scitex 2.7.0__py3-none-any.whl → 2.7.3__py3-none-any.whl

scitex/io/_save.py

@@ -161,7 +161,7 @@ def save(
     auto_crop: bool = True,
     crop_margin_mm: float = 1.0,
     metadata_extra: dict = None,
-    json_schema: str = "recipe",
+    json_schema: str = "editable",
     **kwargs,
 ) -> None:
     """
@@ -206,9 +206,10 @@ def save(
         Default is None.
     json_schema : str, optional
         Schema type for JSON metadata output. Options:
-        - "recipe": Minimal schema with method calls + data refs (default, ~95% smaller)
+        - "editable": Schema v0.3.0 with element geometry for interactive editing (default)
+        - "recipe": Minimal schema with method calls + data refs
         - "verbose": Full schema with all artist details
-        Default is "recipe".
+        Default is "editable".
     **kwargs
         Additional keyword arguments to pass to the underlying save function of the specific format.
 
@@ -218,9 +219,16 @@ def save(
 
     Notes
     -----
-    Supported formats include CSV, NPY, PKL, JOBLIB, PNG, HTML, TIFF, MP4, YAML, JSON, HDF5, PTH, MAT, and CBM.
+    Supported formats include CSV, NPY, PKL, JOBLIB, PNG, HTML, TIFF, MP4, YAML, JSON, HDF5, PTH, MAT, CBM,
+    and SciTeX bundles (.figz, .pltz, .statsz).
     The function dynamically selects the appropriate saving mechanism based on the file extension.
 
+    Bundle Formats:
+    - .figz: Publication figure bundle (panels dict). Default: ZIP archive.
+    - .pltz: Plot bundle (matplotlib figure). Default: directory bundle.
+    - .statsz: Statistics bundle (comparisons list). Default: directory bundle.
+    - Use .d suffix (e.g., "Figure1.figz.d") to force directory format for .figz.
+
     Examples
     --------
     >>> import scitex
@@ -495,6 +503,482 @@ def _symlink_to(spath_final, symlink_to, verbose):
             logger.success(f"Symlinked: {spath_final} -> {symlink_to_full}")
 
