From: schwarze@mdocml.bsd.lv
To: source@mdocml.bsd.lv
Subject: mdocml: After careful gprof(1)ing of the new apropos(1), move the
Date: Wed, 9 Apr 2014 17:50:08 -0400 (EDT) [thread overview]
Message-ID: <201404092150.s39Lo8p1018999@krisdoz.my.domain> (raw)
Log Message:
-----------
After careful gprof(1)ing of the new apropos(1), move the descriptions
back from the keys table to the mpages table: I found a good way
to still use them in searches, without complication of the code.
On my notebook, this reduces typical apropos(1) search times by about 40%,
it reduces /usr/share/man database size by 6% in makewhatis(8) -Q mode
and by 2% in standard mode (less overhead storing pointers to mpages),
and it doesn't measurably change database build times (may even be
going down by a percent or so because less data is being copied
around in ohashes).
Modified Files:
--------------
mdocml:
mandocdb.c
mansearch.c
mansearch.h
mansearch_const.c
Revision Data
-------------
Index: mansearch.h
===================================================================
RCS file: /usr/vhosts/mdocml.bsd.lv/cvs/mdocml/mansearch.h,v
retrieving revision 1.10
retrieving revision 1.11
diff -Lmansearch.h -Lmansearch.h -u -p -r1.10 -r1.11
--- mansearch.h
+++ mansearch.h
@@ -22,45 +22,45 @@
#define TYPE_NAME 0x0000000000000001ULL
#define TYPE_Nm 0x0000000000000002ULL
-#define TYPE_Nd 0x0000000000000004ULL
-#define TYPE_arch 0x0000000000000008ULL
-#define TYPE_sec 0x0000000000000010ULL
-#define TYPE_Xr 0x0000000000000020ULL
-#define TYPE_Ar 0x0000000000000040ULL
-#define TYPE_Fa 0x0000000000000080ULL
-#define TYPE_Fl 0x0000000000000100ULL
-#define TYPE_Dv 0x0000000000000200ULL
-#define TYPE_Fn 0x0000000000000400ULL
-#define TYPE_Ic 0x0000000000000800ULL
-#define TYPE_Pa 0x0000000000001000ULL
-#define TYPE_Cm 0x0000000000002000ULL
-#define TYPE_Li 0x0000000000004000ULL
-#define TYPE_Em 0x0000000000008000ULL
-#define TYPE_Cd 0x0000000000010000ULL
-#define TYPE_Va 0x0000000000020000ULL
-#define TYPE_Ft 0x0000000000040000ULL
-#define TYPE_Tn 0x0000000000080000ULL
-#define TYPE_Er 0x0000000000100000ULL
-#define TYPE_Ev 0x0000000000200000ULL
-#define TYPE_Sy 0x0000000000400000ULL
-#define TYPE_Sh 0x0000000000800000ULL
-#define TYPE_In 0x0000000001000000ULL
-#define TYPE_Ss 0x0000000002000000ULL
-#define TYPE_Ox 0x0000000004000000ULL
-#define TYPE_An 0x0000000008000000ULL
-#define TYPE_Mt 0x0000000010000000ULL
-#define TYPE_St 0x0000000020000000ULL
-#define TYPE_Bx 0x0000000040000000ULL
-#define TYPE_At 0x0000000080000000ULL
-#define TYPE_Nx 0x0000000100000000ULL
-#define TYPE_Fx 0x0000000200000000ULL
-#define TYPE_Lk 0x0000000400000000ULL
-#define TYPE_Ms 0x0000000800000000ULL
-#define TYPE_Bsx 0x0000001000000000ULL
-#define TYPE_Dx 0x0000002000000000ULL
-#define TYPE_Rs 0x0000004000000000ULL
-#define TYPE_Vt 0x0000008000000000ULL
-#define TYPE_Lb 0x0000010000000000ULL
+#define TYPE_arch 0x0000000000000004ULL
+#define TYPE_sec 0x0000000000000008ULL
+#define TYPE_Xr 0x0000000000000010ULL
