scitex 2.3.0__py3-none-any.whl → 2.4.0__py3-none-any.whl

scitex/plt/__init__.py

@@ -1,18 +1,112 @@
 #!/usr/bin/env python3
 # -*- coding: utf-8 -*-
-# Timestamp: "2025-06-13 22:42:39 (ywatanabe)"
-# File: /ssh:sp:/home/ywatanabe/proj/SciTeX-Code/src/scitex/plt/__init__.py
+# Timestamp: "2025-12-02 12:30:00 (ywatanabe)"
+# File: /home/ywatanabe/proj/scitex-code/src/scitex/plt/__init__.py
 # ----------------------------------------
 import os
 __FILE__ = __file__
 __DIR__ = os.path.dirname(__FILE__)
 # ----------------------------------------
-"""Scitex plt module."""
+"""
+SciTeX plt module - Publication-quality plotting.
+
+Simply importing this module automatically configures matplotlib with
+SciTeX publication defaults from SCITEX_STYLE.yaml:
+
+    import scitex.plt as splt
+    fig, ax = splt.subplots()
+    ax.plot([1, 2, 3], [1, 4, 9])
+    fig.savefig("figure.png")
+
+No need to call scitex.session() or configure_mpl() - it's automatic.
+
+Style values can be customized by:
+1. Editing SCITEX_STYLE.yaml
+2. Setting environment variables (SCITEX_PLT_FONTS_AXIS_LABEL_PT=8)
+3. Passing parameters directly to subplots()
+"""
 
-# Register Arial fonts eagerly (before lazy imports)
 import matplotlib.font_manager as fm
 import matplotlib as mpl
-
+import matplotlib.pyplot as plt
+
+# =============================================================================
+# Auto-configure matplotlib with SciTeX style on import
+# =============================================================================
+
+def _auto_configure_mpl():
+    """Apply SciTeX style configuration automatically on import."""
+    from .styles import resolve_style_value
+
+    # mm to pt conversion factor
+    mm_to_pt = 2.83465
+
+    # Load all style values from YAML (with env override support)
+    font_size = resolve_style_value("fonts.axis_label_pt", None, 7)
+    title_size = resolve_style_value("fonts.title_pt", None, 8)
+    tick_size = resolve_style_value("fonts.tick_label_pt", None, 7)
+    legend_size = resolve_style_value("fonts.legend_pt", None, 6)
+
+    trace_mm = resolve_style_value("lines.trace_mm", None, 0.2)
+    line_width = trace_mm * mm_to_pt
+
+    axes_thickness_mm = resolve_style_value("axes.thickness_mm", None, 0.2)
+    axes_linewidth = axes_thickness_mm * mm_to_pt
+
+    hide_top = resolve_style_value("behavior.hide_top_spine", None, True, bool)
+    hide_right = resolve_style_value("behavior.hide_right_spine", None, True, bool)
+
+    dpi = int(resolve_style_value("output.dpi", None, 300))
+
+    # Calculate figure size from axes + margins
+    axes_w = resolve_style_value("axes.width_mm", None, 40)
+    axes_h = resolve_style_value("axes.height_mm", None, 28)
+    margin_l = resolve_style_value("margins.left_mm", None, 20)
+    margin_r = resolve_style_value("margins.right_mm", None, 20)
+    margin_b = resolve_style_value("margins.bottom_mm", None, 20)
+    margin_t = resolve_style_value("margins.top_mm", None, 20)
+    fig_w_mm = axes_w + margin_l + margin_r
+    fig_h_mm = axes_h + margin_b + margin_t
+    figsize_inch = (fig_w_mm / 25.4, fig_h_mm / 25.4)
+
+    # Apply rcParams
+    mpl_config = {
+        # Resolution
+        "figure.dpi": max(100, dpi // 3),
+        "savefig.dpi": dpi,
+        # Figure Size
+        "figure.figsize": figsize_inch,
