Initial import

.gitignore

@@ -0,0 +1,2 @@
+accept-language
+accept-language.vcl

Makefile

@@ -0,0 +1,19 @@
+# Language preferences
+DEFAULT_LANGUAGE=en
+SUPPORTED_LANGUAGES=en es it ja ru zh-cn
+
+CC=cc
+DEBUG=
+#DEBUG=-g3
+
+all: accept-language accept-language.vcl
+
+accept-language: accept-language.c
+	$(CC) -Wall -pedantic $(DEBUG) -o accept-language accept-language.c
+
+accept-language.vcl: Makefile accept-language.c gen_vcl.pl
+	./gen_vcl.pl $(DEFAULT_LANGUAGE) $(SUPPORTED_LANGUAGES) > accept-language.vcl
+
+test:
+	prove -I./t -v ./t
+

README

@@ -0,0 +1,67 @@
+Varnish Accept-Language VCL
+===========================
+
+Last updated: 21/01/2010
+	Cosimo Streppone <cosimo@opera.com>
+	Opera Software ASA
+
+Here you will find a VCL config file for Varnish (http://varnish-cache.org)
+This VCL allows you to normalize and filter all the incoming requests
+Accept-Language headers and reduce them to just the languages your site supports.
+
+*** WARNING ***
+This VCL consists of C code. Your Varnish might explode. YMMV.
+Don't use it in production if you don't know what you're doing.
+We are using it in production, but we don't know what we're doing.
+
+Why would you want this?
+------------------------
+Your site supports English and Japanese languages.
+Your client browsers will send every possible Accept-Language header on Earth.
+If you enable "Vary: Accept-Language" on Varnish or on your backends,
+the cache hit ratio will rapidly drop, because of the huge variations
+in Accept-Language contents.
+
+With this VCL, the Accept-Language header will be "rewritten" to just
+"en" or "ja", depending on your client settings. If no match occurs,
+you can select a default language.
+
+The rewritten header is "X-Varnish-Accept-Language".
+You can choose to put this header back in "Accept-Language" if you wish.
+In this way, the normalization will be completely transparent.
+
+Requirements
+------------
+
+- gcc, make
+- a recent perl, with `prove'
+- File::Slurp CPAN module
+
+Instructions
+-------------
+
+1) Configure the list of languages your site supports
+   and the default fallback in the Makefile, *NOT* in the C code.
+
+2) Run 'make && make test'
+   You should see "All tests successful" at the end of the execution
+
+3) Install the generated accept-language.vcl in /etc/varnish/
+
+4) Add the following include directives at the top of your main VCL file
+   (default.vcl or the name you're using).
+
+   #include <string.h>
+   #include <stdio.h>
+   #include <stdlib.h>
+
+5) Inside your vcl_recv(), add the following line:
+
+   include "/etc/varnish/accept-language.vcl"
+
+6) Restart Varnish
+
+7) Cross your fingers
+
+8) Profit !!
+

