Initial update to pdfbox2.0

tpaskhalis · Mar 30, 2018 · d848d40 · d848d40
1 parent 4e3eae1
commit d848d40
Show file tree

Hide file tree

Showing 6 changed files with 27 additions and 19 deletions.
diff --git a/R/extract_tables.R b/R/extract_tables.R
@@ -70,7 +70,7 @@ function(file,
         pages <- as.integer(pages)
         pageIterator <- oe$extract(make_pages(pages))
     }
-    npages <- pdfDocument$getDocumentCatalog()$getAllPages()$size()
+    npages <- pdfDocument$getNumberOfPages()
     area <- make_area(area = area, pages = pages, npages = npages)
     columns <- make_columns(columns = columns, pages = pages, npages = npages)
 

diff --git a/R/extract_text.R b/R/extract_text.R
@@ -25,7 +25,7 @@ extract_text <- function(file, pages = NULL, password = NULL, encoding = NULL) {
     pdfDocument <- load_doc(file, password = password)
     on.exit(pdfDocument$close())
 
-    stripper <- new(J("org.apache.pdfbox.util.PDFTextStripper"))
+    stripper <- new(J("org.apache.pdfbox.text.PDFTextStripper"))
 
     if (!is.null(pages)) {
         pages <- as.integer(pages)

diff --git a/R/get_page_dims.R b/R/get_page_dims.R
@@ -33,7 +33,7 @@ get_page_dims <- function(file, doc, pages = NULL, password = NULL) {
         pages <- 1L:(get_n_pages(doc = doc))
     }
 
-    allpages <- doc$getDocumentCatalog()$getAllPages()
+    allpages <- doc$getDocumentCatalog()$getPages()
     lapply(pages, function(x) {
         thispage <- allpages$get(x-1L)
         c(thispage$getMediaBox()$getWidth(), thispage$getMediaBox()$getHeight())
@@ -47,5 +47,5 @@ get_n_pages <- function(file, doc, password = NULL) {
         doc <- load_doc(file, password = password, copy = FALSE)
         on.exit(doc$close())
     }
-    doc$getDocumentCatalog()$getAllPages()$size()
+    doc$getNumberOfPages()
 }
diff --git a/R/make_thumbnails.R b/R/make_thumbnails.R
@@ -22,8 +22,12 @@
 #' @importFrom rJava J new
 #' @seealso \code{\link{extract_tables}}, \code{\link{extract_text}}, \code{\link{make_thumbnails}}
 #' @export
-make_thumbnails <- 
-function(file, outdir = NULL, pages = NULL, format = c("png", "jpeg", "bmp", "gif"), resolution = 72L, password = NULL) {
+make_thumbnails <- function(file,
+                            outdir = NULL,
+                            pages = NULL,
+                            format = c("png", "jpeg", "bmp", "gif"),
+                            resolution = 72L,
+                            password = NULL) {
     file <- localize_file(file)
     pdfDocument <- load_doc(file, password = password)
     on.exit(pdfDocument$close())
@@ -37,17 +41,22 @@ function(file, outdir = NULL, pages = NULL, format = c("png", "jpeg", "bmp", "gi
     format <- match.arg(format)
     fileseq <- formatC(pages, width = max(nchar(pages)), flag = 0)
     if (is.null(outdir)) {
-        prefix <- basename(file_path_sans_ext(file))
         outfile <- paste0(file_path_sans_ext(file), fileseq, ".", format)
     } else {
-        prefix <- file.path(outdir, basename(file_path_sans_ext(file)))
-        outfile <- file.path(outdir, paste0(basename(file_path_sans_ext(file)), fileseq, ".", format))
+        filename <- paste0(basename(file_path_sans_ext(file)), fileseq, ".", format)
+        outfile <- file.path(outdir, filename)
     }
     for (i in seq_along(pages)) {
-        PDFImageWriter <- new(J("org.apache.pdfbox.util.PDFImageWriter"))
-        PDFImageWriter$writeImage(pdfDocument, format, "", pages[i], pages[i], 
-                                  prefix, 1L, as.integer(resolution))
+        pageIndex <- pages[i] - 1L
+        PDFRenderer <- new(J("org.apache.pdfbox.rendering.PDFRenderer"),
+                           document = pdfDocument)
+        # BufferedImage <- PDFRenderer$renderImageWithDPI(pages[i],
+        #                                          scale = as.double(resolution))
+        BufferedImage <- PDFRenderer$renderImage(pageIndex)
+        JavaFile <- new(J("java.io.File"), pathname = outfile[i])
+        J("javax.imageio.ImageIO")$write(BufferedImage,
+                                         format,
+                                         JavaFile)
     }
-    file.rename(from = paste0(prefix, pages, ".", format), to = outfile)
     ifelse(file.exists(outfile), outfile, NA_character_)
 }
diff --git a/R/split_merge.R b/R/split_merge.R
@@ -30,7 +30,7 @@ split_pdf <- function(file, outdir = NULL, password = NULL) {
     file <- localize_file(file, copy = TRUE)
     pdfDocument <- load_doc(file, password = password)
     on.exit(pdfDocument$close())
-    splitter <- new(J("org.apache.pdfbox.util.Splitter"))
+    splitter <- new(J("org.apache.pdfbox.multipdf.Splitter"))
     splitArray <- splitter$split(pdfDocument)
     iterator <- splitArray$iterator()
     p <- 1L
@@ -57,7 +57,7 @@ split_pdf <- function(file, outdir = NULL, password = NULL) {
 merge_pdfs <- function(file, outfile) {
     outfile <- path.expand(outfile)
     file <- unlist(lapply(file, localize_file, copy = TRUE))
-    merger <- new(J("org.apache.pdfbox.util.PDFMergerUtility"))
+    merger <- new(J("org.apache.pdfbox.multipdf.PDFMergerUtility"))
     merger$setDestinationFileName(outfile)
     lapply(file, merger$addSource)
     merger$mergeDocuments()

diff --git a/R/utils.R b/R/utils.R
@@ -19,12 +19,11 @@ localize_file <- function(path, copy = FALSE, quiet = TRUE) {
 load_doc <- function(file, password = NULL, copy = TRUE) {
     file <- localize_file(path = file, copy = copy)
     pdfDocument <- new(J("org.apache.pdfbox.pdmodel.PDDocument"))
-    scratchFileJava <- new(J("java.io.File"), scratchFile <- tempfile())
-    randomAccess <- new(J("org.apache.pdfbox.io.RandomAccessFile"), scratchFileJava, "rw")
+    fileJava <- new(J("java.io.File"), pathname = file)
     if (is.null(password)) {
-        doc <- pdfDocument$load(filename = file, scratchFile = randomAccess)
+        doc <- pdfDocument$load(file = fileJava)
     } else {
-        doc <- pdfDocument$load(filename = file, scratchFile = randomAccess, password = password)
+        doc <- pdfDocument$load(file = fileJava, password = password)
     }
     pdfDocument$close()
     doc