-/* $Id: mandocdb.c,v 1.82 2013/12/27 03:06:17 schwarze Exp $ */
+/* $Id: mandocdb.c,v 1.114 2014/01/22 20:58:39 schwarze Exp $ */
/*
* Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
- * Copyright (c) 2011, 2012, 2013 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 "manpath.h"
#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))
};
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 */
+ 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 inodev {
struct mlink {
char file[PATH_MAX]; /* filename rel. to manpath */
enum form dform; /* format from directory */
- enum form sform; /* format from file name suffix */
+ 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 title {
- char *title; /* name(sec/arch) given inside the file */
- char *file; /* file name in case of mismatch */
+ struct mlink *next; /* singly linked list */
};
enum stmt {
};
static void dbclose(int);
-static void dbindex(const struct mpage *, struct mchars *);
+static void dbadd(const struct mpage *, struct mchars *);
static int dbopen(int);
static void dbprune(void);
-static void fileadd(struct mlink *);
-static int filecheck(const char *);
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 mpage *);
-static int inocheck(const struct stat *);
-static void ofadd(int, const char *, const char *, const char *,
- const char *, const char *, const struct stat *);
+static void mlink_add(struct mlink *, const struct stat *);
+static int mlink_check(struct mpage *, struct mlink *);
static void mlink_free(struct mlink *);
+static void mlinks_undupe(struct mpage *);
static void mpages_free(void);
-static void mpages_merge(struct mchars *, struct mparse *, int);
+static void 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_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_In(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_St(struct mpage *, const struct mdoc_node *);
static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
-static void putkey(const struct mpage *,
- const char *, uint64_t);
+static void 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 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 char *progname;
-static int use_all; /* use all found files */
static int nodb; /* no database changes */
+static int quick; /* 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 char basedir[PATH_MAX]; /* current base directory */
static struct ohash mpages; /* table of distinct manual pages */
-static struct ohash filenames; /* table of filenames */
+static struct ohash mlinks; /* table of directory entries */
static struct ohash strings; /* table of all strings */
static sqlite3 *db = NULL; /* current database */
static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
{ parse_mdoc_Fn, 0 }, /* Fn */
{ NULL, TYPE_Ft }, /* Ft */
{ NULL, TYPE_Ic }, /* Ic */
- { parse_mdoc_In, TYPE_In }, /* In */
+ { NULL, TYPE_In }, /* In */
{ NULL, TYPE_Li }, /* Li */
{ parse_mdoc_Nd, TYPE_Nd }, /* Nd */
{ parse_mdoc_Nm, TYPE_Nm }, /* Nm */
{ NULL, 0 }, /* Ot */
{ NULL, TYPE_Pa }, /* Pa */
{ NULL, 0 }, /* Rv */
- { parse_mdoc_St, 0 }, /* St */
+ { NULL, TYPE_St }, /* St */
{ NULL, TYPE_Va }, /* Va */
{ parse_mdoc_body, TYPE_Va }, /* Vt */
{ parse_mdoc_Xr, 0 }, /* Xr */
struct mchars *mc;
struct manpaths dirs;
struct mparse *mp;
- struct ohash_info mpages_info, filename_info;
+ struct ohash_info mpages_info, mlinks_info;
memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
memset(&dirs, 0, sizeof(struct manpaths));
- mpages_info.alloc = filename_info.alloc = hash_alloc;
- mpages_info.halloc = filename_info.halloc = hash_halloc;
- mpages_info.hfree = filename_info.hfree = hash_free;
+ mpages_info.alloc = mlinks_info.alloc = hash_alloc;
+ mpages_info.halloc = mlinks_info.halloc = hash_halloc;
+ mpages_info.hfree = mlinks_info.hfree = hash_free;
mpages_info.key_offset = offsetof(struct mpage, inodev);
