setiastrosuitepro 1.6.7__py3-none-any.whl

setiastro/saspro/graxpert.py

@@ -0,0 +1,638 @@
+# pro/graxpert.py
+from __future__ import annotations
+import os
+import platform
+import shutil
+import tempfile
+import stat
+import glob
+import subprocess
+import numpy as np
+
+from PyQt6.QtCore import QThread, pyqtSignal, Qt
+from PyQt6.QtWidgets import (
+    QDialog, QVBoxLayout, QTextEdit, QPushButton, QFileDialog,
+    QMessageBox, QInputDialog, QFormLayout, QDialogButtonBox, QDoubleSpinBox,
+    QRadioButton, QLabel, QComboBox, QCheckBox, QWidget
+)
+from setiastro.saspro.config import Config
+
+# Prefer the exact loader you used in SASv2
+try:
+    # adjust this import path if your loader lives elsewhere
+    from setiastro.saspro.legacy.image_manager import load_image as _legacy_load_image
+except Exception:
+    _legacy_load_image = None
+
+
+class GraXpertOperationDialog(QDialog):
+    """Choose operation + parameter (smoothing or strength) + (optional) denoise model."""
+    def __init__(self, parent=None):
+        super().__init__(parent)
+
+        self.setWindowTitle("GraXpert")
+        self.setWindowFlag(Qt.WindowType.Window, True)
+        self.setWindowModality(Qt.WindowModality.NonModal)
+        self.setModal(False)
+
+        root = QVBoxLayout(self)
+
+        # radios
+        self.rb_bg  = QRadioButton("Remove gradient")
+        self.rb_dn  = QRadioButton("Denoise")
+        self.rb_bg.setChecked(True)
+
+        # param widgets
+        self.spin = QDoubleSpinBox()
+        self.spin.setRange(0.0, 1.0)
+        self.spin.setDecimals(2)
+        self.spin.setSingleStep(0.01)
+        self.spin.setValue(0.10)   # default for smoothing
+
+        # dynamic label
+        self.param_label = QLabel("Smoothing (0–1):")
+
+        # denoise model (optional)
+        self.model_label = QLabel("Denoise model:")
+        self.model_combo = QComboBox()
+        # Index 0 = auto/latest (empty payload → omit flag)
+        self.model_combo.addItem("Latest (auto)", "")     # omit -ai_version
+        for v in ["3.0.2", "3.0.1", "3.0.0", "2.0.0", "1.1.0", "1.0.0"]:
+            self.model_combo.addItem(v, v)
+
+        # GPU toggle (persists via QSettings if available)
+        self.cb_gpu = QCheckBox("Use GPU acceleration")
+        use_gpu_default = True
+        try:
+            settings = getattr(parent, "settings", None)
+            if settings is not None:
+                use_gpu_default = settings.value("graxpert/use_gpu", True, type=bool)
+        except Exception:
+            pass
+        self.cb_gpu.setChecked(bool(use_gpu_default))
+
+
+        # layout
+        form = QFormLayout()
+        form.addRow(self.rb_bg)
+        form.addRow(self.rb_dn)
+        form.addRow(self.param_label, self.spin)
+        form.addRow(self.model_label, self.model_combo)
+        form.addRow(self.cb_gpu)
+        root.addLayout(form)
+
+        # switch label/defaults and enable/disable model picker
+        def _to_bg():
+            self.param_label.setText("Smoothing (0–1):")
+            # If param was the denoise default, flip back to smoothing default
+            self.spin.setValue(0.10 if abs(self.spin.value() - 0.50) < 1e-6 else self.spin.value())
+            self.model_label.setEnabled(False)
+            self.model_combo.setEnabled(False)
+
+        def _to_dn():