 
+def _save_pltz_bundle(obj, spath, as_zip=False, data=None, layered=True, **kwargs):
+    """Save a matplotlib figure as a .pltz bundle.
+
+    Bundle structure v2.0 (layered - default):
+        plot.pltz.d/
+            spec.json           # Semantic: WHAT to plot (canonical)
+            style.json          # Appearance: HOW it looks (canonical)
+            data.csv            # Raw data (immutable)
+            exports/            # PNG, SVG, hitmap
+            cache/              # geometry_px.json, render_manifest.json
+
+    Bundle structure v1.0 (legacy):
+        plot.json  - specification (axes, styles, theme, etc.)
+        plot.csv   - raw data (immutable)
+        plot.png   - raster export (required)
+        plot.svg   - vector export (optional)
+        plot.pdf   - publication export (optional)
+
+    Parameters
+    ----------
+    obj : matplotlib.figure.Figure
+        The figure to save.
+    spath : str or Path
+        Output path (e.g., "plot.pltz.d" or "plot.pltz").
+    as_zip : bool
+        If True, save as ZIP archive.
+    data : pandas.DataFrame, optional
+        Data to embed in the bundle as plot.csv.
+    layered : bool
+        If True (default), use new layered format (spec/style/geometry).
+        If False, use legacy single JSON format.
+    **kwargs
+        Additional arguments passed to savefig.
+    """
+    from pathlib import Path
+    import tempfile
+    import json
+    import numpy as np
+    from ._bundle import save_bundle, BundleType
+
+    p = Path(spath)
+
+    # Extract basename from path (e.g., "myplot.pltz" -> "myplot", "myplot.pltz.d" -> "myplot")
+    basename = p.stem  # e.g., "myplot.pltz" or "myplot"
+    if basename.endswith('.pltz'):
+        basename = basename[:-5]  # Remove .pltz suffix
+    elif basename.endswith('.d'):
+        # Handle myplot.pltz.d -> myplot.pltz -> myplot
+        basename = Path(basename).stem
+        if basename.endswith('.pltz'):
+            basename = basename[:-5]
+
+    # Extract figure from various matplotlib object types
+    import matplotlib.figure
+    fig = obj
+    if hasattr(obj, 'figure'):
+        fig = obj.figure
+    elif hasattr(obj, 'fig'):
+        fig = obj.fig
+
+    if not isinstance(fig, matplotlib.figure.Figure):
+        raise TypeError(f"Expected matplotlib Figure, got {type(obj).__name__}")
+
+    dpi = kwargs.pop('dpi', 300)
+
+    # === Always use layered format ===
+    from scitex.plt.io import save_layered_pltz_bundle
+    import shutil
+    import tempfile
+
+    # Determine bundle directory path
+    if as_zip:
+        # For ZIP: save to temp dir, then compress
+        temp_dir = Path(tempfile.mkdtemp())
+        bundle_dir = temp_dir / f"{basename}.pltz.d"
+        zip_path = p if not str(p).endswith('.d') else Path(str(p)[:-2])
+    else:
+        # For directory: save directly
+        bundle_dir = p if str(p).endswith('.d') else Path(str(p) + '.d')
+
+    # Get CSV data from figure if not provided
+    csv_df = data
+    if csv_df is None:
+        csv_source = _get_figure_with_data(obj)
+        if csv_source is not None and hasattr(csv_source, 'export_as_csv'):
+            try:
+                csv_df = csv_source.export_as_csv()
+            except Exception:
+                pass
+
+    save_layered_pltz_bundle(
+        fig=fig,
+        bundle_dir=bundle_dir,
+        basename=basename,
+        dpi=dpi,
+        csv_df=csv_df,
+    )
+
+    # Compress to ZIP if requested
+    if as_zip:
+        from ._bundle import pack_bundle
+        pack_bundle(bundle_dir, zip_path)
+        shutil.rmtree(temp_dir)  # Clean up temp directory
+
+    return  # Done with layered format
+
+    # === Legacy format below (DEPRECATED - kept for reference) ===
+
+    # Calculate size info
+    fig_width_inch, fig_height_inch = fig.get_size_inches()
+    fig_dpi = fig.get_dpi()
+
+    # Build spec according to contract (using basename for file references)
+    spec = {
+        'schema': {'name': 'scitex.plt.plot', 'version': '1.0.0'},
+        'backend': 'mpl',
+        'data': {
+            'source': f'{basename}.csv',
+            'path': f'{basename}.csv',
+            'hash': None,  # Will be computed after data extraction
+            'columns': [],  # Will be populated after data extraction
+        },
+        'size': {
+            'width_inch': round(fig_width_inch, 2),
+            'height_inch': round(fig_height_inch, 2),
+            'width_mm': round(fig_width_inch * 25.4, 2),
+            'height_mm': round(fig_height_inch * 25.4, 2),
+            'width_px': int(fig_width_inch * dpi),
+            'height_px': int(fig_height_inch * dpi),
+            'dpi': dpi,
+            'crop_margin_mm': 1.0,
+        },
+        'axes': [],
+        'theme': {
+            'mode': 'light',
+            'colors': {
+                'background': 'transparent',
+                'axes_bg': 'white',
+                'text': 'black',
+                'spine': 'black',
+                'tick': 'black',
+            }
+        },
+    }
+
+    # Extract data from plot lines if no data provided
+    extracted_data = {}
+
+    # Extract axes metadata
+    for i, ax in enumerate(fig.axes):
+        # Get axes bounding box in figure coordinates (0-1)
+        bbox = ax.get_position()
+
+        ax_info = {
+            'xlabel': ax.get_xlabel() or None,
+            'ylabel': ax.get_ylabel() or None,
+            'title': ax.get_title() or None,
+            'xlim': [round(v, 2) for v in ax.get_xlim()],
+            'ylim': [round(v, 2) for v in ax.get_ylim()],
+            'plot_type': 'line',  # Default, could be detected
+            # Bounding box in normalized figure coordinates (0-1)
+            'bbox': {
+                'x0': round(bbox.x0, 4),
+                'y0': round(bbox.y0, 4),
+                'x1': round(bbox.x1, 4),
+                'y1': round(bbox.y1, 4),
+                'width': round(bbox.width, 4),
+                'height': round(bbox.height, 4),
+            },
+            # Bounding box in mm
+            'bbox_mm': {
+                'x0': round(bbox.x0 * fig_width_inch * 25.4, 2),
+                'y0': round(bbox.y0 * fig_height_inch * 25.4, 2),
+                'x1': round(bbox.x1 * fig_width_inch * 25.4, 2),
+                'y1': round(bbox.y1 * fig_height_inch * 25.4, 2),
+                'width': round(bbox.width * fig_width_inch * 25.4, 2),
+                'height': round(bbox.height * fig_height_inch * 25.4, 2),
+            },
+            # Bounding box in pixels
+            'bbox_px': {
