OCR to rename file

Read a PDF file and rename the file with INFO in the page

What is it?

OCR to rename file is software that allows you to open pdf files, read the contents of those files, and rename the file according to the contents of the document.

What does it do?

The software reads the configuration, then reads the .pdf files within the input folder, turns page x into an image, the various pieces to be read are cropped, an OCR scan is done, and then with the results it had, renames and moves the .pdf file with the read data.

How to clone software and setup

1. Download last software release from: https://github.com/Kekko01/OCR-to-rename-file/releases

2. Extract it and open folder

3. Installs the required libraries with the requirements.txt:

   pip install -r requirements.txt

4. Edit the config-EDITME.json file and rename it config.json, with this structure::

   {
       "configs":
       [
           {
               "name": "Config 1",
               "description": "Description 1",
               "in_dir": "dir\\in\\files",
               "out_dir": "dir\\out\\files",
               "dpi": 150,
               "languages": ["it"],
               "filename": "{FIELD1} {field2} - {Field3} some text.pdf",
               "fields": [
                   {"name": "FIELD1", "left": 194, "top": 138, "right": 548, "bottom": 173},
                   {"name": "field2", "left": 549, "top": 141, "right": 1146, "bottom": 164},
                   {"name": "Field3", "left": 849, "top": 179, "right": 877, "bottom": 205}
               ]
           }
           
           ,
   
           {
               "name": "Config 2",
               "description": "Description 2",
               "in_dir": "dir\\in\\files",
               "out_dir": "dir\\out\\files",
               "dpi": 150,
               "languages": ["fr", "en"],
               "page": 1,
               "filename": "{FIELD1} {field2} - {Field3} some text.pdf",
               "fields": [
                   {"name": "FIELD1", "left": 194, "top": 138, "right": 548, "bottom": 173},
                   {"name": "field2", "left": 549, "top": 141, "right": 1146, "bottom": 164},
                   {"name": "Field3", "left": 849, "top": 179, "right": 877, "bottom": 205}
               ]
           }
       ]
   }

   "name": (string) - Name of the configuration "description": (string) - Description of the configuration

   "in_dir": (string) - Input folder where pdf files are located

   "out_dir": (string) - Output folder where pdf files will be moved

   "dpi": (int) - DPI used to convert the page to be analyzed to jpg (OCR only works with images, so the software will convert the page). By default if this field is not added it is 150 DPI, but this field can be changed. The higher the DPI, the more defined the page will be, but it will be slower to convert and the generated file will be heavier.

   "languages": (array of strings) - Languages to be used for OCR, the software will try to read the text with the first language, if it fails it will try with the second, and so on. Check supported languages here: https://www.jaided.ai/easyocr/#:~:text=languages%20and%20expanding.-,Supported%20Languages,-Language

   "page": (int) - Page to be analyzed, if not specified, the software will analyze all the first page of the pdf files. Important: The first page equals the number 0, second page equals the number 1, and so on.

   "filename": (string) - Name of the file to be generated, the fields to be read must be enclosed in curly brackets, the rest of the text will be copied as it is. Please use names for variables without spaces, the software recognizes whether it is SHIFT, lowercase or Capitalize and then will format the text read as it finds the {variable}. If there are multiple words in the variable, use allattached or all_attached.

   "fields": (array of objects) - Fields to be read, the software will read the text in the coordinates specified in the pdf file, and will save the text read in the variable specified in the name field. The coordinates are expressed in pixels, and are calculated from the top left corner of the page. The coordinates can be found by first converting a test pdf page with Convert to image software, and then with the generated image take the various coordinates with software such as Photopea (free) or Photoshop. With Photopea, once the image is loaded, just click on the i in the upper right corner, and the x and y coordinates will appear. Move the pointer to the top left of the field you want to take and mark the coordinates and x will be "left," y will be "top"; then move the pointer to the bottom right and take the x for "right," and y for "bottom". Then assign a "name" following the rules written before.

5. Start the main.py software:

   python main.py

Compiled software (Beta)

You could skip some of these steps (2, 3, 5) and avoid installing Python by downloading the precompiled .exe package from the Releases page. It doesn't work everywhere and doesn't work as perfectly as the guide explained earlier, however, if you are on Windows you might try it. The package was done with pyinstaller, and I don't know why a virus is detected, probably False Positive.

---

FAQ

How to install Python?

Go here: https://www.python.org/downloads/ and install the verson for yout PC

Have you problems?

Don't worry, please you can found or create a issue here: https://github.com/Kekko01/OCR-to-rename-file/issues

I would like to contribute, how can I do that?

You can fork the repository and then submit a pull request. If you don't know how to do that, you can read the Github forking guide.

To-do:

• Add more pages to analyze in a PDF
• Create a software alongside the main one that helps to create the .json file in an easy and intuitive way.