+            self.param_label.setText("Strength (0–1):")
+            # If param was the smoothing default, flip to denoise default
+            self.spin.setValue(0.50 if abs(self.spin.value() - 0.10) < 1e-6 else self.spin.value())
+            self.model_label.setEnabled(True)
+            self.model_combo.setEnabled(True)
+
+        self.rb_bg.toggled.connect(lambda checked: _to_bg() if checked else None)
+        self.rb_dn.toggled.connect(lambda checked: _to_dn() if checked else None)
+
+        # initialize state
+        _to_bg()
+
+        btns = QDialogButtonBox(QDialogButtonBox.StandardButton.Ok | QDialogButtonBox.StandardButton.Cancel)
+        btns.accepted.connect(self.accept)
+        btns.rejected.connect(self.reject)
+        root.addWidget(btns)
+
+    def result(self):
+        op = "background" if self.rb_bg.isChecked() else "denoise"
+        val = float(self.spin.value())
+        ai_version = self.model_combo.currentData() if not self.rb_bg.isChecked() else ""
+        use_gpu = self.cb_gpu.isChecked()
+        return op, val, (ai_version or None), use_gpu
+
+def _build_graxpert_cmd(
+    exe: str,
+    operation: str,
+    input_path: str,
+    *,
+    smoothing: float | None = None,
+    strength: float | None = None,
+    ai_version: str | None = None,
+    gpu: bool = True,
+    batch_size: int | None = None
+) -> list[str]:
+    op = "denoising" if operation == "denoise" else "background-extraction"
+    cmd = [exe, "-cmd", op, input_path, "-cli", "-gpu", "true" if gpu else "false"]
+    if op == "denoising":
+        if strength is not None:
+            cmd += ["-strength", f"{strength:.2f}"]
+        if batch_size is not None:
+            cmd += ["-batch_size", str(int(batch_size))]
+        # Only include if user chose a specific model
+        if ai_version:
+            cmd += ["-ai_version", ai_version]
+    else:
+        if smoothing is not None:
+            cmd += ["-smoothing", f"{smoothing:.2f}"]
+    return cmd
+
+# ---------- Public entry point (call this from your main window) ----------
+def remove_gradient_with_graxpert(main_window, target_doc=None):
+    """
+    Exactly mirror SASv2 flow:
+      - write input_image.tif
+      - run GraXpert
+      - read input_image_GraXpert.{fits|tif|tiff|png} using legacy loader
+      - apply to target document
+    """
+    if getattr(main_window, "_graxpert_headless_running", False):
+        return
+    if getattr(main_window, "_graxpert_guard", False):   # cool-down guard
+        return
+
+    # 1) pick the document: explicit > fallback
+    doc = target_doc
+
+    if doc is None:
+        # Backwards compatibility: fall back to _active_doc
+        doc = getattr(main_window, "_active_doc", None)
+        if callable(doc):
+            doc = doc()
+
+    if doc is None and hasattr(main_window, "mdi"):
+        # Extra fallback: resolve from active subwindow if possible
+        try:
+            sw = main_window.mdi.activeSubWindow()
+            if sw is not None:
+                view = sw.widget()
+                doc = getattr(view, "document", None)
+        except Exception:
+            pass
+
+    if doc is None or getattr(doc, "image", None) is None:
+        QMessageBox.warning(
+            main_window,
+            "No Image",
+            "Please load an image before removing the gradient."
