New command to generate source distribution based on a manifest file.

command/dist.py

+"""distutils.command.dist
+
+Implements the Distutils 'dist' command (create a source distribution)."""
+
+# created 1999/09/22, Greg Ward
+
+__rcsid__ = "$Id$"
+
+import sys, os, string, re
+import fnmatch
+from types import *
+from glob import glob
+from distutils.core import Command
+from distutils.text_file import TextFile
+
+
+# Possible modes of operation:
+#   - require an explicit manifest that lists every single file (presumably
+#     along with a way to auto-generate the manifest)
+#   - require an explicit manifest, but allow it to have globs or
+#     filename patterns of some kind (and also have auto-generation)
+#   - allow an explict manifest, but automatically augment it at runtime
+#     with the source files mentioned in 'packages', 'py_modules', and
+#     'ext_modules' (and any other such things that might come along)
+
+# I'm liking the third way.  Possible gotchas:
+#   - redundant specification: 'packages' includes 'foo' and manifest
+#     includes 'foo/*.py'
+#   - obvious conflict: 'packages' includes 'foo' and manifest
+#     includes '! foo/*.py' (can't imagine why you'd want this)
+#   - subtle conflict:  'packages' includes 'foo' and manifest
+#     includes '! foo/bar.py' (this could well be desired: eg. exclude
+#     an experimental module from distribution)
+
+# Syntax for the manifest file:
+#   - if a line is just a Unix-style glob by itself, it's a "simple include
+#     pattern": go find all files that match and add them to the list
+#     of files
+#   - if a line is a glob preceded by "!", then it's a "simple exclude
+#     pattern": go over the current list of files and exclude any that
+#     match the glob pattern
+#   - if a line consists of a directory name followed by zero or more
+#     glob patterns, then we'll recursively explore that directory tree
+#     - the glob patterns can be include (no punctuation) or exclude
+#       (prefixed by "!", no space)
+#     - if no patterns given or the first pattern is not an include pattern,
+#       then assume "*" -- ie. find everything (and then start applying
+#       the rest of the patterns)
+#     - the patterns are given in order of increasing precedence, ie.
+#       the *last* one to match a given file applies to it
+# 
+# example (ignoring auto-augmentation!):
+#   distutils/*.py
+#   distutils/command/*.py
+#   ! distutils/bleeding_edge.py
+#   examples/*.py
+#   examples/README
+# 
+# smarter way (that *will* include distutils/command/bleeding_edge.py!)
+#   distutils *.py
+#   ! distutils/bleeding_edge.py
+#   examples !*~ !*.py[co]     (same as: examples * !*~ !*.py[co])
+#   test test_* *.txt !*~ !*.py[co]
+#   README
+#   setup.py
+#
+# The actual Distutils manifest (don't need to mention source files,
+# README, setup.py -- they're automatically distributed!):
+#   examples !*~ !*.py[co]
+#   test !*~ !*.py[co]
+
+# The algorithm that will make it work:
+#   files = stuff from 'packages', 'py_modules', 'ext_modules',
+#     plus README, setup.py, ... ?
+#   foreach pattern in manifest file:
+#     if simple-include-pattern:         # "distutils/*.py"
+#       files.append (glob (pattern))
+#     elif simple-exclude-pattern:       # "! distutils/foo*"
+#       xfiles = glob (pattern)
+#       remove all xfiles from files
+#     elif recursive-pattern:            # "examples" (just a directory name)
+#       patterns = rest-of-words-on-line
+#       dir_files = list of all files under dir
+#       if patterns:
+#         if patterns[0] is an exclude-pattern:
+#           insert "*" at patterns[0]
+#         for file in dir_files:
+#           for dpattern in reverse (patterns):
+#             if file matches dpattern:
+#               if dpattern is an include-pattern:
+#                 files.append (file)
+#               else:
+#                 nothing, don't include it
+#               next file
+#       else:
+#         files.extend (dir_files)    # ie. accept all of them
+
+
+# Anyways, this is all implemented below -- BUT it is largely untested; I
+# know it works for the simple case of distributing the Distutils, but
+# haven't tried it on more complicated examples.  Undoubtedly doing so will
+# reveal bugs and cause delays, so I'm waiting until after I've released
+# Distutils 0.1.
+
+
+# Other things we need to look for in creating a source distribution:
+#   - make sure there's a README
+#   - make sure the distribution meta-info is supplied and non-empty
+#     (*must* have name, version, ((author and author_email) or
+#     (maintainer and maintainer_email)), url
+#
+# Frills:
+#   - make sure the setup script is called "setup.py"
+#   - make sure the README refers to "setup.py" (ie. has a line matching
+#     /^\s*python\s+setup\.py/)
+
+# A crazy idea that conflicts with having/requiring 'version' in setup.py:
+#   - make sure there's a version number in the "main file" (main file
+#     is __init__.py of first package, or the first module if no packages,
+#     or the first extension module if no pure Python modules)
+#   - XXX how do we look for __version__ in an extension module?
