Merge 368856c into cecb734

lib/content_negotiation_filter.js

@@ -0,0 +1,82 @@
+"use strict";
+
+const HyperSwitch = require('hyperswitch');
+const semver = require('semver');
+const HTTPError = HyperSwitch.HTTPError;
+const mwUtil = require('./mwUtil');
+
+module.exports = (hyper, req, next, options, specInfo) => {
+    if (!mwUtil.isHTMLRoute(specInfo.path)) {
+        // The filter is  only supported on HTML routes,
+        // but don't punish the client for our own mistake, just log an error.
+        hyper.logger.log('error/configuration',
+            `Content negotiation filter is not supported on ${specInfo.path}`);
+        return next(hyper, req);
+    }
+
+    if (mwUtil.isNoCacheRequest(req)) {
+        // Ignore for the no-cache requests.
+        return next(hyper, req);
+    }
+
+    let requestedVersion = mwUtil.extractHTMLProfileVersion(req.headers.accept);
+    // TODO: This can happen in many different cases, think how to properly candle it.
+    if  (!requestedVersion) {
+        // 1. Lack of accept header
+        // 2. Malformed accept header
+        // 3. False-positive is possible in case of
+        // content-type mismatch, not just version.
+        return next(hyper, req);
+    }
+    // We ignore the patch version, so if it's specified, replace it with 'x'
+    requestedVersion = requestedVersion.replace(/\d+$/, 'x');
+    return next(hyper, req)
+    .then((res) => {
+        const storedVersion =
+            mwUtil.extractHTMLProfileVersion(res.headers['content-type']);
+        if (!storedVersion) {
+            // TODO: This can happen if sections are requested;
+            // Think what to do with that.
+            // If it's not the sections - something is very wrong on our side.
+            return res;
+        }
+
+        if (semver.satisfies(storedVersion, requestedVersion)) {
+            // Nothing to do
+            return res;
+        }
+
+        if (semver.gt(requestedVersion, storedVersion)) {
+            // The request for the a greater semver version of the content!
+            // Try repeating the request with no-cache.
+            req.headers['cache-control'] = 'no-cache';
+            return next(hyper, req)
+            .then((newRes) => {
+                const newVersion =
+                    mwUtil.extractHTMLProfileVersion(newRes.headers['content-type']);
+
+                // We tried, so no accept the minor semver version different from requested.
+                // TODO: maybe log?
+                const relaxedRequestedVersion = requestedVersion.replace(/.d+\.x$/, 'x.x');
+                if (semver.satisfies(newVersion, relaxedRequestedVersion))  {
+                    return newRes;
+                }
+
+                throw new HTTPError({
+                    status: 406,
+                    body: {
+                        message: 'Failed to provide requested major version',
+                        expected: requestedVersion,
+                        received: newVersion
+                    }
+                });
+            });
+        }
+
+        // The version is less then we expected - downgrade!
+        // TODO: Call parsoid to downgrade the version.
+
+        return res;
+    });
+
+};

lib/language_variants_filter.js

@@ -46,7 +46,7 @@ module.exports = (hyper, req, next, options, specInfo) => {
                 .thenReturn(res));
             }
 
-            if (/\/page\/html\//.test(specInfo.path)) {
+            if (mwUtil.isHTMLRoute(specInfo.path)) {
                 // It's HTML, hit Parsoid for conversion
                 return next(hyper, req)
                 .then(htmlRes => hyper.post({

lib/mwUtil.js

@@ -9,6 +9,7 @@ const gunzip = P.promisify(require('zlib').gunzip);
 const Title = require('mediawiki-title').Title;
 const HyperSwitch = require('hyperswitch');
 const querystring = require('querystring');
+const cType = require('content-type');
 const HTTPError = HyperSwitch.HTTPError;
 const URI = HyperSwitch.URI;
 const mwUtil = {};
@@ -595,4 +596,33 @@ mwUtil.removeDuplicateTitles = (arr, upd) => {
     });
 };
 
+/**
+ * Checks if the provided path is an HTML route.
+ * @param {string} path the path to check.
+ * @return {boolean}
+ */
+mwUtil.isHTMLRoute = path => /\/page\/html\//.test(path);
+
+/**
+ * Extracts the version number from the content-type or accept profile.
+ * @param {string} header the content-type or accept header.
+ * @return {string|undefined}
+ */
+// TODO: in case of the accept header the parsing should be more involved
+// since the accept header can provide waited lists.
+mwUtil.extractHTMLProfileVersion = (header) => {
+    if (!header) {
+        return undefined;
+    }
+    const profile = cType.parse(header).parameters.profile;
+    if (!profile) {
+        return undefined;
+    }
+    const match = profile.match(/\/([^/]+)$/);
+    if  (!match || !Array.isArray(match) || match.length < 2) {
+        return undefined;
+    }
+    return match[1];
+};
+
 module.exports = mwUtil;

package.json

@@ -43,7 +43,8 @@
     "hyperswitch": "^0.10.2",
     "jsonwebtoken": "^8.3.0",
     "mediawiki-title": "^0.6.5",
-    "entities": "^1.1.1"
+    "entities": "^1.1.1",
+    "semver": "latest"
   },
   "devDependencies": {
     "ajv": "^5.1.5",

v1/content.yaml

@@ -126,6 +126,7 @@ paths:
       - path: lib/security_response_header_filter.js
         options:
           allowInlineStyles: true
+      - path: lib/content_negotiation_filter.js
     get:
       x-route-filters:
         - path: lib/ensure_content_type.js
@@ -409,6 +410,7 @@ paths:
       - path: lib/security_response_header_filter.js
         options:
           allowInlineStyles: true
+      - path: lib/content_negotiation_filter.js
     get:
       x-route-filters:
         - path: lib/ensure_content_type.js