
+ close(fd);
+ return(1);
+ }
+ }
+ if (NULL == realpath(targetdir, basedir)) {
+ basedir[0] = '\0';
+ exitcode = (int)MANDOCLEVEL_BADARG;
+ say(targetdir, NULL);
return(0);
- } else if (-1 == chdir(base)) {
- perror(base);
+ } else if (-1 == chdir(basedir)) {
+ exitcode = (int)MANDOCLEVEL_BADARG;
+ say("", NULL);
return(0);
}
return(1);
}
+
+static void
+say(const char *file, const char *format, ...)
+{
+ va_list ap;
+
+ if ('\0' != *basedir)
+ fprintf(stderr, "%s", basedir);
+ if ('\0' != *basedir && '\0' != *file)
+ fputs("//", stderr);
+ if ('\0' != *file)
+ fprintf(stderr, "%s", file);
+ fputs(": ", stderr);
+
+ if (NULL == format) {
+ perror(NULL);
+ return;
+ }
+
+ va_start(ap, format);
+ vfprintf(stderr, format, ap);
+ va_end(ap);
+
+ fputc('\n', stderr);
+}
git.ckatri.com / mandoc.git / blobdiff