+#define TYPE_Ar 0x0000000000000020ULL
+#define TYPE_Fa 0x0000000000000040ULL
+#define TYPE_Fl 0x0000000000000080ULL
+#define TYPE_Dv 0x0000000000000100ULL
+#define TYPE_Fn 0x0000000000000200ULL
+#define TYPE_Ic 0x0000000000000400ULL
+#define TYPE_Pa 0x0000000000000800ULL
+#define TYPE_Cm 0x0000000000001000ULL
+#define TYPE_Li 0x0000000000002000ULL
+#define TYPE_Em 0x0000000000004000ULL
+#define TYPE_Cd 0x0000000000008000ULL
+#define TYPE_Va 0x0000000000010000ULL
+#define TYPE_Ft 0x0000000000020000ULL
+#define TYPE_Tn 0x0000000000040000ULL
+#define TYPE_Er 0x0000000000080000ULL
+#define TYPE_Ev 0x0000000000100000ULL
+#define TYPE_Sy 0x0000000000200000ULL
+#define TYPE_Sh 0x0000000000400000ULL
+#define TYPE_In 0x0000000000800000ULL
+#define TYPE_Ss 0x0000000001000000ULL
+#define TYPE_Ox 0x0000000002000000ULL
+#define TYPE_An 0x0000000004000000ULL
+#define TYPE_Mt 0x0000000008000000ULL
+#define TYPE_St 0x0000000010000000ULL
+#define TYPE_Bx 0x0000000020000000ULL
+#define TYPE_At 0x0000000040000000ULL
+#define TYPE_Nx 0x0000000080000000ULL
+#define TYPE_Fx 0x0000000100000000ULL
+#define TYPE_Lk 0x0000000200000000ULL
+#define TYPE_Ms 0x0000000400000000ULL
+#define TYPE_Bsx 0x0000000800000000ULL
+#define TYPE_Dx 0x0000001000000000ULL
+#define TYPE_Rs 0x0000002000000000ULL
+#define TYPE_Vt 0x0000004000000000ULL
+#define TYPE_Lb 0x0000008000000000ULL
+#define TYPE_Nd 0x0000010000000000ULL
__BEGIN_DECLS
Index: mansearch.c
===================================================================
RCS file: /usr/vhosts/mdocml.bsd.lv/cvs/mdocml/mansearch.c,v
retrieving revision 1.25
retrieving revision 1.26
diff -Lmansearch.c -Lmansearch.c -u -p -r1.25 -r1.26
--- mansearch.c
+++ mansearch.c
@@ -74,6 +74,7 @@ struct expr {
struct match {
uint64_t id; /* identifier in database */
+ char *desc; /* manual page description */
int form; /* 0 == catpage */
};
@@ -220,7 +221,8 @@ mansearch(const struct mansearch *search
SQL_BIND_BLOB(db, s, j, ep->regexp);
} else
SQL_BIND_TEXT(db, s, j, ep->substr);
- SQL_BIND_INT64(db, s, j, ep->bits);
+ if (0 == (TYPE_Nd & ep->bits))
+ SQL_BIND_INT64(db, s, j, ep->bits);
}
memset(&htab, 0, sizeof(struct ohash));
@@ -235,9 +237,9 @@ mansearch(const struct mansearch *search
* distribution of buckets in the table.
*/
while (SQLITE_ROW == (c = sqlite3_step(s))) {
- id = sqlite3_column_int64(s, 1);
+ id = sqlite3_column_int64(s, 2);
idx = ohash_lookup_memory
- (&htab, (char *)&id,
+ (&htab, (char *)&id,
sizeof(uint64_t), (uint32_t)id);
if (NULL != ohash_find(&htab, idx))
@@ -245,7 +247,10 @@ mansearch(const struct mansearch *search
mp = mandoc_calloc(1, sizeof(struct match));
mp->id = id;
- mp->form = sqlite3_column_int(s, 0);
+ mp->form = sqlite3_column_int(s, 1);
+ if (TYPE_Nd == outbit)
+ mp->desc = mandoc_strdup(
+ sqlite3_column_text(s, 0));
ohash_insert(&htab, idx, mp);
}
@@ -279,7 +284,8 @@ mansearch(const struct mansearch *search
mpage->form = mp->form;
buildnames(mpage, db, s, mp->id,
paths->paths[i], mp->form);
- mpage->output = outbit ?
