[MRG] Rename library to sourmash (#374)

* Change library name to sourmash instead of sourmash_lib

  * sourmash_lib is still available for older scripts
  * Remove the local sourmash script
  * Fix tests that expected the local sourmash script

* Fix test discovery

* Fix tox dependencies

* add cpp file to gitignore

* Fix sourmash_lib import in lca

* Tests passing

* avoid using import sourmash inside the module code, use . imports instead

* Avoid changing current tests

* Remove redundant import

.gitignore

@@ -18,3 +18,4 @@ _minhash.so
 *.so
 .coverage
 sourmash_lib/_minhash.cpp
+sourmash/_minhash.cpp

MANIFEST.in

@@ -1,6 +1,7 @@
 include LICENSE Makefile Dockerfile LICENSE Makefile README.md requirements.txt
 include index.ipynb
 recursive-include sourmash_lib *
+recursive-include sourmash *
 recursive-include third-party *.cc *.h
 global-exclude *.orig
 global-exclude *.pyc

README.md

@@ -74,7 +74,7 @@ Development happens on github at
 
 `sourmash` is the main command-line entry point; run it for help.
 
-`sourmash_lib/` contains the library code.
+`sourmash/` contains the library code.
 
 Tests require py.test and can be run with `make test`.
 

doc/api-example.md

@@ -16,14 +16,14 @@ Define two sequences:
 Create two minhashes using 3-mers, and add the sequences:
 
 ```
->>> import sourmash_lib
->>> E1 = sourmash_lib.MinHash(n=20, ksize=3)
+>>> import sourmash
+>>> E1 = sourmash.MinHash(n=20, ksize=3)
 >>> E1.add_sequence(seq1)
 
 ```
 
 ```
->>> E2 = sourmash_lib.MinHash(n=20, ksize=3)
+>>> E2 = sourmash.MinHash(n=20, ksize=3)
 >>> E2.add_sequence(seq2)
 
 ```
@@ -81,7 +81,7 @@ raising an exception.
 >>> import screed
 >>> minhashes = []
 >>> for g in genomes:
-...     E = sourmash_lib.MinHash(n=500, ksize=31)
+...     E = sourmash.MinHash(n=500, ksize=31)
 ...     for record in screed.open(g):
 ...         E.add_sequence(record.sequence[:50000], True)
 ...     minhashes.append(E)
@@ -110,7 +110,7 @@ making the minhashes, which can be saved and loaded easily.
 ## Saving and loading signature files
 
 ```
->>> from sourmash_lib import signature
+>>> from sourmash import signature
 >>> sig1 = signature.SourmashSignature(minhashes[0], name=genomes[0][:20])
 >>> with open('/tmp/genome1.sig', 'wt') as fp:
 ...   signature.save_signatures([sig1], fp)

doc/api.rst

@@ -12,24 +12,24 @@ underlying C++ code, but for now this is undocumented.)
 ``MinHash``: basic MinHash sketch functionality
 ===============================================
 
-.. automodule:: sourmash_lib
+.. automodule:: sourmash
    :members:
 
 ``SourmashSignature``: save and load MinHash sketches in JSON
 =============================================================
 
-.. automodule:: sourmash_lib.signature
+.. automodule:: sourmash.signature
    :members:
 
 ``SBT``: save and load Sequence Bloom Trees in JSON
 =============================================================
 
-.. automodule:: sourmash_lib.sbt
+.. automodule:: sourmash.sbt
    :members: GraphFactory, Node, NodePos, SBT, Leaf
    :undoc-members:
 
-``sourmash_lib.fig``: make plots and figures
+``sourmash.fig``: make plots and figures
 ============================================
 
-.. automodule:: sourmash_lib.fig
+.. automodule:: sourmash.fig
    :members:

doc/release.md

@@ -48,7 +48,7 @@ cd sourmash
         pip uninstall -y sourmash; pip uninstall -y sourmash; make install
         mkdir ../not-sourmash # if there is a subdir named 'sourmash' py.test will execute tests
         # there instead of the installed sourmash module's tests
-        pushd ../not-sourmash; py.test --pyargs sourmash_lib; popd
+        pushd ../not-sourmash; py.test --pyargs sourmash; popd
 
 
         # Secondly we test via pip
@@ -63,7 +63,7 @@ cd sourmash
         cp dist/sourmash*tar.gz ../../../testenv3/
         pip uninstall -y sourmash; pip uninstall -y sourmash; make install
         cd ../.. # no subdir named sourmash here, safe for testing installed sourmash module
-        py.test --pyargs sourmash_lib
+        py.test --pyargs sourmash
 
         # Is the distribution in testenv2 complete enough to build another
         # functional distribution?
@@ -80,7 +80,7 @@ cd sourmash
         make test
         pip uninstall -y sourmash; pip uninstall -y sourmash; make install
         mkdir ../not-sourmash
-        pushd ../not-sourmash ; py.test  --pyargs sourmash_lib ; popd
+        pushd ../not-sourmash ; py.test  --pyargs sourmash ; popd
 ```
 4\. Publish the new release on the testing PyPI server.  You will need
    to change your PyPI credentials as documented here:
@@ -100,7 +100,7 @@ cd sourmash
         # install as much as possible from non-test server!
         pip install screed pytest numpy matplotlib scipy
         pip install -i https://testpypi.python.org/pypi --pre --no-clean sourmash
-        py.test --pyargs sourmash_lib
+        py.test --pyargs sourmash
 ```
 5\. Do any final testing:
 
@@ -144,5 +144,5 @@ so:
 ```
 