+        )
+        return
+
+    # 2) smoothing/denoise prompt
+    op_dlg = GraXpertOperationDialog(main_window)
+    if op_dlg.exec() != QDialog.DialogCode.Accepted:
+        return
+    operation, param, ai_version, use_gpu = op_dlg.result()
+
+    # 3) resolve GraXpert executable
+    exe = _resolve_graxpert_exec(main_window)
+    if not exe:
+        return
+
+    # Persist the checkbox choice for next time
+    try:
+        if hasattr(main_window, "settings"):
+            main_window.settings.setValue("graxpert/use_gpu", bool(use_gpu))
+    except Exception:
+        pass
+
+    # 🔁 NEW: record this as a replayable headless-style command
+    try:
+        remember = getattr(main_window, "remember_last_headless_command", None)
+        if remember is None:
+            remember = getattr(main_window, "_remember_last_headless_command", None)
+
+        if callable(remember):
+            preset = {
+                "op": operation,           # "background" or "denoise"
+                "gpu": bool(use_gpu),
+            }
+            if operation == "background":
+                preset["smoothing"] = float(param)
+                desc = "GraXpert Gradient Removal"
+            else:
+                preset["strength"] = float(param)
+                if ai_version:
+                    preset["ai_version"] = ai_version
+                desc = "GraXpert Denoise"
+
+            remember("graxpert", preset, description=desc)
+
+            # Optional log entry, if you want:
+            if hasattr(main_window, "_log"):
+                try:
+                    main_window._log(
+                        f"[Replay] GraXpert preset stored from dialog: "
+                        f"op={operation}, keys={list(preset.keys())}"
+                    )
+                except Exception:
+                    pass
+    except Exception:
+        # Don't let replay bookkeeping break GraXpert itself
+        pass
+
+    # 4) write input to a temp working dir but KEEP THE SAME BASENAMES as v2
+    workdir = tempfile.mkdtemp(prefix="saspro_graxpert_")
+    input_basename = "input_image"
+    input_path = os.path.join(workdir, f"{input_basename}.tif")
+    try:
+        _write_tiff_float32(doc.image, input_path)
+    except Exception as e:
+        QMessageBox.critical(main_window, "GraXpert", f"Failed to write temporary input:\n{e}")
+        shutil.rmtree(workdir, ignore_errors=True)
+        return
+
+    # 5) build the exact v2 command (now with optional ai_version for denoise)
+    command = _build_graxpert_cmd(
+        exe,
+        operation,
+        input_path,
+        smoothing=param if operation == "background" else None,
+        strength=param if operation == "denoise" else None,
+        ai_version=ai_version if operation == "denoise" else None,
+        gpu=bool(use_gpu),
+        batch_size=(4 if use_gpu else 1)
+    )
+
+    # Label + metadata for history/undo
+    op_label = "GraXpert Denoise" if operation == "denoise" else "GraXpert Gradient Removal"
+    meta_extras = {
+        "graxpert_operation": operation,  # "denoise" | "background"
+        "graxpert_param": float(param),
+        "graxpert_ai_version": (ai_version or "latest") if operation == "denoise" else None,
+        "graxpert_gpu": bool(use_gpu),
+    }
+
+    # 6) run and wait with a small log dialog
+    output_basename = f"{input_basename}_GraXpert"
+    _run_graxpert_command(
+        main_window,
+        command,
+        output_basename,
+        workdir,
+        target_doc=doc,
+        op_label=op_label,
+        meta_extras=meta_extras,
+    )
+
+
+# ---------- helpers ----------
+def _resolve_graxpert_exec(main_window) -> str | None:
+    # prefer QSettings if available (all OS)
+    path = None
+    if hasattr(main_window, "settings"):
+        try:
+            path = main_window.settings.value("paths/graxpert", type=str)
+        except Exception:
+            path = None
+    if path and os.path.exists(path):
+        _ensure_exec_bit(path)
+        return path
+
+    sysname = platform.system()
+    default = Config.get_graxpert_default_path()
+    
+    if sysname == "Windows":
+        # rely on PATH (like v2) or default
+        return default if default else "GraXpert.exe"
+        
+    if sysname == "Darwin":
+        if default and os.path.exists(default):
+            _ensure_exec_bit(default)
+            if hasattr(main_window, "settings"):
+                main_window.settings.setValue("paths/graxpert", default)
+            return default
+        return _pick_graxpert_path_and_store(main_window)
+        
+    if sysname == "Linux":
+        # in v2 you asked user and saved; do the same
+        return _pick_graxpert_path_and_store(main_window)
+
+    QMessageBox.critical(main_window, "GraXpert", f"Unsupported operating system: {sysname}")
+    return None
+
+def _pick_graxpert_path_and_store(main_window) -> str | None:
+    path, _ = QFileDialog.getOpenFileName(main_window, "Select GraXpert Executable")
+    if not path:
+        QMessageBox.warning(main_window, "Cancelled", "GraXpert path selection was cancelled.")
