viewinline 0.2.2__tar.gz → 0.3.0__tar.gz

viewinline-0.2.2/README.md → viewinline-0.3.0/PKG-INFO

@@ -1,10 +1,43 @@
+Metadata-Version: 2.4
+Name: viewinline
+Version: 0.3.0
+Summary: Quick look geospatial viewer for the terminal, with inline image previews
+Project-URL: Homepage, https://github.com/nkeikon/viewinline
+Project-URL: Repository, https://github.com/nkeikon/viewinline
+Project-URL: Issues, https://github.com/nkeikon/viewinline/issues
+Author: Keiko Nomura
+License: Apache-2.0
+License-File: LICENSE
+Requires-Python: >=3.9
+Requires-Dist: geopandas
+Requires-Dist: matplotlib
+Requires-Dist: numpy
+Requires-Dist: pandas
+Requires-Dist: pillow
+Requires-Dist: pyogrio
+Requires-Dist: rasterio
+Provides-Extra: all
+Requires-Dist: duckdb; extra == 'all'
+Requires-Dist: h5py; extra == 'all'
+Requires-Dist: netcdf4; extra == 'all'
+Requires-Dist: pyarrow; extra == 'all'
+Provides-Extra: hdf5
+Requires-Dist: h5py; extra == 'hdf5'
+Provides-Extra: netcdf
+Requires-Dist: netcdf4; extra == 'netcdf'
+Provides-Extra: parquet
+Requires-Dist: pyarrow; extra == 'parquet'
+Provides-Extra: sql
+Requires-Dist: duckdb; extra == 'sql'
+Description-Content-Type: text/markdown
+
 # viewinline
 [![Downloads](https://static.pepy.tech/badge/viewinline)](https://pepy.tech/project/viewinline)
 [![PyPI version](https://img.shields.io/pypi/v/viewinline)](https://pypi.org/project/viewinline/)
 [![Python version](https://img.shields.io/badge/python-%3E%3D3.9-blue.svg)](https://pypi.org/project/viewinline/)
 
 **Quick-look geospatial viewer for compatible terminals.**  
-Displays rasters, vectors, and tabular data directly in the terminal with no GUI and no temporary files.
+Displays rasters, vectors, and tabular data directly in the terminal with no GUI.
 
 <p align="center">
   <a href="viewinline_gif1.gif"><img src="viewinline_gif1.gif" width="49%"></a>
@@ -17,7 +50,7 @@ Particularly useful on HPC systems and remote servers accessed via SSH. Images r
 
 This tool combines the core display logic of `viewtif` and `viewgeom`, but is **non-interactive**: you can't zoom, pan, or switch colormaps on the fly. Instead, you control everything through command-line options (e.g. --display, --color-by, --colormap).
 
-It uses the iTerm2 inline image protocol (OSC 1337) to render previews. In incompatible terminals, the escape codes are silently ignored with no errors or crashes.
+It uses the iTerm2 inline image protocol (OSC 1337) in supported terminals, and falls back to `chafa` in others, which displays real high-res images in terminals like kitty and Ghostty, and colored block-art (ASCII art) previews in Terminal.app, VS Code, and most Linux terminals. Without `chafa` installed, non-iTerm2-family terminals show an info message instead.
 
 ## Installation  
 Requires Python 3.9 or later.  
@@ -31,6 +64,7 @@ pip install viewinline
 # Rasters
 viewinline path/to/file.tif
 viewinline R.tif G.tif B.tif                 # RGB composite (also works with --rgbfiles)
+viewinline hyperspectral.nc --band 50
 viewinline path/to/multiband.tif --rgb 3 2 1
 viewinline path/to/folder --gallery 4x3      # show image gallery (e.g. 4x3 grid)
 
@@ -39,6 +73,7 @@ viewinline file.nc                           # list variables
 viewinline file.nc --subset 2                # display variable 2
 viewinline file.nc --subset 1 --band 10      # variable 1, timestep 10 --band or --timestep
 viewinline temp.nc --subset 1 --colormap plasma --vmin 273 --vmax 310
+viewinline hyperspectral.nc --subset 1 --reduce NumberOfScanlines  # override auto-detected axis
 
 # Vectors
 viewinline path/to/vector.geojson
@@ -63,18 +98,31 @@ viewinline data.geoparquet --table --where "POP > 100000" --sort POP --desc
 
 ## Compatible terminals
 
-The iTerm2 inline image protocol (OSC 1337) is supported by:
+Native (no extra install required): images render via the iTerm2 inline image protocol on:
 
 - **iTerm2** (macOS)
 - **WezTerm** (cross-platform)
 - **Konsole** (Linux/KDE)
 - **Rio**, **Contour** (cross-platform)
 
-**Not compatible:** Mac Terminal, GNOME Terminal, Kitty (uses different protocol), Ghostty, Alacritty
+Via `chafa` (recommended for everyone else): install chafa and viewinline works in nearly every terminal:
+- kitty, Ghostty, foot — real high-resolution images via the kitty graphics protocol or sixel
+- Terminal.app, VS Code, GNOME Terminal, Alacritty, Warp, Hyper — colored block-art previews with 24-bit color
+
+Install chafa once (it's a system binary, available across all conda/virtualenv environments):
+
+```
+brew install chafa          # macOS
+sudo apt install chafa      # Debian/Ubuntu
+sudo dnf install chafa      # Fedora
+scoop install chafa         # Windows
+```
+Without chafa, terminals outside the native list above show an info message instead of an image.
+You can also force the chafa path on any terminal by setting `INLINE_VIEWER_ENGINE=chafa`.
 
