Bug 1403802: Port nsHttpHandler::PrepareAcceptLanguages over to Rust;…

… r=valentin,froydnj

netwerk/base/rust-helper/Cargo.toml

@@ -0,0 +1,8 @@
+[package]
+name = "netwerk_helper"
+version = "0.0.1"
+authors = ["Jeff Hemphill <jthemphill@mozilla.com>"]
+
+[dependencies]
+nserror = { path = "../../../xpcom/rust/nserror" }
+nsstring = { path = "../../../xpcom/rust/nsstring" }

netwerk/base/rust-helper/src/helper.h

@@ -0,0 +1,14 @@
+#ifndef RUST_NS_NET_HELPER
+#define RUST_NS_NET_HELPER
+
+#include "nsError.h"
+#include "nsString.h"
+
+extern "C" {
+
+nsresult
+rust_prepare_accept_languages(const nsACString* i_accept_languages,
+                              nsACString* o_accept_languages);
+}
+
+#endif // RUST_NS_NET_HELPER

netwerk/base/rust-helper/src/lib.rs

@@ -0,0 +1,143 @@
+use std::ascii::AsciiExt;
+
+extern crate nserror;
+use self::nserror::*;
+
+extern crate nsstring;
+use self::nsstring::nsACString;
+
+/// HTTP leading whitespace, defined in netwerk/protocol/http/nsHttp.h
+static HTTP_LWS: &'static [u8] = &[' ' as u8, '\t' as u8];
+
+/// Trim leading whitespace, trailing whitespace, and quality-value
+/// from a token.
+fn trim_token(token: &[u8]) -> &[u8] {
+    // Trim left whitespace
+    let ltrim = token.iter()
+        .take_while(|c| HTTP_LWS.iter().any(|ws| &ws == c))
+        .count();
+
+    // Trim right whitespace
+    // remove "; q=..." if present
+    let rtrim = token[ltrim..]
+        .iter()
+        .take_while(|c| **c != (';' as u8) && HTTP_LWS.iter().all(|ws| ws != *c))
+        .count();
+
+    &token[ltrim..ltrim + rtrim]
+}
+
+#[no_mangle]
+#[allow(non_snake_case)]
+/// Allocates an nsACString that contains a ISO 639 language list
+/// notated with HTTP "q" values for output with an HTTP Accept-Language
+/// header. Previous q values will be stripped because the order of
+/// the langs implies the q value. The q values are calculated by dividing
+/// 1.0 amongst the number of languages present.
+///
+/// Ex: passing: "en, ja"
+///     returns: "en,ja;q=0.5"
+///
+///     passing: "en, ja, fr_CA"
+///     returns: "en,ja;q=0.7,fr_CA;q=0.3"
+pub extern "C" fn rust_prepare_accept_languages<'a, 'b>(i_accept_languages: &'a nsACString,
+                                                        o_accept_languages: &'b mut nsACString)
+                                                        -> nsresult {
+    if i_accept_languages.is_empty() {
+        return NS_OK;
+    }
+
+    let make_tokens = || {
+        i_accept_languages.split(|c| *c == (',' as u8))
+            .map(|token| trim_token(token))
+            .filter(|token| token.len() != 0)
+    };
+
+    let n = make_tokens().count();
+
+    for (count_n, i_token) in make_tokens().enumerate() {
+
+        // delimiter if not first item
+        if count_n != 0 {
+            o_accept_languages.append(",");
+        }
+
+        let token_pos = o_accept_languages.len();
+        o_accept_languages.append(&i_token as &[u8]);
+
+        {
+            let o_token = o_accept_languages.to_mut();
+            canonicalize_language_tag(&mut o_token[token_pos..]);
+        }
+
+        // Divide the quality-values evenly among the languages.
+        let q = 1.0 - count_n as f32 / n as f32;
+
+        let u: u32 = ((q + 0.005) * 100.0) as u32;
+        // Only display q-value if less than 1.00.
+        if u < 100 {
+            // With a small number of languages, one decimal place is
+            // enough to prevent duplicate q-values.
+            // Also, trailing zeroes do not add any information, so
+            // they can be removed.
+            if n < 10 || u % 10 == 0 {
+                let u = (u + 5) / 10;
+                o_accept_languages.append(&format!(";q=0.{}", u));
+            } else {
+                // Values below 10 require zero padding.
+                o_accept_languages.append(&format!(";q=0.{:02}", u));
+            }
+        }
+    }
+
+    NS_OK
+}
+
+/// Defines a consistent capitalization for a given language string.
+///
+/// # Arguments
+/// * `token` - a narrow char slice describing a language.
+///
+/// Valid language tags are of the form
+/// "*", "fr", "en-US", "es-419", "az-Arab", "x-pig-latin", "man-Nkoo-GN"
+///
+/// Language tags are defined in the
+/// [rfc5646](https://tools.ietf.org/html/rfc5646) spec. According to
+/// the spec:
+///
+/// > At all times, language tags and their subtags, including private
+/// > use and extensions, are to be treated as case insensitive: there
+/// > exist conventions for the capitalization of some of the subtags,
+/// > but these MUST NOT be taken to carry meaning.
+///
+/// So why is this code even here? See bug 1108183, I guess.
+fn canonicalize_language_tag(token: &mut [u8]) {
+    for c in token.iter_mut() {
+        *c = AsciiExt::to_ascii_lowercase(c);
+    }
+
+    let sub_tags = token.split_mut(|c| *c == ('-' as u8));
+    for (i, mut sub_tag) in sub_tags.enumerate() {
+        if i == 0 {
+            // ISO 639-1 language code, like the "en" in "en-US"
+            continue;
+        }
+
+        match sub_tag.len() {
+            // Singleton tag, like "x" or "i". These signify a
+            // non-standard language, so we stop capitalizing after
+            // these.
+            1 => break,
+            // ISO 3166-1 Country code, like "US"
+            2 => {
+                sub_tag[0] = AsciiExt::to_ascii_uppercase(&sub_tag[0]);
+                sub_tag[1] = AsciiExt::to_ascii_uppercase(&sub_tag[1]);
+            },
+            // ISO 15924 script code, like "Nkoo"
+            4  => {
+                sub_tag[0] = AsciiExt::to_ascii_uppercase(&sub_tag[0]);
+            },
+            _ => {},
+        };
+    }
+}