+        # Font Sizes
+        "font.size": font_size,
+        "axes.titlesize": title_size,
+        "axes.labelsize": font_size,
+        "xtick.labelsize": tick_size,
+        "ytick.labelsize": tick_size,
+        # Legend
+        "legend.fontsize": legend_size,
+        "legend.frameon": False,
+        "legend.loc": "best",
+        # Auto Layout
+        "figure.autolayout": True,
+        # Spines
+        "axes.spines.top": not hide_top,
+        "axes.spines.right": not hide_right,
+        # Line widths
+        "axes.linewidth": axes_linewidth,
+        "lines.linewidth": line_width,
+        "lines.markersize": 6.0,
+        # Grid
+        "grid.linewidth": axes_linewidth,
+        "grid.alpha": 0.3,
+        # Math text
+        "mathtext.fontset": "dejavusans",
+        "mathtext.default": "regular",
+    }
+
+    mpl.rcParams.update(mpl_config)
+
+
+# Register Arial fonts eagerly (before style configuration)
 _arial_enabled = False
 try:
     fm.findfont("Arial", fallback_to_default=False)
@@ -38,27 +132,38 @@ except Exception:
         except Exception:
             pass
 
-# Configure matplotlib to use Arial if available
+# Configure font family
 if _arial_enabled:
     mpl.rcParams["font.family"] = "Arial"
     mpl.rcParams["font.sans-serif"] = ["Arial", "Helvetica", "DejaVu Sans", "Liberation Sans"]
 else:
-    # Warn about missing Arial
-    try:
-        from scitex.logging import getLogger
-        _logger = getLogger(__name__)
-        _logger.warning(
-            "Arial font not found. Using fallback fonts (Helvetica/DejaVu Sans). "
-            "For publication figures with Arial: sudo apt-get install ttf-mscorefonts-installer && fc-cache -fv"
-        )
-    except:
-        pass  # Skip warning if logging not available
+    mpl.rcParams["font.family"] = "sans-serif"
+    mpl.rcParams["font.sans-serif"] = ["Helvetica", "DejaVu Sans", "Liberation Sans", "sans-serif"]
+    # Suppress font warnings
+    import logging
+    logging.getLogger('matplotlib.font_manager').setLevel(logging.ERROR)
+
+# Apply SciTeX style configuration automatically
+_auto_configure_mpl()
+
+# Set up color cycle from scitex colors
+try:
+    from . import color as _color_module
+    _rgba_norm_cycle = {
+        k: tuple(_color_module.update_alpha(v, 1.0))
+        for k, v in _color_module.PARAMS.get("RGBA_NORM_FOR_CYCLE", {}).items()
+    }
+    if _rgba_norm_cycle:
+        mpl.rcParams["axes.prop_cycle"] = plt.cycler(color=list(_rgba_norm_cycle.values()))
+except Exception:
+    pass  # Use matplotlib default colors if color module fails
 
 from ._tpl import termplot
 from . import color
 from . import utils
 from . import ax
-from . import presets
+from .styles import presets
+from . import styles
 
 # Lazy import for subplots to avoid circular dependencies
 _subplots = None
@@ -125,6 +230,353 @@ def crop(input_path, output_path=None, margin=12, overwrite=False, verbose=False
         _crop = _crop_func
     return _crop(input_path, output_path, margin, overwrite, verbose)
 
+def load(path, apply_manual=True):
+    """
+    Load a figure from saved JSON + CSV files.
+
+    Parameters
+    ----------
+    path : str or Path
+        Path to JSON file, PNG file, or CSV file.
+        Will auto-detect sibling files in same directory or organized subdirectories.
