A lua script for nextcloud files_script to run OCR with tesseract and put the output as comment

ocr_comments.lua

local fileObj = get_input_files()[1]

if (fileObj == nil or not is_file(fileObj) ) then
	abort("No file name was provided")
end

local fileMeta = meta_data(fileObj)
if ( not string.find(fileMeta.mimetype, "image/") ) then
  abort("File ".. fileObj.name .. " is not an image (mimetype:" .. fileMeta.mimetype ..")")
end

local filePath = fileMeta.local_path
local fileOwner = fileMeta.owner_id

-- Run OCR cmd
local tesseractCmd = "tesseract '" .. filePath .. "' stdout"
local tesseractRes = shell_command(tesseractCmd)
if (tesseractRes == nil) then
  abort("Failed to execute command '".. tesseractCmd.."'. Error:".. tesseractRes.errors .." Code:".. tesseractRes.exit_code)
end


-- ----------------------------------------------------------------
-- Clean the comment (remove non-visible & special chars, and trim message)
local comment = tesseractRes.output
comment = comment:gsub("[^A-Za-z0-9_+\\- ]", " ")
comment = comment:sub(0, 1000)

-- only add comment if tesseract found something
if (comment ~= "") then
  -- Workaround when script is run from occ, users_find() returns nil. Using file owner in that case
  local commentUser = users_find()
  if (userCurrent == nil) then
    commentUser = users_find(nil, fileOwner)
  end
  commentUser = commentUser[1]

  commentOpt = { unsafe_impersonate_user = commentUser }
  commentRes = comment_create(comment, fileObj, commentOpt)
  if (commentRes == nil) then
    abort("Failed to add comment '".. comment .."' on file '".. fileObj.name .. "' (cmd: '" .. tesseractCmd .. "', stderr: '".. tesseractRes.errors .."', return:".. tesseractRes.exit_code ..")")
  end
end


-- Remove tag if triggered from it
local ocrTagName = "run:ocrimg"
local fileTags = get_file_tags(fileObj)
if (fileTags ~= nil) then
  for _, tag in ipairs(fileTags) do
    if (tag.name == ocrTagName) then
      if (tag_file_unassign(fileObj, tag) == false) then
        abort("Failed to unassign tag ".. ocrTagName)
      end
    end
  end
end