+                'x0': int(bbox.x0 * fig_width_inch * dpi),
+                'y0': int(bbox.y0 * fig_height_inch * dpi),
+                'x1': int(bbox.x1 * fig_width_inch * dpi),
+                'y1': int(bbox.y1 * fig_height_inch * dpi),
+                'width': int(bbox.width * fig_width_inch * dpi),
+                'height': int(bbox.height * fig_height_inch * dpi),
+            },
+        }
+
+        # SciTeX-specific axis dimensions
+        if hasattr(ax, '_scitex_axes_width_mm'):
+            ax_info['axes_width_mm'] = ax._scitex_axes_width_mm
+        else:
+            ax_info['axes_width_mm'] = round(bbox.width * fig_width_inch * 25.4, 1)
+
+        if hasattr(ax, '_scitex_axes_height_mm'):
+            ax_info['axes_height_mm'] = ax._scitex_axes_height_mm
+        else:
+            ax_info['axes_height_mm'] = round(bbox.height * fig_height_inch * 25.4, 1)
+
+        # Extract line data for CSV and build lines array
+        lines_info = []
+        for j, line in enumerate(ax.get_lines()):
+            label = line.get_label()
+            if label is None or label.startswith('_'):
+                label = f'series_{j}'
+            xdata, ydata = line.get_data()
+            if len(xdata) > 0:
+                col_x = f'{label}_x' if i == 0 else f'ax{i}_{label}_x'
+                col_y = f'{label}_y' if i == 0 else f'ax{i}_{label}_y'
+                extracted_data[col_x] = np.array(xdata)
+                extracted_data[col_y] = np.array(ydata)
+
+                # Get line color (convert RGBA to hex)
+                color = line.get_color()
+                if isinstance(color, (list, tuple)):
+                    import matplotlib.colors as mcolors
+                    color = mcolors.to_hex(color)
+
+                lines_info.append({
+                    'label': label,
+                    'x_col': col_x,
+                    'y_col': col_y,
+                    'color': color,
+                    'linewidth': line.get_linewidth(),
+                })
+
+        if lines_info:
+            ax_info['lines'] = lines_info
+
+        spec['axes'].append(ax_info)
+
+    # Handle theme from figure
+    if hasattr(fig, '_scitex_theme'):
+        theme_mode = fig._scitex_theme
+        spec['theme']['mode'] = theme_mode
+        # Update colors based on theme mode
+        if theme_mode == 'dark':
+            spec['theme']['colors'] = {
+                'background': 'transparent',
+                'axes_bg': 'transparent',
+                'text': '#e8e8e8',
+                'spine': '#e8e8e8',
+                'tick': '#e8e8e8',
+            }
+            # Re-apply theme colors to ensure legends and other elements get the correct colors
+            from scitex.plt.utils._figure_mm import _apply_theme_colors
+            for ax in fig.axes:
+                _apply_theme_colors(ax, theme='dark')
+
+    # Build bundle data (include basename for file naming)
+    bundle_data = {'spec': spec, 'basename': basename}
+
+    # Use provided data or extracted data for CSV
+    # Priority: 1) explicit data param, 2) export_as_csv method, 3) line extraction fallback
+    csv_df = None
+    if data is not None:
+        csv_df = data
+        bundle_data['data'] = data
+    else:
+        # Try to use export_as_csv from SciTeX wrapped objects (handles all plot types)
+        csv_source = _get_figure_with_data(obj)
+        if csv_source is not None and hasattr(csv_source, 'export_as_csv'):
+            try:
+                csv_df = csv_source.export_as_csv()
+                if csv_df is not None and not csv_df.empty:
+                    bundle_data['data'] = csv_df
+                    logger.debug(f"CSV data extracted via export_as_csv: {len(csv_df)} rows, {len(csv_df.columns)} cols")
+            except Exception as e:
+                logger.debug(f"export_as_csv failed: {e}")
+                csv_df = None
+
+        # Fallback to line extraction if export_as_csv didn't work
+        if csv_df is None and extracted_data:
+            try:
+                import pandas as pd
+                # Pad arrays to same length
+                max_len = max(len(v) for v in extracted_data.values())
+                padded = {}
+                for k, v in extracted_data.items():
+                    if len(v) < max_len:
+                        padded[k] = np.pad(v, (0, max_len - len(v)), constant_values=np.nan)
+                    else:
+                        padded[k] = v
+                csv_df = pd.DataFrame(padded)
+                bundle_data['data'] = csv_df
+                logger.debug(f"CSV data extracted via line fallback: {len(csv_df)} rows")
+            except ImportError:
+                pass
+
+    # Compute hash and columns for data section
+    if csv_df is not None:
+        import hashlib
+        # Get CSV string for hash computation
+        csv_str = csv_df.to_csv(index=False)
+        csv_hash = hashlib.sha256(csv_str.encode()).hexdigest()
+        spec['data']['hash'] = f'sha256:{csv_hash[:16]}'
+        spec['data']['columns'] = list(csv_df.columns)
+
+    # Save figure to multiple formats
+    import warnings
+    from PIL import Image as PILImage
+    from scitex.plt.utils._hitmap import (
+        apply_hitmap_colors, restore_original_colors, extract_path_data,
+        extract_selectable_regions, HITMAP_BACKGROUND_COLOR, HITMAP_AXES_COLOR
+    )
+
+    crop_box = None
+    color_map = {}
+
+    with tempfile.TemporaryDirectory() as tmp_dir:
+        tmp_path = Path(tmp_dir)
+
+        # Suppress tight_layout warnings for SciTeX figures with custom axes
+        with warnings.catch_warnings():
+            warnings.filterwarnings('ignore', message='.*tight_layout.*')
+
+            # Always use transparent background for SciTeX figures (both light and dark themes)
+            use_transparent = True
+
+            # Save PNG (raster) - required
+            png_path = tmp_path / "plot.png"
+            fig.savefig(png_path, dpi=dpi, bbox_inches='tight', format='png', transparent=use_transparent)
+
+            # Save SVG (vector) - optional
+            svg_path = tmp_path / "plot.svg"
+            fig.savefig(svg_path, bbox_inches='tight', format='svg')
+
+            # Save PDF (vector) - optional
+            pdf_path = tmp_path / "plot.pdf"
+            fig.savefig(pdf_path, bbox_inches='tight', format='pdf')
+
+            # Now generate hitmap by applying ID colors to data elements ONLY
+            # Keep axes/spines/labels with original colors to preserve bbox_inches='tight' bounds