- filename_info.key_offset = offsetof(struct mlink, file);
+ mlinks_info.key_offset = offsetof(struct mlink, file);
progname = strrchr(argv[0], '/');
if (progname == NULL)
path_arg = NULL;
op = OP_DEFAULT;
- 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;
case ('n'):
nodb = 1;
break;
+ case ('Q'):
+ quick = 1;
+ break;
+ case ('T'):
+ if (strcmp(optarg, "utf8")) {
+ fprintf(stderr, "-T%s: Unsupported "
+ "output format\n", optarg);
+ goto usage;
+ }
+ write_utf8 = 1;
+ break;
case ('t'):
CHECKOP(op, ch);
dup2(STDOUT_FILENO, STDERR_FILENO);
exitcode = (int)MANDOCLEVEL_OK;
mp = mparse_alloc(MPARSE_AUTO,
- MANDOCLEVEL_FATAL, NULL, NULL, NULL);
+ MANDOCLEVEL_FATAL, NULL, NULL, quick);
mc = mchars_alloc();
ohash_init(&mpages, 6, &mpages_info);
- ohash_init(&filenames, 6, &filename_info);
+ ohash_init(&mlinks, 6, &mlinks_info);
if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
/*
if (OP_TEST != op)
dbprune();
if (OP_DELETE != op)
- mpages_merge(mc, mp, 0);
+ mpages_merge(mc, mp);
dbclose(1);
} else {
/*
if (j) {
ohash_init(&mpages, 6, &mpages_info);
- ohash_init(&filenames, 6, &filename_info);
+ ohash_init(&mlinks, 6, &mlinks_info);
}
if (0 == set_basedir(dirs.paths[j]))
if (0 == dbopen(0))
goto out;
- mpages_merge(mc, mp, warnings && !use_all);
+ mpages_merge(mc, mp);
dbclose(0);
if (j + 1 < dirs.sz) {
mpages_free();
ohash_delete(&mpages);
- ohash_delete(&filenames);
+ ohash_delete(&mlinks);
}
}
}
mparse_free(mp);
mpages_free();
ohash_delete(&mpages);
- ohash_delete(&filenames);
+ 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);
{
FTS *f;
FTSENT *ff;
+ struct mlink *mlink;
int dform;
- char *sec;
- const char *dsec, *arch, *cp, *path;
+ char *dsec, *arch, *fsec, *cp;
+ const char *path;
const char *argv[2];
argv[0] = ".";
while (NULL != (ff = fts_read(f))) {
path = ff->fts_path + 2;
/*
- * If we're a regular file, add an mpage 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 (0 == strcmp(path, MANDOC_DB))
if (warnings)
say(path, "Extraneous file");
continue;
- } else if (inocheck(ff->fts_statp)) {
- if (warnings)
- say(path, "Duplicate file");
- continue;
- } else if (NULL == (sec =
+ } else if (NULL == (fsec =
strrchr(ff->fts_name, '.'))) {
if ( ! use_all) {
if (warnings)
"No filename suffix");
continue;
}
- } else if (0 == strcmp(++sec, "html")) {
+ } else if (0 == strcmp(++fsec, "html")) {
if (warnings)
say(path, "Skip html");
continue;
- } else if (0 == strcmp(sec, "gz")) {
+ } else if (0 == strcmp(fsec, "gz")) {
if (warnings)
say(path, "Skip gz");
continue;
- } else if (0 == strcmp(sec, "ps")) {
+ } else if (0 == strcmp(fsec, "ps")) {
if (warnings)
say(path, "Skip ps");
continue;
- } else if (0 == strcmp(sec, "pdf")) {
+ } else if (0 == strcmp(fsec, "pdf")) {
if (warnings)
say(path, "Skip pdf");
continue;
} else if ( ! use_all &&
- ((FORM_SRC == dform && strcmp(sec, dsec)) ||
- (FORM_CAT == dform && strcmp(sec, "0")))) {
+ ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
+ (FORM_CAT == dform && strcmp(fsec, "0")))) {
if (warnings)
say(path, "Wrong filename suffix");
continue;
} else
- sec[-1] = '\0';
- ofadd(dform, path, ff->fts_name, dsec, sec,
- arch, ff->fts_statp);
+ 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 &&
+ } else if (FTS_D != ff->fts_info &&
FTS_DP != ff->fts_info) {
if (warnings)
say(path, "Not a regular file");
* 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;
} else if (0 == strncmp(cp, "cat", 3)) {
dform = FORM_CAT;
dsec = cp + 3;
+ } else {
+ dform = FORM_NONE;
+ dsec = NULL;
}
if (NULL != dsec || use_all)
* Possibly our architecture.
* If we're descending, keep tabs on it.
*/
- arch = NULL;
if (FTS_DP != ff->fts_info && NULL != dsec)
arch = ff->fts_name;
+ else
+ arch = NULL;
break;
default:
if (FTS_DP == ff->fts_info || use_all)
}
/*
- * 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 mpage vector.
* See treescan() for the fts(3) version of this.