+ mpage->output = TYPE_Nd & outbit ?
+ mp->desc : outbit ?
buildoutput(db, s2, mp->id, outbit) : NULL;
free(mp);
@@ -493,11 +499,16 @@ sql_statement(const struct expr *e)
sql_append(&sql, &sz, " OR ", 1);
if (e->open)
sql_append(&sql, &sz, "(", e->open);
- sql_append(&sql, &sz, NULL == e->substr ?
- "id IN (SELECT pageid FROM keys "
- "WHERE key REGEXP ? AND bits & ?)" :
- "id IN (SELECT pageid FROM keys "
- "WHERE key MATCH ? AND bits & ?)", 1);
+ sql_append(&sql, &sz,
+ TYPE_Nd & e->bits
+ ? (NULL == e->substr
+ ? "desc REGEXP ?"
+ : "desc MATCH ?")
+ : (NULL == e->substr
+ ? "id IN (SELECT pageid FROM keys "
+ "WHERE key REGEXP ? AND bits & ?)"
+ : "id IN (SELECT pageid FROM keys "
+ "WHERE key MATCH ? AND bits & ?)"), 1);
if (e->close)
sql_append(&sql, &sz, ")", e->close);
needop = 1;
@@ -554,13 +565,29 @@ exprcomp(const struct mansearch *search,
next = exprterm(search, argv[i], !igncase);
if (NULL == next)
goto fail;
- next->open = toopen;
- next->and = (1 == logic);
- if (NULL != first) {
+ if (NULL == first)
+ first = next;
+ else
cur->next = next;
- cur = next;
+
+ /*
+ * Searching for descriptions must be split out
+ * because they are stored in the mpages table,
+ * not in the keys table.
+ */
+
+ if (TYPE_Nd & next->bits && ~TYPE_Nd & next->bits) {
+ cur = mandoc_calloc(1, sizeof(struct expr));
+ memcpy(cur, next, sizeof(struct expr));
+ next->open = 1;
+ next->bits = TYPE_Nd;
+ next->next = cur;
+ cur->bits &= ~TYPE_Nd;
+ cur->close = 1;
} else
- cur = first = next;
+ cur = next;
+ next->and = (1 == logic);
+ next->open += toopen;
toopen = logic = igncase = 0;
}
if (toopen || logic || igncase || toclose)
Index: mandocdb.c
===================================================================
RCS file: /usr/vhosts/mdocml.bsd.lv/cvs/mdocml/mandocdb.c,v
retrieving revision 1.131
retrieving revision 1.132
diff -Lmandocdb.c -Lmandocdb.c -u -p -r1.131 -r1.132
--- mandocdb.c
+++ mandocdb.c
@@ -224,7 +224,7 @@ static const struct mdoc_handler mdocs[M
{ NULL, TYPE_Ic }, /* Ic */
{ NULL, TYPE_In }, /* In */
{ NULL, TYPE_Li }, /* Li */
- { parse_mdoc_Nd, TYPE_Nd }, /* Nd */
+ { parse_mdoc_Nd, 0 }, /* Nd */
{ parse_mdoc_Nm, 0 }, /* Nm */
{ NULL, 0 }, /* Op */
{ NULL, 0 }, /* Ot */
@@ -1154,8 +1154,9 @@ mpages_merge(struct mchars *mc, struct m
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);
+ if (NULL == mpage->desc)
+ mpage->desc = mandoc_strdup(
+ mpage->mlinks->name);
} else if (NULL != man)
parse_man(mpage, man_node(man));
else
@@ -1298,7 +1299,6 @@ parse_cat(struct mpage *mpage, int fd)
"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;
@@ -1339,7 +1339,6 @@ parse_cat(struct mpage *mpage, int fd)
assert(NULL == mpage->desc);
mpage->desc = mandoc_strdup(p);
- putkey(mpage, mpage->desc, TYPE_Nd);
