From 1552f5dcd8b0abe89f53182c7a495a7b8dc008fc Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Wed, 8 Apr 2020 15:51:47 -0700
Subject: helper bash scripts for PDF derivatives

These are lazy (won't run derive process if output already exists),
which makes the process much faster
---
 bin/make_dir_derivatives.sh | 16 ++++++++++++++++
 bin/make_pdf_derivatives.sh | 22 ++++++++++++++++++++++
 2 files changed, 38 insertions(+)
 create mode 100755 bin/make_dir_derivatives.sh
 create mode 100755 bin/make_pdf_derivatives.sh

(limited to 'bin')

diff --git a/bin/make_dir_derivatives.sh b/bin/make_dir_derivatives.sh
new file mode 100755
index 0000000..c33d75d
--- /dev/null
+++ b/bin/make_dir_derivatives.sh
@@ -0,0 +1,16 @@
+#!/usr/bin/env bash
+
+set -e -u -o pipefail
+
+export FULLTEXTDIR=$1
+
+if [ ! -d $FULLTEXTDIR ]; then
+    echo "Directory does not exist: $FULLTEXTDIR"
+    exit -1
+fi
+
+# make directories
+ls $FULLTEXTDIR/pdf/ | parallel mkdir -p $FULLTEXTDIR/pdftotext/{}
+ls $FULLTEXTDIR/pdf/ | parallel mkdir -p $FULLTEXTDIR/thumbnail/{}
+
+fd -I .pdf $FULLTEXTDIR/pdf/ | sed "s/\.pdf//g" | cut -d/ -f3-4 | parallel -j10 ./bin/make_pdf_derivatives.sh $FULLTEXTDIR {}
diff --git a/bin/make_pdf_derivatives.sh b/bin/make_pdf_derivatives.sh
new file mode 100755
index 0000000..8727119
--- /dev/null
+++ b/bin/make_pdf_derivatives.sh
@@ -0,0 +1,22 @@
+#!/usr/bin/env bash
+
+set -e -u -o pipefail
+
+export FULLTEXTDIR=$1
+export BLOBPATH=$2
+export PDFPATH="$FULLTEXTDIR/pdf/$BLOBPATH.pdf"
+
+if [[ ! -f $PDFPATH ]]; then
+    echo "PDF does not exist: $PDFPATH"
+    exit -1
+fi
+
+echo "processing: $PDFPATH"
+
+if [[ ! -f "$FULLTEXTDIR/pdftotext/$BLOBPATH.txt" ]]; then
+    pdftotext $PDFPATH $FULLTEXTDIR/pdftotext/$BLOBPATH.txt
+fi
+
+if [[ ! -f "$FULLTEXTDIR/thumbnail/$BLOBPATH.png" ]]; then
+    pdftocairo -png -singlefile -scale-to-x 400 -scale-to-y -1 $PDFPATH $FULLTEXTDIR/thumbnail/$BLOBPATH
+fi
-- 
cgit v1.2.3