Convert pdfs to pngs

Signed-off-by: Richard Marston <[email protected]>

Dockerfile

@@ -1,7 +1,7 @@
 from alpine:latest
 
 run apk update
-run apk add python3 py3-pip
+run apk add python3 py3-pip poppler-utils
 run /usr/bin/pip3 install pyxb openpyxl
 copy . /excel2xml
 cmd sh /excel2xml/process_all_xlsx.sh

process_images.sh

@@ -2,9 +2,17 @@
 INPUT_DIR=$1
 OUTPUT_DIR=$2
 
-find $INPUT_DIR -type f -not -name '*.xlsx' -exec sh -c '
+find $INPUT_DIR -type f -not \( -name '*.xlsx' -o -name '*.pdf' \) -exec sh -c '
   FILE="$0"
   OUTPUT_DIR="$1"
   FILE_NO_SPACES=$(basename "${FILE}" | tr '[:blank:]' '_')
   cp "$FILE" "$OUTPUT_DIR/$FILE_NO_SPACES"
 ' {} ${OUTPUT_DIR} ';'
+
+find $INPUT_DIR -type f -name '*.pdf' -exec sh -c '
+  FILE="$0"
+  OUTPUT_DIR="$1"
+  FILE_NO_SPACES=$(basename "${FILE}" | tr '[:blank:]' '_')
+  pdftoppm -png -singlefile "$FILE" "$OUTPUT_DIR/$FILE_NO_SPACES"
+  echo "Converting $FILE to $OUTPUT_DIR/${FILE_NO_SPACES}.png"
+' {} ${OUTPUT_DIR} ';'

xlsx2xml.py

@@ -160,7 +160,11 @@ def switch_uriType(argument):
       name        = cell_str(sheet_ranges, 38, 3+index)
       drawingType = cell_str(sheet_ranges, 39, 3+index)
       uriType     = cell_str(sheet_ranges, 40, 3+index)
-      uri = cell_str(sheet_ranges, 41, 3+index).strip().replace(' ', '_')
+      uri         = cell_str(sheet_ranges, 41, 3+index).strip().replace(' ', '_')
+
+      if uri.endswith('.pdf'):
+        uri = uri + '.png'
+
       if name != 'None' and uri!='None' :
         drawing             = IEC62559.Drawing()
         resourcestr         = IEC62559.Resource_String(uri)