 **SSH/HPC usage:** Works over SSH when connecting from a compatible terminal. Images render on your local machine, not the remote server. No X11 forwarding or VNC required.
 
-**tmux/screen:** Inline images don't work inside tmux or screen sessions, even with `allow-passthrough on`. Use a plain terminal tab.
+**tmux/screen:** Inline images work inside tmux only when the outer terminal is iTerm2 (or WezTerm/Konsole/Rio/Contour). In tmux with other outer terminals (kitty, Terminal.app, etc.), viewinline displays ASCII art previews instead of full-quality images.
 
 **Fallback:** In terminals that do not support inline images, you can fallback to ASCII art by installing [`chafa`](https://hpjansson.org/chafa/) command-line tool. Install `chafa` with your package manager (e.g. `brew install chafa` or `sudo apt install chafa`). You can also force the use of `chafa` by setting the environment variable `INLINE_VIEWER_ENGINE=chafa`.
 
@@ -124,7 +172,9 @@ The iTerm2 inline image protocol (OSC 1337) is supported by:
 
 **NetCDF/HDF notes:**
 - viewinline lists only variables that can be displayed as 2D or 3D arrays
-- Variables with additional dimensions (e.g., vertical levels) may be listed but will fail to display with a clear error message
+- 3D variables with time or known spatial dimensions are auto-handled (slices along the non-spatial axis)
+- For 3D variables with non-standard dimensions (e.g., hyperspectral cubes like PICARD), viewinline auto-detects the band axis by smallest dimension. Use `--reduce DIM_NAME` to override.
+- Variables with 4+ dimensions are not supported
 - For a complete variable list, use `ncdump -h file.nc` or `viewtif`
 
 ## Dependencies
@@ -137,6 +187,7 @@ The iTerm2 inline image protocol (OSC 1337) is supported by:
 - `numpy`, `pandas` — data handling
 