+#   - XXX do we import and look for __version__? or just scan source for
+#     /^__version__\s*=\s*"[^"]+"/ ?
+#   - what about 'version_from' as an alternative to 'version' -- then
+#     we know just where to search for the version -- no guessing about
+#     what the "main file" is
+
+
+
+class Dist (Command):
+
+    options = [('formats=', 'f',
+                "formats for source distribution (tar, ztar, gztar, or zip)"),
+               ('manifest=', 'm',
+                "name of manifest file"),
+              ]
+
+    default_format = { 'posix': 'gztar',
+                       'nt': 'zip' }
+
+    exclude_re = re.compile (r'\s*!\s*(\S+)') # for manifest lines
+
+
+    def set_default_options (self):
+        self.formats = None
+        self.manifest = None
+
+
+    def set_final_options (self):
+        if self.formats is None:
+            try:
+                self.formats = [self.default_format[os.name]]
+            except KeyError:
+                raise DistutilsPlatformError, \
+                      "don't know how to build source distributions on " + \
+                      "%s platform" % os.name
+        elif type (self.formats) is StringType:
+            self.formats = string.split (self.formats, ',')
+
+        if self.manifest is None:
+            self.manifest = "MANIFEST"
+
+
+    def run (self):
+
+        self.check_metadata ()
+
+        self.files = []
+        self.find_defaults ()
+        self.read_manifest ()
+
+        self.make_distribution ()
+
+
+    def check_metadata (self):
+
+        dist = self.distribution
+
+        missing = []
+        for attr in ('name', 'version', 'url'):
+            if not (hasattr (dist, attr) and getattr (dist, attr)):
+                missing.append (attr)
+
+        if missing:
+            self.warn ("missing required meta-data: " +
+                       string.join (missing, ", "))
+
+        if dist.author:
+            if not dist.author_email:
+                self.warn ("missing meta-data: if 'author' supplied, " +
+                           "'author_email' must be supplied too")
+        elif dist.maintainer:
+            if not dist.maintainer_email:
+                self.warn ("missing meta-data: if 'maintainer' supplied, " +
+                           "'maintainer_email' must be supplied too")
+        else:
+            self.warn ("missing meta-data: either author (and author_email) " +
+                       "or maintainer (and maintainer_email) " +
+                       "must be supplied")
+
+    # check_metadata ()
+
+
+    def find_defaults (self):
+
+        standards = ['README', 'setup.py']
+        for fn in standards:
+            if os.path.exists (fn):
+                self.files.append (fn)
+            else:
+                self.warn ("standard file %s not found" % fn)
+
+        optional = ['test/test*.py']
+        for pattern in optional:
+            files = glob (pattern)
+            if files:
+                self.files.extend (files)
+
+        if self.distribution.packages or self.distribution.py_modules:
+            build_py = self.find_peer ('build_py')
+            build_py.ensure_ready ()
+            self.files.extend (build_py.get_source_files ())
+
+        if self.distribution.ext_modules:
+            build_ext = self.find_peer ('build_ext')
+            build_ext.ensure_ready ()
+            self.files.extend (build_ext.get_source_files ())
+
+
+
+    def open_manifest (self, filename):
+        return TextFile (filename,
+                         strip_comments=1,
+                         skip_blanks=1,
+                         join_lines=1,
+                         lstrip_ws=1,
+                         rstrip_ws=1,
+                         collapse_ws=1)
+
+
+    def search_dir (self, dir, patterns):
+
+        allfiles = findall (dir)
+        if patterns:
+            if patterns[0][0] == "!":   # starts with an exclude spec?
+                patterns.insert (0, "*")# then accept anything that isn't
+                                        # explicitly excluded
+
+            act_patterns = []           # "action-patterns": (include,regexp)
+                                        # tuples where include is a boolean
+            for pattern in patterns:
+                if pattern[0] == '!':
+                    act_patterns.append \
+                        ((0, re.compile (fnmatch.translate (pattern[1:]))))
+                else:
+                    act_patterns.append \
+                        ((1, re.compile (fnmatch.translate (pattern))))
+            act_patterns.reverse()
+
+
+            files = []
+            for file in allfiles:
+                for (include,regexp) in act_patterns:
+                    if regexp.match (file):
+                        if include:
+                            files.append (file)
+                        break           # continue to next file
+        else:
+            files = allfiles
+
+        return files
+
+    # search_dir ()
+
+
+    def exclude_files (self, pattern):
+
+        regexp = re.compile (fnmatch.translate (pattern))
+        for i in range (len (self.files)-1, -1, -1):
+            if regexp.match (self.files[i]):
+                del self.files[i]
+
+
+    def read_manifest (self):
+
+        # self.files had better already be defined (and hold the
+        # "automatically found" files -- Python modules and extensions,
+        # README, setup script, ...)
