diff --git a/.gitignore b/.gitignore index 18c1ceb..fdbfbff 100644 --- a/.gitignore +++ b/.gitignore @@ -2,3 +2,5 @@ downloaded_files/ testing/ cropped.png backup/ +__pycache__/ +page.html diff --git a/coin.ui b/coin.ui index ed8eaa5..c7b9668 100644 --- a/coin.ui +++ b/coin.ui @@ -6,7 +6,7 @@ 0 0 - 1280 + 1695 720 @@ -58,7 +58,7 @@ 0 0 - 594 + 802 644 @@ -101,7 +101,14 @@ - + + + Save to file + + + + + Save to database diff --git a/gui.py b/gui.py index e0a1f7e..286d90f 100644 --- a/gui.py +++ b/gui.py @@ -9,20 +9,30 @@ from PyQt6.QtWidgets import * from PyQt6 import uic from PyQt6 import QtCore, QtGui, QtWidgets import glob +import upc +from upc import get_name_from_upc # [!!!] Change this if tesseract or coin can't find any data MODELS_PATH = '/usr/share/tessdata/' # [!!!] Sane values for scaling -SCALE_MIN = 0.3 -SCALE_MAX = 4.0 +SCALE_MIN = 0.3 +SCALE_MAX = 4.0 # How much (in decimal) to change the scale for every degree SCALE_DELTA = 8 * 360 * 1 # [!!!] Sane values for recognition -WIDTH_MIN = 10 +WIDTH_MIN = 10 HEIGHT_MIN = 10 +# Column Information +TABLE_COL_CNT = 5 +TABLE_COL_GET = 0 +TABLE_COL_DEL = 1 +TABLE_COL_UPC = 2 +TABLE_COL_NAME = 3 +TABLE_COL_PRICE = 4 + # Automatically finds all available language models models_list = glob.glob(MODELS_PATH + "*.traineddata") model_names = [] @@ -34,7 +44,6 @@ if len(model_names) == 0: print("No tesseract models found at", MODELS_PATH, "!", file=sys.stderr) sys.exit(0) - class coin_app(QtWidgets.QMainWindow): def __init__(self): QtWidgets.QMainWindow.__init__(self) @@ -42,6 +51,10 @@ class coin_app(QtWidgets.QMainWindow): self.image = None self.ui.openButton.clicked.connect(self.open) + self.ui.getNameButton.clicked.connect(self.get_all) + self.ui.saveFileButton.clicked.connect(self.save_file) + self.ui.saveDbButton.clicked.connect(self.save_database) + self.rubberBand = QRubberBand(QRubberBand.Shape.Rectangle, self) self.ui.photo.setMouseTracking(True) self.ui.photo.installEventFilter(self) @@ -53,23 +66,52 @@ class coin_app(QtWidgets.QMainWindow): self.models.currentTextChanged.connect(self.update_now) self.models.setCurrentIndex(model_names.index(self.model)) - self.items = QStandardItemModel() - self.itemsTable.setModel(self.items) - self.items.setColumnCount(1) - self.scale = 1.0 + self.items = None + def update_now(self, value): self.model = value print("Model selected as:", self.model) + + def set_table_header(self): + item = QStandardItem(" Get ") + item.setTextAlignment(Qt.AlignmentFlag.AlignCenter) + item.setEditable(False) + self.items.setItem(0, TABLE_COL_GET, item) + item = QStandardItem(" Del ") + item.setTextAlignment(Qt.AlignmentFlag.AlignCenter) + item.setEditable(False) + self.items.setItem(0, TABLE_COL_DEL, item) + item = QStandardItem("UPC") + item.setTextAlignment(Qt.AlignmentFlag.AlignCenter) + item.setEditable(False) + self.items.setItem(0, TABLE_COL_UPC, item) + item = QStandardItem("Item Name") + item.setTextAlignment(Qt.AlignmentFlag.AlignCenter) + item.setEditable(False) + self.items.setItem(0, TABLE_COL_NAME, item) + item = QStandardItem("Price") + item.setTextAlignment(Qt.AlignmentFlag.AlignCenter) + item.setEditable(False) + self.items.setItem(0, TABLE_COL_PRICE, item) + self.itemsTable.resizeColumnsToContents() def open(self): filename = QFileDialog.getOpenFileName(self, 'Select File') + if filename[0] == "": + print("No file was selected or opened!") + return None + + if self.items is not None: + self.items.deleteLater() + print("Opening file: ", filename[0]) self.image = cv2.imread(str(filename[0])) - if self.image.size == 0: + if