 **Optional dependencies:**
+- `chafa` — strongly recommended for terminal coverage beyond iTerm2/WezTerm/Konsole/Rio/Contour. System binary, not a Python package. See "Terminal support" above for install instructions.
 - `duckdb` — required for `--where`, `--sort`, `--sql`, `--limit` with filtering
   ```bash
   pip install duckdb
@@ -152,7 +203,7 @@ The iTerm2 inline image protocol (OSC 1337) is supported by:
 
 **Note on HDF support:**
 - **HDF5** (.h5, .hdf5): Supported via rasterio if GDAL has HDF5 support (most installations)
-- **HDF4** (.hdf): Requires GDAL compiled with HDF4 support (common in MODIS data)
+- **HDF4** (.hdf): Requires GDAL compiled with HDF4 support (the legacy format used by MODIS and older NASA products)
 - **NetCDF** (.nc): Supported via rasterio (uses GDAL's NetCDF driver)
 
 ## Available options
@@ -164,6 +215,7 @@ Raster:
   --band BAND           Band number to display (single raster), or slice number for NetCDF. (default: 1)
   --timestep INTEGER    Alias for --band when working with NetCDF files.
   --subset INTEGER      Variable index for NetCDF/HDF files (e.g., --subset 1).
+  --reduce DIM_NAME     For 3D NetCDF variables, specify which dimension to use as the band/slider axis.  Auto-detected if omitted.
   --colormap            Apply colormap to single-band rasters. Flag without the color scheme → 'terrain'.
   --rgb R G B           Three band numbers for RGB display (e.g., --rgb 4 3 2). Overrides default 1 2 3.
   --rgbfiles R G B      Three single-band rasters for RGB composite. Can also provide as positional arguments.

viewinline-0.2.2/PKG-INFO → viewinline-0.3.0/README.md

@@ -1,40 +1,10 @@
-Metadata-Version: 2.4
-Name: viewinline
-Version: 0.2.2
-Summary: Quick look geospatial viewer for iTerm2 compatible terminals
-Project-URL: Homepage, https://github.com/nkeikon/viewinline
-Project-URL: Repository, https://github.com/nkeikon/viewinline
-Project-URL: Issues, https://github.com/nkeikon/viewinline/issues
-Author: Keiko Nomura
-License: Apache-2.0
-License-File: LICENSE
-Requires-Python: >=3.9
-Requires-Dist: geopandas
-Requires-Dist: matplotlib
-Requires-Dist: numpy
-Requires-Dist: pandas
-Requires-Dist: pillow
-Requires-Dist: pyogrio
-Requires-Dist: rasterio
-Provides-Extra: all
-Requires-Dist: duckdb; extra == 'all'
-Requires-Dist: h5py; extra == 'all'
-Requires-Dist: pyarrow; extra == 'all'
-Provides-Extra: hdf5
-Requires-Dist: h5py; extra == 'hdf5'
-Provides-Extra: parquet
-Requires-Dist: pyarrow; extra == 'parquet'
-Provides-Extra: sql
-Requires-Dist: duckdb; extra == 'sql'
-Description-Content-Type: text/markdown
-
 # viewinline
 [![Downloads](https://static.pepy.tech/badge/viewinline)](https://pepy.tech/project/viewinline)
 [![PyPI version](https://img.shields.io/pypi/v/viewinline)](https://pypi.org/project/viewinline/)
 [![Python version](https://img.shields.io/badge/python-%3E%3D3.9-blue.svg)](https://pypi.org/project/viewinline/)
 
 **Quick-look geospatial viewer for compatible terminals.**  
-Displays rasters, vectors, and tabular data directly in the terminal with no GUI and no temporary files.
+Displays rasters, vectors, and tabular data directly in the terminal with no GUI.
 
 <p align="center">
   <a href="viewinline_gif1.gif"><img src="viewinline_gif1.gif" width="49%"></a>
@@ -47,7 +17,7 @@ Particularly useful on HPC systems and remote servers accessed via SSH. Images r
 
 This tool combines the core display logic of `viewtif` and `viewgeom`, but is **non-interactive**: you can't zoom, pan, or switch colormaps on the fly. Instead, you control everything through command-line options (e.g. --display, --color-by, --colormap).
 
-It uses the iTerm2 inline image protocol (OSC 1337) to render previews. In incompatible terminals, the escape codes are silently ignored with no errors or crashes.
+It uses the iTerm2 inline image protocol (OSC 1337) in supported terminals, and falls back to `chafa` in others, which displays real high-res images in terminals like kitty and Ghostty, and colored block-art (ASCII art) previews in Terminal.app, VS Code, and most Linux terminals. Without `chafa` installed, non-iTerm2-family terminals show an info message instead.
 
 ## Installation  
 Requires Python 3.9 or later.  
@@ -61,6 +31,7 @@ pip install viewinline
 # Rasters
 viewinline path/to/file.tif
 viewinline R.tif G.tif B.tif                 # RGB composite (also works with --rgbfiles)
+viewinline hyperspectral.nc --band 50
 viewinline path/to/multiband.tif --rgb 3 2 1
 viewinline path/to/folder --gallery 4x3      # show image gallery (e.g. 4x3 grid)
 
@@ -69,6 +40,7 @@ viewinline file.nc                           # list variables
 viewinline file.nc --subset 2                # display variable 2
 viewinline file.nc --subset 1 --band 10      # variable 1, timestep 10 --band or --timestep
 viewinline temp.nc --subset 1 --colormap plasma --vmin 273 --vmax 310
+viewinline hyperspectral.nc --subset 1 --reduce NumberOfScanlines  # override auto-detected axis
 
 # Vectors
 viewinline path/to/vector.geojson
@@ -93,18 +65,31 @@ viewinline data.geoparquet --table --where "POP > 100000" --sort POP --desc
 
 ## Compatible terminals
 
-The iTerm2 inline image protocol (OSC 1337) is supported by:
+Native (no extra install required): images render via the iTerm2 inline image protocol on:
 
 - **iTerm2** (macOS)
 - **WezTerm** (cross-platform)
 - **Konsole** (Linux/KDE)
 - **Rio**, **Contour** (cross-platform)
 
-**Not compatible:** Mac Terminal, GNOME Terminal, Kitty (uses different protocol), Ghostty, Alacritty
+Via `chafa` (recommended for everyone else): install chafa and viewinline works in nearly every terminal:
+- kitty, Ghostty, foot — real high-resolution images via the kitty graphics protocol or sixel
+- Terminal.app, VS Code, GNOME Terminal, Alacritty, Warp, Hyper — colored block-art previews with 24-bit color
+
+Install chafa once (it's a system binary, available across all conda/virtualenv environments):
+
+```
+brew install chafa          # macOS
+sudo apt install chafa      # Debian/Ubuntu
+sudo dnf install chafa      # Fedora
+scoop install chafa         # Windows
+```
+Without chafa, terminals outside the native list above show an info message instead of an image.
+You can also force the chafa path on any terminal by setting `INLINE_VIEWER_ENGINE=chafa`.
 
 **SSH/HPC usage:** Works over SSH when connecting from a compatible terminal. Images render on your local machine, not the remote server. No X11 forwarding or VNC required.
 
-**tmux/screen:** Inline images don't work inside tmux or screen sessions, even with `allow-passthrough on`. Use a plain terminal tab.
+**tmux/screen:** Inline images work inside tmux only when the outer terminal is iTerm2 (or WezTerm/Konsole/Rio/Contour). In tmux with other outer terminals (kitty, Terminal.app, etc.), viewinline displays ASCII art previews instead of full-quality images.
 
 **Fallback:** In terminals that do not support inline images, you can fallback to ASCII art by installing [`chafa`](https://hpjansson.org/chafa/) command-line tool. Install `chafa` with your package manager (e.g. `brew install chafa` or `sudo apt install chafa`). You can also force the use of `chafa` by setting the environment variable `INLINE_VIEWER_ENGINE=chafa`.
 
@@ -154,7 +139,9 @@ The iTerm2 inline image protocol (OSC 1337) is supported by:
 
 **NetCDF/HDF notes:**
 - viewinline lists only variables that can be displayed as 2D or 3D arrays
-- Variables with additional dimensions (e.g., vertical levels) may be listed but will fail to display with a clear error message
+- 3D variables with time or known spatial dimensions are auto-handled (slices along the non-spatial axis)
+- For 3D variables with non-standard dimensions (e.g., hyperspectral cubes like PICARD), viewinline auto-detects the band axis by smallest dimension. Use `--reduce DIM_NAME` to override.
+- Variables with 4+ dimensions are not supported
 - For a complete variable list, use `ncdump -h file.nc` or `viewtif`
 
 ## Dependencies
@@ -167,6 +154,7 @@ The iTerm2 inline image protocol (OSC 1337) is supported by:
 - `numpy`, `pandas` — data handling
 
 **Optional dependencies:**
+- `chafa` — strongly recommended for terminal coverage beyond iTerm2/WezTerm/Konsole/Rio/Contour. System binary, not a Python package. See "Terminal support" above for install instructions.
 - `duckdb` — required for `--where`, `--sort`, `--sql`, `--limit` with filtering
   ```bash
   pip install duckdb
@@ -182,7 +170,7 @@ The iTerm2 inline image protocol (OSC 1337) is supported by:
 
