Files
zulip/web/src/copy_and_paste.js
Anders Kaseorg d6be353299 compose_state: Make get_or_set accept a full selector.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
2023-11-02 14:06:04 -07:00

541 lines
21 KiB
JavaScript

import isUrl from "is-url";
import $ from "jquery";
import TurndownService from "turndown";
import * as compose_ui from "./compose_ui";
import * as message_lists from "./message_lists";
import {page_params} from "./page_params";
import * as rows from "./rows";
function find_boundary_tr($initial_tr, iterate_row) {
let j;
let skip_same_td_check = false;
let $tr = $initial_tr;
// If the selection boundary is somewhere that does not have a
// parent tr, we should let the browser handle the copy-paste
// entirely on its own
if ($tr.length === 0) {
return undefined;
}
// If the selection boundary is on a table row that does not have an
// associated message id (because the user clicked between messages),
// then scan downwards until we hit a table row with a message id.
// To ensure we can't enter an infinite loop, bail out (and let the
// browser handle the copy-paste on its own) if we don't hit what we
// are looking for within 10 rows.
for (j = 0; !$tr.is(".message_row") && j < 10; j += 1) {
$tr = iterate_row($tr);
}
if (j === 10) {
return undefined;
} else if (j !== 0) {
// If we updated tr, then we are not dealing with a selection
// that is entirely within one td, and we can skip the same td
// check (In fact, we need to because it won't work correctly
// in this case)
skip_same_td_check = true;
}
return [rows.id($tr), skip_same_td_check];
}
function construct_recipient_header($message_row) {
const message_header_content = rows
.get_message_recipient_header($message_row)
.text()
.replaceAll(/\s+/g, " ")
.replace(/^\s/, "")
.replace(/\s$/, "");
return $("<p>").append($("<strong>").text(message_header_content));
}
/*
The techniques we use in this code date back to
2013 and may be obsolete today (and may not have
been even the best workaround back then).
https://github.com/zulip/zulip/commit/fc0b7c00f16316a554349f0ad58c6517ebdd7ac4
The idea is that we build a temp div, let jQuery process the
selection, then restore the selection on a zero-second timer back
to the original selection.
Do not be afraid to change this code if you understand
how modern browsers deal with copy/paste. Just test
your changes carefully.
*/
function construct_copy_div($div, start_id, end_id) {
const copy_rows = rows.visible_range(start_id, end_id);
const $start_row = copy_rows[0];
const $start_recipient_row = rows.get_message_recipient_row($start_row);
const start_recipient_row_id = rows.id_for_recipient_row($start_recipient_row);
let should_include_start_recipient_header = false;
let last_recipient_row_id = start_recipient_row_id;
for (const $row of copy_rows) {
const recipient_row_id = rows.id_for_recipient_row(rows.get_message_recipient_row($row));
// if we found a message from another recipient,
// it means that we have messages from several recipients,
// so we have to add new recipient's bar to final copied message
// and wouldn't forget to add start_recipient's bar at the beginning of final message
if (recipient_row_id !== last_recipient_row_id) {
$div.append(construct_recipient_header($row));
last_recipient_row_id = recipient_row_id;
should_include_start_recipient_header = true;
}
const message = message_lists.current.get(rows.id($row));
const $content = $(message.content);
$content.first().prepend(message.sender_full_name + ": ");
$div.append($content);
}
if (should_include_start_recipient_header) {
$div.prepend(construct_recipient_header($start_row));
}
}
function select_div($div, selection) {
$div.css({
position: "absolute",
left: "-99999px",
// Color and background is made according to "light theme"
// exclusively here because when copying the content
// into, say, Gmail compose box, the styles come along.
// This is done to avoid copying the content with dark
// background when using the app in dark theme.
// We can avoid other custom styles since they are wrapped
// inside another parent such as `.message_content`.
color: "#333",
background: "#FFF",
}).attr("id", "copytempdiv");
$("body").append($div);
selection.selectAllChildren($div[0]);
}
function remove_div(_div, ranges, selection) {
window.setTimeout(() => {
selection = window.getSelection();
selection.removeAllRanges();
for (const range of ranges) {
selection.addRange(range);
}
$("#copytempdiv").remove();
}, 0);
}
export function copy_handler() {
// This is the main handler for copying message content via
// `Ctrl+C` in Zulip (note that this is totally independent of the
// "select region" copy behavior on Linux; that is handled
// entirely by the browser, our HTML layout, and our use of the
// no-select CSS classes). We put considerable effort
// into producing a nice result that pastes well into other tools.