self.image is None: print("Error: image is empty!", file=sys.stderr) + return None frame = cv2.cvtColor(self.image, cv2.COLOR_BGR2RGB) image = QImage(frame, frame.shape[1], frame.shape[0], @@ -81,20 +123,123 @@ class coin_app(QtWidgets.QMainWindow): print("Updated photo!") print("Photo scale set to:", self.scale) + self.items = QStandardItemModel() + self.itemsTable.setModel(self.items) + self.items.setColumnCount(TABLE_COL_CNT) + + self.set_table_header() + print("Initialized new table!") + + self.statusBar().showMessage("Opened file: " + filename[0]) + + return None + + def save_file(self): + filename = QFileDialog.getSaveFileName(self, 'Select File') + + if filename[0] == "": + print("No file was selected or opened!") + return None + + out = open(filename[0], "w") + + for i in range(self.items.rowCount()): + print(self.items.item(i,TABLE_COL_UPC).text(), file=out, end=", ") + name = self.items.item(i,TABLE_COL_NAME).text() + name = name.replace("\n", " ").replace('"', '""') + print(name, file=out, end=", ") + print(self.items.item(i,TABLE_COL_PRICE).text().replace("\n", " "), + file=out) + out.close() + + self.statusBar().showMessage("Saved to: " + filename[0]) + + def save_database(self): + print("Not implemented!") + def scale_update(self): self.ui.photo.setPixmap(self.pixmap.scaled( int(self.scale * self.pixmap.width()), int(self.scale * self.pixmap.height()))) print("Photo scale set to:", self.scale) + self.statusBar().showMessage("Photo scale set to: " + str(self.scale)) + def image_to_text(self, cropped_image): gray = cv2.cvtColor(cropped_image, cv2.COLOR_BGR2GRAY) gray = cv2.medianBlur(gray, 1) crop = Image.fromarray(gray) - text = pytesseract.image_to_string(crop, lang=self.model) + text = pytesseract.image_to_string(crop, lang=self.model).strip() print("Text selected:", text) return text + def get_row(self): + button = self.sender() + row = self.itemsTable.indexAt(button.pos()).row() + upc = self.items.item(row, TABLE_COL_UPC).text() + print("Started getting name of UPC:", upc) + name = get_name_from_upc(upc, max_tries=3) + print("Got name", name, "from UPC:", upc) + item = QStandardItem(name) + item.setTextAlignment(Qt.AlignmentFlag.AlignCenter) + self.items.setItem(row, TABLE_COL_NAME, item) + self.itemsTable.resizeColumnToContents(TABLE_COL_NAME) + self.itemsTable.resizeRowToContents(row) + + def get_all(self): + for i in range(1, self.items.rowCount()): + + if self.items.item(i, TABLE_COL_NAME) is None or self.items.item(i, TABLE_COL_NAME).text() == "N/A": + upc = self.items.item(i, TABLE_COL_UPC).text() + print("Started getting name of UPC:", upc) + self.statusBar().showMessage("Started getting name of UPC: " + upc) + name = get_name_from_upc(upc, wait_interval=1, max_tries=30) + print("Got name", name, "from UPC:", upc) + item = QStandardItem(name) + item.setTextAlignment(Qt.AlignmentFlag.AlignCenter) + self.items.setItem(i, TABLE_COL_NAME, item) + self.itemsTable.resizeColumnToContents(TABLE_COL_NAME) + self.itemsTable.resizeRowsToContents() + + def del_row(self): + button = self.sender() + index = self.itemsTable.indexAt(button.pos()) + self.items.removeRow(index.row()) + print("Removed row", index.row()) + + def populate_buttons(self): + del_button = QPushButton("Del") + self.itemsTable.setIndexWidget( + self.items.index(self.items.rowCount() - 1, TABLE_COL_DEL), + del_button) + del_button.clicked.connect(self.del_row) + + get_button = QPushButton("Get") + self.itemsTable.setIndexWidget( + self.items.index(self.items.rowCount() - 1, TABLE_COL_GET), + get_button) + get_button.clicked.connect(self.get_row) + + def