accept-language.c

@@ -0,0 +1,126 @@
+/*
+ * Accept-language header normalization
+ *
+ * Cosimo, 21/01/2010
+ *
+ */
+
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h> /* qsort */
+
+#define DEFAULT_LANGUAGE "en"
+#define SUPPORTED_LANGUAGES ":bg:cs:da:en:fi:fy:hu:it:ja:no:pl:ru:tr:uk:xx-lol:vn:zh-cn:"
+
+#define vcl_string char
+#define LANG_LIST_SIZE 10
+#define LANG_MAXLEN 10
+#define RETURN_LANG(x) { \
+    strncpy(lang, x, LANG_MAXLEN); \
+    return; \
+}
+#define RETURN_DEFAULT_LANG RETURN_LANG(DEFAULT_LANGUAGE)
+#define PUSH_LANG(x,y) { \
+    pl[curr_lang].lang = x;    \
+    pl[curr_lang].q = y;       \
+    curr_lang++;               \
+}
+
+struct lang_list {
+    vcl_string *lang;
+    float q;
+};
+
+/* Checks if a given language is in the static list of the ones we support */
+int is_supported(vcl_string *lang) {
+    vcl_string *supported_languages = SUPPORTED_LANGUAGES;
+    vcl_string match_str[LANG_MAXLEN + 3] = "";  /* :, :, \0 = 3 */
+    int is_supported = 0;
+
+    /* Search ":<lang>:" in supported languages string */
+    strncpy(match_str, ":", 1);
+    strncat(match_str, lang, LANG_MAXLEN);
+    strncat(match_str, ":\0", 2);
+
+    if (strstr(supported_languages, match_str)) {
+        is_supported = 1;
+    }
+
+    return is_supported;
+}
+
+/* Used by qsort() below */
+int sort_by_q(const void *x, const void *y) {
+    struct lang_list *a = (struct lang_list *)x;
+    struct lang_list *b = (struct lang_list *)y;
+    if (a->q > b->q) return -1;
+    if (a->q < b->q) return 1;
+    return 0;
+}
+
+/* Reads Accept-Language, parses it, and finds the first match
+   among the supported languages. In case of no match,
+   returns the default language.
+*/
+void select_language(const vcl_string *incoming_header, char *lang) {
+
+    struct lang_list pl[LANG_LIST_SIZE];
+    vcl_string *lang_tok = NULL;
+    vcl_string *header;
+    vcl_string *pos = NULL;
+    vcl_string *q_spec = NULL;
+    unsigned int curr_lang = 0, i = 0;
+    float q;
+
+    /* Empty string, return default language immediately */
+    if (! incoming_header || (0 == strcmp(incoming_header, ""))) {
+        RETURN_DEFAULT_LANG;
+    }
+
+    /* Tokenize Accept-Language */
+    header = (vcl_string *) incoming_header;
+
+    while ((lang_tok = strtok_r(header, " ,", &pos))) {
+        q = 1.0;
+        if ((q_spec = strstr(lang_tok, ";q="))) {
+            /* Truncate language name before ';' */
+            *q_spec = '\0';
+            /* Get q value */
+            sscanf(q_spec + 3, "%f", &q);
+        }
+        /* Wildcard language '*' should be last in list */
+        if ((*lang_tok) == '*') q = 0.0;
+        /* Push in the prioritized list */
+        PUSH_LANG(lang_tok, q);
+        /* For strtok_r() to proceed from where it left off */
+        header = NULL;
+        /* Break out if stored max no. of languages */
+        if (curr_lang >= LANG_MAXLEN) break;
+    }
+
+    /* Sort by priority */
+    qsort(pl, curr_lang, sizeof(struct lang_list), &sort_by_q);
+
+    /* Match with supported languages */
+    for (i = 0; i < curr_lang; i++) {
+        if (! is_supported(pl[i].lang)) continue;
+        RETURN_LANG(pl[i].lang);
+    }
+
+    RETURN_DEFAULT_LANG;
+}
+
+int main(int argc, char **argv) {
+    vcl_string lang[LANG_MAXLEN] = "";
+    if (argc != 2 || ! argv[1]) {
+        strncpy(lang, "en", 2);
+    }
+    else {
+        select_language(argv[1], lang);
+    }
+    printf("%s\n", lang);
+    return 0;
+}
+
+/* vim: syn=c ts=4 et sts=4 sw=4 tw=0
+*/

