markup: support rectangle and circle annotations

markup/img.py

@@ -9,6 +9,8 @@ import dumper
 import random as rng
 from pathlib import Path
 
+from pdfminer.psparser import LIT
+
 from .utils import cv2_rect, ensure_dir, set_file_perms, WORKDIR
 
 # https://www.pyimagesearch.com/2014/10/20/finding-shapes-images-using-python-opencv/
@@ -112,10 +114,10 @@ def write_debug_image(workdir, page_num, prods, scribbles):
     set_file_perms(path)
 
 
-def write_inklist(obj, path):
+def write_inklist(obj, mediabox, path):
     """Draw an image of the inklist."""
-    pagew = int(11*72)
-    pageh = int(8.5*72)
+    pagew = mediabox[2] - mediabox[0]
+    pageh = mediabox[3] - mediabox[1]
 
     img = Image.new('RGBA', (pagew, pageh), (0, 0, 0, 0))
     draw = ImageDraw.Draw(img, 'RGBA')
@@ -129,3 +131,24 @@ def write_inklist(obj, path):
 
     img.save(path)
     set_file_perms(path)
+
+
+def write_square_or_circle(obj, mediabox, path):
+    """Draw an image of the inklist."""
+    pagew = mediabox[2] - mediabox[0]
+    pageh = mediabox[3] - mediabox[1]
+
+    img = Image.new('RGBA', (pagew, pageh), (0, 0, 0, 0))
+    draw = ImageDraw.Draw(img, 'RGBA')
+
+    if obj["Subtype"] == LIT('Square'):
+        draw.rectangle(obj['Rect'], fill=None, outline='black', width=3)
+    else:
+        draw.ellipse(*obj['Rect'], fill=None, outline='black', width=3)
+
+    # account for the difference in coordinate systems
+    # between pdf and images.
+    img = img.transpose(Image.FLIP_TOP_BOTTOM)
+
+    img.save(path)
+    set_file_perms(path)

markup/pdf.py

@@ -10,7 +10,7 @@ from pdfminer.pdfdocument import PDFDocument
 from pdfminer.pdftypes import PDFObjRef, resolve1
 
 from .utils import Rect, pdf_rect, ensure_dir, set_file_perms
-from .img import write_inklist
+from .img import write_inklist, write_square_or_circle
 
 
 def make_product_box(obj, pagenum, mediabox):
@@ -44,7 +44,19 @@ def make_ink_scribble(obj, pagenum, mediabox, workdir):
     oid = obj['NM'].decode('utf-8')
     png_path = os.path.join(workdir, f"export-page{pagenum:03d}-nm{oid}.png")
 
-    write_inklist(obj, png_path)
+    write_inklist(obj, mediabox, png_path)
+
+    return { 'page': pagenum,
+             'rect': Rect(*mediabox),
+             'objid': oid,
+             'image': png_path }
+
+
+def make_square_or_circle_scribble(obj, pagenum, mediabox, workdir):
+    oid = obj['NM'].decode('utf-8')
+    png_path = os.path.join(workdir, f"export-page{pagenum:03d}-nm{oid}.png")
+
+    write_square_or_circle(obj, mediabox, png_path)
 
     return { 'page': pagenum,
              'rect': Rect(*mediabox),
@@ -160,6 +172,13 @@ def is_inklist_annotation(anno):
     return 'Subtype' in anno and anno["Subtype"] == LIT('Ink')
 
 
+def is_square_or_circle_annotation(anno):
+    if 'Subtype' in anno:
+        if anno["Subtype"] == LIT('Square') or anno["Subtype"] == LIT('Circle'):
+            return True
+    return False
+
+
 def parse_pdf(fname, workdir, debug=0):
     PDFDocument.debug = debug
     PDFParser.debug = debug
@@ -192,10 +211,16 @@ def parse_pdf(fname, workdir, debug=0):
             anno = resolve1(anno)
             if is_inklist_annotation(anno):
                 scribbles.append(make_ink_scribble(anno, pagenum, mediabox, workdir))
+            elif is_square_or_circle_annotation(anno):
+                scribbles.append(make_square_or_circle_scribble(anno, pagenum, mediabox, workdir))
             elif 'AAPL:AKExtras' in anno:
                 scribbles.append(make_aapl_scribble(anno, pagenum, mediabox, workdir))
             elif 'ProCatName' in anno:
                 prod_boxes.append(make_product_box(anno, pagenum, mediabox))
+            elif anno['Subtype'] == LIT('FreeText'):
+                print('ignoring FreeText annotation')
+            elif anno['Subtype'] == LIT('Highlight'):
+                print('ignoring Highlight annotation')
             else:
                 print('ignoring other annotation:')
                 print(anno)