    apt-cache update && apt-get -y install python-dev libfreetype6-dev && \
-   pip install sourmash[test] && py.test --pyargs sourmash_lib
+   pip install sourmash[test] && py.test --pyargs sourmash
 ```

pytest.ini

@@ -1,4 +1,4 @@
 [pytest]
-addopts = --doctest-glob='doc/*.md'
-python_files = sourmash_lib/*.py tests/*.py
-norecursedirs = utils build buildenv .tox .asv
+addopts = --doctest-glob='doc/*.md' --ignore=setup.py
+python_files = sourmash/*.py tests/*.py
+norecursedirs = utils build buildenv .tox .asv .eggs

setup.py

@@ -1,12 +1,12 @@
 from __future__ import print_function
 import sys
-from setuptools import setup
+from setuptools import setup, find_packages
 from setuptools import Extension
 import os
 
-# retrieve VERSION from sourmash_lib/VERSION.
+# retrieve VERSION from sourmash/VERSION.
 thisdir = os.path.dirname(__file__)
-version_file = open(os.path.join(thisdir, 'sourmash_lib', 'VERSION'))
+version_file = open(os.path.join(thisdir, 'sourmash', 'VERSION'))
 VERSION = version_file.read().strip()
 
 EXTRA_COMPILE_ARGS = ['-std=c++11', '-pedantic']
@@ -51,16 +51,16 @@
     "author": "C. Titus Brown",
     "author_email": "[email protected]",
     "license": "BSD 3-clause",
-    "packages": ["sourmash_lib"],
+    "packages": find_packages(),
     "entry_points": {'console_scripts': [
-        'sourmash = sourmash_lib.__main__:main'
+        'sourmash = sourmash.__main__:main'
         ]
     },
-    "ext_modules": [Extension("sourmash_lib._minhash",
-                               sources=["sourmash_lib/_minhash.pyx",
+    "ext_modules": [Extension("sourmash._minhash",
+                               sources=["sourmash/_minhash.pyx",
                                         "third-party/smhasher/MurmurHash3.cc"],
-                               depends=["sourmash_lib/kmer_min_hash.hh"],
-                               include_dirs=["./sourmash_lib",
+                               depends=["sourmash/kmer_min_hash.hh"],
+                               include_dirs=["./sourmash",
                                              "./third-party/smhasher/"],
                                language="c++",
                                extra_compile_args=EXTRA_COMPILE_ARGS,
@@ -74,7 +74,7 @@
         },
     "include_package_data": True,
     "package_data": {
-        "sourmash_lib": ['*.pxd']
+        "sourmash": ['*.pxd']
     },
     "classifiers": CLASSIFIERS
     }

sourmash/__init__.py

@@ -0,0 +1,26 @@
+#! /usr/bin/env python
+"""
+An implementation of a MinHash bottom sketch, applied to k-mers in DNA.
+"""
+from __future__ import print_function
+import re
+import math
+import os
+
+from ._minhash import (MinHash, get_minhash_default_seed, get_minhash_max_hash)
+DEFAULT_SEED = get_minhash_default_seed()
+MAX_HASH = get_minhash_max_hash()
+
+from .signature import (load_signatures, load_one_signature, SourmashSignature,
+                        save_signatures)
+from .sbtmh import load_sbt_index, search_sbt_index, create_sbt_index
+from . import lca
+from . import sbt
+from . import sbtmh
+from . import sbt_storage
+from . import signature
+
+# retrieve VERSION from sourmash/VERSION.
+thisdir = os.path.dirname(__file__)
+version_file = open(os.path.join(thisdir, 'VERSION'))
+VERSION = version_file.read().strip()

sourmash_lib/commands.py → sourmash/commands.py

@@ -8,10 +8,11 @@
 import random
 
 import screed
-import sourmash_lib
+from . import DEFAULT_SEED, MinHash, load_sbt_index, create_sbt_index
