1 files changed, 25 insertions, 0 deletions
diff --git a/scraper/pdf_dump_first_page.sh b/scraper/pdf_dump_first_page.sh
new file mode 100644
index 00000000..6277d40c
--- /dev/null
+++ b/scraper/pdf_dump_first_page.sh
@@ -0,0 +1,25 @@
+#!/bin/bash
+
+for i in datasets/s2/pdf/*/*/*.pdf
+  do
+    OUTPUT="${i%.*}.txt"
+    if [[ ! -e $OUTPUT ]]
+    then
+      pdf2txt.py -p 1 $i > $OUTPUT
+      if [ -s $OUTPUT ]
+      then
+        echo "found $OUTPUT"
+      else
+        echo "rm empty $OUTPUT"
+        rm -f $OUTPUT
+      fi
+    else
+      if [ -s $OUTPUT ]
+      then
+        echo "found $OUTPUT"
+      else
+        echo "rm empty $OUTPUT"
+        rm -f $OUTPUT
+      fi
+    fi
+  done