*/
static void
filescan(const char *file)
{
char buf[PATH_MAX];
- const char *sec, *arch, *name, *dsec;
- char *p, *start;
- int dform;
struct stat st;
+ struct mlink *mlink;
+ char *p, *start;
assert(use_all);
exitcode = (int)MANDOCLEVEL_BADARG;
say(file, NULL);
return;
- } else if (OP_TEST != op && strstr(buf, basedir) != buf) {
+ }
+
+ 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 (-1 == stat(buf, &st)) {
+ }
+
+ if (-1 == stat(buf, &st)) {
exitcode = (int)MANDOCLEVEL_BADARG;
say(file, NULL);
return;
exitcode = (int)MANDOCLEVEL_BADARG;
say(file, "Not a regular file");
return;
- } else if (inocheck(&st)) {
- if (warnings)
- say(file, "Duplicate file");
- return;
}
- start = buf + strlen(basedir);
- 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(dform, file, name, dsec, sec, arch, &st);
-}
-
-/*
- * See fileadd().
- */
-static int
-filecheck(const char *name)
-{
-
- return(NULL != ohash_find(&filenames,
- ohash_qlookup(&filenames, name)));
+ }
+ mlink_add(mlink, &st);
}
-/*
- * Use the standard hashing mechanism (K&R) to see if the given filename
- * already exists.
- */
static void
-fileadd(struct mlink *mlink)
-{
- unsigned int slot;
-
- slot = ohash_qlookup(&filenames, mlink->file);
- assert(NULL == ohash_find(&filenames, slot));
- ohash_insert(&filenames, slot, mlink);
-}
-
-/*
- * See inoadd().
- */
-static int
-inocheck(const struct stat *st)
+mlink_add(struct mlink *mlink, const struct stat *st)
{
struct inodev inodev;
- uint32_t hash;
-
- memset(&inodev, 0, sizeof(inodev));
- inodev.st_ino = hash = st->st_ino;
- inodev.st_dev = st->st_dev;
-
- return(NULL != ohash_find(&mpages, ohash_lookup_memory(
- &mpages, (char *)&inodev, sizeof(inodev), hash)));
-}
-
-/*
- * 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.
- */
-static void
-inoadd(const struct stat *st, struct mpage *mpage)
-{
+ struct mpage *mpage;
unsigned int slot;
- mpage->inodev.st_ino = st->st_ino;
- mpage->inodev.st_dev = st->st_dev;
- slot = ohash_lookup_memory(&mpages, (char *)&mpage->inodev,
- sizeof(struct inodev), st->st_ino);
+ assert(NULL != mlink->file);
- assert(NULL == ohash_find(&mpages, slot));
- ohash_insert(&mpages, slot, mpage);
-}
+ 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 : "");
-static void
-ofadd(int dform, const char *file, const char *name, const char *dsec,
- const char *sec, const char *arch, const struct stat *st)
-{
- struct mpage *mpage;
- struct mlink *mlink;
- int sform;
-
- assert(NULL != file);
-
- if (NULL == name)
- name = "";
- if (NULL == sec)
- sec = "";
- if (NULL == dsec)
- dsec = "";
- if (NULL == arch)
- arch = "";
-
- if ('0' == *sec) {
- sec = dsec;
- sform = FORM_CAT;
- } else if ('1' <= *sec && '9' >= *sec)
- sform = FORM_SRC;
+ 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
- sform = FORM_NONE;
+ mlink->fform = FORM_NONE;
- mlink = mandoc_calloc(1, sizeof(struct mlink));
- strlcpy(mlink->file, file, PATH_MAX);
- mlink->dform = dform;
- mlink->sform = sform;
- mlink->dsec = mandoc_strdup(dsec);
- mlink->arch = mandoc_strdup(arch);
- mlink->name = mandoc_strdup(name);
- mlink->fsec = mandoc_strdup(sec);
-
- mpage = mandoc_calloc(1, sizeof(struct mpage));
- mpage->mlinks = mlink;
+ slot = ohash_qlookup(&mlinks, mlink->file);
+ assert(NULL == ohash_find(&mlinks, slot));
+ ohash_insert(&mlinks, slot, mlink);
- /*
- * 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.