 **Note on HDF support:**
 - **HDF5** (.h5, .hdf5): Supported via rasterio if GDAL has HDF5 support (most installations)
-- **HDF4** (.hdf): Requires GDAL compiled with HDF4 support (common in MODIS data)
+- **HDF4** (.hdf): Requires GDAL compiled with HDF4 support (the legacy format used by MODIS and older NASA products)
 - **NetCDF** (.nc): Supported via rasterio (uses GDAL's NetCDF driver)
 
 ## Available options
@@ -194,6 +182,7 @@ Raster:
   --band BAND           Band number to display (single raster), or slice number for NetCDF. (default: 1)
   --timestep INTEGER    Alias for --band when working with NetCDF files.
   --subset INTEGER      Variable index for NetCDF/HDF files (e.g., --subset 1).
+  --reduce DIM_NAME     For 3D NetCDF variables, specify which dimension to use as the band/slider axis.  Auto-detected if omitted.
   --colormap            Apply colormap to single-band rasters. Flag without the color scheme → 'terrain'.
   --rgb R G B           Three band numbers for RGB display (e.g., --rgb 4 3 2). Overrides default 1 2 3.
   --rgbfiles R G B      Three single-band rasters for RGB composite. Can also provide as positional arguments.

{viewinline-0.2.2 → viewinline-0.3.0}/pyproject.toml

@@ -4,8 +4,8 @@ build-backend = "hatchling.build"
 
 [project]
 name = "viewinline"
-version = "0.2.2"
-description = "Quick look geospatial viewer for iTerm2 compatible terminals"
+version = "0.3.0"
+description = "Quick look geospatial viewer for the terminal, with inline image previews"
 readme = "README.md"
 license = { text = "Apache-2.0" }
 authors = [
@@ -22,12 +22,12 @@ dependencies = [
   "pyogrio",
   "pandas"
 ]
-
 [project.optional-dependencies]
 sql = ["duckdb"]
 parquet = ["pyarrow"]
 hdf5 = ["h5py"]
-all = ["duckdb", "pyarrow", "h5py"]
+netcdf = ["netCDF4"]
+all = ["duckdb", "pyarrow", "h5py", "netCDF4"]
 
 [project.scripts]
 viewinline = "viewinline.viewinline:main"

{viewinline-0.2.2 → viewinline-0.3.0}/src/viewinline/viewinline.py

@@ -28,12 +28,19 @@ from matplotlib import colormaps
 import matplotlib as mpl
 import subprocess
 
+try:
+    import netCDF4
+    HAS_NETCDF4 = True
+except ImportError:
+    HAS_NETCDF4 = False
+
 import warnings
 
 warnings.filterwarnings("ignore", message="More than one layer found", category=UserWarning)
 warnings.filterwarnings("ignore", message="Dataset has no geotransform", category=UserWarning)
+warnings.filterwarnings("ignore", message="invalid scale_factor or add_offset attribute", category=UserWarning)
 
-__version__ = "0.2.2"
+__version__ = "0.3.0"
 