examples/accept-language.vcl

@@ -0,0 +1,134 @@
+C{
+
+/* ------------------------------------------------------------------ */
+/* THIS FILE IS AUTOMATICALLY GENERATED BY ./gen_vcl.pl. DO NOT EDIT. */
+
+/*
+ * Accept-language header normalization
+ *
+ * Cosimo, 21/01/2010
+ *
+ */
+
+
+#define DEFAULT_LANGUAGE "en"
+#define SUPPORTED_LANGUAGES ":en:ja:"
+
+#define vcl_string char
+#define LANG_LIST_SIZE 10
+#define LANG_MAXLEN 10
+#define RETURN_LANG(x) { \
+    strncpy(lang, x, LANG_MAXLEN); \
+    return; \
+}
+#define RETURN_DEFAULT_LANG RETURN_LANG(DEFAULT_LANGUAGE)
+#define PUSH_LANG(x,y) { \
+    pl[curr_lang].lang = x;    \
+    pl[curr_lang].q = y;       \
+    curr_lang++;               \
+}
+
+struct lang_list {
+    vcl_string *lang;
+    float q;
+};
+
+/* Checks if a given language is in the static list of the ones we support */
+int is_supported(vcl_string *lang) {
+    vcl_string *supported_languages = SUPPORTED_LANGUAGES;
+    vcl_string match_str[LANG_MAXLEN + 3] = "";  /* :, :, \0 = 3 */
+    int is_supported = 0;
+
+    /* Search ":<lang>:" in supported languages string */
+    strncpy(match_str, ":", 1);
+    strncat(match_str, lang, LANG_MAXLEN);
+    strncat(match_str, ":\0", 2);
+
+    if (strstr(supported_languages, match_str)) {
+        is_supported = 1;
+    }
+
+    return is_supported;
+}
+
+/* Used by qsort() below */
+int sort_by_q(const void *x, const void *y) {
+    struct lang_list *a = (struct lang_list *)x;
+    struct lang_list *b = (struct lang_list *)y;
+    if (a->q > b->q) return -1;
+    if (a->q < b->q) return 1;
+    return 0;
+}
+
+/* Reads Accept-Language, parses it, and finds the first match
+   among the supported languages. In case of no match,
+   returns the default language.
+*/
+void select_language(const vcl_string *incoming_header, char *lang) {
+
+    struct lang_list pl[LANG_LIST_SIZE];
+    vcl_string *lang_tok = NULL;
+    vcl_string *header;
+    vcl_string *pos = NULL;
+    vcl_string *q_spec = NULL;
+    unsigned int curr_lang = 0, i = 0;
+    float q;
+
+    /* Empty string, return default language immediately */
+    if (! incoming_header || (0 == strcmp(incoming_header, ""))) {
+        RETURN_DEFAULT_LANG;
+    }
+
+    /* Tokenize Accept-Language */
+    header = (vcl_string *) incoming_header;
+
+    while ((lang_tok = strtok_r(header, " ,", &pos))) {
+        q = 1.0;
+        if ((q_spec = strstr(lang_tok, ";q="))) {
+            /* Truncate language name before ';' */
+            *q_spec = '\0';
+            /* Get q value */
+            sscanf(q_spec + 3, "%f", &q);
+        }
+        /* Wildcard language '*' should be last in list */
+        if ((*lang_tok) == '*') q = 0.0;
+        /* Push in the prioritized list */
+        PUSH_LANG(lang_tok, q);
+        /* For strtok_r() to proceed from where it left off */
+        header = NULL;
+        /* Break out if stored max no. of languages */
+        if (curr_lang >= LANG_MAXLEN) break;
+    }
+
+    /* Sort by priority */
+    qsort(pl, curr_lang, sizeof(struct lang_list), &sort_by_q);
+
+    /* Match with supported languages */
+    for (i = 0; i < curr_lang; i++) {
+        if (! is_supported(pl[i].lang)) continue;
+        RETURN_LANG(pl[i].lang);
+    }
+
+    RETURN_DEFAULT_LANG;
+}
+
+
+
+/* Get Accept-Language header from client */
+incoming_header = VRT_GetHdr(sp, HDR_REQ, "\020Accept-Language:");
+/* syslog(LOG_INFO, "accept-language.vcl: incoming header \"%s\"", incoming_header); */
+
+/* Normalize and filter out by list of supported languages */
+select_language(incoming_header, lang);
+
+/* Set another header back, so it can be consumed */
+VRT_SetHdr(sp, HDR_REQ, "\032X-Varnish-Accept-Language:", lang);
+/* syslog(LOG_INFO, "accept-language.vcl: selected lang   \"%s\"", lang); */
+
+
+/* vim: syn=c ts=4 et sts=4 sw=4 tw=0
+*/
+
+/* THIS FILE IS AUTOMATICALLY GENERATED BY ./gen_vcl.pl. DO NOT EDIT. */
+/* ------------------------------------------------------------------ */
+}C

examples/sample.vcl

@@ -0,0 +1,39 @@
+# Sample VCL file
+
+# For the accept-language.vcl "plugin" to work, we
+# need these includes here. I couldn't manage to get them
+# working in accept-language.vcl
+
+C{
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h>
+}C
+
+# Everything proceeds as normal
+sub vcl_recv {
+
+	# ...
+
+    include "/etc/varnish/accept-language.vcl";
+
+	# ...
+	# lookup;
+
+    pass;
+}
+
+sub vcl_fetch {
+
+	# ...
+
+	# Store different versions of the resource by the
+	# content of the Accept-Language header
+	set obj.http.Vary = "Accept-Language";
+
+	# ...
+	# deliver;
+
+    pass;
+}
+