feat: Remove OCR Support (#2838)

* remove ocr package

* remove tesseract

* remove OCR from app

* remove OCR from tests

* fix docs

mealie/routes/__init__.py

@@ -7,7 +7,6 @@ from . import (
     comments,
     explore,
     groups,
-    ocr,
     organizers,
     parser,
     recipe,
@@ -32,4 +31,3 @@ router.include_router(unit_and_foods.router)
 router.include_router(admin.router)
 router.include_router(validators.router)
 router.include_router(explore.router)
-router.include_router(ocr.router)

mealie/routes/ocr/__init__.py

@@ -1,7 +0,0 @@
-from fastapi import APIRouter
-
-from . import pytesseract
-
-router = APIRouter(prefix="/ocr")
-
-router.include_router(pytesseract.router)

mealie/routes/ocr/pytesseract.py

@@ -1,37 +0,0 @@
-from fastapi import APIRouter, File
-
-from mealie.routes._base import BaseUserController, controller
-from mealie.schema.ocr.ocr import OcrAssetReq, OcrTsvResponse
-from mealie.services.ocr.pytesseract import OcrService
-from mealie.services.recipe.recipe_data_service import RecipeDataService
-from mealie.services.recipe.recipe_service import RecipeService
-
-router = APIRouter()
-
-
-@controller(router)
-class OCRController(BaseUserController):
-    def __init__(self):
-        self.ocr_service = OcrService()
-
-    @router.post("/", response_model=str)
-    def image_to_string(self, file: bytes = File(...)):
-        return self.ocr_service.image_to_string(file)
-
-    @router.post("/file-to-tsv", response_model=list[OcrTsvResponse])
-    def file_to_tsv(self, file: bytes = File(...)):
-        tsv = self.ocr_service.image_to_tsv(file)
-        return self.ocr_service.format_tsv_output(tsv)
-
-    @router.post("/asset-to-tsv", response_model=list[OcrTsvResponse])
-    def asset_to_tsv(self, req: OcrAssetReq):
-        recipe_service = RecipeService(self.repos, self.user, self.group)
-        recipe = recipe_service._get_recipe(req.recipe_slug)
-        if recipe.id is None:
-            return []
-        data_service = RecipeDataService(recipe.id, recipe.group_id)
-        asset_path = data_service.dir_assets.joinpath(req.asset_name)
-        file = open(asset_path, "rb")
-        tsv = self.ocr_service.image_to_tsv(file.read())
-
-        return self.ocr_service.format_tsv_output(tsv)

mealie/routes/recipe/recipe_crud_routes.py

@@ -27,10 +27,7 @@ from mealie.schema.make_dependable import make_dependable
 from mealie.schema.recipe import Recipe, RecipeImageTypes, ScrapeRecipe
 from mealie.schema.recipe.recipe import CreateRecipe, CreateRecipeByUrlBulk, RecipeLastMade, RecipeSummary
 from mealie.schema.recipe.recipe_asset import RecipeAsset
-from mealie.schema.recipe.recipe_ingredient import RecipeIngredient
 from mealie.schema.recipe.recipe_scraper import ScrapeRecipeTest
-from mealie.schema.recipe.recipe_settings import RecipeSettings
-from mealie.schema.recipe.recipe_step import RecipeStep
 from mealie.schema.recipe.request_helpers import RecipeDuplicate, RecipeZipTokenResponse, UpdateImageResponse
 from mealie.schema.response import PaginationBase, PaginationQuery
 from mealie.schema.response.pagination import RecipeSearchQuery
@@ -489,37 +486,3 @@ class RecipeController(BaseRecipeController):
         self.mixins.update_one(recipe, slug)
 
