hooks: rewrite pygraphviz hook

Rewrite `pygraphviz` hook to fix discovery and collection of `graphviz`
files under various Linux distributions, in Anaconda environments
(Windows, Linux, and macOS), and msys2 environments (Windows).

We now discover active `graphviz` installation on all OSes by looking
for `dot` program in `PATH`. To find the location of shared libraries
and the plugin directory on non-Windows, we now perform binary
dependency analysis on the `dot` executable and look for known
`graphviz` shared libraries in the results.

This improves the discovery on various platforms, and also fixes the
issue where system-installed (on Linux) or Homebrew-installed
(on macOS) `graphviz` shared libraries and plugins would end up being
collected  instead of Anaconda-packaged ones when using Anaconda
environment with Anaconda-packaged `graphviz`.

_pyinstaller_hooks_contrib/stdhooks/hook-pygraphviz.py

@@ -9,58 +9,137 @@
 #
 # SPDX-License-Identifier: GPL-2.0-or-later
 # ------------------------------------------------------------------
-
-import glob
 import os
+import pathlib
 import shutil
 
-from PyInstaller.compat import is_win, is_darwin
-from PyInstaller.depend.bindepend import findLibrary
-
-binaries = []
-datas = []
-
-# List of binaries agraph.py may invoke.
-progs = [
-    "neato",
-    "dot",
-    "twopi",
-    "circo",
-    "fdp",
-    "nop",
-    "acyclic",
-    "gvpr",
-    "gvcolor",
-    "ccomps",
-    "sccmap",
-    "tred",
-    "sfdp",
-    "unflatten",
-]
-
-if is_win:
-    for prog in progs:
-        for binary in glob.glob("c:/Program Files/Graphviz*/bin/" + prog + ".exe"):
-            binaries.append((binary, "."))
-    for binary in glob.glob("c:/Program Files/Graphviz*/bin/*.dll"):
-        binaries.append((binary, "."))
-    for data in glob.glob("c:/Program Files/Graphviz*/bin/config*"):
-        datas.append((data, "."))
-else:
-    # The dot binary in PATH is typically a symlink, handle that.
-    # graphviz_bindir is e.g. /usr/local/Cellar/graphviz/2.46.0/bin
-    graphviz_bindir = os.path.dirname(os.path.realpath(shutil.which("dot")))
-    for binary in progs:
-        binaries.append((graphviz_bindir + "/" + binary, "."))
-    if is_darwin:
-        suffix = "dylib"
-        # graphviz_libdir is e.g. /usr/local/Cellar/graphviz/2.46.0/lib/graphviz
-        graphviz_libdir = os.path.realpath(graphviz_bindir + "/../lib/graphviz")
+from PyInstaller import compat
+from PyInstaller.depend import bindepend
+from PyInstaller.utils.hooks import logger
+
+
+def _collect_graphviz_files():
+    binaries = []
+    datas = []
+
+    # A working `pygraphviz` installation requires graphviz programs in PATH. Attempt to resolve the `dot` executable to
+    # see if this is the case.
+    dot_binary = shutil.which('dot')
+    if not dot_binary:
+        logger.warning(
+            "hook-pygraphviz: 'dot' program not found in PATH!"
+        )
+        return binaries, datas
+    logger.info("hook-pygraphviz: found 'dot' program: %r", dot_binary)
+    bin_dir = pathlib.Path(dot_binary).parent
+
+    # Collect graphviz programs that might be called from `pygaphviz.agraph.AGraph`:
+    # https://github.com/pygraphviz/pygraphviz/blob/pygraphviz-1.14/pygraphviz/agraph.py#L1330-L1348
+    # On macOS and on Linux, most of these are symbolic links to a single executable.
+    progs = (
+        "neato",
+        "dot",
+        "twopi",
+        "circo",
+        "fdp",
+        "nop",
+        "osage",
+        "patchwork",
+        "gc",
+        "acyclic",
+        "gvpr",
+        "gvcolor",
+        "ccomps",
+        "sccmap",
+        "tred",
+        "sfdp",
+        "unflatten",
+    )
+
+    logger.debug("hook-pygraphviz: collecting graphviz program executables...")
+    for program_name in progs:
+        program_binary = shutil.which(program_name)
+        if not program_binary:
+            logger.debug("hook-pygaphviz: graphviz program %r not found!", program_name)
+            continue
+
+        # Ensure that the program executable was found in the same directory as the `dot` executable. This should
+        # prevent us from falling back to other graphviz installations that happen to be in PATH.
+        if pathlib.Path(program_binary).parent != bin_dir:
+            logger.debug(
+                "hook-pygraphviz: found program %r (%r) outside of directory %r - ignoring!",
