pdf-table-extractor/pdf_table_extractor.py

import sys
import traceback
from typing import Dict, List, Optional, Sequence, Tuple, TypeAlias

import PIL.Image
import gi

gi.require_version("Gtk", "3.0")
from gi.repository import GLib, Gio, Gtk, GObject

gi.require_foreign("cairo")
import cairo

import fitz

Coords: TypeAlias = Tuple[float, float]


class Selection:
    def __init__(self, bounds: Tuple[Coords, Coords], columns: Optional[Sequence[float]] = None):
        self.bounds = bounds
        self.columns: list[float] = list(columns or [])

class Page:
    def __init__(self, index: int, raw: fitz.Page):
        self.index = index
        self.raw = raw
        self.selections: list[Selection] = []

class Document:
    def __init__(self, filename: str):
        self.filename = filename
        self.raw = fitz.Document(filename)
        self.pages = []

        for i, p in enumerate(self.raw.pages()):  # type: ignore
            page = Page(index=i, raw=p)
            self.pages.append(page)


class PdfPage(Gtk.DrawingArea):
    def __init__(self, page, *args, **kwargs):
        super().__init__(*args, **kwargs)

        self.page: Page = page
        pix: fitz.Pixmap = self.page.raw.get_pixmap(dpi=96)  # type: ignore
        self.set_size_request(pix.width, pix.height)
        self.connect("draw", self.on_draw, {})

        if self.page.index == 4:
            self.page.selections.append(Selection(
                ((0.1, 0.18), (0.9, 0.72)),
                columns=[0.08, 0.24, 0.34, 0.42, 0.51, 0.59, 0.67, 0.73, 0.91]
            ))

    def on_draw(self, widget: Gtk.DrawingArea, cr: cairo.Context, data: GObject.GPointer):
        width = widget.get_allocated_width()
        height = widget.get_allocated_height()

        sctx = widget.get_style_context()
        Gtk.render_background(sctx, cr, 0, 0, width, height)

        pix: fitz.Pixmap = self.page.raw.get_pixmap(dpi=96)  # type: ignore
        img = PIL.Image.frombytes("RGBA" if pix.alpha else "RGB", [pix.width, pix.height], pix.samples)
        img.putalpha(1)
        img = PIL.Image.merge("RGBA", (lambda r, g, b, a: (b, g, r, a))(*img.split()))  # type: ignore
        mv: memoryview = memoryview(bytearray(img.tobytes()))
        ims = cairo.ImageSurface.create_for_data(mv, cairo.Format.RGB24, pix.width, pix.height)
        cr.set_source_surface(ims, 0, 0)
        cr.paint()

        for sel in self.page.selections:
            sel_x1 = sel.bounds[0][0]*pix.width
            sel_y1 = sel.bounds[0][1]*pix.height
            sel_x2 = sel.bounds[1][0]*pix.width
            sel_y2 = sel.bounds[1][1]*pix.height

            # Base settings
            cr.set_line_cap(cairo.LINE_CAP_BUTT)
            cr.set_line_width(2)

            # Columns (draw first - below selection)
            cr.set_dash([5])
            cr.set_source_rgba(1, 0, 0)
            for col in sel.columns:
                col_x = sel_x1 + (sel_x2 - sel_x1) * col
                cr.move_to(col_x, sel_y1)
                cr.line_to(col_x, sel_y2)
                cr.stroke()

            # Selection
            cr.rectangle(sel_x1, sel_y1, sel_x2 - sel_x1, sel_y2 - sel_y1)

            # White part of the pattern
            cr.set_source_rgba(1, 1, 1)
            cr.set_dash([5], 5)
            cr.stroke_preserve() # important preserve - reuse rectangle