// Our user-facing specification is the following:
//
// * If the selection is contained within a single message, we
// want to just copy the portion that was selected, which we
// implement by letting the browser handle the Ctrl+C event.
//
// * Otherwise, we want to copy the bodies of all messages that
// were partially covered by the selection.
const selection = window.getSelection();
const analysis = analyze_selection(selection);
const ranges = analysis.ranges;
const start_id = analysis.start_id;
const end_id = analysis.end_id;
const skip_same_td_check = analysis.skip_same_td_check;
const $div = $("<div>");
if (start_id === undefined || end_id === undefined) {
// In this case either the starting message or the ending
// message is not defined, so this is definitely not a
// multi-message selection and we can let the browser handle
// the copy.
document.execCommand("copy");
return;
}
if (!skip_same_td_check && start_id === end_id) {
// Check whether the selection both starts and ends in the
// same message. If so, Let the browser handle this.
document.execCommand("copy");
return;
}
// We've now decided to handle the copy event ourselves.
//
// We construct a temporary div for what we want the copy to pick up.
// We construct the div only once, rather than for each range as we can
// determine the starting and ending point with more confidence for the
// whole selection. When constructing for each `Range`, there is a high
// chance for overlaps between same message ids, avoiding which is much
// more difficult since we can get a range (start_id and end_id) for
// each selection `Range`.
construct_copy_div($div, start_id, end_id);
// Select div so that the browser will copy it
// instead of copying the original selection
select_div($div, selection);
document.execCommand("copy");
remove_div($div, ranges, selection);
}
export function analyze_selection(selection) {
// Here we analyze our selection to determine if part of a message
// or multiple messages are selected.
//
// Firefox and Chrome handle selection of multiple messages
// differently. Firefox typically creates multiple ranges for the
// selection, whereas Chrome typically creates just one.
//
// Our goal in the below loop is to compute and be prepared to
// analyze the combined range of the selections, and copy their
// full content.
let i;
let range;
const ranges = [];
let $startc;
let $endc;
let $initial_end_tr;
let start_id;
let end_id;
let start_data;
let end_data;
// skip_same_td_check is true whenever we know for a fact that the
// selection covers multiple messages (and thus we should no
// longer consider letting the browser handle the copy event).
let skip_same_td_check = false;
for (i = 0; i < selection.rangeCount; i += 1) {
range = selection.getRangeAt(i);
ranges.push(range);
$startc = $(range.startContainer);
start_data = find_boundary_tr(
$startc.parents(".selectable_row, .message_header").first(),
($row) => $row.next(),
);
if (start_data === undefined) {
// Skip any selection sections that don't intersect a message.
continue;
}
if (start_id === undefined) {
// start_id is the Zulip message ID of the first message
// touched by the selection.
start_id = start_data[0];
}
$endc = $(range.endContainer);
$initial_end_tr = get_end_tr_from_endc($endc);
end_data = find_boundary_tr($initial_end_tr, ($row) => $row.prev());
if (end_data === undefined) {
// Skip any selection sections that don't intersect a message.
continue;
}
if (end_data[0] !== undefined) {
end_id = end_data[0];
}
if (start_data[1] || end_data[1]) {
// If the find_boundary_tr call for either the first or
// the last message covered by the selection
skip_same_td_check = true;
}
}
return {
ranges,
start_id,
end_id,
skip_same_td_check,
};
}
function get_end_tr_from_endc($endc) {
if ($endc.attr("id") === "bottom_whitespace" || $endc.attr("id") === "compose_close") {
// If the selection ends in the bottom whitespace, we should
// act as though the selection ends on the final message.
// This handles the issue that Chrome seems to like selecting
// the compose_close button when you go off the end of the
// last message
return $(".message_row").last();
}
// Sometimes (especially when three click selecting in Chrome) the selection
// can end in a hidden element in e.g. the next message, a date divider.
// We can tell this is the case because the selection isn't inside a
// `messagebox-content` div, which is where the message text itself is.
// TODO: Ideally make it so that the selection cannot end there.