- */
- inoadd(st, mpage);
- fileadd(mpage->mlinks);
+ 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;
}
static void
mpage = ohash_first(&mpages, &slot);
while (NULL != mpage) {
while (NULL != (mlink = mpage->mlinks)) {
- mpage->mlinks = NULL;
+ mpage->mlinks = mlink->next;
mlink_free(mlink);
}
free(mpage->sec);
}
}
+/*
+ * 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
+mlinks_undupe(struct mpage *mpage)
+{
+ char buf[PATH_MAX];
+ struct mlink **prev;
+ struct mlink *mlink;
+ char *bufp;
+
+ mpage->form = FORM_CAT;
+ prev = &mpage->mlinks;
+ while (NULL != (mlink = *prev)) {
+ if (FORM_CAT != mlink->dform) {
+ mpage->form = FORM_NONE;
+ goto nextlink;
+ }
+ if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) {
+ if (warnings)
+ say(mlink->file, "Filename too long");
+ goto nextlink;
+ }
+ bufp = strstr(buf, "cat");
+ assert(NULL != bufp);
+ memcpy(bufp, "man", 3);
+ if (NULL != (bufp = strrchr(buf, '.')))
+ *++bufp = '\0';
+ strlcat(buf, mlink->dsec, PATH_MAX);
+ if (NULL == ohash_find(&mlinks,
+ ohash_qlookup(&mlinks, buf)))
+ goto nextlink;
+ if (warnings)
+ say(mlink->file, "Man source exists: %s", buf);
+ if (use_all)
+ goto nextlink;
+ *prev = mlink->next;
+ mlink_free(mlink);
+ continue;
+nextlink:
+ prev = &(*prev)->next;
+ }
+}
+
+static int
+mlink_check(struct mpage *mpage, struct mlink *mlink)
+{
+ int match;
+
+ match = 1;
+
+ /*
+ * 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).
+ */
+
+ 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);
+ }
+
+ /*
+ * 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 "mpages"
* and add them to the database specified in "basedir".
* and filename to determine whether the file is parsable or not.
*/
static void
-mpages_merge(struct mchars *mc, struct mparse *mp, int check_reachable)
+mpages_merge(struct mchars *mc, struct mparse *mp)
{
- struct ohash title_table;
- struct ohash_info title_info, str_info;
- char buf[PATH_MAX];
+ char any[] = "any";
+ struct ohash_info str_info;
struct mpage *mpage;
+ struct mlink *mlink;
struct mdoc *mdoc;
struct man *man;
- struct title *title_entry;
- char *bufp, *title_str;
- const char *cp;
- size_t sz;
+ char *cp;
int match;
- unsigned int pslot, tslot;
+ unsigned int pslot;
enum mandoclevel lvl;
str_info.alloc = hash_alloc;
str_info.hfree = hash_free;
str_info.key_offset = offsetof(struct str, key);
- if (check_reachable) {
- title_info.alloc = hash_alloc;
- title_info.halloc = hash_halloc;
- title_info.hfree = hash_free;
- title_info.key_offset = offsetof(struct title, title);
- ohash_init(&title_table, 6, &title_info);
- }
+ if (0 == nodb)
+ SQL_EXEC("BEGIN TRANSACTION");
mpage = ohash_first(&mpages, &pslot);
while (NULL != mpage) {
- /*
- * 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 == mpage->mlinks->dform) {
- sz = strlcpy(buf, mpage->mlinks->file, PATH_MAX);
- if (sz >= PATH_MAX) {
- if (warnings)
- say(mpage->mlinks->file,
- "Filename too long");
- mpage = ohash_next(&mpages, &pslot);
- continue;
- }
- bufp = strstr(buf, "cat");
- assert(NULL != bufp);
- memcpy(bufp, "man", 3);
- if (NULL != (bufp = strrchr(buf, '.')))
- *++bufp = '\0';
- strlcat(buf, mpage->mlinks->dsec, PATH_MAX);
- if (filecheck(buf)) {
- if (warnings)
- say(mpage->mlinks->file, "Man "
- "source exists: %s", buf);
- mpage = ohash_next(&mpages, &pslot);
- continue;
- }
+ mlinks_undupe(mpage);
+ if (NULL == mpage->mlinks) {
+ mpage = ohash_next(&mpages, &pslot);
+ continue;
}
ohash_init(&strings, 6, &str_info);
mparse_reset(mp);
mdoc = NULL;
man = NULL;
- match = 1;
/*
* Try interpreting the file as mdoc(7) or man(7)
* formatted. Fall back to formatted mode.