+        assert self.files is not None
+
+        manifest = self.open_manifest (self.manifest)
+        while 1:
+
+            pattern = manifest.readline()
+            if pattern is None:            # end of file
+                break
+
+            # Cases:
+            #   1) simple-include: "*.py", "foo/*.py", "doc/*.html", "FAQ"
+            #   2) simple-exclude: same, prefaced by !
+            #   3) recursive: multi-word line, first word a directory
+
+            exclude = self.exclude_re.match (pattern)
+            if exclude:
+                pattern = exclude.group (1)
+
+            words = string.split (pattern)
+            assert words                # must have something!
+            if os.name != 'posix':
+                words[0] = apply (os.path.join, string.split (words[0], '/'))
+
+            # First word is a directory, possibly with include/exclude
+            # patterns making up the rest of the line: it's a recursive
+            # pattern
+            if os.path.isdir (words[0]):
+                if exclude:
+                    file.warn ("exclude (!) doesn't apply to " +
+                               "whole directory trees")
+                    continue
+
+                dir_files = self.search_dir (words[0], words[1:])
+                self.files.extend (dir_files)
+
+            # Multiple words in pattern: that's a no-no unless the first
+            # word is a directory name
+            elif len (words) > 1:
+                file.warn ("can't have multiple words unless first word " +
+                           "('%s') is a directory name" % words[0])
+                continue
+
+            # Single word, no bang: it's a "simple include pattern"
+            elif not exclude:
+                matches = glob (pattern)
+                if matches:
+                    self.files.extend (matches)
+                else:
+                    manifest.warn ("no matches for '%s' found" % pattern)
+
+
+            # Single word prefixed with a bang: it's a "simple exclude pattern"
+            else:
+                if self.exclude_files (pattern) == 0:
+                    file.warn ("no files excluded by '%s'" % pattern)
+
+            # if/elif/.../else on 'pattern'
+
+        # loop over lines of 'manifest'
+
+    # read_manifest ()
+
+
+    def make_release_tree (self, base_dir, files):
+
+        # XXX this is Unix-specific
+
+        # First get the list of directories to create
+        need_dir = {}
+        for file in files:
+            need_dir[os.path.join (base_dir, os.path.dirname (file))] = 1
+        need_dirs = need_dir.keys()
+        need_dirs.sort()
+
+        # Now create them
+        for dir in need_dirs:
+            self.mkpath (dir)
+
+        # And walk over the list of files, making a hard link for
+        # each one that doesn't already exist in its corresponding
+        # location under 'base_dir'
+    
+        self.announce ("making hard links in %s..." % base_dir)
+        for file in files:
+            dest = os.path.join (base_dir, file)
+            if not os.path.exists (dest):
+                self.execute (os.link, (file, dest),
+                              "linking %s -> %s" % (file, dest))
+    # make_release_tree ()
+
+
+    def make_tarball (self, base_dir):
+
+        # XXX GNU tar 1.13 has a nifty option to add a prefix directory.
+        # It's pretty new, though, so we certainly can't require it -- but
+        # it would be nice to take advantage of it to skip the "create a
+        # tree of hardlinks" step!
+
+        # But I am a lazy bastard, so I require GNU tar anyways.
+
+        archive_name = base_dir + ".tar.gz"
+        self.spawn (["tar", "-czf", archive_name, base_dir])
+
+
+    def make_zipfile (self, base_dir):
+
+        # This assumes the Unix 'zip' utility -- it could be easily recast
+        # to use pkzip (or whatever the command-line zip creation utility
+        # on Redmond's archaic CP/M knockoff is nowadays), but I'll let
+        # someone who can actually test it do that.
+
+        self.spawn (["zip", "-r", base_dir, base_dir])
+
+
+    def make_distribution (self):
+
+        # Don't warn about missing meta-data here -- should be done
+        # elsewhere.
+        name = self.distribution.name or "UNKNOWN"
+        version = self.distribution.version
+
+        if version:
+            base_dir = "%s-%s" % (name, version)
+        else:
+            base_dir = name
+
+        # Remove any files that match "base_dir" from the fileset -- we
+        # don't want to go distributing the distribution inside itself!
+        self.exclude_files (base_dir + "*")
+ 
+        self.make_release_tree (base_dir, self.files)
+        if 'gztar' in self.formats:
+            self.make_tarball (base_dir)
+        if 'zip' in self.formats:
+            self.make_zipfile (base_dir)
+
+# class Dist
+
+
+# ----------------------------------------------------------------------
+# Utility functions
+
+def findall (dir = os.curdir):
+    """Find all files under 'dir' and return the sorted list of full
+       filenames (relative to 'dir')."""