+                program_name, program_binary, str(bin_dir)
+            )
+            continue
+
+        logger.debug("hook-pygraphviz: collecting graphviz program %r: %r", program_name, program_binary)
+        binaries += [(program_binary, '.')]
+
+    # Graphviz shared libraries should be automatically collected when PyInstaller performs binary dependency
+    # analysis of the collected program executables. However, we need to collect plugins and their accompanying
+    # config file.
+    logger.debug("hook-pygraphviz: looking for graphviz plugin directory...")
+    if compat.is_win:
+        # Under Windows, we have several installation variants:
+        #  - official installer
+        #  - chocolatey
+        #  - msys2
+        #  - Anaconda
+        # In all variants, the plugins and the config file are located in the `bin`  directory, next to the program
+        # executables.
+        plugin_dir = bin_dir
+        plugin_dest_dir = '.'  # Collect into top-level application directory
+        # Official installer and Anaconda use unversioned `gvplugin-{name}.dll` plugin names, while msys2 uses
+        # versioned `libgvplugin-{name}-{version}.dll` plugin names.
+        plugin_pattern = '*gvplugin*.dll'
     else:
-        suffix = "so"
-        # graphviz_libdir is e.g. /usr/lib64/graphviz
-        graphviz_libdir = os.path.join(os.path.dirname(findLibrary('libcdt')), 'graphviz')
-    for binary in glob.glob(graphviz_libdir + "/*." + suffix):
-        binaries.append((binary, "graphviz"))
-    for data in glob.glob(graphviz_libdir + "/config*"):
-        datas.append((data, "graphviz"))
+        # Perform binary dependency analysis on the `dot` executable to obtain path to the graphiz shared libraries.
+        # These need to be in library search path for the programs to work, or discoverable via run-paths (e.g.,
+        # Anaconda on Linux and macOS, Homebrew on macOS).
+        graphviz_lib_candidates = ['cdt', 'gvc', 'cgraph']
+
+        if hasattr(bindepend, 'get_imports'):
+            # PyInstaller >= 6.0
+            dot_imports = [path for name, path in bindepend.get_imports(dot_binary) if path is not None]
+        else:
+            # PyInstaller < 6.0
+            dot_imports = bindepend.getImports(dot_binary)
+
+        graphviz_lib_paths = [
+            path for path in dot_imports
+            if any(candidate in os.path.basename(path) for candidate in graphviz_lib_candidates)
+        ]
+
+        if not graphviz_lib_paths:
+            logger.warning("hook-pygraphviz: could not determine location of graphviz shared libraries!")
+            return binaries, datas
+
+        graphviz_lib_dir = pathlib.Path(graphviz_lib_paths[0]).parent
+        logger.debug("hook-pygraphviz: location of graphviz shared libraries: %r", str(graphviz_lib_dir))
+
+        # Plugins should be located in `graphviz` directory next to shared libraries.
+        plugin_dir = graphviz_lib_dir / 'graphviz'
+        plugin_dest_dir = 'graphviz'  # Colelct into graphviz sub-directory
+
+        if compat.is_darwin:
+            plugin_pattern = '*gvplugin*.dylib'
+        else:
+            # Collect only versioned .so library files (for example, `/lib64/graphviz/libgvplugin_core.so.6` and
+            # `/lib64/graphviz/libgvplugin_core.so.6.0.0`). The unversioned .so library files (for example`,
+            # lib64/graphviz/libgvplugin_core.so`), if available, are meant for linking (and are usually installed
+            # as part of development package).
+            plugin_pattern = '*gvplugin*.so.*'
+
+    if not plugin_dir.is_dir():
+        logger.warning("hook-pygraphviz: could not determine location of graphviz plugins!")
+        return binaries, datas
+
+    logger.info("hook-pygraphviz: collecting graphviz plugins directory: %r", str(plugin_dir))
+
+    binaries += [(str(file), plugin_dest_dir) for file in plugin_dir.glob(plugin_pattern)]
+    datas += [(str(file), plugin_dest_dir) for file in plugin_dir.glob("config*")]
+
+    return binaries, datas
+
+
+binaries, datas = _collect_graphviz_files()

news/849.update.rst

@@ -0,0 +1,3 @@
+Rewrite ``pygraphviz`` hook to fix discovery and collection of ``graphviz``
+files under various Linux distributions, in Anaconda environments
+(Windows, Linux, and macOS), and msys2 environments (Windows).