         return asset_in
-
-    # ==================================================================================================================
-    # OCR
-    @router.post("/create-ocr", status_code=201, response_model=str)
-    def create_recipe_ocr(
-        self, extension: str = Form(...), file: UploadFile = File(...), makefilerecipeimage: bool = Form(...)
-    ):
-        """Takes an image and creates a recipe based on the image"""
-        slug = self.service.create_one(
-            Recipe(
-                name="New OCR Recipe",
-                recipe_ingredient=[RecipeIngredient(note="", title=None, unit=None, food=None, original_text=None)],
-                recipe_instructions=[RecipeStep(text="")],
-                is_ocr_recipe=True,
-                settings=RecipeSettings(show_assets=True),
-                id=None,
-                image=None,
-                recipe_yield=None,
-                rating=None,
-                orgURL=None,
-                date_added=None,
-                date_updated=None,
-                created_at=None,
-                update_at=None,
-                nutrition=None,
-            )
-        ).slug
-        RecipeController.upload_recipe_asset(self, slug, "Original recipe image", "", extension, file)
-        if makefilerecipeimage:
-            # Get the pointer to the beginning of the file to read it once more
-            file.file.seek(0)
-            self.update_recipe_image(slug, file.file.read(), extension)
-
-        return slug

mealie/schema/ocr/__init__.py

@@ -1,7 +0,0 @@
-# This file is auto-generated by gen_schema_exports.py
-from .ocr import OcrAssetReq, OcrTsvResponse
-
-__all__ = [
-    "OcrAssetReq",
-    "OcrTsvResponse",
-]

mealie/schema/ocr/ocr.py

@@ -1,21 +0,0 @@
-from mealie.schema._mealie import MealieModel
-
-
-class OcrTsvResponse(MealieModel):
-    level: int = 0
-    page_num: int = 0
-    block_num: int = 0
-    par_num: int = 0
-    line_num: int = 0
-    word_num: int = 0
-    left: int = 0
-    top: int = 0
-    width: int = 0
-    height: int = 0
-    conf: float = 0.0
-    text: str = ""
-
-
-class OcrAssetReq(MealieModel):
-    recipe_slug: str
-    asset_name: str

mealie/schema/recipe/recipe.py

@@ -128,7 +128,6 @@ class Recipe(RecipeSummary):
     assets: list[RecipeAsset] | None = []
     notes: list[RecipeNote] | None = []
     extras: dict | None = {}
-    is_ocr_recipe: bool | None = False
 
     comments: list[RecipeCommentOut] | None = []
 

mealie/services/ocr/pytesseract.py

@@ -1,56 +0,0 @@
-from io import BytesIO
-
-import pytesseract
-from PIL import Image
-
-from mealie.schema.ocr.ocr import OcrTsvResponse
-from mealie.services._base_service import BaseService
-
-
-class OcrService(BaseService):
-    """
-    Class for ocr engines.
-    """
-
-    def image_to_string(self, image_data):
-        """
-        Returns a plain text translation of an image
-        """
-        return pytesseract.image_to_string(Image.open(image_data))
-
-    def image_to_tsv(self, image_data, lang=None):
-        """
-        Returns the pytesseract default tsv output
-        """
-        if lang is not None:
-            return pytesseract.image_to_data(Image.open(BytesIO(image_data)), lang=lang)
-
-        return pytesseract.image_to_data(Image.open(BytesIO(image_data)))
-
-    def format_tsv_output(self, tsv: str) -> list[OcrTsvResponse]:
-        """
-        Returns a OcrTsvResponse from a default pytesseract tsv output
-        """
-        lines = tsv.split("\n")
-        titles = [t.strip() for t in lines[0].split("\t")]
-        response: list[OcrTsvResponse] = []
-
-        for i in range(1, len(lines)):
-            if lines[i] == "":
-                continue
-
-            line = OcrTsvResponse()
-            for key, value in zip(titles, lines[i].split("\t"), strict=False):
-                if key == "text":
-                    setattr(line, key, value.strip())
-                elif key == "conf":
-                    setattr(line, key, float(value.strip()))
-                elif key in OcrTsvResponse.__fields__:
-                    setattr(line, key, int(value.strip()))
-                else:
-                    continue
-
-            if isinstance(line, OcrTsvResponse):
-                response.append(line)
-
-        return response