Make more strings translatable #685

pamelafox · 2018-09-19T11:53:02Z

High-level description of change

The strings in Handlebars templates aren't translatable by webapp, so this moves the rest of them out. I did not move them out of live-editor.handlebars, as webapp does not use that file, it overrides the rendering with a React output.

I also fixed an issue with an untranslatable SQL error message.

Are there performance implications for this change?

No

Have you added tests to cover this new/updated code?

No tests added. I ran existing tests, which includes a test for SQL error messages. No tests for the UI, so I manually tested those.

Risks involved

I'm most nervous about the SQL change, as it assumes the format of the string. However, I checked the C code that it's compiled from, and the format seems consistent.

Are there any dependencies or blockers for merging this?

No

How can we verify that this change works?

Open the SQL demo, see results, remove a semi-colon, see error
Open the webpage demo, type an , click for file picker, see "Pick file", click for modal, see "OK" and close" in modal
Open the PJS demo, type playSound(getSound("rpg/metal-clink"));, click for file picker, see "Pick file", click for modal, see "OK" and close" in modal

Note that the modal doesn't work in this repo, it only works in webapp. Apparently we never included bootstrap-modal here.

pamelafox · 2018-09-19T11:54:49Z

js/output/sql/sql-output.js

@@ -65,95 +65,97 @@ window.SQLOutput = Backbone.View.extend({
        statement = statement || "";
        statement = statement.toUpperCase();

+        // Error messages take form: 'near \"%T\": syntax error'


I grepped the SQLite C source, it looks like they all take this form unless it's in fulltextsearch mode, and sqlite.js doesn't seem to enable that by default (https://github.com/kripken/sql.js/pull/199)

pamelafox · 2018-09-19T11:55:07Z

js/output/sql/sql-output.js

        }

        // Possible SELECT with missing FROM
        if (errorMessage.indexOf("no such column:") !== -1 &&
                statement.indexOf("SELECT") !== -1 &&
                statement.indexOf("FROM") === -1) {
-            errorMessage += ". " + i18n._("Are you missing a FROM clause?");
+            errorMessage += " " + i18n._("Are you missing a FROM clause?");


I moved the period into the original sentence, seems better?

pamelafox · 2018-09-19T11:55:28Z

js/output/sql/sql-output.js

-            results: results
+            results: results,
+            databaseMsg: i18n._("Database Schema"),
+            resultsMsg: i18n._("Query results"),


I changed from "Results" to "Query results" for more uniqueness in Crowdin

Oh, this is great, thanks for thinking about this. <3

pamelafox · 2018-09-19T11:56:16Z

Some screenshots:

davidflanagan

I've got one suggestion for handling the SQL error message, and have one concern about how the existing code searches the already-translated error message to add more details about the error.

But neither one is a blocker for me.

davidflanagan · 2018-09-19T16:48:47Z

js/output/sql/sql-output.js

        var isSyntaxError = errorMessage.indexOf(": syntax error") > -1;
        if (isSyntaxError) {
-            errorMessage = i18n._("There's a syntax error " +
-                errorMessage.split(":")[0]);
+            const nearPhrase = errorMessage.split(":")[0];


If you're concerned about future-proofing this, you could use a regexp to match the expected form of the error message and extract the phrase, and then if it doesn't match just pass the entire message.

davidflanagan · 2018-09-19T16:57:41Z

js/output/sql/sql-output.js

-                errorMessage.split(":")[0]);
+            const nearPhrase = errorMessage.split(":")[0];
+            errorMessage = i18n._("There's a syntax error near %(nearThing)s.",
+                {nearThing: nearPhrase.substr(5)});
        }

        // Possible SELECT with missing FROM
        if (errorMessage.indexOf("no such column:") !== -1 &&


It seems unlikely to cause a problem, but this line is now searching through a translated string for the error message. Would it be better here (and a few places below as well) to use your new nearPhrase value? (Though I don't understand how "no such column" is works with "near", so maybe that is in some other part of the error message? It almost looks, actually, like this "no such column" thing will never match.

danielhollas · 2018-09-19T23:33:16Z

Thanks so much to both!