netwerk/protocol/http/nsHttpHandler.cpp

@@ -9,6 +9,7 @@
 
 #include "prsystem.h"
 
+#include "nsError.h"
 #include "nsHttp.h"
 #include "nsHttpHandler.h"
 #include "nsHttpChannel.h"
@@ -58,6 +59,7 @@
 #include "nsIXULRuntime.h"
 #include "nsCharSeparatedTokenizer.h"
 #include "nsRFPService.h"
+#include "rust-helper/src/helper.h"
 
 #include "mozilla/net/NeckoChild.h"
 #include "mozilla/net/NeckoParent.h"
@@ -1908,51 +1910,6 @@ nsHttpHandler::PrefsChanged(nsIPrefBranch *prefs, const char *pref)
 #undef MULTI_PREF_CHANGED
 }
 
-
-/**
- * Currently, only regularizes the case of subtags.
- */
-static void
-CanonicalizeLanguageTag(char *languageTag)
-{
-    char *s = languageTag;
-    while (*s != '\0') {
-        *s = nsCRT::ToLower(*s);
-        s++;
-    }
-
-    s = languageTag;
-    bool isFirst = true;
-    bool seenSingleton = false;
-    while (*s != '\0') {
-        char *subTagEnd = strchr(s, '-');
-        if (subTagEnd == nullptr) {
-            subTagEnd = strchr(s, '\0');
-        }
-
-        if (isFirst) {
-            isFirst = false;
-        } else if (seenSingleton) {
-            // Do nothing
-        } else {
-            size_t subTagLength = subTagEnd - s;
-            if (subTagLength == 1) {
-                seenSingleton = true;
-            } else if (subTagLength == 2) {
-                *s = nsCRT::ToUpper(*s);
-                *(s + 1) = nsCRT::ToUpper(*(s + 1));
-            } else if (subTagLength == 4) {
-                *s = nsCRT::ToUpper(*s);
-            }
-        }
-
-        s = subTagEnd;
-        if (*s != '\0') {
-            s++;
-        }
-    }
-}
-
 /**
  *  Allocates a C string into that contains a ISO 639 language list
  *  notated with HTTP "q" values for output with a HTTP Accept-Language
@@ -1972,78 +1929,9 @@ PrepareAcceptLanguages(const char *i_AcceptLanguages, nsACString &o_AcceptLangua
     if (!i_AcceptLanguages)
         return NS_OK;
 