+
+    list = []
+    stack = [dir]
+    pop = stack.pop
+    push = stack.append
+
+    while stack:
+        dir = pop()
+        names = os.listdir (dir)
+
+        for name in names:
+            fullname = os.path.join (dir, name)
+            list.append (fullname)
+            if os.path.isdir (fullname) and not os.path.islink(fullname):
+                push (fullname)
+
+    list.sort()
+    return list
+
+
+
+
+
+# ======================================================================
+# Here follows some extensive mental masturbation about how to
+# make the manifest file and search algorithm even more complex.
+# I think this is all gratuitous, really.
+
+# Hmm, something extra: want to apply an exclude pattern over a whole
+# subtree without necessarily having to explicitly include files from it,
+# ie. it should apply after gathering files by other means (simple
+# include pattern)
+#   . !*~ !*.bak !#*#
+# and we also want to prune at certain directories:
+#   . !RCS !CVS
+# which again should apply globally.
+#
+# possible solution:
+#   - exclude pattern in a directory applies to all files found under that
+#     directory
+#   - subdirectories that match an exclude pattern will be pruned
+#   - hmmm, to be consistent, subdirectories that match an include
+#     pattern should be recursively included
+#   - and this should apply to "simple" patterns too
+#
+# thus:
+#
+#     examples/
+#
+# means get everything in examples/ and all subdirs;
+#
+#     examples/ !*~ !#*# !*.py[co]
+#
+# means get everything under examples/ except files matching those three globs;
+#
+#     ./ !RCS !CVS
+#
+# means get everything under current dir, but prune RCS/CVS directories;
+#
+#     ./ !*~ !#*# !*.py[co] !RCS !CVS
+#     ! build/
+#     ! experimental/
+#
+# means get everything under the distribution directory except the usual
+# excludes at all levels; exclude "build" and "experimental" under the
+# distribution dir only.
+#
+# Do the former examples still work?
+#
+#     distutils/ *.py
+#     ! distutils/bleeding_edge.py
+#
+# means all .py files recursively found under distutils, except for the one
+# explicitly named.
+#
+#     distutils/ *.py !bleeding_edge.py
+#
+# means the same, except bleeding_edge.py will be excluded wherever it's
+# found -- thus this can exclude up to one file per directory under
+# distutils.
+#
+#     distutils/*.py
+#     ! distutils/bleeding_edge.py
+#
+# gets exactly distutils/*.py, minus the one explicitly mentioned exclude, and
+#
+#     distutils/*.py
+#     distutils/ !bleeding_edge.py
+#
+# coincidentally does the same, but only because there can only be one file
+# that matches the exclude pattern.  Oh, we'd still like
+#
+#     distutils *.py !bleeding*.py
+#     distutils/bleeding_ledge.py
+#
+# to include distutils/bleeding_ledge.py -- i.e. it should override the
+# earlier exclude pattern by virtue of appearing later in the manifest.  Does
+# this conflict with the above requirements, ie. that "!RCS" and "!*~" should
+# apply everywhere?  Hmm, I think it doesn't have to, as long as we're smart
+# about it.  Consequence:
+#
+#     . !RCS !CVS
+#     distutils *
+#
+# will go ahead and include RCS and CVS files under distutils, but
+#
+#     distutils *
+#     . !RCS !CVS
+#
+# will do the right thing.  Hmmm.  I think that's OK, and an inevitable
+# consequence of the ability to override exclusions.
+
+# OK, new crack at the search algorithm.
+#
+#   for pattern in manifest:
+#     if dir-pattern:             # ie. first word is a directory (incl. "."!)
+#       dir = first word on line
+#       patterns = rest of line
+#       if patterns:
+#         for dpattern in patterns:
+#           if exclude-pattern:
+#             remove from files anything matching dpattern (including pruning
+#             subtrees rooted at directories that match dpattern)
+#           else:
+#             files.append (recursive_glob (dir, dpattern))
+#       else:
+#         files.append (recursive_glob (dir, '*')
+#
+#     elif include-pattern:       # it's a "simple include pattern"
+#       files.append (glob (pattern))
+#
+#     else:                    # it's a "simple exclude pattern"
+#       remove from files anything matching pattern
+
+# The two removal algorithms might be a bit tricky:
+#
+#   "remove simple exclude pattern":
+#     for f in files:
+#       if f matches pattern:
+#         delete it
+# 
+#   "remove recursive exclude pattern":
+#     for f in files:
+#
+#       t = tail (f)
+#       while t:
+#         if t matches pattern:
+#           delete current file
+#           continue
+#         t = tail (t)
+#
+# Well, that was an interesting mental exercise.  I'm not completely
+# convinced it will work, nor am I convinced this level of complexity
+# is necessary.  If you want to exclude RCS or CVS directories, just
+# don't bloody include them!
+
+