 AVAILABLE_COLORMAPS = [
     "viridis", "inferno", "magma", "plasma",
@@ -41,12 +48,61 @@ AVAILABLE_COLORMAPS = [
     "Spectral", "cubehelix", "tab10", "turbo"
 ]
 
+# Terminals that don't natively support the iTerm2 OSC 1337 inline image
+# protocol. Output for these is routed through chafa instead.
+#
+# Note: presence in this list does NOT mean "no images." Chafa auto-detects
+# the terminal and picks the best output:
+#   - kitty (xterm-kitty) → real images via kitty graphics protocol
+#   - some others (e.g. foot, Ghostty) → may render real images via sixel
+#     or kitty protocol depending on chafa's detection
+#   - most others → Unicode block-art preview with 24-bit color
+#     (Terminal.app, VS Code, GNOME Terminal, Alacritty, Warp, etc.)
+# Only terminals without chafa installed see no rendering at all.
+
 _TERMINALS_WITHOUT_IMAGES = [
-    'unknown', 'cygwin', 'tmux', 'screen',
-    'vscode', 'xterm', 'rxvt', 'alacritty', 'foot', 'st', 'kitty',
-    'gnome-terminal', 'xfce4-terminal', 'lxterminal', 'terminator',
-    'tilix', 'hyper', 'windows terminal', 'putty', 'sakura',
-    'terminology', 'eterm', 'guake', 'tilda', 'deepin-terminal',
+    # macOS
+    'Apple_Terminal',         # $TERM_PROGRAM for Terminal.app
+
+    # kitty (renders real images via chafa → kitty graphics protocol)
+    'xterm-kitty',            # $TERM in kitty
+
+    # tmux / screen (TERM strings; $TMUX env var also signals tmux)
+    'screen', 'screen-256color',
+    'tmux', 'tmux-256color',
+
+    # Editors / IDE terminals
+    'vscode',                 # $TERM_PROGRAM in VS Code integrated terminal
+
+    # Cross-platform terminals known not to support OSC 1337
+    'alacritty',
+    'foot',                   # supports sixel → chafa renders real images
+    'ghostty', 'xterm-ghostty',
+    'WarpTerminal',           # $TERM_PROGRAM in Warp
+    'Hyper',                  # $TERM_PROGRAM in Hyper
+
+    # Generic / legacy
+    'unknown',
+    'cygwin',
+    'rxvt', 'rxvt-unicode', 'rxvt-unicode-256color',
+    'st-256color',            # suckless st
+
+    # Linux desktop terminals (most are VTE-based, no OSC 1337)
+    'gnome-terminal',
+    'xfce4-terminal',
+    'lxterminal',
+    'terminator',
+    'tilix',
+    'sakura',
+    'terminology',
+    'guake',
+    'tilda',
+    'deepin-terminal',
+    'eterm',
+
+    # Windows
+    'putty',
+    'Windows Terminal',
 ]
 
 def detect_terminal() -> dict[str, str]:
@@ -135,41 +191,51 @@ def show_inline_image(image_array: np.ndarray, display_scale = None, is_vector:
 
     if _TERMINAL_SUPPORTS_IMAGES:
         sys.stdout.write(f"\033]1337;File=inline=1;width={width_pct}%:{encoded}\a\n")
-    else:
+    else: 
         if is_chafa_available():
+            # Inside tmux, force chafa to use block-art symbols instead of
+            # graphics protocols. Tmux mangles kitty graphics and sixel
+            # protocols, producing dot-character garbage on screen. Block-art
+            # passes through tmux reliably on every outer terminal.
+            chafa_args = ["chafa", "-"]
+            if os.environ.get("TMUX"):
+                chafa_args = ["chafa", "-f", "symbols", "-"]
             chafa_output = subprocess.check_output(
-                ["chafa", "-"],  
+                chafa_args,
                 input=image_bytes
             ).decode()
             
             sys.stdout.write(f"\n{chafa_output}\a\n")
+            
         else:
             sys.stdout.write(f"[INFO] Use supported terminal or install 'chafa' for ascii art fallback. Detected: {_TERMINAL_INFO}\n")
     
     sys.stdout.flush()
 
-
-def show_image_auto(img: np.ndarray, display_scale = None, is_vector: bool = False) -> None:
-    """Attempt inline image display. No fallbacks, no detection.
+def show_image_auto(img: np.ndarray, display_scale=None, is_vector: bool = False) -> None:
+    """Render an image inline, with chafa fallback for non-iTerm2 terminals.
     
-    Just sends the iTerm2 inline image escape sequence. If the terminal supports it,
-    great. If not, the escape codes are ignored and nothing happens.
+    Cascade:
+      1. If terminal supports OSC 1337 → emit iTerm2 inline image sequence.
+      2. Else if chafa is installed → pipe through chafa (which auto-detects
+         and emits the terminal's native graphics protocol or block-art).
+      3. Else → print an info message suggesting chafa installation.
     