I've just had a look into computer.pot and some of these strings are already there (I guess that is expected). However, there are also a couple of strings that seem like they should not be available for translations. They are all visible here:

https://crowdin.com/translate/khanacademy/26342/enus-cs#q=debugger.handlebars

For example:

Debug mode, coming from third_party/javascript-khansrc/live-editor/tmpl/debugger.handlebars:2

I am not sure how much this is related to this PR but just FYI.

danielhollas · 2018-09-19T23:38:39Z

Oh actually, in string Step over, Caroline Charrow wrote:

@MelliPo I asked our CS content person, and she said "That feature isn't currently enabled on the site, but it's for debugging code, "Step Over" means to step over a bit of code. It's like what's described in this Chrome debugging tutorial: https://developers.google.com/web/tools/chrome-devtools/javascript/step-code#stepping_in_action" I asked her to go through the CS section of the TP and hide any strings that aren't actually enabled on the site, so people know they don't need to be translated.)

It seems that at least "Debug Mode" could be hidden in this way as well. The other words/phrases are perhaps too generic and could potentially appear elsewhere in the future...

(Sorry for out-of-place comments :-/ )

pamelafox · 2018-09-21T06:17:03Z

@danielhollas I've gone and hidden the other strings from debugger.handlebars. I also have a separate PR that removes that from the code path.

@davidflanagan Thanks for your helpful comments, I'll address them shortly.

pamelafox · 2018-09-21T15:58:05Z

@davidflanagan I've made a few changes:

I store the original error message in sqliteError and the translated message in errorMessage, defaulted to the value of sqliteError
I start off by translating the sqlite error messages into friendly i18n versions, updating errorMessage
I then add on the additional helper messages to errorMessage

To check my changes, I went through the SQL tests and added the full error message to every failingTest, where they previously only checked the helper messages. I found a number of messages that needed to be i18n'ed that way.

Regex matching would be more future-proof, but given the many tests, it feels relatively safe to use the indexOf/split.

There are likely still some error messages that aren't translated. It isn't particularly easy to go through the C code to find all the messages/situations to cover. This should cover a huge % though.

davidflanagan

I like the new organization of the error message parsing a lot. I don't think you need another approval, but here it is since I took a look.

davidflanagan · 2018-09-21T16:18:18Z

js/output/sql/sql-output.js

                i18n._("Do you have a semi-colon after each statement?");
        } else if (isSyntaxError &&
            statement.indexOf("INSERT") !== -1 &&
-            statement.search(/[^INSERT],\d*\s*[a-zA-Z]+/) > -1 
+            statement.search(/[^INSERT],\d*\s*[a-zA-Z]+/) > -1


Not related to this patch, but I can't imagine that [^INSERT] is actually what is wanted in this regexp

Hm, it matches the desired text cases, when INSERT is at the start of the line (even if its the second line). I don't want to mess with it in case there was a reason I used the ^ .

But the square brackets change the caret from a start of line anchor to a NOT operator and the result is that you are matching any one character that is not one of I, N, S, E, R, or T followed by a comma, optional digits, optional spaces and letters. So "blah INSERT Q,a" matches but "blah INSERT R,a" does not match

davidflanagan · 2018-09-21T16:22:18Z

js/output/sql/sql-output.js

+        }
+        const uniqError = sqliteError.indexOf("UNIQUE constraint failed:") > -1;
+        if (uniqError) {
+            const colName = sqliteError.split(":")[1].trim();


One more thought about split... In most or all of these cases you could split on the string that you just matched with indexOf, and then you'd be sure you were getting the right part of the string even if sqllite adds some kind of "ERROR:" prefix to the string. That seems unlikely and you have a lot of tests, so I don't feel like you need to make this change, and am just mentioning it because the idea of splitting on a relatively long string is not something I've done before and I thought it was interesting :-)

pamelafox · 2018-09-21T17:23:58Z

@davidflanagan Changed the split to split on the same string that indexOf is checking.

pamelafox · 2018-09-21T18:11:13Z

Huh, OK, I need to spend more quality time with regexpal.com. Thanks for the observation! I'll file a bug.

…

On Fri, Sep 21, 2018 at 11:10 AM David Flanagan ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In js/output/sql/sql-output.js <#685 (comment)>: > i18n._("Do you have a semi-colon after each statement?"); } else if (isSyntaxError && statement.indexOf("INSERT") !== -1 && - statement.search(/[^INSERT],\d*\s*[a-zA-Z]+/) > -1 + statement.search(/[^INSERT],\d*\s*[a-zA-Z]+/) > -1 :-) — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <#685 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AASIUnqmaHO0IjDCroudhyssTuFjgz0kks5udSt7gaJpZM4WwACl> .

danielhollas · 2018-09-22T00:15:55Z

:D I had no idea that xkcd is a valid review comment :D

Thanks again to both for your work, much appreciated!