 from . import signature as sig
 from . import sourmash_args
 from .logging import notify, error, print_results, set_quiet
+from .sbtmh import SearchMinHashesFindBest, SigLeaf
 
 from .sourmash_args import DEFAULT_LOAD_K
 DEFAULT_COMPUTE_K = '21,31,51'
@@ -90,7 +91,7 @@ def compute(args):
                         help='choose number of hashes as 1 in FRACTION of input k-mers')
     parser.add_argument('--seed', type=int,
                         help='seed used by MurmurHash (default: 42)',
-                        default=sourmash_lib.DEFAULT_SEED)
+                        default=DEFAULT_SEED)
     parser.add_argument('--randomize', action='store_true',
                         help='shuffle the list of input filenames randomly')
     parser.add_argument('--license', default='CC0', type=str,
@@ -173,18 +174,18 @@ def make_minhashes():
         Elist = []
         for k in ksizes:
             if args.protein:
-                E = sourmash_lib.MinHash(ksize=k, n=args.num_hashes,
-                                            is_protein=True,
-                                    track_abundance=args.track_abundance,
-                                            scaled=args.scaled,
-                                            seed=seed)
+                E = MinHash(ksize=k, n=args.num_hashes,
+                            is_protein=True,
+                            track_abundance=args.track_abundance,
+                            scaled=args.scaled,
+                            seed=seed)
                 Elist.append(E)
             if args.dna:
-                E = sourmash_lib.MinHash(ksize=k, n=args.num_hashes,
-                                            is_protein=False,
-                                    track_abundance=args.track_abundance,
-                                            scaled=args.scaled,
-                                            seed=seed)
+                E = MinHash(ksize=k, n=args.num_hashes,
+                            is_protein=False,
+                            track_abundance=args.track_abundance,
+                            scaled=args.scaled,
+                            seed=seed)
                 Elist.append(E)
         return Elist
 
@@ -551,7 +552,7 @@ def import_csv(args):
             hashes = hashes.strip()
             hashes = list(map(int, hashes.split(' ' )))
 
-            e = sourmash_lib.MinHash(len(hashes), ksize)
+            e = MinHash(len(hashes), ksize)
             e.add_many(hashes)
             s = sig.SourmashSignature(e, filename=name)
             siglist.append(s)
@@ -595,10 +596,10 @@ def sbt_combine(args):
     inp_files = list(args.sbts)
     notify('combining {} SBTs', len(inp_files))
 
-    tree = sourmash_lib.load_sbt_index(inp_files.pop(0))
+    tree = load_sbt_index(inp_files.pop(0))
 
     for f in inp_files:
-        new_tree = sourmash_lib.load_sbt_index(f)
+        new_tree = load_sbt_index(f)
         # TODO: check if parameters are the same for both trees!
         tree.combine(new_tree)
 
@@ -610,8 +611,6 @@ def index(args):
     """
     Build an Sequence Bloom Tree index of the given signatures.
     """
-    import sourmash_lib.sbtmh
-
     parser = argparse.ArgumentParser()
     parser.add_argument('sbt_name', help='name to save SBT into')
     parser.add_argument('signatures', nargs='+',
@@ -639,10 +638,9 @@ def index(args):
     moltype = sourmash_args.calculate_moltype(args)
 
     if args.append:
