Merge pull request #67 from quanteda/issue-66

Issue 66

.github/workflows/test-coverage.yaml

@@ -31,7 +31,7 @@ jobs:
           covr::codecov(
             quiet = FALSE,
             clean = FALSE,
-            install_path = file.path(Sys.getenv("RUNNER_TEMP"), "package")
+            install_path = file.path(normalizePath(Sys.getenv("RUNNER_TEMP"), winslash = "/"), "package")
           )
         shell: Rscript {0}
 

DESCRIPTION

@@ -1,5 +1,5 @@
 Package: quanteda.textstats
-Version: 0.96.4
+Version: 0.96.5
 Title: Textual Statistics for the Quantitative Analysis of Textual Data
 Description: Textual statistics functions formerly in the 'quanteda' package.
     Textual statistics for characterizing and comparing textual data. Includes 

NEWS.md

@@ -3,6 +3,7 @@
 * Fixes for C++ header compatibility for existing **quanteda** 3.x and the forthcoming 4.0 version.
 * Fixes for compatibility with Matrix >= 1.5. (#54)
 * Fixed how subsetting (`[`) works for textstat outputs, to fix #50.
+* Updated the C++ code generally and for better calling the tbb library for multi-threading.
 
 # quanteda.textstats 0.95
 

R/RcppExports.R

@@ -1,11 +1,11 @@
 # Generated by using Rcpp::compileAttributes() -> do not edit by hand
 # Generator token: 10BE3573-1514-4C36-9D1C-5A225CD40393
 
-qatd_cpp_collocations <- function(texts_, types_, words_ignore_, count_min, sizes_, method, smoothing) {
-    .Call(`_quanteda_textstats_qatd_cpp_collocations`, texts_, types_, words_ignore_, count_min, sizes_, method, smoothing)
+cpp_collocations <- function(texts_, types_, words_ignore_, count_min, sizes_, method, smoothing, thread = 1L) {
+    .Call(`_quanteda_textstats_cpp_collocations`, texts_, types_, words_ignore_, count_min, sizes_, method, smoothing, thread)
 }
 
-qatd_cpp_keyness <- function(mt, measure, correct) {
-    .Call(`_quanteda_textstats_qatd_cpp_keyness`, mt, measure, correct)
+cpp_keyness <- function(mt, measure, correct, thread = 1L) {
+    .Call(`_quanteda_textstats_cpp_keyness`, mt, measure, correct, thread)
 }
 

R/textstat_collocations.R

@@ -140,7 +140,6 @@ textstat_collocations.tokens <- function(x, method = "lambda",
     x <- as.tokens(x)
     method <- match.arg(method, c("lambda"))
 
-
     if (any(size == 1))
         stop("Collocation sizes must be larger than 1")
     if (any(size > 5))
@@ -159,9 +158,9 @@ textstat_collocations.tokens <- function(x, method = "lambda",
     types <- types(x)
     id_ignore <- unlist(pattern2id("^\\p{P}+$", types, "regex", FALSE), use.names = FALSE)
     if (is.null(id_ignore)) id_ignore <- integer()
-    result <- qatd_cpp_collocations(x, types, id_ignore, min_count, size,
-                                    if (method == "lambda1") "lambda1" else "lambda",
-                                    smoothing)
+    result <- cpp_collocations(x, types, id_ignore, min_count, size,
+                               if (method == "lambda1") "lambda1" else "lambda",
+                               smoothing, get_threads())
 
     # compute z for lambda methods
     result$z <- result$lambda / result$sigma

R/textstat_keyness.R

@@ -136,8 +136,8 @@ textstat_keyness.dfm <- function(x, target = 1L, measure = c("chi2", "exact", "l
             label <- c("target", "reference")
         }
     }
-    grouping <- factor(target, levels = c(TRUE, FALSE), labels = label)
-    temp <- dfm_group(x, groups = grouping)
+    g <- factor(target, levels = c(TRUE, FALSE), labels = label)
+    temp <- dfm_group(x, groups = g)
 
     if (measure == "exact") {
         if (measure == "exact" && !correction %in% c("default", "none"))
@@ -148,7 +148,7 @@ textstat_keyness.dfm <- function(x, target = 1L, measure = c("chi2", "exact", "l
             warning("correction is always none for pmi")
         result <- data.frame(
             feature = featnames(temp),
-            stat = qatd_cpp_keyness(temp, measure, correction),
+            stat = cpp_keyness(temp, measure, correction, get_threads()),
             p = NA,
             n_target = as.vector(temp[1, ]),
             n_reference = as.vector(temp[2, ]),

R/utils.R

@@ -21,3 +21,11 @@ check_dots <- function(..., method = NULL) {
         warning(arg, " argument is not used.", call. = FALSE)
     }
 }
+
+get_threads <- function() {
+    value <- getOption("quanteda_threads", -1L)
+    if (!is.integer(value) || length(value) != 1L)
+        stop("Invalid value of threads in quanteda options")
+    return(value)
+}
+

README.md

@@ -5,7 +5,7 @@
 
 [![CRAN
 Version](https://www.r-pkg.org/badges/version/quanteda.textstats)](https://CRAN.R-project.org/package=quanteda.textstats)
-[![](https://img.shields.io/badge/devel%20version-0.96.4-royalblue.svg)](https://github.com/quanteda/quanteda.textstats)
+[![](https://img.shields.io/badge/devel%20version-0.96.5-royalblue.svg)](https://github.com/quanteda/quanteda.textstats)
 [![Downloads](https://cranlogs.r-pkg.org/badges/quanteda.textstats)](https://CRAN.R-project.org/package=quanteda.textstats)
 [![Total
 Downloads](https://cranlogs.r-pkg.org/badges/grand-total/quanteda.textstats?color=orange)](https://CRAN.R-project.org/package=quanteda.textstats)

cran-comments.md

@@ -2,10 +2,7 @@
 
 Purpose:
 
-* To fix a C++ issue with TBB that was causing an installation failure on 
-  downstream packages on ubuntu-devel environments.
-* To remove some C++ code for similarity computations that is not needed 
-  because it is contained in the **proxy** library.
+* To update the C++ code to better call the tbb library for parallel computing.
 
