mobile sharing: Extract shared/js/typeahead.js.

This extracts get_emoji_matcher and all the
functions it depended on, most of which were
in composebox_typeahead.js.

We also move remove_diacritics out of the people
module.

This is the first major step for #13728.

static/shared/js/typeahead.js

@@ -0,0 +1,80 @@
+const unicode_marks = /\p{M}/gu;
+
+exports.remove_diacritics = function (s) {
+    return s.normalize("NFKD").replace(unicode_marks, "");
+};
+
+function query_matches_string(query, source_str, split_char) {
+    source_str = exports.remove_diacritics(source_str);
+
+    // If query doesn't contain a separator, we just want an exact
+    // match where query is a substring of one of the target characters.
+    if (query.indexOf(split_char) > 0) {
+        // If there's a whitespace character in the query, then we
+        // require a perfect prefix match (e.g. for 'ab cd ef',
+        // query needs to be e.g. 'ab c', not 'cd ef' or 'b cd
+        // ef', etc.).
+        const queries = query.split(split_char);
+        const sources = source_str.split(split_char);
+        let i;
+
+        for (i = 0; i < queries.length - 1; i += 1) {
+            if (sources[i] !== queries[i]) {
+                return false;
+            }
+        }
+
+        // This block is effectively a final iteration of the last
+        // loop.  What differs is that for the last word, a
+        // partial match at the beginning of the word is OK.
+        if (sources[i] === undefined) {
+            return false;
+        }
+        return sources[i].indexOf(queries[i]) === 0;
+    }
+
+    // For a single token, the match can be anywhere in the string.
+    return source_str.indexOf(query) !== -1;
+}
+
+// This function attempts to match a query with source's attributes.
+// * query is the user-entered search query
+// * Source is the object we're matching from, e.g. a user object
+// * match_attrs are the values associated with the target object that
+// the entered string might be trying to match, e.g. for a user
+// account, there might be 2 attrs: their full name and their email.
+// * split_char is the separator for this syntax (e.g. ' ').
+exports.query_matches_source_attrs = (query, source, match_attrs, split_char) => {
+    return _.any(match_attrs, function (attr) {
+        const source_str = source[attr].toLowerCase();
+        return query_matches_string(query, source_str, split_char);
+    });
+};
+
+function clean_query(query) {
+    query = exports.remove_diacritics(query);
+    // When `abc ` with a space at the end is typed in a
+    // contenteditable widget such as the composebox PM section, the
+    // space at the end was a `no break-space (U+00A0)` instead of
+    // `space (U+0020)`, which lead to no matches in those cases.
+    query = query.replace(/\u00A0/g, String.fromCharCode(32));
+
+    return query;
+}
+
+exports.clean_query_lowercase = function (query) {
+    query = query.toLowerCase();
+    query = clean_query(query);
+    return query;
+};
+
+exports.get_emoji_matcher = (query) => {
+    // replaces spaces with underscores for emoji matching
+    query = query.split(" ").join("_");
+    query = exports.clean_query_lowercase(query);
+
+    return function (emoji) {
+        return exports.query_matches_source_attrs(query, emoji, ["emoji_name"], "_");
+    };
+};
+