v0.2.1 updates

Fix #131, Fix #129, Fix #127, Fix #126, Fix #119, Fix #118, Fix #114.

.travis.yml

@@ -1,4 +1,8 @@
+env:
+  - NOT_CRAN=TRUE _R_CHECK_LENGTH_1_LOGIC2_=TRUE
+
 language: r
+
 r:
   - 3.1
   - oldrel

CRAN.md

@@ -1,28 +1,16 @@
 This is a minor release primarily intended
-to address the undeclared dependencies in 
-tests raised by Professor Hornik.
-
-I have removed the tests in question completely
-from the built package.  Previously, these tests
-were run conditionally.
+to add R 3.1 support.
 
 ## R CMD check --as-cran
 
 Completes with 'Status: OK'
 
-## Test Environments
+On a tarball built with R 3.5.1
 
-I have tested this package against the following
-environments:
+## Test Environments
 
-* Travis Ubuntu 14.04.5 LTS
-    * R devel (2018-06-10 r74877)
-    * R version 3.5.0 (2017-01-27)
-    * R version 3.4.4 (2017-01-27)
-    * R version 3.2.5 (2017-01-27)
-* Winbuilder
-    * R devel (2018-06-07 r74865)
-      https://win-builder.r-project.org/T0GxmEvBfRne
-* Locally Mac OS 10.12.6
-    * R Version 3.5.0 (2017-01-27)
+I have tested this package against R-devel on
+Winbuilder and on Travis (Linux).  Additionally,
+I have tested against release and older
+versions on Linux and OSx.
 

DESCRIPTION

@@ -3,8 +3,8 @@ Type: Package
 Title: Diffs for R Objects
 Description: Generate a colorized diff of two R objects for an intuitive
     visualization of their differences.
-Version: 0.1.11
-Date: 2018-07-28
+Version: 0.2.1
+Date: 2019-01-02
 Authors@R: c(
     person(
     "Brodie", "Gaslam", email="brodie.gaslam@yahoo.com",
@@ -19,14 +19,16 @@ License: GPL (>= 2)
 LazyData: true
 URL: https://github.com/brodieG/diffobj
 BugReports: https://github.com/brodieG/diffobj/issues
-RoxygenNote: 6.0.1
+RoxygenNote: 6.1.0.9000
 VignetteBuilder: knitr
+Encoding: UTF-8
 Suggests:
     knitr,
     rmarkdown,
     testthat
 Collate:
     'capt.R'
+    'options.R'
     'pager.R'
     'check.R'
     'finalizer.R'

DEVNOTES.md

@@ -1,3 +1,45 @@
+## Crayon or Not Crayon
+
+### ANSI detection And Settings
+
+Do we want our substring functions to be ansi blind if we detect that the system
+does not support ansi?
+
+One problem right now is that when in 'raw' mode, we set `crayon.enabled=FALSE`,
+but we might still be in a system that supports ANSI and the input itself may
+have ANSI in it.
+
+Ideally we move away from crayon for ansi detection and rely solely on the
+`style` parameter.  So probably do that now.
+
+But what about the special case where we are on a system that does not support
+ANSI escapes, and we want to see the diffs in the ANSI escapes?  Do we just not
+support this?
+
+Probably better to add a separate "ansi.sgr.supported" that for now can default
+to crayon::has_color()?  Then what do we do when someone uses and ANSI style but
+this is off?  Just use it anyway?  And provide global nchar and substr functions
+that are aware of this setting?  Yes.
+
+## Fansi Problems
+
+Would be great to use fansi instead of crayon for substr as it will be much
+faster, and also support wide characters, but there are two major issues:
+
+* We still need a terminal ansi-support test, which is well developed and
+  tested in crayon.
+* The fansi tests use `unitizer`, which suggests `diffobj`; need to check if
+  this actually would cause a circular dependency or not.  So diffobj would
+  import fansi, which suggests unitizer.  Currently unitizer imports diffobj.
+  So there does seem to be build time circular dependency.  We could either
+  change all the fansi tests to not be unitizer (or move the unitizer tests to
+  `extra`, and have another version of them that tests output with .Rout files).
+  This seems like a massive PITA, and also annoying because we'll need to make
+  sure both sets of tests are run on travis, but not on CRAN, which then makes
+  it hard to make sure we're submitting the same tarball.  The more realistic
+  solution might be to remove `diffobj` as an `imports` and `suggests` so that
+  it is possible to build unitizer without diffobj.
+
 ## Compiled Testing
 
 Run the 'ses' tests with valgrind.  They should have 100% coverage of the C

NAMESPACE

@@ -1,6 +1,8 @@
 # Generated by roxygen2: do not edit by hand
 
+S3method(print,diffobj_ogewlhgiadfl)
 export(AlignThreshold)
+export(Pager)
 export(PagerBrowser)
 export(PagerOff)
 export(PagerSystem)