+    apply_manual : bool, optional
+        If True, apply .manual.json overrides if exists (default: True)
+
+    Returns
+    -------
+    tuple
+        (fig, axes) where fig is FigWrapper and axes is AxisWrapper or array
+
+    Raises
+    ------
+    FileNotFoundError
+        If required JSON file is not found
+    ValueError
+        If manual.json hash doesn't match (stale manual edits)
+
+    Examples
+    --------
+    >>> # Load from JSON (sibling pattern)
+    >>> fig, axes = stx.plt.load("output/figure.json")
+
+    >>> # Load from PNG (finds sibling JSON + CSV)
+    >>> fig, axes = stx.plt.load("output/figure.png")
+
+    >>> # Load from organized directory pattern
+    >>> fig, axes = stx.plt.load("output/json/figure.json")
+
+    >>> # Skip manual overrides
+    >>> fig, axes = stx.plt.load("figure.json", apply_manual=False)
+
+    Notes
+    -----
+    Supports two directory patterns:
+
+    Pattern 1 (flat/sibling):
+        output/figure.png
+        output/figure.json
+        output/figure.csv
+
+    Pattern 2 (organized):
+        output/png/figure.png
+        output/json/figure.json
+        output/csv/figure.csv
+
+    Manual overrides (.manual.json) are applied if:
+    - apply_manual=True
+    - figure.manual.json exists alongside figure.json
+    - Hash validation passes (warns if stale)
+    """
+    from pathlib import Path
+    import hashlib
+    import warnings
+    import scitex as stx
+
+    path = Path(path)
+
+    # Resolve JSON path from any input (png, csv, or json)
+    json_path, csv_path = _resolve_figure_paths(path)
+
+    if not json_path.exists():
+        raise FileNotFoundError(f"JSON file not found: {json_path}")
+
+    # Load JSON metadata
+    metadata = stx.io.load(json_path)
+
+    # Load CSV data if exists
+    csv_data = None
+    if csv_path and csv_path.exists():
+        csv_data = stx.io.load(csv_path)
+
+    # Check for manual overrides
+    manual_path = json_path.with_suffix('.manual.json')
+    manual_overrides = None
+    if apply_manual and manual_path.exists():
+        manual_data = stx.io.load(manual_path)
+
+        # Validate hash
+        if 'base_hash' in manual_data:
+            current_hash = _compute_file_hash(json_path)
+            if manual_data['base_hash'] != current_hash:
+                warnings.warn(
+                    f"Manual overrides may be stale: base data changed since manual edits.\n"
+                    f"  Expected hash: {manual_data['base_hash'][:16]}...\n"
+                    f"  Current hash:  {current_hash[:16]}...\n"
+                    f"  Review: {manual_path}"
+                )
+        manual_overrides = manual_data.get('overrides', {})
+
+    # Reconstruct figure
+    fig, axes = _reconstruct_figure(metadata, csv_data, manual_overrides)
+
+    return fig, axes
+
+
+def _resolve_figure_paths(path):
+    """
+    Resolve JSON and CSV paths from any input file path.
+
+    Supports both flat (sibling) and organized (subdirectory) patterns.
+    """
+    from pathlib import Path
+
+    path = Path(path)
+    stem = path.stem
+    suffix = path.suffix.lower()
+    parent = path.parent
+
+    # Determine base name (remove .manual if present)
+    if stem.endswith('.manual'):
+        stem = stem[:-7]
+
+    json_path = None
+    csv_path = None
+
+    if suffix == '.json':
+        json_path = path
+        # Try sibling CSV first
+        csv_sibling = parent / f"{stem}.csv"
+        if csv_sibling.exists():
+            csv_path = csv_sibling
+        # Try organized pattern (../csv/)
+        elif parent.name == 'json':
+            csv_organized = parent.parent / 'csv' / f"{stem}.csv"
+            if csv_organized.exists():
+                csv_path = csv_organized
+
+    elif suffix in ('.png', '.jpg', '.jpeg', '.pdf', '.svg'):
+        # Look for sibling JSON
+        json_sibling = parent / f"{stem}.json"
+        csv_sibling = parent / f"{stem}.csv"
+
+        if json_sibling.exists():
+            json_path = json_sibling
+            if csv_sibling.exists():
+                csv_path = csv_sibling
+        # Try organized pattern (parent has png/, look for json/)
+        elif parent.name in ('png', 'jpg', 'jpeg', 'pdf', 'svg'):
+            json_organized = parent.parent / 'json' / f"{stem}.json"
+            csv_organized = parent.parent / 'csv' / f"{stem}.csv"
+            if json_organized.exists():
+                json_path = json_organized
+            if csv_organized.exists():
+                csv_path = csv_organized
+
+    elif suffix == '.csv':
+        csv_path = path
+        # Try sibling JSON
+        json_sibling = parent / f"{stem}.json"
+        if json_sibling.exists():
+            json_path = json_sibling
+        # Try organized pattern (../json/)
+        elif parent.name == 'csv':
+            json_organized = parent.parent / 'json' / f"{stem}.json"
+            if json_organized.exists():
+                json_path = json_organized
+
+    # Fallback: assume it's the JSON path
+    if json_path is None:
+        json_path = path if suffix == '.json' else path.with_suffix('.json')
+
+    return json_path, csv_path
+
+
+def _compute_file_hash(path):
+    """Compute SHA256 hash of a file."""
