Skip to content

Commit c2dd9be

Browse files
gagolewsgagolews
committed
close #128 
1 parent ad24131 commit c2dd9be

File tree

5 files changed

+38
-29
lines changed

5 files changed

+38
-29
lines changed

‎NEWS

+2-1
Original file line numberDiff line numberDiff line change
@@ -35,7 +35,8 @@
3535
(...)
3636

3737
* [NEW FEATURE] #118: `stri_wrap()` gained `indent`, `exdent`, `initial`,
38-
and `prefix` arguments.
38+
and `prefix` arguments. Moreover Knuth's dynamic word wrapping algorithm
39+
now assumes that the cost of printing the last line is zero, see #128.
3940

4041
* [NEW FEATURE] #122: `stri_subset()` gained an `omit_na` argument.
4142

‎R/wrap.R

+3-2
Original file line numberDiff line numberDiff line change
@@ -34,7 +34,7 @@
3434
#'
3535
#' @description
3636
#' This function breaks text paragraphs into lines,
37-
#' each consisting of at most \code{width} code points.
37+
#' each consisting - if it is possible - of at most \code{width} code points.
3838
#'
3939
#' @details
4040
#' Vectorized over \code{str}.
@@ -72,6 +72,7 @@
7272
#' in a more aesthetic way. This method minimizes the squared
7373
#' (by default, see \code{cost_exponent}) number of spaces (raggedness)
7474
#' at the end of each line, so the text is mode arranged evenly.
75+
#' Note that the cost of printing the last line is always zero.
7576
#'
7677
#' @param str character vector of strings to reformat
7778
#' @param width single positive integer giving the desired
@@ -115,7 +116,7 @@
115116
#' Breaking paragraphs into lines, \emph{Software: Practice and Experience} 11(11),
116117
#' 1981, pp. 1119--1184
117118
stri_wrap <- function(str, width=floor(0.9*getOption("width")),
118-
cost_exponent=2.0, simplify=TRUE, normalize=FALSE, indent=0, exdent=0,
119+
cost_exponent=2.0, simplify=TRUE, normalize=TRUE, indent=0, exdent=0,
119120
prefix="", initial=prefix, locale=NULL)
120121
{
121122
simplify <- as.logical(simplify)

‎devel/testthat/test-wrap.R

+21-20
Original file line numberDiff line numberDiff line change
@@ -4,16 +4,17 @@ context("test-wrap.R")
44

55
test_that("stri_wrap", {
66
expect_error(stri_wrap("whatever", 0))
7-
expect_error(stri_wrap("what\never"))
7+
expect_error(stri_wrap("what\never", normalize=FALSE))
88
expect_identical(stri_wrap(c("", "singleword", NA), cost=0.0), c("", "singleword", NA))
99
expect_identical(stri_wrap("a12345 b123456 c1234567", 5, 0.0), c("a12345", "b123456", "c1234567"))
1010
expect_identical(stri_wrap("a12345 b123456 c1234567", 5, 1.0), c("a12345", "b123456", "c1234567"))
1111
expect_identical(stri_wrap("a12345 b123456 c1234567", 5, 2.0), c("a12345", "b123456", "c1234567"))
1212
expect_identical(stri_wrap("a12345 b123456 c1234567", 5, 3.0), c("a12345", "b123456", "c1234567"))
1313

1414
expect_identical(stri_wrap(stri_paste(rep("\u0105\u0105\u0105\u0105\u0105", 5), collapse=" "), 12),
15-
c("\u0105\u0105\u0105\u0105\u0105", "\u0105\u0105\u0105\u0105\u0105 \u0105\u0105\u0105\u0105\u0105",
16-
"\u0105\u0105\u0105\u0105\u0105 \u0105\u0105\u0105\u0105\u0105"))
15+
c("\u0105\u0105\u0105\u0105\u0105 \u0105\u0105\u0105\u0105\u0105",
16+
"\u0105\u0105\u0105\u0105\u0105 \u0105\u0105\u0105\u0105\u0105",
17+
"\u0105\u0105\u0105\u0105\u0105"))
1718
expect_identical(stri_wrap(stri_paste(rep("\u0105\u0105\u0105\u0105\u0105", 5), collapse=" "), 12, cost=-1),
1819
c("\u0105\u0105\u0105\u0105\u0105 \u0105\u0105\u0105\u0105\u0105",
1920
"\u0105\u0105\u0105\u0105\u0105 \u0105\u0105\u0105\u0105\u0105",
@@ -79,23 +80,23 @@ test_that("stri_wrap", {
7980
"quis augue ut massa pellentesque tincidunt. In sed pretium eros.")
8081
)
8182

