# Latest changes from mainline fast\_xs **URL:** \ **Category:** Ruby\ **Created:** [August 22, 2009, 10:25pm UTC](https://www.ruby-forum.com/t/latest-changes-from-mainline-fast-xs/174225 "2009-08-22T22:25:55Z")\ **Posts on this page:** 1\ **Page:** 1

**Author:** ![ericw](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/e/97f17d/32.png) [@ericw](https://www.ruby-forum.com/u/ericw)\ **Post date:** [August 22, 2009, 10:25pm UTC](https://www.ruby-forum.com/t/latest-changes-from-mainline-fast-xs/174225/1 "2009-08-22T22:25:55Z")

Hi all, Hopefully the new maintainer of Hpricot will see this because I have no idea who to send it to now… I originally had a version of this patch posted in \_why’s Trac sometime in 2008 until he took Trac down. I’ve tried emailing him directly on several occasions since and never got a response, last was February 2009. Come to think of it, the only time I ever got any response from him was my initial email to him about fast\_xs back in October 2007. So here it is, I’ve rebased my patch against 2c961095954d5aaa5c046f4c773c62c3d5902ef4 on git://github.com/whymirror/hpricot.git Also I have an entire repo up on git://git.bogomips.org/hpricot and viewable from [http://git.bogomips.org/cgit/hpricot.git](http://git.bogomips.org/cgit/hpricot.git) From 636e3f453b2bbc0c7486b91eda452fe4767e4bbc Mon Sep 17 00:00:00 2001 From: Eric W. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) Date: Sat, 7 Feb 2009 21:31:15 -0800 Subject: [PATCH] \* ext/fast\_xs: latest changes from mainline fast\_xs - alloca() dependency removed. An extra memory allocation and memory copy is saved by writing directly to the string object returned by rb\_str\_new(0, len) (both the Ruby 1.8.6 and 1.9 code has this usage in it, so it should be safe 🙂 This also allows fast\_xs to work on strings larger than the stack size. - general readability and cleanups - fast\_xs works with Ruby 1.9 - Don’t rely on ruby\_digitmap being a valid symbol * * * ext/fast\_xs/fast\_xs.c | 101 +++++++++++++++++++++++±------------------------ 1 files changed, 50 insertions(+), 51 deletions(-) ## diff --git a/ext/fast\_xs/fast\_xs.c b/ext/fast\_xs/fast\_xs.c index 4a30a6c…04b175f 100644 — a/ext/fast\_xs/fast\_xs.c +++ b/ext/fast\_xs/fast\_xs.c @@ -1,8 +1,5 @@ -#define VERSION “0.1” #include \ #include \ -/\* #include \ \*/ #ifndef RARRAY\_LEN #define RARRAY\_LEN(arr) RARRAY(arr)-\>len @@ -72,11 +69,6 @@ static const int cp\_1252[] = { n = cp\_1252[n - 128]; } while(0) -#define return\_const\_len(x) do { \ - memcpy(buf, x, sizeof(x) - 1); \ - return (sizeof(x) - 1); -} while (0) - static inline size\_t bytes\_for(int n) { if (n \< 1000) @@ -91,18 +83,24 @@ static inline size\_t bytes\_for(int n) return sizeof(“�”) - 1; } -static long escape(char \*buf, int n) +static size\_t escape(char \*buf, int n) { + +#define return\_const\_len(x) do { \ - memcpy(buf, x, sizeof(x) - 1); \ - return (sizeof(x) - 1); +} while (0) - /\* handle ASCII first \*/ if (likely(n \< 128)) { - if (likely(n \>= 0x20 || n == 0x9 || n == 0xA || n == 0xD)) { - ``` if (unlikely(n == 34)) ``` - if (likely(n \>= 0x20 || n == ‘\t’ || n == ‘\n’ || n == ‘\r’)) { - ``` if (unlikely(n == '"')) return_const_len("""); ``` - ``` if (unlikely(n == 38)) ``` - ``` if (unlikely(n == '&')) return_const_len("&"); ``` - ``` if (unlikely(n == 60)) ``` - ``` if (unlikely(n == '<')) return_const_len("<"); ``` - ``` if (unlikely(n == 62)) ``` - ``` if (unlikely(n == '>')) return_const_len(">"); buf[0] = (char)n; return 1; ``` @@ -112,16 +110,18 @@ static long escape(char \*buf, int n) return 1; } +#undef return\_const\_len + CP\_1252\_ESCAPE(n); if (VALID\_VALUE(n)) { /\* return snprintf(buf, sizeof(“�”), “&#%i;”, n); \*/ - RUBY\_EXTERN const char ruby\_digitmap[]; - int rv = 3; /\* &#; \*/ - static const char digitmap[] = “0123456789”; - size\_t rv = sizeof(“&#;”) - 1; buf += bytes\_for(n); \*–buf = ‘;’; do { - ``` *--buf = ruby_digitmap[(int)(n % 10)]; ``` - ``` *--buf = digitmap[(int)(n % 10)]; ++rv; ``` } while (n /= 10); \*–buf = ‘#’; @@ -132,27 +132,6 @@ static long escape(char \*buf, int n) return 1; } ## -#undef return\_const\_len -static long escaped\_len(int n) -{ - if (likely(n \< 128)) { - if (unlikely(n == 34)) - ``` return (sizeof(""") - 1); ``` - if (unlikely(n == 38)) - ``` return (sizeof("&") - 1); ``` - if (unlikely(n == 60 || n == 62)) - ``` return (sizeof(">") - 1); ``` - return 1; - } - - CP\_1252\_ESCAPE(n); - - if (VALID\_VALUE(n)) - return bytes\_for(n); - return 1; -} - static VALUE unpack\_utf8(VALUE self) { return rb\_funcall(self, unpack\_id, 1, U\_fmt); @@ -163,28 +142,48 @@ static VALUE unpack\_uchar(VALUE self) return rb\_funcall(self, unpack\_id, 1, C\_fmt); } -VALUE fast\_xs(VALUE self) +/\* - - escapes strings for XML - - The double-quote (") character is translated to “"” - \*/ +static VALUE fast\_xs(VALUE self) { long i; - struct RArray \*array; - char \*s, \*c; - long s\_len = 0; - VALUE array; - char \*c; - size\_t s\_len; VALUE \*tmp; - VALUE rv; - - array = rb\_rescue(unpack\_utf8, self, unpack\_uchar, self); - - for (tmp = RARRAY\_PTR(array), s\_len = i = RARRAY\_LEN(array); - ``` --i >= 0; ``` - ``` tmp++) { ``` - int n = NUM2INT(\*tmp); - if (likely(n \< 128)) { - ``` if (unlikely(n == '"')) ``` - ``` s_len += (sizeof(""") - 2); ``` - ``` if (unlikely(n == '&')) ``` - ``` s_len += (sizeof("&") - 2); ``` - ``` if (unlikely(n == '>' || n == '<')) ``` - ``` s_len += (sizeof(">") - 2); ``` - ``` continue; ``` - } - array = RARRAY(rb\_rescue(unpack\_utf8, self, unpack\_uchar, self)); - CP\_1252\_ESCAPE(n); - tmp = RARRAY\_PTR(array); - for (i = RARRAY\_LEN(array); --i \>= 0; tmp++) - s\_len += escaped\_len(NUM2INT(\*tmp)); - if (VALID\_VALUE(n)) - ``` s_len += bytes_for(n) - 1; ``` - } - c = s = alloca(s\_len + 1); - rv = rb\_str\_new(NULL, s\_len); - c = RSTRING\_PTR(rv); - tmp = RARRAY\_PTR(array); - for (i = RARRAY\_LEN(array); --i \>= 0; tmp++) - for (tmp = RARRAY\_PTR(array), i = RARRAY\_LEN(array); --i \>= 0; tmp++) c += escape(c, NUM2INT(\*tmp)); - \*c = ‘\0’; - return rb\_str\_new(s, s\_len); - return rv; } void Init\_fast\_xs(void)