chrome_print URL regeneration

[r2evans]

Migrated from #97.

I'm sorry for hijacked a completed PR @cderv, I was wondering if the PR was either prematurely merged or something else. I notice a difference in the app's URL (and # handling) when an app is loaded from the dashboard (still with the right nav-bar with info/access/runtime/.../logs) and loaded full-browser ("normal" users). I wonder if there's a way for the app to know this in real-time so that the URL can be generated correctly. Perhaps session$request$SCRIPT_NAME?

Below is the original comment, thanks.

---

As an aside, @RLesur, the sample app is ignoring leading path information (such as on a reverse-proxied RStudio Connect (RSC)). Something like:

--- app.R.orig  2019-07-03 16:10:34.538559600 -0700
+++ app.R       2019-07-03 15:52:33.365021700 -0700
@@ -14,10 +14,11 @@
   ## browser()
   # get the url
   url <- sprintf(
-    "%s//%s:%s",
+    "%s//%s:%s%s",
     session$clientData$url_protocol,
     session$clientData$url_hostname,
-    session$clientData$url_port
+    session$clientData$url_port,
+    session$clientData$url_pathname
   )
   # Get the inputs
   ## ipt <- reactiveValuesToList(input)

is a start. Unfortunately, in an RSC app it produces the first, when it should be producing the second:

# wrong
https://rsc.server.com/subdir/content/11/?_inputs_&bins=30 
# better
https://rsc.server.com/subdir/#/content/11/?_inputs_&bins=30 

I don't have a fully-functioning RSC-version yet. (I've been hijacking this PR and #91 to learn more about pagedown in shiny. Thanks!)

[RLesur]

The only goal of #97 was to enhance chrome_print() to create an async argument which can be used in any context . async = TRUE is required if the user wants to use chrome_print() in a Shiny app.

For instance, one can use this argument in the console

pagedown::chrome_print('https://pagedown.rbind.io/', async = TRUE)
#> <Promise [pending]>

When async = FALSE, the R session holds until chrome_print() terminates.

The chrome_print() function does not handle URLs. If an URL is provided through its input argument, this URL is taken as-is.
IMO, this point should be addressed in a dedicated vignette as @cderv suggests in #114

[r2evans]

I see what you mean, the code above is specific to the example you posted in the previous issue and pull. I understand, and 114 is the right place to discuss it. I understand the purpose of async= and interpreted the shiny implications the same way.

I believe my use-case is just a little different from your previous example, so it might not be as immediate for me: in your case, you want a PDF version of the current app, widgets/controls and all; in my case, I believe that for each app.R, I will likely have a report.Rmd (parameterized, of course). Whether I use rmarkdown::render or rely on RSC to handle serving it is up for trial-and-debate.

Thanks for the discussion, I don't know that I need to keep this issue open, feel free to close if you think it best.