+    The branching happens inside show_inline_image(); this wrapper handles
+    status messaging and exception safety.
     """
-    if os.environ.get("TMUX"):
-        print("[WARN] Inside tmux — inline images won't display even with allow-passthrough on (known iTerm2/tmux issue). Use a plain terminal tab.")
-        return
-    
     try:
         show_inline_image(img, display_scale, is_vector)
-        print("[VIEW] Image sent — visible in compatible terminals")
+        if _TERMINAL_SUPPORTS_IMAGES:
+            print("[VIEW] Inline render complete")
+        elif is_chafa_available():
+            print("[VIEW] Inline render complete via chafa")
+        # If neither path applies, show_inline_image already printed the info message
     except Exception as e:
-        # If image encoding fails, print error but don't crash
-        print(f"[ERROR] Failed to encode image: {e}")
+        print(f"[ERROR] Failed to render image: {e}")
         import traceback
         traceback.print_exc()
 
-
 def resize_to_terminal(img: np.ndarray) -> tuple[np.ndarray, float]:
     """Resize image to fit terminal window (approx 8x16 pixel cells)."""
     cols, rows = shutil.get_terminal_size((100, 40))
@@ -196,7 +262,6 @@ def load_csv_to_df(path: str) -> pd.DataFrame:
         print(f"[ERROR] Failed to read CSV: {e}")
         return pd.DataFrame()
 
-
 # =============================================================
 # Preview
 # =============================================================
@@ -577,6 +642,177 @@ def render_simple_image(filepath: str, args) -> None:
     except Exception as e:
         print(f"[ERROR] Failed to load image: {e}")
 
