X-Git-Url: https://git.squeep.com/?a=blobdiff_plain;f=src%2Flink-helper.js;h=f35aca9419ab082a7d7f8370ebe506e6fa82278d;hb=b806715f9288323cce7b0ab437ee78b01d26c548;hp=1f3cd0992810f482717f0448781968afb0298723;hpb=28de4364128a4b03918a8cbe868009b5d427220a;p=websub-hub

diff --git a/src/link-helper.js b/src/link-helper.js
index 1f3cd09..f35aca9 100644
--- a/src/link-helper.js
+++ b/src/link-helper.js
@@ -11,9 +11,16 @@ const Enum = require('./enum');
 const FeedParser = require('feedparser');
 const { Readable } = require('stream');
 const htmlparser2 = require('htmlparser2');
+const { Iconv } = require('iconv');
 
 const _fileScope = common.fileScope(__filename);
 
+/**
+ * Helper for accessing headers
+ * @param {object} headers header object
+ * @param {string} header header name
+ * @returns {string} header content
+ */
 function getHeader(headers, header) {
   return headers[header.toLowerCase()];
 }
@@ -29,10 +36,10 @@ class LinkHelper {
 
   /**
    * Determine if this hub is listed in response data from url.
-   * @param {String} url
-   * @param {Object} headers
-   * @param {String|Buffer} body
-   * @returns {Boolean}
+   * @param {string} url content url
+   * @param {object} headers headers from accessing url
+   * @param {string | Buffer} body body from accessing url
+   * @returns {Promise<boolean>} url lists this hub
    */
   async validHub(url, headers, body) {
     const _scope = _fileScope('validHub');
@@ -41,19 +48,34 @@ class LinkHelper {
     // Add Link headers first, as they take priority over link elements in body.
     const linkHeader = getHeader(headers, Enum.Header.Link);
     const links = [];
-    try {
-      links.push(...parseLinkHeader(linkHeader));
-    } catch (e) {
-      if (e instanceof ParseSyntaxError) {
-        this.logger.debug(_scope, 'failed to parse link header, bad syntax', { error: e, linkHeader });
-      } else {
-        this.logger.error(_scope, 'failed to parse link header', { error: e, linkHeader });
+    if (linkHeader) {
+      try {
+        links.push(...parseLinkHeader(linkHeader));
+      } catch (e) {
+        /* istanbul ignore else */
+        if (e instanceof ParseSyntaxError) {
+          this.logger.debug(_scope, 'failed to parse link header, bad syntax', { error: e, linkHeader });
+        } else {
+          this.logger.error(_scope, 'failed to parse link header', { error: e, linkHeader });
+        }
+      }
+    }
+
+    const contentType = LinkHelper.parseContentType(getHeader(headers, Enum.Header.ContentType));
+    const nonUTF8Charset = !/utf-*8/i.test(contentType.params.charset) && contentType.params.charset;
+    if (nonUTF8Charset) {
+      const iconv = new Iconv(nonUTF8Charset, 'utf-8//translit//ignore');
+      try {
+        body = iconv.convert(body).toString('utf8');
+      } catch (e) {
+        /* istanbul ignore next */
+        this.logger.error(_scope, 'iconv conversion error', { error: e, contentType, url });
+        // But try to carry on, anyhow.
       }
     }
 
-    const contentType = getHeader(headers, Enum.Header.ContentType);
     let bodyLinks = [];
-    switch (contentType) {
+    switch (contentType.mediaType) {
       case Enum.ContentType.ApplicationAtom:
       case Enum.ContentType.ApplicationRDF:
       case Enum.ContentType.ApplicationRSS:
@@ -75,15 +97,44 @@ class LinkHelper {
     // Fetch all hub relation targets from headers, resolving relative URIs.
     const hubs = LinkHelper.locateHubTargets(links).map((link) => this.absoluteURI(link, url));
 
+    this.logger.debug(_scope, 'valid hubs for url', { url, hubs });
+
     return hubs.includes(this.selfUrl);
   }
 
 
+  /**
+   * @typedef {object} ContentType
+   * @property {string} mediaType media type
+   * @property {object} params map of parameters
+   */
+  /**
+   * Convert a Content-Type string to normalized components.
+   * RFC7231 §3.1.1
+   * N.B. this non-parser implementation will not work if a parameter
+   * value for some reason includes a ; or = within a quoted-string.
+   * @param {string} contentTypeHeader content type header
+   * @returns {ContentType} contentType
+   */
+  static parseContentType(contentTypeHeader) {
+    const [ mediaType, ...params ] = (contentTypeHeader || '').split(/ *; */);
+    return {
+      mediaType: mediaType.toLowerCase() || Enum.ContentType.ApplicationOctetStream,
+      params: params.reduce((obj, param) => {
+        const [field, value] = param.split('=');
+        const isQuoted = value.startsWith('"') && value.endsWith('"');
+        obj[field.toLowerCase()] = isQuoted ? value.slice(1, value.length - 1) : value;
+        return obj;
+      }, {}),
+    };
+  }
+
+
   /**
    * Parse XML-ish feed content, extracting link elements into our own format.
-   * @param {String} feedurl
-   * @param {String} body
-   * @returns {Object[]}
+   * @param {string} feedurl feed url
+   * @param {string} body feed body
+   * @returns {Promise<object[]>} array of link elements
    */
   async linksFromFeedBody(feedurl, body) {
     const _scope = _fileScope('linksFromFeedBody');
@@ -106,7 +157,11 @@ class LinkHelper {
       });
       feedParser.on('meta', (meta) => {
         this.logger.debug(_scope, 'FeedParser meta', { meta });
-        const feedLinks = meta['atom:link'] || [];
+        let feedLinks = meta['atom:link'] || [];
+        if (!Array.isArray(feedLinks)) {
+          // Parsing RSS seems to return a single entry for this rather than a list.
+          feedLinks = [feedLinks];
+        }
         feedLinks
           .map((l) => l['@'])
           .forEach((l) => {
@@ -133,7 +188,8 @@ class LinkHelper {
 
   /**
    * Parse HTML-ish content, extracting link elements into our own format.
-   * @param {String} body
+   * @param {string} body html body
+   * @returns {object[]} array of link elements
    */
   linksFromHTMLBody(body) {
     const _scope = _fileScope('linksFromHTMLBody');
@@ -161,18 +217,18 @@ class LinkHelper {
 
   /**
    * Attempt to resolve a relative target URI
-   * @param {String} uri
-   * @param {String} context
-   * @returns {String}
+   * @param {string} uri target
+   * @param {string} context base
+   * @returns {string} uri
    */
   absoluteURI(uri, context) {
     const _scope = _fileScope('absoluteURI');
     try {
       new URL(uri);
-    } catch (e) {
+    } catch (e) { // eslint-disable-line no-unused-vars
       try {
         uri = new URL(uri, context).href;
-      } catch (e) {
+      } catch (e) { // eslint-disable-line no-unused-vars
         this.logger.debug(_scope, 'could not resolve link URI', { uri, context });
       }
     }
@@ -182,8 +238,8 @@ class LinkHelper {
 
   /**
    * Return all link targets with a hub relation.
-   * @param {Object[]} links
-   * @returns {String[]}
+   * @param {object[]} links array of link objects
+   * @returns {string[]} array of hub targets
    */
   static locateHubTargets(links) {
     return links