+    import hashlib
+    from pathlib import Path
+
+    path = Path(path)
+    sha256 = hashlib.sha256()
+    with open(path, 'rb') as f:
+        for chunk in iter(lambda: f.read(8192), b''):
+            sha256.update(chunk)
+    return f"sha256:{sha256.hexdigest()}"
+
+
+def _reconstruct_figure(metadata, csv_data, manual_overrides=None):
+    """
+    Reconstruct figure from metadata and CSV data.
+
+    Parameters
+    ----------
+    metadata : dict
+        JSON metadata from stx.io.save()
+    csv_data : DataFrame or None
+        CSV data with plot values
+    manual_overrides : dict or None
+        Manual style/annotation overrides
+
+    Returns
+    -------
+    tuple
+        (fig, axes)
+    """
+    import numpy as np
+
+    # Extract dimensions from metadata
+    dims = metadata.get('dimensions', {})
+    fig_size_mm = dims.get('figure_size_mm', [80, 68])
+    dpi = dims.get('dpi', 300)
+
+    # Get style from metadata
+    scitex_meta = metadata.get('scitex', {})
+    style_mm = scitex_meta.get('style_mm', {})
+
+    # Create figure with same dimensions
+    fig, axes = subplots(
+        axes_width_mm=style_mm.get('axes_width_mm', dims.get('axes_size_mm', [40, 28])[0]),
+        axes_height_mm=style_mm.get('axes_height_mm', dims.get('axes_size_mm', [40, 28])[1]),
+        dpi=dpi,
+    )
+
+    # Handle single vs multiple axes
+    ax = axes if not hasattr(axes, 'flat') else list(axes.flat)[0]
+
+    # Set axis labels from metadata
+    axes_meta = metadata.get('axes', {})
+    x_meta = axes_meta.get('x', {})
+    y_meta = axes_meta.get('y', {})
+
+    xlabel = x_meta.get('label', '')
+    ylabel = y_meta.get('label', '')
+    x_unit = x_meta.get('unit', '')
+    y_unit = y_meta.get('unit', '')
+
+    if xlabel:
+        full_xlabel = f"{xlabel} [{x_unit}]" if x_unit else xlabel
+        ax.set_xlabel(full_xlabel)
+    if ylabel:
+        full_ylabel = f"{ylabel} [{y_unit}]" if y_unit else ylabel
+        ax.set_ylabel(full_ylabel)
+
+    # Reconstruct plots from CSV data
+    if csv_data is not None and not csv_data.empty:
+        _reconstruct_plots_from_csv(ax, csv_data, metadata)
+
+    # Apply manual overrides
+    if manual_overrides:
+        _apply_manual_overrides(fig, axes, manual_overrides)
+
+    return fig, axes
+
+
+def _reconstruct_plots_from_csv(ax, csv_data, metadata):
+    """
+    Reconstruct plot elements from CSV data.