+def render_netcdf_via_netcdf4(path, args):
+    """Read a NetCDF file via netCDF4 (bypassing GDAL). Handles hierarchical
+    groups and hyperspectral cubes where GDAL aborts or interprets axes wrong.
+    """
+    if not HAS_NETCDF4:
+        print("[ERROR] netCDF4 not installed. Install with:")
+        print("        pip install netCDF4")
+        print("        or:  pip install viewinline[netcdf]")
+        return
+
+    try:
+        nc = netCDF4.Dataset(path)
+    except Exception as e:
+        print(f"[ERROR] Could not open NetCDF file: {e}")
+        return
+
+    # Recursively collect (path, variable) pairs across all groups
+    def collect_vars(group, prefix=""):
+        out = []
+        for name, var in group.variables.items():
+            full_name = f"{prefix}{name}"
+            out.append((full_name, var))
+        for sub_name, sub in group.groups.items():
+            out.extend(collect_vars(sub, f"{prefix}{sub_name}/"))
+        return out
+
+    all_vars = collect_vars(nc)
+
+    if not all_vars:
+        print("[ERROR] No variables found in file.")
+        nc.close()
+        return
+
+    # If no --subset, list all variables and exit
+    if not args.subset:
+        print(f"Found {len(all_vars)} variables in {os.path.basename(path)}:")
+        for i, (name, var) in enumerate(all_vars, 1):
+            shape_str = "x".join(str(s) for s in var.shape)
+            print(f"  [{i}] {name}  ({shape_str}, {var.dtype})")
+        print(f"\nUse --subset <N> to display a specific variable.")
+        nc.close()
+        return
+
+    # Validate --subset
+    if args.subset < 1 or args.subset > len(all_vars):
+        print(f"[ERROR] --subset must be between 1 and {len(all_vars)}")
+        nc.close()
+        return
+
+    var_name, var = all_vars[args.subset - 1]
+    print(f"[INFO] Displaying variable {args.subset}: {var_name}")
+    print(f"[DATA] Shape: {var.shape}  dtype: {var.dtype}  dims: {var.dimensions}")
+
+    # Detect dimensionality and read the right slice
+    if var.ndim == 2:
+        data = np.asarray(var[:, :], dtype=np.float64)
+        slice_info = "2D variable"
+
+    elif var.ndim == 3:
+        spatial_dims = {'lat', 'lon', 'latitude', 'longitude', 'y', 'x'}
+        
+        spectral_axis = None
+        
+        # 1. User override via --reduce
+        if args.reduce_dim is not None:
+            if args.reduce_dim in var.dimensions:
+                spectral_axis = list(var.dimensions).index(args.reduce_dim)
+                print(f"[INFO] Using user-specified --reduce '{args.reduce_dim}'")
+            else:
+                print(f"[ERROR] --reduce '{args.reduce_dim}' is not a dimension of this variable.")
+                print(f"[INFO] Available dimensions: {list(var.dimensions)}")
+                nc.close()
+                return
+        
+        # 2. Standard convention: reduce along the non-spatial dim
+        if spectral_axis is None:
+            has_standard_spatial = any(d in spatial_dims for d in var.dimensions)
+            if has_standard_spatial:
+                for i, dim_name in enumerate(var.dimensions):
+                    if dim_name not in spatial_dims:
+                        spectral_axis = i
+                        break
+        
+        # 3. Fallback heuristic: smallest dim is typically the band axis
+        if spectral_axis is None:
+            sizes = [(i, var.shape[i]) for i in range(3)]
+            spectral_axis = min(sizes, key=lambda x: x[1])[0]
+            print(f"[INFO] Non-standard dimensions detected: {list(var.dimensions)}")
+            print(f"[INFO] Reducing along '{var.dimensions[spectral_axis]}' (size {var.shape[spectral_axis]}, assumed band/spectral axis)")
+            print(f"[INFO] If this is not correct, use --reduce DIM_NAME to override.")
+        
+        # Slice along chosen axis
+        band_count = var.shape[spectral_axis]
+        band_num = args.band if args.band is not None else 1
+        band_idx = max(0, min(band_num - 1, band_count - 1))
+        slicer = [slice(None)] * 3
+        slicer[spectral_axis] = band_idx
+        data = np.asarray(var[tuple(slicer)], dtype=np.float64)
+        slice_info = f"slice along axis {spectral_axis} ({var.dimensions[spectral_axis]}), band {band_idx + 1} of {band_count}"
+
+    else:
+        print(f"[ERROR] viewinline only supports 2D or 3D variables. This one is {var.ndim}D.")
+        nc.close()
+        return
+
+    print(f"[DATA] {slice_info}")
+    # Apply fill value
+    fill = getattr(var, '_FillValue', None)
+    if fill is not None:
+        data = np.where(data == fill, np.nan, data)
+    # Flip vertically if data is stored south-to-north so north appears at top.
+    # Determine which dims remain after slicing — for the 2D result, figure out
+    # which axis (0 or 1) corresponds to latitude, and check that dim's coord values.
+    if var.ndim == 2:
+        remaining_dims = list(var.dimensions)
+    elif var.ndim == 3:
+        if spectral_axis is not None:
+            remaining_dims = [d for i, d in enumerate(var.dimensions) if i != spectral_axis]
+        else:
+            remaining_dims = list(var.dimensions[1:])  # axis 0 was reduced
+    else:
+        remaining_dims = []
+    lat_names = {'lat', 'latitude', 'y'}
+    for axis_in_2d, dim_name in enumerate(remaining_dims):
+        if dim_name in lat_names and dim_name in nc.variables:
+            lat_vals = nc[dim_name][:]
+            if len(lat_vals) > 1 and lat_vals[0] < lat_vals[-1]:
+                data = np.flip(data, axis=axis_in_2d)
+                print(f"[INFO] Flipped along '{dim_name}' for display (data stored south-to-north).")
+            break
+    nc.close()
+
+    # Normalize and display
+    band_u8 = normalize_to_uint8(data, vmin=args.vmin, vmax=args.vmax,
+                                 nodata=args.nodata)
+
+    if args.colormap:
+        cmap = colormaps[args.colormap]
+        colored = cmap(band_u8 / 255.0)
+        img = (colored[:, :, :3] * 255).astype(np.uint8)
+        print(f"[INFO] Applying colormap: {args.colormap}")
+    else:
+        img = np.stack([band_u8] * 3, axis=-1)
+        print("[INFO] Displaying grayscale")
+
+    # Resize to terminal
+    H, W = img.shape[:2]
+    if args.display:
+        new_w, new_h = max(1, int(W * args.display)), max(1, int(H * args.display))
+        img = np.array(Image.fromarray(img).resize((new_w, new_h), Image.BILINEAR))
+        print(f"[VIEW] Manual resize ×{args.display:.2f} → {new_w}×{new_h}px")
+    # else:
+    #     img, scale = resize_to_terminal(img)
+    #     print(f"[VIEW] Rendered image size → {img.shape[1]}×{img.shape[0]}px (size={scale:.2f})")
+    else:
+            max_dim = 2000
+            if max(img.shape[:2]) > max_dim:
+                scale = max_dim / max(img.shape[:2])
+                new_w = int(img.shape[1] * scale)
+                new_h = int(img.shape[0] * scale)
+                img = np.array(Image.fromarray(img).resize((new_w, new_h), Image.BILINEAR))
+                print(f"[VIEW] Downsampled from {W}×{H}px to {new_w}×{new_h}px (scale={scale:.2f})")
+                print(f"[INFO] Use --display 1 for full resolution.")
+            else:
+            # (matches the width_pct logic in show_inline_image)
+                display_pct = args.display if args.display is not None else 0.33
+            
+                print(f"[VIEW] Rendered image size → {img.shape[1]}×{img.shape[0]}px (size={display_pct:.2f})")  
+       
+    show_image_auto(img, getattr(args, "display", None), is_vector=False)
+
 def render_raster(paths: list[str], args) -> None:
     try:
         import rasterio
@@ -589,10 +825,16 @@ def render_raster(paths: list[str], args) -> None:
 
         if len(paths) == 1:
             path = paths[0]
+
+            if path.lower().endswith('.nc'):
+                render_netcdf_via_netcdf4(path, args)
+                return
             