82-
for (s in strings) {
83-
for (i in c(12,20,30,40)) {
84-
exponents <- c(0, 1, 2, 3)
85-
res <- vector('list', length(exponents))
86-
for (j in seq_along(exponents))
87-
res[[j]] <- stri_wrap(s, i, cost_exponent=exponents[j])
88-
89-
for (j in seq_along(exponents))
90-
expect_true(all(stri_length(res[[j]]) <= i))
91-
92-
for (j in seq_along(exponents)[-1]) {
93-
cost_greedy <- sum((i-stri_length(res[[1]]))^exponents[j])
94-
cost_dynamic <- sum((i-stri_length(res[[j]]))^exponents[j])
95-
expect_true(cost_greedy >= cost_dynamic)
96-
}
97-
}
98-
}
83+
# for (s in strings) { # to do: cost of the last line is zero since stringi_0.4-1
84+
# for (i in c(12,20,30,40)) {
85+
# exponents <- c(0, 1, 2, 3)
86+
# res <- vector('list', length(exponents))
87+
# for (j in seq_along(exponents))
88+
# res[[j]] <- stri_wrap(s, i, cost_exponent=exponents[j])
89+
#
90+
# for (j in seq_along(exponents))
91+
# expect_true(all(stri_length(res[[j]]) <= i))
92+
#
93+
# for (j in seq_along(exponents)[-1]) {
94+
# cost_greedy <- sum((i-stri_length(res[[1]]))^exponents[j])
95+
# cost_dynamic <- sum((i-stri_length(res[[j]]))^exponents[j])
96+
# expect_true(cost_greedy >= cost_dynamic)
97+
# }
98+
# }
99+
# }
99100
})
100101

101102

‎man/stri_wrap.Rd

+3-2
Original file line numberDiff line numberDiff line change
@@ -4,7 +4,7 @@
44
\title{Word Wrap Text to Format Paragraphs}
55
\usage{
66
stri_wrap(str, width = floor(0.9 * getOption("width")), cost_exponent = 2,
7-
simplify = TRUE, normalize = FALSE, indent = 0, exdent = 0,
7+
simplify = TRUE, normalize = TRUE, indent = 0, exdent = 0,
88
prefix = "", initial = prefix, locale = NULL)
99
}
1010
\arguments{
@@ -42,7 +42,7 @@ Otherwise, you will get a list of \code{length(str)} character vectors.
4242
}
4343
\description{
4444
This function breaks text paragraphs into lines,
45-
each consisting of at most \code{width} code points.
45+
each consisting - if it is possible - of at most \code{width} code points.
4646
}
4747
\details{
4848
Vectorized over \code{str}.
@@ -80,6 +80,7 @@ is more complex, but it returns text wrapped
8080
in a more aesthetic way. This method minimizes the squared
8181
(by default, see \code{cost_exponent}) number of spaces (raggedness)
8282
at the end of each line, so the text is mode arranged evenly.
83+
Note that the cost of printing the last line is always zero.
8384
}
8485
\examples{
8586
s <- stri_paste(

‎src/stri_wrap.cpp

+9-4
Original file line numberDiff line numberDiff line change
@@ -96,7 +96,8 @@ void stri__wrap_greedy(std::deque<R_len_t>& wrap_after,
9696
* BreakIterator usage mods
9797
*
9898
* @version 0.4-1 (Marek Gagolewski, 2014-12-06)
99-
* new args: add_para_1, add_para_n
99+
* new args: add_para_1, add_para_n,
100+
* cost of the last line is zero
100101
*/
101102
void stri__wrap_dynamic(std::deque<R_len_t>& wrap_after,
102103
R_len_t nwords, int width_val, double exponent_val,
@@ -108,8 +109,6 @@ void stri__wrap_dynamic(std::deque<R_len_t>& wrap_after,
108109
vector<double> cost(nwords*nwords);
109110
// where cost[IDX(i,j)] == cost of printing words i..j in a single line, i<=j
110111

111-
// @TODO: we may wish not to include the cost of the last line...
112-
113112
// calculate costs:
114113
// there is some "punishment" for leaving blanks at the end of each line
115114
// (number of "blank" codepoints ^ exponent_val)
@@ -131,7 +130,13 @@ void stri__wrap_dynamic(std::deque<R_len_t>& wrap_after,
131130
if (i == 0) ct -= add_para_1;
132131
else ct -= add_para_n;
133132

134-
if (j==i)
133+
if (j == nwords-1) { // last line == cost 0
134+
if (j == i || ct >= 0)
135+
cost[IDX(i,j)] = 0.0;
136+
else
137+
cost[IDX(i,j)] = -1.0/*Inf*/;
138+
}
139+
else if (j == i)
135140
// some words don't fit in a line at all -> cost 0.0
136141
cost[IDX(i,j)] = (ct < 0) ? 0.0 : pow((double)ct, exponent_val);
137142
else

0 commit comments

Comments
 (0)