NEWS.md

@@ -1,5 +1,33 @@
 # diffobj
 
+## v0.2.0-1
+
+### Features
+
+* [#129](https://github.com/brodieG/diffobj/issues/129): Allow pager
+  specification via lists rather than full `Pager` objects for easier changes to
+  defaults.  As part of this we changed `StyleRaw` objects to use default
+  pager instead of `PagerOff`.
+* [#126](https://github.com/brodieG/diffobj/issues/126): Add embedding diffs in
+  Shiny to vignette.
+* [#119](https://github.com/brodieG/diffobj/issues/119): `ignore.whitespace` now
+  also ignores white space differences adjoining punctuation.
+* [#118](https://github.com/brodieG/diffobj/issues/118): New option to preserve
+  temporary diff file output when using pager (see `?Pager`).
+* [#114](https://github.com/brodieG/diffobj/issues/114): New options `strip.sgr`
+  and `sgr.supported` allow finer control of what happens when input already
+  contains ANSI CSI SGR and how ANSI CSI SGR is handled in string manipulations.
+  Related to this, `options(crayon.enabled=TRUE)` is no longer set when
+  capturing output prior to diff as it used to be.  By default pre-existing ANSI
+  CSI SGR is stripped with a warning prior to comparison.
+
+### Bugs
+
+* [#131](https://github.com/brodieG/diffobj/issues/131): Fix missing slot in S4
+  class definition (discovered by Bill Dunlap).
+* [#127](https://github.com/brodieG/diffobj/issues/127): Width CSS conflicts
+  with bootstrap (reported by @eckyu, debugged/fixed by @cpsievert).
+
 ## v0.1.11
 
 * [#123](https://github.com/brodieG/diffobj/issues/123): Compatibility with R3.1

R/capt.R

@@ -1,4 +1,4 @@
-# Copyright (C) 2018  Brodie Gaslam
+# Copyright (C) 2019 Brodie Gaslam
 #
 # This file is part of "diffobj - Diffs for R Objects"
 #
@@ -57,7 +57,7 @@ capt_print <- function(target, current, etc, err, extra){
   if(getRversion() >= "3.2.0") {
     print.match <- try(
       match.call(
-        get("print", envir=etc@frame),
+        get("print", envir=etc@frame, mode='function'),
         as.call(c(list(quote(print), x=NULL), dots)),
         envir=etc@frame
     ) )
@@ -124,22 +124,17 @@ capt_str <- function(target, current, etc, err, extra){
 
   if(getRversion() < "3.2.0") {
     # nocov start
-    str.match <- try(
-      match.call(
-        str_tpl,
-        call=as.call(c(list(quote(str), object=NULL), dots))
-    ) )
+    str.match <- match.call(
+      str_tpl,
+      call=as.call(c(list(quote(str), object=NULL), dots))
+    )
     # nocov end
   } else {
-    str.match <- try(
-      match.call(
-        str_tpl,
-        call=as.call(c(list(quote(str), object=NULL), dots)), envir=etc@frame
-    ) )
+    str.match <- match.call(
+      str_tpl,
+      call=as.call(c(list(quote(str), object=NULL), dots)), envir=etc@frame
+    )
   }
-  if(inherits(str.match, "try-error"))
-    err("Unable to compose `str` call")
-
   names(str.match)[[2L]] <- ""
 
   # Handle auto mode (side by side always for `str`)
@@ -234,10 +229,12 @@ capt_str <- function(target, current, etc, err, extra){
   if(!max.level.supplied) {
     repeat{
       if((safety <- safety + 1L) > max.depth && !first.loop)
+        # nocov start
         stop(
           "Logic Error: exceeded list depth when comparing structures; contact ",
           "maintainer."
         )
+        # nocov end
       if(!first.loop) {
         tar.str <- tar.capt[tar.lvls <= lvl]
         cur.str <- cur.capt[cur.lvls <= lvl]
@@ -381,11 +378,11 @@ capt_csv <- function(target, current, etc, err, extra){
   tar.df <- try(do.call(read.csv, c(list(target), extra), quote=TRUE))
   if(inherits(tar.df, "try-error")) err("Unable to read `target` file.")
   if(!is.data.frame(tar.df))
-    err("`target` file did not produce a data frame when read")
+    err("`target` file did not produce a data frame when read")   # nocov
   cur.df <- try(do.call(read.csv, c(list(current), extra), quote=TRUE))
   if(inherits(cur.df, "try-error")) err("Unable to read `current` file.")
   if(!is.data.frame(cur.df))
-    err("`current` file did not produce a data frame when read")
+    err("`current` file did not produce a data frame when read")  # nocov
 
   capt_print(tar.df, cur.df, etc, err, extra)
 }
@@ -394,11 +391,14 @@ capt_csv <- function(target, current, etc, err, extra){
 
 set_mode <- function(etc, tar.capt, cur.capt) {
   stopifnot(is(etc, "Settings"), is.character(tar.capt), is.character(cur.capt))
-  nc_fun <- etc@style@nchar.fun
   if(etc@mode == "auto") {
     if(
-      any(nc_fun(cur.capt) > etc@text.width.half) ||
-      any(nc_fun(tar.capt) > etc@text.width.half)
+      any(
+        nchar2(cur.capt, sgr.supported=etc@sgr.supported) > etc@text.width.half
+      ) ||
+      any(
+        nchar2(tar.capt, sgr.supported=etc@sgr.supported) > etc@text.width.half
+      )
     ) {
       etc@mode <- "unified"
   } }