             # Handle NetCDF/HDF with subdatasets
+        
             if path.lower().endswith(('.nc', '.hdf', '.hdf5', '.h5')):
                 try:
+                
                     with rasterio.open(path) as src:
                         subdatasets = src.subdatasets
                     
@@ -665,7 +907,7 @@ def render_raster(paths: list[str], args) -> None:
                         resampling=rasterio.enums.Resampling.bilinear
                     )
 
-                    print(f"[PROC] Downsampled for preview → {out_w}×{out_h}px (scale={scale:.3f})")
+                    print(f"[VIEW] Downsampled for preview → {out_w}×{out_h}px (scale={scale:.3f})")
                 else:
                     data = ds.read()
 
@@ -677,8 +919,11 @@ def render_raster(paths: list[str], args) -> None:
                     print(f"[INFO] Multi-band raster detected ({band_count} bands)")
 
             # MULTI BAND RGB (skip for NetCDF - treat as slices/timesteps, not RGB)
-            if band_count >= 3 and not paths[0].lower().endswith('.nc'):
-
+            # if band_count >= 3 and not paths[0].lower().endswith('.nc'):
+            # Auto-composite to RGB only when user didn't explicitly ask for a single band
+            # user_specified_band = args.band is not None and args.band != 1
+            user_specified_band = args.band is not None
+            if band_count >= 3 and not paths[0].lower().endswith('.nc') and not user_specified_band:
 
                 if getattr(args, "rgb", None):
                     try:
@@ -707,7 +952,8 @@ def render_raster(paths: list[str], args) -> None:
             # SINGLE BAND
             else:
 
-                band_idx = max(0, min(args.band - 1, band_count - 1))
+                band_num = args.band if args.band is not None else 1
+                band_idx = max(0, min(band_num - 1, band_count - 1))
                 # print(f"[INFO] Displaying band {band_idx + 1} of {band_count}")
                 raw_band = data[band_idx].astype(float)
 
@@ -784,7 +1030,7 @@ def render_raster(paths: list[str], args) -> None:
             print(f"[ERROR] Cannot display this variable.")
             print("[INFO] viewinline only supports 2D or 3D NetCDF variables")
         else:
-            print(f"[ERROR] Raster rendering failed: {e}")
+            print(f"[ERROR] Inline render failed: {e}")
 
 
 def render_gallery(folder: str, grid: str = "4x4", display_scale=None, is_vector=False) -> None:
@@ -889,13 +1135,6 @@ def render_vector(path, args):
     except Exception as e:
         print(f"[WARN] Could not list layers: {e}")
 
-    # try:
-    #     gdf = gpd.read_file(path, layer=getattr(args, "layer", None))
-    #     print(f"[DATA] Vector loaded: {os.path.basename(path)} ({len(gdf)} features)")
-    # except Exception as e:
-    #     print(f"[ERROR] Failed to read vector: {e}")
-    #     return
-
     try:
         # Use read_parquet for parquet/geoparquet files
         if path.lower().endswith(('.parquet', '.geoparquet')):
@@ -1237,7 +1476,7 @@ def main() -> None:
 
     # Raster options
     parser.add_argument(
-        "--band", type=int, default=1,
+        "--band", type=int, default=None,
         help="Band number to display (single raster case), or slice number for NetCDF."
     )
     parser.add_argument(
@@ -1253,6 +1492,10 @@ def main() -> None:
         "--rgb", nargs=3, type=int, metavar=('R', 'G', 'B'), default=None,
         help="Three band numbers for RGB display (e.g., --rgb 4 3 2). Overrides default 1 2 3."
     )
+    parser.add_argument(
+        "--rgbfiles", nargs=3, type=str, metavar=('R', 'G', 'B'),
+        help="Three single-band rasters for RGB composite (e.g., --rgbfiles R.tif G.tif B.tif). Can also provide as positional arguments without the flag."
+    )
     parser.add_argument(
         "--vmin", type=float, default=None,
         help="Minimum pixel value for raster display scaling."
@@ -1274,8 +1517,9 @@ def main() -> None:
         help="Variable index for NetCDF files (e.g. --subset 1)."
     )
     parser.add_argument(
-        "--rgbfiles", nargs=3, type=str, metavar=('R', 'G', 'B'),
-        help="Three single-band rasters for RGB composite (e.g., --rgbfiles R.tif G.tif B.tif). Can also provide as positional arguments without the flag."
+        "--reduce", dest="reduce_dim", type=str, default=None,
+        metavar="DIM_NAME",
+        help="For 3D NetCDF variables, specify which dimension to use as the band axis (auto-detected if omitted)."
     )
 
     # CSV options
@@ -1359,7 +1603,7 @@ def main() -> None:
     parser.add_argument(
     "--table", action="store_true",
     help="Display vector/parquet file as tabular data instead of rendering geometry."
-)
+    ) 
 
     parser.add_argument("--version", action="version", version=f"%(prog)s {__version__}")