+            # Also detects logical groups (histogram, bar_series, etc.)
+            original_props, color_map, groups = apply_hitmap_colors(fig)
+
+            # Store original background colors and set hitmap colors
+            original_fig_facecolor = fig.patch.get_facecolor()
+            original_ax_facecolors = []
+            original_ax_props = []
+            for ax in fig.axes:
+                original_ax_facecolors.append(ax.get_facecolor())
+                # Store axis element colors for restoration
+                ax_props = {
+                    'ax': ax,
+                    'spine_colors': {k: v.get_edgecolor() for k, v in ax.spines.items()},
+                    'tick_colors': ax.tick_params,  # Will restore later
+                    'xlabel_color': ax.xaxis.label.get_color(),
+                    'ylabel_color': ax.yaxis.label.get_color(),
+                    'title_color': ax.title.get_color(),
+                }
+                original_ax_props.append(ax_props)
+                # Set hitmap colors for non-data elements
+                ax.set_facecolor(HITMAP_BACKGROUND_COLOR)
+                for spine in ax.spines.values():
+                    spine.set_color(HITMAP_AXES_COLOR)
+                ax.tick_params(colors=HITMAP_AXES_COLOR, labelcolor=HITMAP_AXES_COLOR)
+                ax.xaxis.label.set_color(HITMAP_AXES_COLOR)
+                ax.yaxis.label.set_color(HITMAP_AXES_COLOR)
+                ax.title.set_color(HITMAP_AXES_COLOR)
+
+            fig.patch.set_facecolor(HITMAP_BACKGROUND_COLOR)
+
+            # Save hitmap PNG with same bbox_inches='tight'
+            hitmap_path = tmp_path / "plot_hitmap.png"
+            fig.savefig(hitmap_path, dpi=dpi, bbox_inches='tight', format='png', facecolor=HITMAP_BACKGROUND_COLOR)
+
+            # Optimize hitmap PNG size using zlib compression
+            try:
+                hitmap_img = PILImage.open(hitmap_path).convert('RGB')
+                hitmap_img.save(hitmap_path, format='PNG', optimize=True, compress_level=9)
+            except Exception:
+                pass  # Keep original if optimization fails
+
+            # Save hitmap SVG with same bbox_inches='tight'
+            hitmap_svg_path = tmp_path / "plot_hitmap.svg"
+            fig.savefig(hitmap_svg_path, bbox_inches='tight', format='svg')
+
+            # Restore original colors (data elements)
+            restore_original_colors(original_props)
+
+            # Restore original figure and axes colors
+            fig.patch.set_facecolor(original_fig_facecolor)
+            for i, ax in enumerate(fig.axes):
+                ax.set_facecolor(original_ax_facecolors[i])
+                if i < len(original_ax_props):
+                    props = original_ax_props[i]
+                    for spine_name, color in props['spine_colors'].items():
+                        ax.spines[spine_name].set_edgecolor(color)
+                    ax.xaxis.label.set_color(props['xlabel_color'])
+                    ax.yaxis.label.set_color(props['ylabel_color'])
+                    ax.title.set_color(props['title_color'])
+
+            # Now apply auto-crop to BOTH PNG and hitmap with same parameters
+            try:
+                from scitex.plt.utils._crop import crop
+
+                # Crop PNG and get crop coordinates
+                _, crop_offset = crop(
+                    str(png_path),
+                    output_path=str(png_path),
+                    overwrite=True,
+                    margin=12,  # ~1mm at 300 DPI
+                    verbose=False,
+                    return_offset=True,
+                )
+                crop_box = (crop_offset['left'], crop_offset['upper'],
+                           crop_offset['right'], crop_offset['lower'])
+
+                # Apply SAME crop to hitmap PNG
+                crop(
+                    str(hitmap_path),
+                    output_path=str(hitmap_path),
+                    overwrite=True,
+                    crop_box=crop_box,
+                    verbose=False,
+                )
+            except Exception as e:
+                crop_box = None
+                logger.debug(f"Crop failed: {e}")
+
+            # Validate sizes match
+            with PILImage.open(png_path) as png_img, PILImage.open(hitmap_path) as hm_img:
+                if png_img.size != hm_img.size:
+                    logger.warning(f"Size mismatch: PNG={png_img.size}, Hitmap={hm_img.size}")
+
+            with open(png_path, 'rb') as f:
+                bundle_data['png'] = f.read()
+
+            with open(hitmap_path, 'rb') as f:
+                bundle_data['hitmap_png'] = f.read()
+
+            with open(svg_path, 'rb') as f:
+                bundle_data['svg'] = f.read()
+
+            with open(hitmap_svg_path, 'rb') as f:
+                bundle_data['hitmap_svg'] = f.read()
+
+            with open(pdf_path, 'rb') as f:
+                bundle_data['pdf'] = f.read()
+
+    # Add hit_regions to spec
+    try:
+        path_data = extract_path_data(fig)
+
+        spec['hit_regions'] = {
+            'strategy': 'hybrid',
+            'hit_map': f'{basename}_hitmap.png',
+            'hit_map_svg': f'{basename}_hitmap.svg',
+            'color_map': {str(k): v for k, v in color_map.items()},
+            'groups': groups,  # Logical groups (histogram, bar_series, etc.)
+            'path_data': path_data,
+        }
+
+        if crop_box is not None:
+            spec['hit_regions']['crop_box'] = {
+                'left': int(crop_box[0]),
+                'upper': int(crop_box[1]),
+                'right': int(crop_box[2]),
+                'lower': int(crop_box[3]),
+            }
+
+        # Extract selectable regions (bounding boxes for axis/annotation elements)
+        # This complements hitmap color-based selection with bbox-based selection
+        selectable_regions = extract_selectable_regions(fig)
+        if selectable_regions and selectable_regions.get('axes'):
+            spec['selectable_regions'] = selectable_regions
+
+    except Exception as e:
+        logger.debug(f"Hit regions spec failed: {e}")
+
+    # Save the bundle
+    save_bundle(bundle_data, p, bundle_type=BundleType.PLTZ, as_zip=as_zip)
+
+
 def _save(
     obj,
     spath,
@@ -506,7 +990,7 @@ def _save(
     auto_crop=False,
     crop_margin_mm=1.0,
     metadata_extra=None,
-    json_schema="recipe",
+    json_schema="editable",
     **kwargs,
 ):
     # Don't use object's own save method - use consistent handlers
@@ -521,6 +1005,50 @@ def _save(
         save_canvas(obj, spath, **kwargs)
         return
 
