commit fb804916039bfde1628b08ea7a8bc835e7216f06
parent aade19988b7f9c74ef4643b254b8752d692b4b9c
Author: Drew DeVault <sir@cmpwn.com>
Date: Sat, 10 Apr 2021 11:00:49 -0400
format::xml: reuse strio buffers
This reduces the number of allocations required to parse a document,
improving performance.
Diffstat:
3 files changed, 35 insertions(+), 40 deletions(-)
diff --git a/format/xml/+test.ha b/format/xml/+test.ha
@@ -2,6 +2,8 @@ use io;
use bufio;
use strings;
+use fmt;
+
@test fn basic() void = {
xmltest("<?xml version='1.0' encoding='utf-8' ?>
<root>
@@ -68,6 +70,8 @@ fn xmltest(input: str, expected: []token) void = {
},
at: attribute => {
let ex = expected[i] as attribute;
+ fmt::errorfln("want: '{}'='{}', got: '{}'='{}'",
+ ex.0, ex.1, at.0, at.1);
assert(at.0 == ex.0 && at.1 == ex.1);
},
tx: text => {
diff --git a/format/xml/parser.ha b/format/xml/parser.ha
@@ -40,6 +40,9 @@ export fn parse(in: *io::stream) (*parser | error) = {
let par = alloc(parser {
orig = in,
in = in,
+ namebuf = strio::dynamic(),
+ entbuf = strio::dynamic(),
+ textbuf = strio::dynamic(),
...
});
if (!bufio::isbuffered(in)) {
@@ -55,11 +58,15 @@ export fn parser_free(par: *parser) void = {
if (par.in != par.orig) {
io::close(par.in);
};
+ io::close(par.namebuf);
+ io::close(par.entbuf);
+ io::close(par.textbuf);
free(par);
};
-// Scans for and returns the next [token]. The caller must pass the returned
-// token to [token_free] when they're done with it.
+// Scans for and returns the next [token]. Tokens are borrowed from the parser
+// and are not valid on subsequent calls to [scan]; use [token_dup] to extend
+// their lifetime.
export fn scan(par: *parser) (token | void | error) = {
switch (par.state) {
state::ROOT, state::ATTRS => want(par, OPTWS)?,
@@ -117,10 +124,10 @@ export fn scan(par: *parser) (token | void | error) = {
};
fn scan_attr(par: *parser) (token | error) = {
- let name = scan_name(par)?;
+ let name = scan_name(par, par.namebuf)?;
want(par, OPTWS, '=', OPTWS);
let quot = quote(par)?;
- let val = strio::dynamic();
+ strio::reset(par.textbuf);
for (true) match (bufio::scanrune(par.in)?) {
io::EOF => return syntaxerr,
rn: rune => {
@@ -133,10 +140,10 @@ fn scan_attr(par: *parser) (token | error) = {
* => rn,
};
if (rn == quot) break;
- strio::appendrune(val, rn);
+ strio::appendrune(par.textbuf, rn);
},
};
- return (name, strio::finish(val)): attribute;
+ return (name, strio::string(par.textbuf)): attribute;
};
fn scan_comment(par: *parser) (token | void | error) = {
@@ -175,7 +182,7 @@ fn scan_comment(par: *parser) (token | void | error) = {
};
fn scan_content(par: *parser) (text | error) = {
- let content = strio::dynamic();
+ strio::reset(par.textbuf);
for (true) match (bufio::scanrune(par.in)?) {
io::EOF => break,
rn: rune => {
@@ -189,10 +196,10 @@ fn scan_content(par: *parser) (text | error) = {
scan_entity(par)?;
},
};
- strio::appendrune(content, rn);
+ strio::appendrune(par.textbuf, rn);
},
};
- return strio::finish(content);
+ return strio::string(par.textbuf);
};
fn scan_element(par: *parser) (token | error) = {
@@ -205,13 +212,8 @@ fn scan_element(par: *parser) (token | error) = {
* => bufio::unreadrune(par.in, rn),
},
};
- let name = scan_name(par)?;
- if (close) {
- free(name);
- return elementend;
- } else {
- return name: elementstart;
- };
+ let name = scan_name(par, par.namebuf)?;
+ return if (close) elementend else name: elementstart;
};
fn scan_entity(par: *parser) (rune | error) = {
@@ -239,33 +241,31 @@ fn scan_charref(par: *parser) (rune | error) = {
} else bufio::unreadrune(par.in, rn),
};
- let buf = strio::dynamic();
- defer io::close(buf);
+ strio::reset(par.entbuf);
for (true) {
let rn = match (bufio::scanrune(par.in)?) {
io::EOF => return syntaxerr,
rn: rune => rn,
};
if (ascii::isdigit(rn)) {
- strio::appendrune(buf, rn);
+ strio::appendrune(par.entbuf, rn);
} else if (rn == ';') {
break;
} else {
return syntaxerr;
};
};
- if (len(strio::string(buf)) == 0) {
+ if (len(strio::string(par.entbuf)) == 0) {
return syntaxerr;
};
- return match (strconv::stou32b(strio::string(buf), base)) {
+ return match (strconv::stou32b(strio::string(par.entbuf), base)) {
u: u32 => u: rune,
(strconv::invalid | strconv::overflow) => syntaxerr,
};
};
fn scan_namedent(par: *parser) (rune | error) = {
- let name = scan_name(par)?;
- defer free(name);
+ let name = scan_name(par, par.entbuf)?;
want(par, ';')?;
let map = [
("lt", '<'),
@@ -284,8 +284,8 @@ fn scan_namedent(par: *parser) (rune | error) = {
return syntaxerr;
};
-fn scan_name(par: *parser) (str | error) = {
- let buf = strio::dynamic();
+fn scan_name(par: *parser, buf: *io::stream) (str | error) = {
+ strio::reset(buf);
const rn = match (bufio::scanrune(par.in)?) {
io::EOF => return syntaxerr,
@@ -306,7 +306,7 @@ fn scan_name(par: *parser) (str | error) = {
},
};
- return strio::finish(buf);
+ return strio::string(buf);
};
fn scan_pi(par: *parser) (void | error) = {
@@ -341,7 +341,6 @@ fn prolog(par: *parser) (void | error) = {
};
if (encoding) {
let attr = scan_attr(par)? as attribute;
- defer token_free(attr);
if (attr.0 != "encoding") {
return syntaxerr;
};
@@ -362,7 +361,6 @@ fn prolog(par: *parser) (void | error) = {
};
if (standalone) {
let attr = scan_attr(par)? as attribute;
- defer token_free(attr);
if (attr.0 != "standalone") {
return syntaxerr;
};
diff --git a/format/xml/types.ha b/format/xml/types.ha
@@ -7,6 +7,10 @@ export type parser = struct {
in: *io::stream,
buf: [4096]u8,
state: state,
+ // strio buffers:
+ namebuf: *io::stream,
+ entbuf: *io::stream,
+ textbuf: *io::stream,
};
export type state = enum {
@@ -45,15 +49,4 @@ export fn strerror(err: error) const str = {
};
};
-// Frees resources associated with a [token].
-export fn token_free(tok: token) void = {
- match (tok) {
- el: elementstart => free(el),
- attr: attribute => {
- free(attr.0);
- free(attr.1);
- },
- tx: text => free(tx),
- elementend => void,
- };
-};
+export fn token_dup(tok: token) void = abort(); // TODO