// For now, we find the message row directly above wherever the
// selection ended.
if ($endc.closest(".messagebox-content").length === 0) {
// If the selection ends within the message following the selected
// messages, go back to use the actual last message.
if ($endc.parents(".message_row").length > 0) {
const $parent_msg = $endc.parents(".message_row").first();
return $parent_msg.prev(".message_row");
}
// If it's not in a .message_row, it's probably in a .message_header and
// we can use the last message from the previous recipient_row.
if ($endc.parents(".message_header").length > 0) {
const $overflow_recipient_row = $endc.parents(".recipient_row").first();
return $overflow_recipient_row.prev(".recipient_row").last(".message_row");
}
// If somehow we get here, do the default return.
}
return $endc.parents(".selectable_row").first();
}
function deduplicate_newlines(attribute) {
// We replace any occurrences of one or more consecutive newlines followed by
// zero or more whitespace characters with a single newline character.
return attribute ? attribute.replaceAll(/(\n+\s*)+/g, "\n") : "";
}
function image_to_zulip_markdown(_content, node) {
if (node.nodeName === "IMG" && node.classList.contains("emoji") && node.hasAttribute("alt")) {
// For Zulip's custom emoji
return node.getAttribute("alt");
}
const src = node.getAttribute("src") || node.getAttribute("href") || "";
const title = deduplicate_newlines(node.getAttribute("title")) || "";
// Using Zulip's link like syntax for images
return src ? "[" + title + "](" + src + ")" : node.getAttribute("alt") || "";
}
export function paste_handler_converter(paste_html) {
// turning off escaping (for now) to remove extra `/`
TurndownService.prototype.escape = (string) => string;
const turndownService = new TurndownService({
emDelimiter: "*",
codeBlockStyle: "fenced",
headingStyle: "atx",
});
turndownService.addRule("strikethrough", {
filter: ["del", "s", "strike"],
replacement(content) {
return "~~" + content + "~~";
},
});
turndownService.addRule("links", {
filter: ["a"],
replacement(content, node) {
if (node.href === content) {
// Checks for raw links without custom text.
return content;
}
if (node.childNodes.length === 1 && node.firstChild.nodeName === "IMG") {
// ignore link's url if it only has an image
return content;
}
return "[" + content + "](" + node.href + ")";
},
});
turndownService.addRule("listItem", {
// We override the original upstream implementation of this rule
// to have a custom indent of 2 spaces for list items, instead of
// the default 4 spaces. Everything else is the same as upstream.
filter: "li",
replacement(content, node) {
content = content
.replace(/^\n+/, "") // remove leading newlines
.replace(/\n+$/, "\n") // replace trailing newlines with just a single one
.replaceAll(/\n/gm, "\n "); // custom 2 space indent
let prefix = "* ";
const parent = node.parentNode;
if (parent.nodeName === "OL") {
const start = parent.getAttribute("start");
const index = Array.prototype.indexOf.call(parent.children, node);
prefix = (start ? Number(start) + index : index + 1) + ". ";
}
return prefix + content + (node.nextSibling && !/\n$/.test(content) ? "\n" : "");
},
});
turndownService.addRule("zulipCodeBlock", {
// We create a new rule to exclusively handle code blocks in Zulip messages since
// the `fencedCodeBlock` rule in upstream won't work for them. The reason is that
// `fencedCodeBlock` only works for `pre` elements that have `code` elements as
// their 1st child, while Zulip code blocks have an empty span as the 1st child
// of the `pre` element, and then the `code` element. This new rule is a variation
// of upstream's `fencedCodeBlock` rule.
// We modify the filter of upstream's `fencedCodeBlock` rule to only apply to
// Zulip code blocks with the Zulip specific class of `zulip-code-block`.
filter(node, options) {
return (
options.codeBlockStyle === "fenced" &&
node.nodeName === "CODE" &&
node.parentElement?.nodeName === "PRE" &&
node.parentElement.parentElement?.classList.contains("zulip-code-block")
);
},
// We modify the replacement of upstream's `fencedCodeBlock` rule only slightly
// to extract and add the language of the code block (if any) to the fence.