+    # Handle bundle formats (.figz, .pltz, .statsz and their .d variants)
+    # These use special naming: file.figz (ZIP) or file.figz.d (directory)
+    # Note: .figz defaults to ZIP (as_zip=True), .pltz/.statsz default to directory
+    bundle_extensions = (".figz", ".pltz", ".statsz")
+    for bext in bundle_extensions:
+        if spath.endswith(bext) or spath.endswith(f"{bext}.d"):
+            # Remove as_zip from kwargs if present to avoid duplicate
+            bundle_kwargs = {k: v for k, v in kwargs.items() if k != 'as_zip'}
+            as_zip = kwargs.get('as_zip', not spath.endswith(".d"))
+            if bext == ".figz":
+                import scitex.fig as sfig
+                # figz defaults to ZIP, so always pass as_zip explicitly
+                sfig.save_figz(obj, spath, as_zip=as_zip, **bundle_kwargs)
+            elif bext == ".pltz":
+                _save_pltz_bundle(obj, spath, as_zip=as_zip, **bundle_kwargs)
+            elif bext == ".statsz":
+                import scitex.stats as sstats
+                sstats.save_statsz(obj, spath, as_zip=as_zip, **bundle_kwargs)
+
+            # Log "Saved to:" for bundle formats (consistent with other formats)
+            # For bundles, determine the actual saved path (zip or directory)
+            bundle_path = spath if as_zip else f"{spath}.d" if not spath.endswith(".d") else spath
+
+            if verbose and _os.path.exists(bundle_path):
+                file_size = getsize(bundle_path)
+                file_size = readable_bytes(file_size)
+                try:
+                    rel_path = _os.path.relpath(bundle_path, _os.getcwd())
+                except ValueError:
+                    rel_path = bundle_path
+                logger.success(f"Saved to: ./{rel_path} ({file_size})")
+
+            # Handle symlinks for bundle formats (consistent with other formats)
+            if symlink_from_cwd and _os.path.exists(bundle_path):
+                # Create symlink from cwd to bundle path
+                bundle_basename = _os.path.basename(bundle_path)
+                bundle_cwd = _os.path.join(_os.getcwd(), bundle_basename)
+                _symlink(bundle_path, bundle_cwd, symlink_from_cwd, verbose)
+
+            if symlink_to and _os.path.exists(bundle_path):
+                _symlink_to(bundle_path, symlink_to, verbose)
+
+            return
+
     # Try dispatch dictionary first for O(1) lookup
     if ext in _FILE_HANDLERS:
         # Check if handler needs special parameters
@@ -560,7 +1088,7 @@ def _save(
     elif spath.endswith(".pkl.gz"):
         save_pickle_compressed(obj, spath, **kwargs)
     else:
-        warnings.warn(f"Unsupported file format. {spath} was not saved.")
+        logger.warning(f"Unsupported file format. {spath} was not saved.")
 
