pdf_text: add raw argument for page::raw_order_layout (#138)

Author: brownag

R/RcppExports.R

@@ -17,8 +17,8 @@ poppler_pdf_data <- function(x, get_font_info, opw, upw) {
     .Call('_pdftools_poppler_pdf_data', PACKAGE = 'pdftools', x, get_font_info, opw, upw)
 }
 
-poppler_pdf_text <- function(x, opw, upw) {
-    .Call('_pdftools_poppler_pdf_text', PACKAGE = 'pdftools', x, opw, upw)
+poppler_pdf_text <- function(x, opw, upw, raw = FALSE) {
+    .Call('_pdftools_poppler_pdf_text', PACKAGE = 'pdftools', x, opw, upw, raw)
 }
 
 poppler_pdf_pagesize <- function(x, opw, upw) {

R/tools.R

@@ -41,9 +41,10 @@ pdf_info <- function(pdf, opw = "", upw = "") {
 }
 
 #' @rdname pdftools
+#' @param raw if TRUE text is kept in content stream order. Default: FALSE.
 #' @export
-pdf_text <- function(pdf, opw = "", upw = "") {
-  poppler_pdf_text(loadfile(pdf), opw, upw)
+pdf_text <- function(pdf, opw = "", upw = "", raw = FALSE) {
+  poppler_pdf_text(loadfile(pdf), opw, upw, raw)
 }
 
 #' @rdname pdftools

man/pdftools.Rd

@@ -59,15 +59,16 @@ BEGIN_RCPP
 END_RCPP
 }
 // poppler_pdf_text
-CharacterVector poppler_pdf_text(RawVector x, std::string opw, std::string upw);
-RcppExport SEXP _pdftools_poppler_pdf_text(SEXP xSEXP, SEXP opwSEXP, SEXP upwSEXP) {
+CharacterVector poppler_pdf_text(RawVector x, std::string opw, std::string upw, bool raw);
+RcppExport SEXP _pdftools_poppler_pdf_text(SEXP xSEXP, SEXP opwSEXP, SEXP upwSEXP, SEXP rawSEXP) {
 BEGIN_RCPP
     Rcpp::RObject rcpp_result_gen;
     Rcpp::RNGScope rcpp_rngScope_gen;
     Rcpp::traits::input_parameter< RawVector >::type x(xSEXP);
     Rcpp::traits::input_parameter< std::string >::type opw(opwSEXP);
     Rcpp::traits::input_parameter< std::string >::type upw(upwSEXP);
-    rcpp_result_gen = Rcpp::wrap(poppler_pdf_text(x, opw, upw));
+    Rcpp::traits::input_parameter< bool >::type raw(rawSEXP);
+    rcpp_result_gen = Rcpp::wrap(poppler_pdf_text(x, opw, upw, raw));
     return rcpp_result_gen;
 END_RCPP
 }
@@ -175,7 +176,7 @@ static const R_CallMethodDef CallEntries[] = {
     {"_pdftools_get_poppler_config", (DL_FUNC) &_pdftools_get_poppler_config, 0},
     {"_pdftools_poppler_pdf_info", (DL_FUNC) &_pdftools_poppler_pdf_info, 3},
     {"_pdftools_poppler_pdf_data", (DL_FUNC) &_pdftools_poppler_pdf_data, 4},
-    {"_pdftools_poppler_pdf_text", (DL_FUNC) &_pdftools_poppler_pdf_text, 3},
+    {"_pdftools_poppler_pdf_text", (DL_FUNC) &_pdftools_poppler_pdf_text, 4},
     {"_pdftools_poppler_pdf_pagesize", (DL_FUNC) &_pdftools_poppler_pdf_pagesize, 3},
     {"_pdftools_poppler_pdf_fonts", (DL_FUNC) &_pdftools_poppler_pdf_fonts, 3},
     {"_pdftools_poppler_pdf_files", (DL_FUNC) &_pdftools_poppler_pdf_files, 3},

src/RcppExports.cpp

@@ -260,13 +260,13 @@ List poppler_pdf_data (RawVector x, bool get_font_info, std::string opw, std::st
 }
 
 // [[Rcpp::export]]
-CharacterVector poppler_pdf_text (RawVector x, std::string opw, std::string upw) {
+CharacterVector poppler_pdf_text (RawVector x, std::string opw, std::string upw, bool raw = false) {
   std::unique_ptr<poppler::document> doc(read_raw_pdf(x, opw, upw));
   CharacterVector out(doc->pages());
   for(int i = 0; i < doc->pages(); i++){
     std::unique_ptr<poppler::page> p(doc->create_page(i));
     if(!p) continue; //missing page
-    page::text_layout_enum show_text_layout = page::physical_layout;
+    page::text_layout_enum show_text_layout = raw ? page::raw_order_layout : page::physical_layout;
 
     /* media_box includes text in margins: https://github.com/ropensci/pdftools/issues/67 */
     rectf target(p->page_rect(media_box));

src/bindings.cpp

@@ -12,6 +12,7 @@ test_that("reading password protected pdf", {
 
   # Get text with password
   expect_equal(4, length(pdf_text("pdf-example-password.original.pdf", upw = "test")))
+  expect_equal(4, length(pdf_text("pdf-example-password.original.pdf", upw = "test", raw = TRUE)))
   expect_false(pdf_info("pdf-example-password.original.pdf", upw = "test")$locked)
 
   # Reading 'encrypted' file