+        return None
+    try:
+        _ensure_exec_bit(path)
+    except Exception as e:
+        QMessageBox.critical(main_window, "GraXpert", f"Failed to set execute permissions:\n{e}")
+        return None
+    if hasattr(main_window, "settings"):
+        main_window.settings.setValue("paths/graxpert", path)
+    return path
+
+
+def _ensure_exec_bit(path: str) -> None:
+    if platform.system() == "Windows":
+        return
+    try:
+        st = os.stat(path)
+        os.chmod(path, st.st_mode | stat.S_IXUSR | stat.S_IXGRP | stat.S_IXOTH)
+    except Exception:
+        pass
+
+
+def _write_tiff_float32(image, path: str, *, clip01: bool = True):
+    """
+    Always write a 32-bit floating-point TIFF for GraXpert.
+    - Mono stays 2D; RGB stays HxWx3.
+    - Values are clipped to [0,1] by default to avoid weird HDR ranges.
+    """
+    import numpy as np
+
+    arr = np.asarray(image)
+    if arr.ndim == 3 and arr.shape[2] == 1:
+        arr = arr[..., 0]
+
+    # Convert to float32 in [0,1]
+    if np.issubdtype(arr.dtype, np.floating):
+        a32 = arr.astype(np.float32, copy=False)
+        if clip01:
+            a32 = np.clip(a32, 0.0, 1.0)
+    elif np.issubdtype(arr.dtype, np.integer):
+        # Scale integers to [0,1] float32
+        maxv = np.float32(np.iinfo(arr.dtype).max)
+        a32 = (arr.astype(np.float32) / maxv)
+    else:
+        a32 = arr.astype(np.float32)
+
+    if clip01:
+        a32 = np.clip(a32, 0.0, 1.0)
+
+    # Prefer tifffile to guarantee float32 TIFFs
+    try:
+        import tifffile as tiff
+        # Write a plain, contiguous, uncompressed float32 TIFF
+        # (GraXpert doesn't need ImageJ tags; photometric=minisblack is fine)
+        tiff.imwrite(
+            path,
+            a32,
+            dtype=np.float32,
+            photometric='minisblack' if a32.ndim == 2 else None,
+            planarconfig='contig',
+            compression=None,
+            imagej=False,
+        )
+        return
+    except Exception as e1:
+        pass
+
+    # Fallback: imageio (uses tifffile under the hood in many installs)
+    try:
+        import imageio.v3 as iio
+        iio.imwrite(path, a32.astype(np.float32))
+        return
+    except Exception as e2:
+        raise RuntimeError(
+            "Could not write 32-bit TIFF for GraXpert. "
+            "Please install 'tifffile' or 'imageio'.\n"
+            f"tifffile error: {e1}\nimageio error: {e2}"
+        )
+
+
+
+# ---------- runner + dialog ----------
+class _GraXpertThread(QThread):
+    stdout_signal = pyqtSignal(str)
+    finished_signal = pyqtSignal(int)
+
+    def __init__(self, command: list[str], cwd: str | None = None, parent=None):
+        super().__init__(parent)
+        self.command = command
+        self.cwd = cwd
+
+    def run(self):
+        env = os.environ.copy()
+        for k in ("PYTHONHOME", "PYTHONPATH", "DYLD_LIBRARY_PATH",
+                  "DYLD_FALLBACK_LIBRARY_PATH", "PYTHONEXECUTABLE"):
+            env.pop(k, None)
+        try:
+            p = subprocess.Popen(
+                self.command,
+                cwd=self.cwd,
+                stdout=subprocess.PIPE,
+                stderr=subprocess.STDOUT,  # merge; avoids ResourceWarning + deadlocks
+                text=True,