fclose(stream);
free(title);
}
@@ -1480,7 +1479,6 @@ parse_man(struct mpage *mpage, const str
assert(NULL == mpage->desc);
mpage->desc = mandoc_strdup(start);
- putkey(mpage, mpage->desc, TYPE_Nd);
free(title);
return;
}
@@ -1928,16 +1926,7 @@ dbadd(struct mpage *mpage, struct mchars
if (NULL != mlink)
fputs(", ", stdout);
}
- for (key = ohash_first(&strings, &slot); NULL != key;
- key = ohash_next(&strings, &slot)) {
- if (TYPE_Nd & key->mask) {
- if (NULL == key->rendered)
- render_key(mc, key);
- printf(" - %s", key->rendered);
- break;
- }
- }
- putchar('\n');
+ printf(" - %s\n", mpage->desc);
return;
}
@@ -1945,6 +1934,7 @@ dbadd(struct mpage *mpage, struct mchars
say(mlink->file, "Adding to database");
i = 1;
+ SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->desc);
SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
SQL_STEP(stmts[STMT_INSERT_PAGE]);
mpage->recno = sqlite3_last_insert_rowid(db);
@@ -2157,6 +2147,7 @@ dbopen(int real)
create_tables:
sql = "CREATE TABLE \"mpages\" (\n"
+ " \"desc\" TEXT NOT NULL,\n"
" \"form\" INTEGER NOT NULL,\n"
" \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
");\n"
@@ -2189,7 +2180,7 @@ prepare_statements:
"sec=? AND arch=? AND name=?)";
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
sql = "INSERT INTO mpages "
- "(form) VALUES (?)";
+ "(desc,form) VALUES (?,?)";
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
sql = "INSERT INTO mlinks "
"(sec,arch,name,pageid) VALUES (?,?,?,?)";
Index: mansearch_const.c
===================================================================
RCS file: /usr/vhosts/mdocml.bsd.lv/cvs/mdocml/mansearch_const.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -Lmansearch_const.c -Lmansearch_const.c -u -p -r1.2 -r1.3
--- mansearch_const.c
+++ mansearch_const.c
@@ -23,10 +23,10 @@
const int mansearch_keymax = 41;
const char *const mansearch_keynames[41] = {
- "NAME", "Nm", "Nd", "arch", "sec", "Xr", "Ar", "Fa",
- "Fl", "Dv", "Fn", "Ic", "Pa", "Cm", "Li", "Em",
- "Cd", "Va", "Ft", "Tn", "Er", "Ev", "Sy", "Sh",
- "In", "Ss", "Ox", "An", "Mt", "St", "Bx", "At",
- "Nx", "Fx", "Lk", "Ms", "Bsx", "Dx", "Rs", "Vt",
- "Lb"
+ "NAME", "Nm", "arch", "sec", "Xr", "Ar", "Fa", "Fl",
+ "Dv", "Fn", "Ic", "Pa", "Cm", "Li", "Em", "Cd",
+ "Va", "Ft", "Tn", "Er", "Ev", "Sy", "Sh", "In",
+ "Ss", "Ox", "An", "Mt", "St", "Bx", "At", "Nx",
+ "Fx", "Lk", "Ms", "Bsx", "Dx", "Rs", "Vt", "Lb",
+ "Nd"
};
--
To unsubscribe send an email to source+unsubscribe@mdocml.bsd.lv
reply other threads:[~2014-04-09 21:50 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=201404092150.s39Lo8p1018999@krisdoz.my.domain \
--to=schwarze@mdocml.bsd.lv \
--cc=source@mdocml.bsd.lv \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).