Check all the PDF's pages for oversized page dimensions

lib/pdf_info.rb

@@ -18,7 +18,7 @@ def self.read(file_or_path)
 
     def initialize(path)
       @stdout = []
-      Open3.popen2e(Settings.binaries.pdfinfo, path) do |_stdin, stdout, wait|
+      Open3.popen2e(Settings.binaries.pdfinfo, '-l', '-1', path) do |_stdin, stdout, wait|
         stdout.each_line do |line|
           @stdout.push(force_utf8_encoding(line))
         end
@@ -43,7 +43,7 @@ def pages
     end
 
     def page_size
-      width, height = self['Page size'].scan(/\d+/).map(&:to_i)
+      width, height = self['Page    1 size'].scan(/\d+/).map(&:to_i)
       { width:, height: }
     end
 
@@ -55,6 +55,20 @@ def page_size_inches
       }
     end
 
+    def oversized_pages(max_width, max_height)
+      results = []
+      (1..pages).each do |page_num|
+        page_key = page_num < 10_000 ? format('%4d', page_num) : page_num
+        pw_pts, ph_pts = self["Page #{page_key} size"].scan(/\d+/).map(&:to_i)
+        pw_inches = convert_pts_to_inches(pw_pts)
+        ph_inches = convert_pts_to_inches(ph_pts)
+        if pw_inches > max_width || ph_inches > max_height
+          results << "Page #{page_num}, Width: #{pw_inches}, Height: #{ph_inches}"
+        end
+      end
+      results
+    end
+
     def file_size
       self['File size'].scan(/\d+/).first.to_i
     end

lib/pdf_utilities/pdf_validator.rb

@@ -97,14 +97,15 @@ def check_encryption
       end
 
       def check_page_size
-        if @pdf_metadata.present?
-          dimensions = @pdf_metadata.page_size_inches
-          width_limit = @options[:width_limit_in_inches]
-          height_limit = @options[:height_limit_in_inches]
-
-          if dimensions[:width] > width_limit || dimensions[:height] > height_limit
-            @result.add_error("#{PAGE_SIZE_LIMIT_EXCEEDED_MSG} of #{width_limit} in. x #{height_limit} in.")
-          end
+        return if @pdf_metadata.blank?
+
+        width_limit = @options[:width_limit_in_inches]
+        height_limit = @options[:height_limit_in_inches]
+
+        oversized_pages = @pdf_metadata.oversized_pages(width_limit, height_limit)
+
+        if oversized_pages.present?
+          @result.add_error("#{PAGE_SIZE_LIMIT_EXCEEDED_MSG} of #{width_limit} in. x #{height_limit} in.")
         end
       end
     end

modules/vba_documents/lib/vba_documents/pdf_inspector.rb

@@ -91,7 +91,7 @@ def pdf_metadata(pdf)
         dimensions: {
           height: dimensions[:height].round(2),
           width: dimensions[:width].round(2),
-          oversized_pdf: dimensions[:height] > max_height || dimensions[:width] > max_width
+          oversized_pdf: metadata.oversized_pages(max_width, max_height).any?
         },
         file_size: metadata.file_size,
         sha256_checksum: Digest::SHA256.file(pdf).hexdigest

spec/lib/pdf_info/metadata_spec.rb

@@ -20,7 +20,7 @@
       JavaScript:     no
       Pages:          4
       Encrypted:      no
-      Page size:      612 x 792 pts (letter)
+      Page    1 size: 612 x 792 pts (letter)
       Page rot:       0
       File size:      1099807 bytes
       Optimized:      no
@@ -51,7 +51,7 @@
   describe '::read' do
     context 'when passed a string' do
       it 'shells out with the string as the file path' do
-        expect(Open3).to receive(:popen2e).with('pdfinfo', '/tmp/file.pdf').and_yield('', result, good_exit)
+        expect(Open3).to receive(:popen2e).with('pdfinfo', '-l', '-1', '/tmp/file.pdf').and_yield('', result, good_exit)
         described_class.read('/tmp/file.pdf')
       end
     end
@@ -60,14 +60,14 @@
       it 'shells out with the file object path' do
         file = double(File)
         allow(file).to receive(:path).and_return('/tmp/file.pdf')
-        expect(Open3).to receive(:popen2e).with('pdfinfo', '/tmp/file.pdf').and_yield('', result, good_exit)
+        expect(Open3).to receive(:popen2e).with('pdfinfo', '-l', '-1', '/tmp/file.pdf').and_yield('', result, good_exit)
         described_class.read(file)
       end
     end
 
     context 'when the command errors' do
       it 'raises a PdfInfo::MetadataReadError' do
-        expect(Open3).to receive(:popen2e).with('pdfinfo', '/tmp/file.pdf').and_yield('', result, bad_exit)
+        expect(Open3).to receive(:popen2e).with('pdfinfo', '-l', '-1', '/tmp/file.pdf').and_yield('', result, bad_exit)
         expect { described_class.read('/tmp/file.pdf') }.to raise_error(PdfInfo::MetadataReadError, /pdfinfo exited/)
       end
     end