            # Black part of the pattern
            cr.set_source_rgba(0, 0, 0)
            cr.set_dash([5])
            cr.stroke()


@Gtk.Template.from_file("MainWindow.glade")
class MainWindow(Gtk.ApplicationWindow):
    __gtype_name__ = "main_window"

    open_button: Gtk.Button = Gtk.Template.Child()  # type: ignore
    header_bar: Gtk.HeaderBar = Gtk.Template.Child()  # type: ignore
    main_paned: Gtk.Paned = Gtk.Template.Child()  # type: ignore

    pdf_list_box: Gtk.ListBox = Gtk.Template.Child()  # type: ignore

    pdfFileFilter: Gtk.FileFilter = Gtk.Template.Child()  # type: ignore

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)

        self.app: Application = self.get_application()  # type: ignore
        assert self.app is not None

        self.app.connect("notify::document", self.on_document_updated)

    # @Gtk.Template.Callback()
    # def example_button_released_cb(self, widget: Gtk.Button, **kwargs):
    #    assert self.example_button == widget
    #    print(widget.get_label())
    #    widget.set_label("woah")

    @Gtk.Template.Callback()
    def on_open_button_clicked(self, widget, *args, **kwargs):
        dialog = Gtk.FileChooserDialog(
            title="Choose PDF File to open",
            transient_for=self,  # equivalent to parent=
            action=Gtk.FileChooserAction.OPEN,
            filter=self.pdfFileFilter,
            modal=True,
        )
        dialog.add_button("Cancel", Gtk.ResponseType.CANCEL)
        dialog.add_button("Open", Gtk.ResponseType.ACCEPT)

        response = dialog.run()  # type: ignore
        if response == Gtk.ResponseType.ACCEPT:
            filename: str = dialog.get_filename()  # type: ignore
            try:
                self.app.set_property("document", Document(filename))
            except Exception as e:
                message_dialog = Gtk.MessageDialog(
                    title="An error has occured.",
                    transient_for=self,
                    modal=True,
                    message_type=Gtk.MessageType.ERROR,
                    text=repr(e),
                    secondary_text=traceback.format_exc(),
                    buttons=Gtk.ButtonsType.OK,
                )
                message_dialog.run()  # type: ignore
                message_dialog.destroy()
        dialog.destroy()

    # Note: this won't run unless a new document object is put into place
    #       editing an existing one won't trigger it
    def on_document_updated(self, recvobj, gparamstring):
        document: Document = self.app.get_property("document")
        self.header_bar.set_title(document.filename.split("/")[-1])
        self.header_bar.set_subtitle(document.filename)

        for child in self.pdf_list_box.get_children():
            if type(child) is Gtk.ListBoxRow:
                child.destroy()

        for i in document.pages:
            self.pdf_list_box.add(PdfPage(i))

        self.pdf_list_box.show_all()

    @Gtk.Template.Callback()
    def on_open_button_small_clicked(self, widget, **kwargs):
        TEST_FILENAME = "/home/luna/Documents/Resources/Praca Licencjacka/sources/2018_Torres-Benitez_Metabolomic analysis Parmotrema.pdf"
        try:
            self.app.set_property("document", Document(TEST_FILENAME))
        except Exception as e:
            message_dialog = Gtk.MessageDialog(
                title="An error has occured.",
                transient_for=self,
                modal=True,
                message_type=Gtk.MessageType.ERROR,
                text=repr(e),
                secondary_text=traceback.format_exc(),
                buttons=Gtk.ButtonsType.OK,
            )
            message_dialog.run()  # type: ignore
            message_dialog.destroy()


class Application(Gtk.Application):
    document = GObject.Property(type=GObject.TYPE_PYOBJECT, flags=GObject.ParamFlags.READWRITE)

    def __init__(self, *args, **kwargs):
        super().__init__(
            *args,
            application_id="zone.lunareclipse.pdf_table_extractor",
            flags=Gio.ApplicationFlags.FLAGS_NONE,
            # flags=Gio.ApplicationFlags.HANDLES_COMMAND_LINE, # TODO
            **kwargs
        )
        self.window = None

    def do_activate(self):
        self.window = self.window or MainWindow(application=self)
        self.window.show_all()  # type: ignore


if __name__ == "__main__":
    app = Application()
    app.run(sys.argv)