 ## Test environments
 

inst/WORDLIST

@@ -125,7 +125,6 @@ Zeitschrift
 Zusammenhang
 anderen
 angewandte
-codecov
 cruft
 des
 deutsche
@@ -134,18 +133,15 @@ dfm's
 dfms
 dist
 docvars
-doi
 eines
 für
 gfortran
 hamann
 hamman
-io
 keyness
 koRpus
 macOS
 nsyllable
-org
 proxyC
 puncts
 quanteda
@@ -154,6 +150,7 @@ rix
 sents
 simil
 sy
+tbb
 textstat
 textstats
 th

src/RcppExports.cpp

@@ -11,9 +11,9 @@ Rcpp::Rostream<true>&  Rcpp::Rcout = Rcpp::Rcpp_cout_get();
 Rcpp::Rostream<false>& Rcpp::Rcerr = Rcpp::Rcpp_cerr_get();
 #endif
 
-// qatd_cpp_collocations
-DataFrame qatd_cpp_collocations(const List& texts_, const CharacterVector& types_, const IntegerVector& words_ignore_, const unsigned int count_min, const IntegerVector sizes_, const String& method, const double smoothing);
-RcppExport SEXP _quanteda_textstats_qatd_cpp_collocations(SEXP texts_SEXP, SEXP types_SEXP, SEXP words_ignore_SEXP, SEXP count_minSEXP, SEXP sizes_SEXP, SEXP methodSEXP, SEXP smoothingSEXP) {
+// cpp_collocations
+DataFrame cpp_collocations(const List& texts_, const CharacterVector& types_, const IntegerVector& words_ignore_, const unsigned int count_min, const IntegerVector sizes_, const String& method, const double smoothing, const int thread);
+RcppExport SEXP _quanteda_textstats_cpp_collocations(SEXP texts_SEXP, SEXP types_SEXP, SEXP words_ignore_SEXP, SEXP count_minSEXP, SEXP sizes_SEXP, SEXP methodSEXP, SEXP smoothingSEXP, SEXP threadSEXP) {
 BEGIN_RCPP
     Rcpp::RObject rcpp_result_gen;
     Rcpp::RNGScope rcpp_rngScope_gen;
@@ -24,27 +24,29 @@ BEGIN_RCPP
     Rcpp::traits::input_parameter< const IntegerVector >::type sizes_(sizes_SEXP);
     Rcpp::traits::input_parameter< const String& >::type method(methodSEXP);
     Rcpp::traits::input_parameter< const double >::type smoothing(smoothingSEXP);
-    rcpp_result_gen = Rcpp::wrap(qatd_cpp_collocations(texts_, types_, words_ignore_, count_min, sizes_, method, smoothing));
+    Rcpp::traits::input_parameter< const int >::type thread(threadSEXP);
+    rcpp_result_gen = Rcpp::wrap(cpp_collocations(texts_, types_, words_ignore_, count_min, sizes_, method, smoothing, thread));
     return rcpp_result_gen;
 END_RCPP
 }
-// qatd_cpp_keyness
-Rcpp::NumericVector qatd_cpp_keyness(arma::sp_mat& mt, const std::string measure, const std::string correct);
-RcppExport SEXP _quanteda_textstats_qatd_cpp_keyness(SEXP mtSEXP, SEXP measureSEXP, SEXP correctSEXP) {
+// cpp_keyness
+Rcpp::NumericVector cpp_keyness(arma::sp_mat& mt, const std::string measure, const std::string correct, const int thread);
+RcppExport SEXP _quanteda_textstats_cpp_keyness(SEXP mtSEXP, SEXP measureSEXP, SEXP correctSEXP, SEXP threadSEXP) {
 BEGIN_RCPP
     Rcpp::RObject rcpp_result_gen;
     Rcpp::RNGScope rcpp_rngScope_gen;
     Rcpp::traits::input_parameter< arma::sp_mat& >::type mt(mtSEXP);
     Rcpp::traits::input_parameter< const std::string >::type measure(measureSEXP);
     Rcpp::traits::input_parameter< const std::string >::type correct(correctSEXP);
-    rcpp_result_gen = Rcpp::wrap(qatd_cpp_keyness(mt, measure, correct));
+    Rcpp::traits::input_parameter< const int >::type thread(threadSEXP);
+    rcpp_result_gen = Rcpp::wrap(cpp_keyness(mt, measure, correct, thread));
     return rcpp_result_gen;
 END_RCPP
 }
 
 static const R_CallMethodDef CallEntries[] = {
-    {"_quanteda_textstats_qatd_cpp_collocations", (DL_FUNC) &_quanteda_textstats_qatd_cpp_collocations, 7},
-    {"_quanteda_textstats_qatd_cpp_keyness", (DL_FUNC) &_quanteda_textstats_qatd_cpp_keyness, 3},
+    {"_quanteda_textstats_cpp_collocations", (DL_FUNC) &_quanteda_textstats_cpp_collocations, 8},
+    {"_quanteda_textstats_cpp_keyness", (DL_FUNC) &_quanteda_textstats_cpp_keyness, 4},
     {NULL, NULL, 0}
 };