+                universal_newlines=True,
+                env=env,
+                start_new_session=True
+            )
+            for line in iter(p.stdout.readline, ""):
+                if not line:
+                    break
+                self.stdout_signal.emit(line.rstrip())
+            try:
+                p.stdout.close()
+            except Exception:
+                pass
+            rc = p.wait()
+        except Exception as e:
+            self.stdout_signal.emit(str(e))
+            rc = -1
+        self.finished_signal.emit(rc)
+
+
+def _run_graxpert_command(parent, command: list[str], output_basename: str,
+                          working_dir: str, target_doc,
+                          op_label: str | None = None,
+                          meta_extras: dict | None = None):
+    dlg = QDialog(parent)
+    dlg.setWindowTitle("GraXpert Progress")
+    dlg.setMinimumSize(600, 420)
+    lay = QVBoxLayout(dlg)
+    log = QTextEdit(readOnly=True)
+    lay.addWidget(log)
+    btn_cancel = QPushButton("Cancel")
+    lay.addWidget(btn_cancel)
+
+    thr = _GraXpertThread(command, cwd=working_dir)
+    thr.stdout_signal.connect(lambda s: log.append(s))
+    thr.finished_signal.connect(
+        lambda code: _on_graxpert_finished(
+            parent,
+            code,
+            output_basename,
+            working_dir,
+            target_doc,
+            dlg,
+            op_label,
+            meta_extras,
+        )
+    )
+    btn_cancel.clicked.connect(thr.terminate)
+
+    thr.start()
+    dlg.exec()
+
+
+
+# ---------- finish: import EXACT base like v2, via legacy loader ----------
+def _persist_output_file(src_path: str) -> str | None:
+    """Optional: move/copy GraXpert output to an app cache we control."""
+    try:
+        from PyQt6.QtCore import QStandardPaths
+        cache_root = QStandardPaths.writableLocation(QStandardPaths.StandardLocation.CacheLocation)
+    except Exception:
+        cache_root = None
+    try:
+        base = os.path.join(cache_root or os.path.expanduser("~/.saspro_cache"), "graxpert")
+        os.makedirs(base, exist_ok=True)
+        dst = os.path.join(base, os.path.basename(src_path))
+        # prefer move (cheaper); fall back to copy if cross-device issues
+        try:
+            shutil.move(src_path, dst)
+        except Exception:
+            shutil.copy2(src_path, dst)
+        return dst
+    except Exception:
+        return None
+
+
+def _on_graxpert_finished(parent,
+                          return_code: int,
+                          output_basename: str,
+                          working_dir: str,
+                          target_doc,
+                          dlg,
+                          op_label: str | None = None,
+                          meta_extras: dict | None = None):
+    try:
+        dlg.close()
+    except Exception:
+        pass
+
+    if return_code != 0:
+        QMessageBox.critical(parent, "GraXpert", "GraXpert process failed.")
+        shutil.rmtree(working_dir, ignore_errors=True)
+        return
+
+    # 1) find output file in the temp working dir
+    output_file = _pick_exact_output(working_dir, output_basename)
+    if not output_file:
+        QMessageBox.critical(parent, "GraXpert", "GraXpert output file not found.")