+
+    CSV columns follow pattern: ax_00_<type>_<name>
+    """
+    import pandas as pd
+    import numpy as np
+
+    # Group columns by plot type
+    plot_type = metadata.get('plot_type', metadata.get('method', 'line'))
+
+    # Parse column names to find plot data
+    columns = csv_data.columns.tolist()
+
+    # Find x/y data columns
+    x_cols = [c for c in columns if '_x' in c.lower() and 'text' not in c.lower()]
+    y_cols = [c for c in columns if '_y' in c.lower() and 'text' not in c.lower()]
+
+    if plot_type == 'line' or plot_type == 'plot':
+        # For line plots, look for paired x/y or just y with index
+        if x_cols and y_cols:
+            for x_col, y_col in zip(x_cols, y_cols):
+                x = csv_data[x_col].dropna().values
+                y = csv_data[y_col].dropna().values
+                if len(x) > 0 and len(y) > 0:
+                    ax.plot(x[:len(y)], y[:len(x)])
+        elif y_cols:
+            for y_col in y_cols:
+                y = csv_data[y_col].dropna().values
+                if len(y) > 0:
+                    ax.plot(y)
+
+    elif plot_type == 'scatter':
+        if x_cols and y_cols:
+            x = csv_data[x_cols[0]].dropna().values
+            y = csv_data[y_cols[0]].dropna().values
+            ax.scatter(x[:min(len(x), len(y))], y[:min(len(x), len(y))])
+
+    # Add text annotations
+    text_cols = [c for c in columns if 'text' in c.lower() and 'content' in c.lower()]
+    for text_col in text_cols:
+        # Find corresponding x, y columns
+        prefix = text_col.rsplit('_content', 1)[0]
+        x_col = f"{prefix}_x"
+        y_col = f"{prefix}_y"
+
+        if x_col in columns and y_col in columns:
+            x_vals = csv_data[x_col].dropna()
+            y_vals = csv_data[y_col].dropna()
+            text_vals = csv_data[text_col].dropna()
+
+            for i in range(min(len(x_vals), len(y_vals), len(text_vals))):
+                ax.text(
+                    x_vals.iloc[i], y_vals.iloc[i], str(text_vals.iloc[i]),
+                    transform=ax.transAxes, fontsize=6,
+                    verticalalignment='top', horizontalalignment='right'
+                )
+
+
+def _apply_manual_overrides(fig, axes, overrides):
+    """
+    Apply manual style/annotation overrides to figure.
+
+    Parameters
+    ----------
+    fig : FigWrapper
+        Figure to modify
+    axes : AxisWrapper or array
+        Axes to modify
+    overrides : dict
+        Override specifications like {"axes[0].style.linewidth": 0.5}
+    """
+    # Simple override application - can be extended
+    for key, value in overrides.items():
+        # Parse key like "axes[0].title" or "style.linewidth"
+        parts = key.split('.')
+
+        if parts[0].startswith('axes['):
+            # Extract axis index
+            import re
+            match = re.match(r'axes\[(\d+)\]', parts[0])
+            if match:
+                idx = int(match.group(1))
+                ax = axes if not hasattr(axes, 'flat') else list(axes.flat)[idx]
+
+                if len(parts) > 1:
+                    attr = parts[1]
+                    if attr == 'title':
+                        ax.set_title(value)
+                    elif attr == 'xlabel':
+                        ax.set_xlabel(value)
+                    elif attr == 'ylabel':
+                        ax.set_ylabel(value)
+
+
 def tight_layout(**kwargs):
     """
     Wrapper for matplotlib.pyplot.tight_layout that handles colorbar layout compatibility.
@@ -202,6 +654,7 @@ def colorbar(mappable=None, cax=None, ax=None, **kwargs):
 __all__ = [
     "termplot",
     "subplots",
+    "load",
     "figure",
     "tight_layout",
     "utils",