replacement(content, node, options) {
const language = node.closest(".codehilite")?.dataset?.codeLanguage || "";
const fenceChar = options.fence.charAt(0);
let fenceSize = 3;
const fenceInCodeRegex = new RegExp("^" + fenceChar + "{3,}", "gm");
let match;
while ((match = fenceInCodeRegex.exec(content))) {
if (match[0].length >= fenceSize) {
fenceSize = match[0].length + 1;
}
}
const fence = fenceChar.repeat(fenceSize);
return (
"\n\n" +
fence +
language +
"\n" +
content.replace(/\n$/, "") +
"\n" +
fence +
"\n\n"
);
},
});
turndownService.addRule("zulipImagePreview", {
filter(node) {
// select image previews in Zulip messages
return (
node.classList.contains("message_inline_image") && node.firstChild.nodeName === "A"
);
},
replacement(content, node) {
// We parse the copied html to then check if the generating link (which, if
// present, always comes before the preview in the copied html) is also there.
// If the preview has an aria-label, it means it does have a named link in the
// message, and if the 1st element with the same image link in the copied html
// does not have the `message_inline_image` class, it means it is the generating
// link, and not the preview, meaning the generating link is copied as well.
const copied_html = new DOMParser().parseFromString(paste_html, "text/html");
if (
node.firstChild.hasAttribute("aria-label") &&
!copied_html
.querySelector("a[href='" + node.firstChild.getAttribute("href") + "']")
?.parentNode?.classList.contains("message_inline_image")
) {
// We skip previews which have their generating link copied too, to avoid
// double pasting the same link.
return "";
}
return image_to_zulip_markdown(content, node.firstChild);
},
});
turndownService.addRule("images", {
filter: "img",
replacement: image_to_zulip_markdown,
});
turndownService.addRule("math", {
// We don't have a way to get the original LaTeX code from the rendered
// `math` so we drop it to avoid pasting gibberish.
// In the future, we could have a data-original-latex feature in Zulip HTML
// if we wanted to paste the original LaTeX for Zulip messages.
filter: "math",
replacement() {
return "";
},
});
let markdown_text = turndownService.turndown(paste_html);
// Checks for escaped ordered list syntax.
markdown_text = markdown_text.replaceAll(/^(\W* {0,3})(\d+)\\\. /gm, "$1$2. ");
// Removes newlines before the start of a list and between list elements.
markdown_text = markdown_text.replaceAll(/\n+([*+-])/g, "\n$1");
return markdown_text;
}
function is_safe_url_paste_target($textarea) {
const range = $textarea.range();
if (!range.text) {
// No range is selected
return false;
}
if (isUrl(range.text.trim())) {
// Don't engage our URL paste logic over existing URLs
return false;
}
if (range.start <= 2) {
// The range opens too close to the start of the textarea
// to have to worry about Markdown link syntax
return true;
}
// Look at the two characters before the start of the original
// range in search of the tell-tale `](` from existing Markdown
// link syntax
const possible_markdown_link_markers = $textarea[0].value.slice(range.start - 2, range.start);
if (possible_markdown_link_markers === "](") {
return false;
}
return true;
}
export function paste_handler(event) {
const clipboardData = event.originalEvent.clipboardData;
if (!clipboardData) {
// On IE11, ClipboardData isn't defined. One can instead
// access it with `window.clipboardData`, but even that
// doesn't support text/html, so this code path couldn't do
// anything special anyway. So we instead just let the
// default paste handler run on IE11.
return;
}
if (clipboardData.getData) {
const $textarea = $(event.currentTarget);
const paste_text = clipboardData.getData("text");
const paste_html = clipboardData.getData("text/html");
// Trim the paste_text to accommodate sloppy copying
const trimmed_paste_text = paste_text.trim();
// Only intervene to generate formatted links when dealing
// with a URL and a URL-safe range selection.
if (isUrl(trimmed_paste_text) && is_safe_url_paste_target($textarea)) {
event.preventDefault();
event.stopPropagation();
const url = trimmed_paste_text;
compose_ui.format_text($textarea, "linked", url);
return;
}
if (paste_html && page_params.development_environment) {
event.preventDefault();
event.stopPropagation();
const text = paste_handler_converter(paste_html);
compose_ui.insert_syntax_and_focus(text);
}
}
}
export function initialize() {
$("textarea#compose-textarea").on("paste", paste_handler);
$("body").on("paste", ".message_edit_content", paste_handler);
}