Merge pull request #5 from datascouting/all-text-select-on-pdf

allow add selectable text on pdf

Author: Iordanis Kostelidis

.gitignore

@@ -4,4 +4,4 @@
 /dist/
 /*.egg-info/
 .idea
-
+*.pdf

hocr-pdf

@@ -49,7 +49,7 @@ class StdoutWrapper:
         sys.stdout.write(data)
 
 
-def export_pdf(playground, default_dpi, savefile=False):
+def export_pdf(playground, default_dpi, savefile=False, selectable_text=False):
     """Create a searchable PDF from a pile of HOCR + JPEG"""
     images = sorted(glob.glob(os.path.join(playground, '*.jpg')))
     images = Tcl().call('lsort', '-dict', images)
@@ -58,6 +58,7 @@ def export_pdf(playground, default_dpi, savefile=False):
               "\nScript cannot proceed without them and will terminate now.\n")
         sys.exit(0)
     load_invisible_font()
+    load_default_font()
     pdf = Canvas(savefile if savefile else StdoutWrapper(), pageCompression=1)
     pdf.setCreator('hocr-tools')
     pdf.setTitle(os.path.basename(playground))
@@ -73,13 +74,17 @@ def export_pdf(playground, default_dpi, savefile=False):
         height = h * 72 / dpi
         pdf.setPageSize((width, height))
         pdf.drawImage(image, 0, 0, width=width, height=height)
-        add_text_layer(pdf, image, height, dpi)
+        add_text_layer(pdf, image, height, dpi, selectable_text)
         pdf.showPage()
     pdf.save()
 
 
-def add_text_layer(pdf, image, height, dpi):
+def add_text_layer(pdf, image, height, dpi, selectable_text=False):
     """Draw an invisible text layer for OCR data"""
+    font = 'invisible'
+    if selectable_text:
+        font = "DejaVuSans"
+
     p1 = re.compile(r'bbox((\s+\d+){4})')
     p2 = re.compile(r'baseline((\s+[\d\.\-]+){2})')
     hocrfile = os.path.splitext(image)[0] + ".hocr"
@@ -101,19 +106,14 @@ def add_text_layer(pdf, image, height, dpi):
             rawtext = word.text_content().strip()
             if rawtext == '':
                 continue
-            font_width = pdf.stringWidth(rawtext, 'invisible', 8)
-            if font_width <= 0:
-                continue
             box = p1.search(word.attrib['title']).group(1).split()
             box = [float(i) for i in box]
             b = polyval(baseline,
                         (box[0] + box[2]) / 2 - linebox[0]) + linebox[3]
             text = pdf.beginText()
             text.setTextRenderMode(3)  # double invisible
-            text.setFont('invisible', 8)
+            text.setFont(font, 8)
             text.setTextOrigin(box[0] * 72 / dpi, height - b * 72 / dpi)
-            box_width = (box[2] - box[0]) * 72 / dpi
-            text.setHorizScale(100.0 * box_width / font_width)
             text.textLine(rawtext)
             pdf.drawText(text)
 
@@ -154,6 +154,10 @@ CMGjwvxTsr74/f/F95m3TH9x8o0/TU//N+7/D/ScVcA=
     pdfmetrics.registerFont(TTFont('invisible', ttf))
 
 
+def load_default_font():
+    pdfmetrics.registerFont(TTFont('DejaVuSans', 'DejaVuSans.ttf'))
+
+
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(
         description="Create a searchable PDF from a pile of hOCR and JPEG")
@@ -167,7 +171,9 @@ if __name__ == "__main__":
         "--savefile",
         help="Save to this file instead of outputting to stdout"
     )
+    parser.add_argument("--selectable-text", action="store_true",
+                        help="allow text selection on pdf")
     args = parser.parse_args()
     if not os.path.isdir(args.imgdir):
         sys.exit("ERROR: Given path '" + args.imgdir + "' is not a directory")
-    export_pdf(args.imgdir, 300, args.savefile)
+    export_pdf(args.imgdir, 300, args.savefile, args.selectable_text)

test/hocr-pdf/test-hocr-pdf.tsht

@@ -14,4 +14,9 @@ file_not_empty "${work}.pdf"
 exec_ok pdfgrep 'tribunali' "${work}.pdf"
 hocr-pdf . --savefile "${work}-saved.pdf"
 file_not_empty "${work}-saved.pdf"
+hocr-pdf --selectable-text . > "${work}.pdf"
+file_not_empty "${work}.pdf"
+exec_ok pdfgrep 'tribunali' "${work}.pdf"
+hocr-pdf --selectable-text . --savefile "${work}-saved.pdf"
+file_not_empty "${work}-saved.pdf"
 rm $work*