Can I ask how many new strings can we expect to appear in Crowdin? I am just wondering whether this deserves an announcement on the forum to notify others.

pamelafox · 2018-10-02T01:28:06Z

I posted in the forum, thanks for the note about that, Daniel. This change is now deployed, so I imagine they show up in Crowdin sometime soon.

…

On Fri, Sep 21, 2018 at 5:16 PM Daniel Hollas ***@***.***> wrote: :D I had no idea that xkcd is a valid review comment :D Thanks again to both for your work, much appreciated! Can I ask how many new strings can we expect to appear in Crowdin? I am just wondering whether this deserves an announcement on the forum to notify others. — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <#685 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AASIUkr9dXjqH-BHhekDFW6xv1oWMJ4Oks5udYFVgaJpZM4WwACl> .

danielhollas · 2018-10-30T00:19:45Z

@pamelafox I think there's one more string that is currently untranslatable in SQL output. "row" or "rows" in the database header.

I think it sits in here:

live-editor/tmpl/sql-results.handlebars

Line 95 in a4edcb1

    
           <th><a href="javascript:void(0)">{{name}}</a> <span class="row-count">{{rowCount}} row</span></th>

Let me know if I should open a new issue here or a JIRA ticket.

Also, many thanks for taking part in today's advocate sessions and for your generous offer! ❤️
(I think you had a chance to speak with Markéta from our Czech team. I was unfortunately attending only remotely).

pamelafox added 2 commits September 19, 2018 04:39

Making more strings translatable

35e6544

Pass msg in to audio

b3d0c6b

pamelafox requested a review from davidflanagan September 19, 2018 11:53

pamelafox commented Sep 19, 2018

View reviewed changes

davidflanagan approved these changes Sep 19, 2018

View reviewed changes

Translate the messages better, add in tests

9d9260b

davidflanagan approved these changes Sep 21, 2018

View reviewed changes

Splitting on the full strings

383b505

Merge branch 'master' into pamela-i18nmore

a4edcb1

pamelafox merged commit 676d0f3 into master Sep 21, 2018

pamelafox deleted the pamela-i18nmore branch September 21, 2018 18:07

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Make more strings translatable #685

Make more strings translatable #685

pamelafox commented Sep 19, 2018

pamelafox Sep 19, 2018

pamelafox Sep 19, 2018

pamelafox Sep 19, 2018

danielhollas Sep 19, 2018

pamelafox commented Sep 19, 2018

davidflanagan left a comment

davidflanagan Sep 19, 2018

davidflanagan Sep 19, 2018

danielhollas commented Sep 19, 2018 •

edited

Loading

danielhollas commented Sep 19, 2018

pamelafox commented Sep 21, 2018

pamelafox commented Sep 21, 2018

davidflanagan left a comment

davidflanagan Sep 21, 2018

pamelafox Sep 21, 2018

davidflanagan Sep 21, 2018

davidflanagan Sep 21, 2018

davidflanagan Sep 21, 2018

davidflanagan Sep 21, 2018

pamelafox commented Sep 21, 2018

pamelafox commented Sep 21, 2018 via email

danielhollas commented Sep 22, 2018

pamelafox commented Oct 2, 2018 via email

danielhollas commented Oct 30, 2018

Make more strings translatable #685

Make more strings translatable #685

Conversation

pamelafox commented Sep 19, 2018

High-level description of change

Are there performance implications for this change?

Have you added tests to cover this new/updated code?

Risks involved

Are there any dependencies or blockers for merging this?

How can we verify that this change works?

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

pamelafox commented Sep 19, 2018

davidflanagan left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

danielhollas commented Sep 19, 2018 • edited Loading

danielhollas commented Sep 19, 2018

pamelafox commented Sep 21, 2018

pamelafox commented Sep 21, 2018

davidflanagan left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

pamelafox commented Sep 21, 2018

pamelafox commented Sep 21, 2018 via email

danielhollas commented Sep 22, 2018

pamelafox commented Oct 2, 2018 via email

danielhollas commented Oct 30, 2018

danielhollas commented Sep 19, 2018 •

edited

Loading