+        shutil.rmtree(working_dir, ignore_errors=True)
+        return
+
+    # 2) read pixels (we *do not* want its header to replace ours)
+    arr, header = None, None
+    if _legacy_load_image is not None:
+        try:
+            out = _legacy_load_image(output_file, return_metadata=True)
+            if out and len(out) == 5:
+                a, h, bit_depth, is_mono, out_meta = out
+            else:
+                a, h, bit_depth, is_mono = out
+                out_meta = {}
+            arr, header = a, h
+        except Exception:
+            arr = None
+            header = None
+            bit_depth = "32-bit floating point"
+            is_mono = None
+            out_meta = {}
+    else:
+        out_meta = {}
+        bit_depth = "32-bit floating point"
+        is_mono = None
+    # Decide how it appears in history/undo
+    step_label = op_label or "GraXpert Gradient Removal"
+
+    # 3) base metadata: START FROM EXISTING DOC METADATA
+    base_meta = dict(getattr(target_doc, "metadata", {}) or {})
+
+    # Keep original_header / wcs_header from the doc.
+    # If you want to keep GraXpert's header for debugging, store separately:
+    from astropy.io import fits as _fits_mod
+    if header is not None and isinstance(header, _fits_mod.Header):
+        base_meta.setdefault("graxpert_header", header)
+
+    # Basic fields we do want to update
+    base_meta["step_name"]   = step_label
+    base_meta["description"] = step_label
+    base_meta["bit_depth"]   = "32-bit floating point"
+    if is_mono is not None:
+        base_meta["is_mono"] = bool(is_mono)
+
+    # Copy over any interesting fields from GraXpert's own metadata that are SAFE
+    # but do NOT overwrite original_header / wcs_header.
+    for k, v in (out_meta or {}).items():
+        if k in ("original_header", "fits_header", "wcs_header"):
+            continue
+        base_meta.setdefault(k, v)
+
+    # Operation-specific extras
+    if meta_extras:
+        # these are non-header fields like graxpert_operation, etc.
+        base_meta.update(meta_extras)
+
+    # 4) apply to the target doc
+    try:
+        target_doc.apply_edit(
+            arr.astype(np.float32, copy=False),
+            metadata=base_meta,
+            step_name=step_label,
+        )
+    except Exception as e:
+        QMessageBox.critical(parent, "GraXpert", f"Failed to apply result:\n{e}")
+    finally:
+        shutil.rmtree(working_dir, ignore_errors=True)
+
+
+def _pick_exact_output(folder: str, base: str) -> str | None:
+    # exact filenames only, like v2 did
+    exts = ("fits", "tif", "tiff", "png")
+    for ext in exts:
+        p = os.path.join(folder, f"{base}.{ext}")
+        if os.path.exists(p):
+            return p
+        # also try case-variants just in case
+        for q in glob.glob(os.path.join(folder, f"{base}.*")):
+            if q.lower().endswith("." + ext):
+                return q
+    return None
+
+
+def _fallback_read_float01(path: str) -> np.ndarray | None:
+    """Basic loader: return float32 in [0,1], mono or RGB, without being too clever."""
+    try:
+        import imageio.v3 as iio
+        arr = iio.imread(path)
+    except Exception:
+        try:
+            import tifffile as tiff
+            arr = tiff.imread(path)
+        except Exception:
+            try:
+                from astropy.io import fits
+                with fits.open(path, memmap=False) as hdul:
+                    arr = hdul[0].data
+            except Exception:
+                try:
+                    import cv2
+                    arr = cv2.imread(path, cv2.IMREAD_UNCHANGED)
+                    if arr is not None and arr.ndim == 3:
+                        arr = arr[..., ::-1]  # BGR->RGB
+                except Exception:
+                    arr = None
+    if arr is None:
+        return None
+
+    arr = np.asarray(arr)
+    if arr.ndim == 3 and arr.shape[2] == 1:
+        arr = arr[..., 0]
+    if arr.dtype.kind in "ui":
+        scale = 65535.0 if arr.dtype.itemsize >= 2 else 255.0
+        arr = arr.astype(np.float32) / scale
+    else:
+        arr = arr.astype(np.float32, copy=False)
+        mx = float(arr.max()) if arr.size else 1.0
+        if mx > 5.0:
+            arr = arr / mx
+    return arr