*/
if (FORM_CAT != mpage->mlinks->dform ||
- FORM_CAT != mpage->mlinks->sform) {
+ FORM_CAT != mpage->mlinks->fform) {
lvl = mparse_readfd(mp, -1, mpage->mlinks->file);
if (lvl < MANDOCLEVEL_FATAL)
mparse_result(mp, &mdoc, &man);
mpage->sec =
mandoc_strdup(mdoc_meta(mdoc)->msec);
mpage->arch = mdoc_meta(mdoc)->arch;
- if (NULL == mpage->arch)
- mpage->arch = "";
- mpage->arch = mandoc_strdup(mpage->arch);
+ mpage->arch = mandoc_strdup(
+ NULL == mpage->arch ? "" : mpage->arch);
mpage->title =
mandoc_strdup(mdoc_meta(mdoc)->title);
} else if (NULL != man) {
mpage->title =
mandoc_strdup(mpage->mlinks->name);
}
-
- /*
- * 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 (warnings && !use_all && FORM_SRC == mpage->form &&
- strcasecmp(mpage->sec, mpage->mlinks->dsec)) {
- match = 0;
- say(mpage->mlinks->file, "Section \"%s\" "
- "manual in %s directory",
- mpage->sec, mpage->mlinks->dsec);
+ 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);
}
- /*
- * 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 (warnings && !use_all &&
- strcasecmp(mpage->arch, mpage->mlinks->arch)) {
- match = 0;
- say(mpage->mlinks->file, "Architecture \"%s\" "
- "manual in \"%s\" directory",
- mpage->arch, mpage->mlinks->arch);
- }
- if (warnings && !use_all &&
- strcasecmp(mpage->title, mpage->mlinks->name))
+ if (warnings && !use_all) {
match = 0;
-
- putkey(mpage, mpage->mlinks->name, TYPE_Nm);
+ 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))
else
parse_cat(mpage);
- /*
- * Build a title string for the file. If it matches
- * the location of the file, remember the title as
- * found; else, remember it as missing.
- */
-
- if (check_reachable) {
- if (-1 == asprintf(&title_str, "%s(%s%s%s)",
- mpage->title, mpage->sec,
- '\0' == *mpage->arch ? "" : "/",
- mpage->arch)) {
- perror(NULL);
- exit((int)MANDOCLEVEL_SYSERR);
- }
- tslot = ohash_qlookup(&title_table, title_str);
- title_entry = ohash_find(&title_table, tslot);
- if (NULL == title_entry) {
- title_entry = mandoc_malloc(
- sizeof(struct title));
- title_entry->title = title_str;
- title_entry->file = mandoc_strdup(
- match ? "" : mpage->mlinks->file);
- ohash_insert(&title_table, tslot,
- title_entry);
- } else {
- if (match)
- *title_entry->file = '\0';
- free(title_str);
- }
- }
-
- dbindex(mpage, mc);
+ dbadd(mpage, mc);
ohash_delete(&strings);
mpage = ohash_next(&mpages, &pslot);
}
- if (check_reachable) {
- title_entry = ohash_first(&title_table, &tslot);
- while (NULL != title_entry) {
- if ('\0' != *title_entry->file)
- say(title_entry->file,
- "Probably unreachable, title is %s",
- title_entry->title);
- free(title_entry->title);
- free(title_entry->file);
- free(title_entry);
- title_entry = ohash_next(&title_table, &tslot);
- }
- ohash_delete(&title_table);
- }
+ if (0 == nodb)
+ SQL_EXEC("END TRANSACTION");
}
static void
* Put a type/word pair into the word database for this particular file.