add_upc(self, text: str): + if text == "": + print("No text was recognized, will not add a new item!") + return None + self.statusBar().showMessage("Text selected: " + text) + item = QStandardItem(text) + item.setTextAlignment(Qt.AlignmentFlag.AlignCenter) + self.items.setItem(self.items.rowCount(), TABLE_COL_UPC, item) + + item = QStandardItem("N/A") + item.setTextAlignment(Qt.AlignmentFlag.AlignCenter) + self.items.setItem(self.items.rowCount() - 1, TABLE_COL_NAME, item) + item = QStandardItem("N/A") + item.setTextAlignment(Qt.AlignmentFlag.AlignCenter) + self.items.setItem(self.items.rowCount() - 1, TABLE_COL_PRICE, item) + + self.populate_buttons() + self.itemsTable.resizeColumnToContents(TABLE_COL_UPC) + return None + def eventFilter(self, source, event): # Handling the selection box if event.type() == QEvent.Type.MouseButtonPress and source is self.ui.photo: @@ -132,8 +277,9 @@ class coin_app(QtWidgets.QMainWindow): if width >= WIDTH_MIN and height >= HEIGHT_MIN and self.image is not None: print("Cropped image:", self.x1, self.y1, self.x2, self.y2) self.crop = self.image[self.y1:self.y2, self.x1:self.x2] - cv2.imwrite("cropped.png", self.crop) + # cv2.imwrite("cropped.png", self.crop) self.text = self.image_to_text(self.crop) + self.add_upc(self.text) ##self.ui.textEdit() return True diff --git a/upc.py b/upc.py index 18a0939..c73488d 100644 --- a/upc.py +++ b/upc.py @@ -4,27 +4,37 @@ from seleniumbase import Driver driver = Driver(uc=True) driver.implicitly_wait(5) -def get_name_from_upc(upc: str, wait_interval=10) -> str: +def get_name_from_upc(upc: str, wait_interval=3, max_tries=10) -> str: url = "https://stocktrack.ca/wm/index.php?s=wm&upc=" + upc driver.get(url) + upc = upc.lstrip("0") + + print("Removed leading 0's:", upc) + # Change the cookies here to match that of yours when you lookup # any item on stocktrack.ca to bypass the Cloudflare checks driver.add_cookie({ "name": "PHPSESSID", - "value": "835ttgvbgncf4uq82dpagmk688", + "value": "9n1ic479r1bteiv758gm9hk65p", "path": "/", "domain": "stocktrack.ca" }) driver.add_cookie({ "name": "cf_chl_3", - "value": "1fad2e046cb9148", + "value": "1d706187484b25c", "path": "/", "domain": "stocktrack.ca" }) driver.add_cookie({ "name": "cf_clearance", - "value": "LUZoWRwn3n1wBEv7bY1XinLQbX0ZTAnggLIfkoy2S60-1708745781-1.0-AXqZtJSIQrq/6/H03NLlHwbpLA1tJmPjgjMJhzaczZjGabhR6ow16wMK5pTwilZiqK7C7fdcYvM0qsbAARkLGEU=", + "value": "Wp8tAMUKLdS3a4Y9AT09BIlZKx4x120uC1QzBQTUluQ-1710517775-1.0.1.1-hMEP8oeggZHBkylkwkQfi2p57H6zUUvGG40d_M4vGqOqg2Zh7wZsg6KrGl3XkDUn3mXAqyZrTqlQfd5pgHCZWQ", + "path": "/", + "domain": "stocktrack.ca" + }) + driver.add_cookie({ + "name": "fp", + "value": "26f4acb9b23415f921bba6977b68d55f", "path": "/", "domain": "stocktrack.ca" }) @@ -32,25 +42,29 @@ def get_name_from_upc(upc: str, wait_interval=10) -> str: driver.refresh() name = "" str_s = "target=\"_blank\">" - str_t = "
UPC: " + upc + "
" + str_t = "
UPC: " + upc + "
" + str_tt = "
SKU:" times = 0 - while True: + while times < max_tries: if __debug__: print("Iteration No. ", times) - times = times + 1 + + times = times + 1 time.sleep(wait_interval) page = str(driver.execute_script( "return document.getElementsByTagName('html')[0].innerHTML")) t = page.find(str_t) s = page.rfind(str_s, 0, t) + tt = page.rfind(str_tt, 0, t) if __debug__: - print(page[s + len(str_s) : t - 1]) + p = open("page.html", "w") + print(page, file=p) if t == -1 or s == -1: continue else: - name = page[s + len(str_s) : t - 1] + name = page[s + len(str_s) : tt] break - return name + return name.replace(" ", "\n")