-    uint32_t n, count_n, size, wrote;
-    double q, dec;
-    char *p, *p2, *token, *q_Accept, *o_Accept;
-    const char *comma;
-    int32_t available;
-
-    o_Accept = strdup(i_AcceptLanguages);
-    if (!o_Accept)
-        return NS_ERROR_OUT_OF_MEMORY;
-    for (p = o_Accept, n = size = 0; '\0' != *p; p++) {
-        if (*p == ',') n++;
-            size++;
-    }
-
-    available = size + ++n * 11 + 1;
-    q_Accept = new char[available];
-    if (!q_Accept) {
-        free(o_Accept);
-        return NS_ERROR_OUT_OF_MEMORY;
-    }
-    *q_Accept = '\0';
-    q = 1.0;
-    dec = q / (double) n;
-    count_n = 0;
-    p2 = q_Accept;
-    for (token = nsCRT::strtok(o_Accept, ",", &p);
-         token != nullptr;
-         token = nsCRT::strtok(p, ",", &p))
-    {
-        token = net_FindCharNotInSet(token, HTTP_LWS);
-        char* trim;
-        trim = net_FindCharInSet(token, ";" HTTP_LWS);
-        if (trim != nullptr)  // remove "; q=..." if present
-            *trim = '\0';
-
-        if (*token != '\0') {
-            CanonicalizeLanguageTag(token);
-
-            comma = count_n++ != 0 ? "," : ""; // delimiter if not first item
-            uint32_t u = QVAL_TO_UINT(q);
-
-            // Only display q-value if less than 1.00.
-            if (u < 100) {
-                const char *qval_str;
-
-                // With a small number of languages, one decimal place is enough to prevent duplicate q-values.
-                // Also, trailing zeroes do not add any information, so they can be removed.
-                if ((n < 10) || ((u % 10) == 0)) {
-                    u = (u + 5) / 10;
-                    qval_str = "%s%s;q=0.%u";
-                } else {
-                    // Values below 10 require zero padding.
-                    qval_str = "%s%s;q=0.%02u";
-                }
-
-                wrote = snprintf(p2, available, qval_str, comma, token, u);
-            } else {
-                wrote = snprintf(p2, available, "%s%s", comma, token);
-            }
-
-            q -= dec;
-            p2 += wrote;
-            available -= wrote;
-            MOZ_ASSERT(available > 0, "allocated string not long enough");
-        }
-    }
-    free(o_Accept);
-
-    o_AcceptLanguages.Assign((const char *) q_Accept);
-    delete [] q_Accept;
-
-    return NS_OK;
+    const nsAutoCString ns_accept_languages(i_AcceptLanguages);
+    return rust_prepare_accept_languages(&ns_accept_languages,
+                                         &o_AcceptLanguages);
 }
 
 nsresult

netwerk/test/unit/test_header_Accept-Language_case.js

@@ -22,6 +22,8 @@ function run_test() {
     ["de,en-us,en",     "de,en-US;q=0.7,en;q=0.3"],
     ["en-US, en",       "en-US,en;q=0.5"],
     ["EN-US;q=0.2, EN", "en-US,en;q=0.5"],
+    ["en ;q=0.8, de  ", "en,de;q=0.5"],
+    [",en,",            "en"],
   ];
 
   for (let i = 0; i < testData.length; i++) {

toolkit/library/rust/shared/Cargo.toml

@@ -10,6 +10,7 @@ geckoservo = { path = "../../../../servo/ports/geckolib", optional = true }
 mp4parse_capi = { path = "../../../../media/libstagefright/binding/mp4parse_capi" }
 nsstring = { path = "../../../../xpcom/rust/nsstring" }
 nserror = { path = "../../../../xpcom/rust/nserror" }
+netwerk_helper = { path = "../../../../netwerk/base/rust-helper" }
 rust_url_capi = { path = "../../../../netwerk/base/rust-url-capi" }
 webrender_bindings = { path = "../../../../gfx/webrender_bindings", optional = true }
 cubeb-pulse = { path = "../../../../media/libcubeb/cubeb-pulse-rs", optional = true, features=["pulse-dlopen"] }

toolkit/library/rust/shared/lib.rs

@@ -9,6 +9,7 @@ extern crate mp4parse_capi;
 extern crate nsstring;
 extern crate nserror;
 extern crate rust_url_capi;
+extern crate netwerk_helper;
 #[cfg(feature = "quantum_render")]
 extern crate webrender_bindings;
 #[cfg(feature = "cubeb_pulse_rust")]