-        tree = sourmash_lib.load_sbt_index(args.sbt_name)
+        tree = load_sbt_index(args.sbt_name)
     else:
-        tree = sourmash_lib.create_sbt_index(args.bf_size,
-                                             n_children=args.n_children)
+        tree = create_sbt_index(args.bf_size, n_children=args.n_children)
 
     if args.traverse_directory:
         inp_files = list(sourmash_args.traverse_find_sigs(args.signatures))
@@ -670,7 +668,7 @@ def index(args):
             nums.add(ss.minhash.num)
             scaleds.add(ss.minhash.scaled)
 
-            leaf = sourmash_lib.sbtmh.SigLeaf(ss.md5sum(), ss)
+            leaf = SigLeaf(ss.md5sum(), ss)
             tree.add_node(leaf)
             n += 1
 
@@ -833,7 +831,7 @@ def categorize(args):
             for row in r:
                 already_names.add(row[0])
 
-    tree = sourmash_lib.load_sbt_index(args.sbt_name)
+    tree = load_sbt_index(args.sbt_name)
 
     if args.traverse_directory:
         inp_files = set(sourmash_args.traverse_find_sigs(args.queries))
@@ -852,7 +850,7 @@ def categorize(args):
                query_ksize, query_moltype)
 
         results = []
-        search_fn = sourmash_lib.sbtmh.SearchMinHashesFindBest().search
+        search_fn = SearchMinHashesFindBest().search
 
         for leaf in tree.find(search_fn, query, args.threshold):
             if leaf.data.md5sum() != query.md5sum(): # ignore self.
@@ -1006,16 +1004,14 @@ def gather(args):
             outname = args.output_unassigned.name
             notify('saving unassigned hashes to "{}"', outname)
 
-            e = sourmash_lib.MinHash(ksize=query.minhash.ksize, n=0,
-                                     max_hash=new_max_hash)
+            e = MinHash(ksize=query.minhash.ksize, n=0, max_hash=new_max_hash)
             e.add_many(next_query.minhash.get_mins())
             sig.save_signatures([ sig.SourmashSignature(e) ],
                                 args.output_unassigned)
 
 
 def watch(args):
     "Build a signature from raw FASTA/FASTQ coming in on stdin, search."
-    from sourmash_lib.sbtmh import SearchMinHashesFindBest
 
     parser = argparse.ArgumentParser()
     parser.add_argument('sbt_name', help='name of SBT to search')
@@ -1053,7 +1049,7 @@ def watch(args):
         moltype = 'protein'
         is_protein = True
 
-    tree = sourmash_lib.load_sbt_index(args.sbt_name)
+    tree = load_sbt_index(args.sbt_name)
 
     # check ksize from the SBT we are loading
     ksize = args.ksize
@@ -1062,8 +1058,7 @@ def watch(args):
         tree_mh = leaf.data.minhash
         ksize = tree_mh.ksize
 
-    E = sourmash_lib.MinHash(ksize=ksize, n=args.num_hashes,
-                             is_protein=is_protein)
+    E = MinHash(ksize=ksize, n=args.num_hashes, is_protein=is_protein)
     streamsig = sig.SourmashSignature(E, filename='stdin', name=args.name)
 
     notify('Computing signature for k={}, {} from stdin', ksize, moltype)

sourmash_lib/lca/__main__.py → sourmash/lca/__main__.py

@@ -1,13 +1,13 @@
 """
-Command-line entry point for 'python -m sourmash_lib.lca'
+Command-line entry point for 'python -m sourmash.lca'
 """
 
 import sys
 import argparse
 
 from . import classify, index, summarize_main, rankinfo_main, gather_main
 from .command_compare_csv import compare_csv
-from sourmash_lib.logging import set_quiet, error
+from ..logging import set_quiet, error
 
 usage='''
 sourmash lca <command> [<args>] - work with taxonomic information.