     if verbose:
         if _os.path.exists(spath):
@@ -657,7 +1185,7 @@ def _handle_image_with_csv(
     auto_crop=True,
     crop_margin_mm=1.0,
     metadata_extra=None,
-    json_schema="recipe",
+    json_schema="editable",
     **kwargs,
 ):
     """Handle image file saving with optional CSV export and auto-cropping."""
@@ -701,7 +1229,10 @@ def _handle_image_with_csv(
 
                 # Collect metadata using scitex's metadata collector
                 try:
-                    if json_schema == "recipe":
+                    if json_schema == "editable":
+                        from scitex.plt.utils.metadata import export_editable_figure
+                        auto_metadata = export_editable_figure(fig_mpl)
+                    elif json_schema == "recipe":
                         from scitex.plt.utils import collect_recipe_metadata
                         auto_metadata = collect_recipe_metadata(
                             fig_mpl, ax,
@@ -716,15 +1247,14 @@ def _handle_image_with_csv(
                         kwargs["metadata"] = auto_metadata
                         collected_metadata = auto_metadata  # Save for JSON export
                         if verbose:
-                            schema_name = "recipe" if json_schema == "recipe" else "verbose"
+                            schema_names = {"editable": "editable v0.3", "recipe": "recipe", "verbose": "verbose"}
+                            schema_name = schema_names.get(json_schema, json_schema)
                             logger.info(f"  • Auto-collected metadata ({schema_name} schema)")
                 except ImportError:
                     pass  # collect_figure_metadata not available
                 except Exception as e:
                     if verbose:
-                        import warnings
-
-                        warnings.warn(f"Could not auto-collect metadata: {e}")
+                        logger.warning(f"Could not auto-collect metadata: {e}")
         except Exception:
             pass  # Silently continue if auto-collection fails
     else:
@@ -820,9 +1350,7 @@ def _handle_image_with_csv(
                     )
 