*/
static void
-putkey(const struct mpage *mpage, const char *value, uint64_t type)
+putkey(const struct mpage *mpage, char *value, uint64_t type)
{
+ char *cp;
assert(NULL != value);
+ 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);
}
byte = start[sz];
start[sz] = '\0';
+ /*
+ * 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) {
if (end > start)
putkeys(mpage, start, end - start + 1, TYPE_In);
- return(1);
-}
-
-static int
-parse_mdoc_In(struct mpage *mpage, const struct mdoc_node *n)
-{
-
- if (NULL != n->child && MDOC_TEXT == n->child->type)
- return(0);
-
- putkey(mpage, n->child->string, TYPE_In);
- return(1);
+ return(0);
}
static int
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);
return(0);
}
-static int
-parse_mdoc_St(struct mpage *mpage, const struct mdoc_node *n)
-{
-
- if (NULL == n->child || MDOC_TEXT != n->child->type)
- return(0);
-
- putkey(mpage, n->child->string, TYPE_St);
- return(1);
-}
-
static int
parse_mdoc_Xr(struct mpage *mpage, 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
const char *cp, size_t sz, uint64_t v)
{
struct str *s;
- unsigned int slot;
const char *end;
+ uint64_t mask;
+ unsigned int slot;
+ int i;
if (0 == sz)
return;
+ 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);
+ }
+
end = cp + sz;
slot = ohash_qlookupi(&strings, cp, &end);
s = ohash_find(&strings, slot);
}
/*
- * 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(const struct mpage *mpage, struct mchars *mc)
+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;
if (verb)
- say(mpage->mlinks->file, "Adding to index");
+ say(mpage->mlinks->file, "Adding to database");
if (nodb)
return;
- desc = "";
- if (NULL != mpage->desc && '\0' != *mpage->desc) {
- key = ohash_find(&strings,
- ohash_qlookup(&strings, mpage->desc));
- assert(NULL != key);
- if (NULL == key->utf8)
- utf8key(mc, key);
- desc = key->utf8;
- }
-
- SQL_EXEC("BEGIN TRANSACTION");
-
i = 1;
- /*
- * XXX The following three lines are obsolete
- * and only kept for backward compatibility
- * until apropos(1) and friends have caught up.
- */
- SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->mlinks->file);
- SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->mlinks->dsec);
- SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->mlinks->arch);
- SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, desc);
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]);
- i = 1;
- SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mpage->mlinks->dsec);
- SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mpage->mlinks->arch);
- SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mpage->mlinks->file);
- SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, recno);
- SQL_STEP(stmts[STMT_INSERT_LINK]);
- sqlite3_reset(stmts[STMT_INSERT_LINK]);
+ 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]);
+ }
for (key = ohash_first(&strings, &slot); NULL != key;
key = ohash_next(&strings, &slot)) {
assert(key->mpage == mpage);
- if (NULL == key->utf8)
- utf8key(mc, key);
+ 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->utf8);
+ 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->utf8 != key->key)
- free(key->utf8);
+ if (key->rendered != key->key)
+ free(key->rendered);
free(key);
}
-
- SQL_EXEC("END TRANSACTION");
}
static void
size_t i;
unsigned int slot;
- if (nodb)
- return;
+ if (0 == nodb)
+ SQL_EXEC("BEGIN TRANSACTION");
- mpage = ohash_first(&mpages, &slot);
- while (NULL != mpage) {
+ for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
+ mpage = ohash_next(&mpages, &slot)) {
mlink = mpage->mlinks;
- i = 1;
- SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE], i, mlink->file);
- SQL_STEP(stmts[STMT_DELETE_PAGE]);
- sqlite3_reset(stmts[STMT_DELETE_PAGE]);
if (verb)
- say(mlink->file, "Deleted from index");
- mpage = ohash_next(&mpages, &slot);
+ 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");
}
/*
return(0);
}
- /*
- * XXX The first three columns in table mpages are obsolete
- * and only kept for backward compatibility
- * until apropos(1) and friends have caught up.
- */
sql = "CREATE TABLE \"mpages\" (\n"
- " \"file\" TEXT NOT NULL,\n"
- " \"sec\" TEXT NOT NULL,\n"
- " \"arch\" TEXT NOT NULL,\n"
- " \"desc\" TEXT NOT NULL,\n"
" \"form\" INTEGER NOT NULL,\n"
" \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
");\n"
" \"arch\" TEXT NOT NULL,\n"
" \"name\" TEXT NOT NULL,\n"
" \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
- "ON DELETE CASCADE,\n"
- " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
+ "ON DELETE CASCADE\n"
");\n"
"\n"
"CREATE TABLE \"keys\" (\n"
" \"bits\" INTEGER NOT NULL,\n"
" \"key\" TEXT NOT NULL,\n"
" \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
- "ON DELETE CASCADE,\n"
- " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
- ");\n"
- "\n"
- "CREATE INDEX \"key_index\" ON keys (key);\n";
+ "ON DELETE CASCADE\n"
+ ");\n";
if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
exitcode = (int)MANDOCLEVEL_SYSERR;
prepare_statements:
SQL_EXEC("PRAGMA foreign_keys = ON");
- sql = "DELETE FROM mpages where file=?";
+ 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 "
- "(file,sec,arch,desc,form) VALUES (?,?,?,?,?)";
+ "(form) VALUES (?)";
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
sql = "INSERT INTO mlinks "
"(sec,arch,name,pageid) VALUES (?,?,?,?)";
git.ckatri.com / mandoc.git / blobdiff