             except Exception as e:
-                import warnings
-
-                warnings.warn(f"Auto-crop failed: {e}. Image saved without cropping.")
+                logger.warning(f"Auto-crop failed: {e}. Image saved without cropping.")
 
     # Handle separate legend saving
     _save_separate_legends(
@@ -1057,12 +1585,7 @@ def _handle_image_with_csv(
                             )
                             _symlink(csv_sigmaplot_path, csv_cwd, True, True)
         except Exception as e:
-            import warnings
-
-            warnings.warn(f"CSV export failed: {e}")
-            import traceback
-
-            traceback.print_exc()
+            logger.warning(f"CSV export failed: {e}")
 
     # Save metadata as JSON if collected
     if collected_metadata is not None and not dry_run:
@@ -1109,8 +1632,8 @@ def _handle_image_with_csv(
             )
 
             # Verify CSV/JSON consistency (data_ref must match columns_actual)
-            # Only check for verbose schema - recipe schema uses different data_ref structure
-            if csv_path and not dry_run and json_schema != "recipe":
+            # Only check for verbose schema - recipe/editable schemas use different data_ref structure
+            if csv_path and not dry_run and json_schema == "verbose":
                 from scitex.plt.utils._collect_figure_metadata import (
                     assert_csv_json_consistency,
                 )
@@ -1181,17 +1704,12 @@ def _handle_image_with_csv(
             # Re-raise assertion errors - these are validation failures that should stop execution
             raise
         except Exception as e:
-            import warnings
-
-            warnings.warn(f"JSON metadata export failed: {e}")
-            import traceback
-
-            traceback.print_exc()
+            logger.warning(f"JSON metadata export failed: {e}")
 
 
 # Dispatch dictionary for O(1) file format lookup
 _FILE_HANDLERS = {
-    # Canvas directory format (scitex.vis)
+    # Canvas directory format (scitex.fig)
     ".canvas": save_canvas,
     # Excel formats
     ".xlsx": save_excel,