.travis.yml

 language: python
 sudo: false
 python:
-- '2.7'
-- '3.4'
-- '3.5'
+  - '2.7'
+  - '3.4'
+  - '3.5'
+  - '3.6'
 branches:
   except:
   - "/^feature.*$/"

CHANGES.txt

+3.0.19 [unreleased]:
+  Add ftps and directftps protocols
+  #4 Harden hardlinks
+  copy_files_with_regexp: remove duplicates
+3.0.18:
+  Allow to use hardlinks in copy_files and copy_files_with_regexp
+3.0.17:
+  Fix --log option
 3.0.16:
   Add some warnings if some file is missing
 3.0.15:

README.md

@@ -8,3 +8,5 @@ Properties can be overriden by environment variables with pattern BIOMAJ_X_Y_Z f
 Example:
 
 use BIOMAJ_LDAP_HOST for ldap.host
+
+[![PyPI version](https://badge.fury.io/py/biomaj-core.svg)](https://badge.fury.io/py/biomaj-core)

biomaj_core/config.py

@@ -8,6 +8,7 @@ import os
 import time
 import sys
 
+from biomaj_core.utils import Utils
 from biomaj_core.bmajindex import BmajIndex
 
 if sys.version < '3':
@@ -40,8 +41,10 @@ class BiomajConfig(object):
         'db.type': '',
         'db.formats': '',
         'keep.old.version': 1,
+        'keep.old.sessions': 0,
         'docker.sudo': '1',
-        'auto_publish': 0
+        'auto_publish': 0,
+        'use_hardlinks': 0
     }
 
     # Old biomaj level compatibility
@@ -191,9 +194,11 @@ class BiomajConfig(object):
             if options is not None and options.get_option('log') is not None:
                 hdlr.setLevel(BiomajConfig.LOGLEVEL[options.get_option('log')])
                 self.log_level = BiomajConfig.LOGLEVEL[options.get_option('log')]
+                logger.setLevel(self.log_level)
             else:
                 hdlr.setLevel(BiomajConfig.LOGLEVEL[self.get('historic.logfile.level')])
                 self.log_level = BiomajConfig.LOGLEVEL[self.get('historic.logfile.level')]
+                logger.setLevel(self.log_level)
             formatter = logging.Formatter('%(asctime)s %(levelname)-5.5s [%(name)s][%(threadName)s] %(message)s')
             hdlr.setFormatter(formatter)
             logger.addHandler(hdlr)
@@ -281,19 +286,9 @@ class BiomajConfig(object):
         """
         Get a boolean property from bank or general configration. Optionally in section.
         """
-        value = None
+        value = self.get(prop, section, escape, default)
 
-        if self._in_env(prop):
-            value = self._in_env(prop)
-        else:
-            value = self.get(prop, section, escape, default)
-
-        if value is None:
-            return False
-        if value is True or value == 'true' or value == '1':
-            return True
-        else:
-            return False
+        return Utils.to_bool(value)
 
     def get(self, prop, section='GENERAL', escape=True, default=None):
         """
@@ -402,8 +397,8 @@ class BiomajConfig(object):
             status = False
         else:
             protocol = self.get('protocol')
-            allowed_protocols = ['none', 'multi', 'local', 'ftp', 'sftp', 'http',
-                                 'https', 'directftp', 'directhttp', 'directhttps', 'rsync', 'irods']
+            allowed_protocols = ['none', 'multi', 'local', 'ftp', 'ftps', 'http', 'https',
+                                 'directftp', 'directftps', 'directhttp', 'directhttps', 'rsync', 'irods']
             if protocol not in allowed_protocols:
                 logging.error('Protocol not supported: ' + protocol)
                 status = False
@@ -417,7 +412,7 @@ class BiomajConfig(object):
                 elif not self.get('remote.dir').endswith('/'):
                     logging.error('remote.dir must end with a /')
                     return False
-                if protocol not in ['direcftp', 'directhttp', 'directhttps'] and\
+                if protocol not in ['directftp', 'directftps', 'directhttp', 'directhttps'] and\
                         not self.get('remote.files') and\
                         not self.get('remote.list'):
                     logging.error('remote.files not set')

biomaj_core/utils.py

 import os
+import errno
 import re
 import logging
 import shutil
@@ -152,17 +153,18 @@ class Utils(object):
 
         Each file is a dict like with (at least) parameters: year, month, day
         """
-        release = None
+        if not files:
+            return None
+        # release = None
+        rfile = files[0]
+        release = {'year': rfile['year'], 'month': rfile['month'], 'day': rfile['day']}
         for rfile in files:
-            if release is None:
-                release = {'year': rfile['year'], 'month': rfile['month'], 'day': rfile['day']}
-            else:
-                rel_date = datetime.date(int(release['year']), int(release['month']), int(release['day']))
-                file_date = datetime.date(int(rfile['year']), int(rfile['month']), int(rfile['day']))
-                if file_date > rel_date:
-                    release['year'] = rfile['year']
-                    release['month'] = rfile['month']
-                    release['day'] = rfile['day']
+            rel_date = datetime.date(int(release['year']), int(release['month']), int(release['day']))
+            file_date = datetime.date(int(rfile['year']), int(rfile['month']), int(rfile['day']))
+            if file_date > rel_date:
+                release['year'] = rfile['year']
+                release['month'] = rfile['month']
+                release['day'] = rfile['day']
         return release
 
     @staticmethod
@@ -195,7 +197,8 @@ class Utils(object):
             }[date]
 
     @staticmethod
-    def copy_files(files_to_copy, to_dir, move=False, lock=None):
+    def copy_files(files_to_copy, to_dir, move=False, lock=None,
+                   use_hardlinks=False):
         """
         Copy or move files to to_dir, keeping directory structure.
 
@@ -214,6 +217,8 @@ class Utils(object):
         :type move: bool
         :param lock: thread lock object for multi-threads
         :type lock: Lock
+        :param use_hardlinks: use hard links (if possible)
+        :type link: bool
         """
         logger = logging.getLogger('biomaj')
         nb_files = len(files_to_copy)
@@ -244,14 +249,48 @@ class Utils(object):
             else:
                 start_time = datetime.datetime.now()
                 start_time = time.mktime(start_time.timetuple())
-                shutil.copyfile(from_file, to_file)
+                if use_hardlinks:
+                    try:
+                        os.link(from_file, to_file)
+                        logger.debug("Using hardlinks to copy %s",
+                                     file_to_copy['name'])
+                    except OSError as e:
+                        if e.errno in (errno.ENOSYS, errno.ENOTSUP):
+                            msg = "Your system doesn't support hard links. Using regular copy."
+                            logger.warn(msg)
+                            # Copy this file (the stats are copied at the end
+                            # of the function)
+                            shutil.copyfile(from_file, to_file)
+                            # Don't try links anymore
+                            use_hardlinks = False
+                        elif e.errno == errno.EPERM:
+                            msg = "The FS at %s doesn't support hard links. Using regular copy."
+                            logger.warn(msg, to_dir)
+                            # Copy this file (the stats are copied at the end
+                            # of the function)
+                            shutil.copyfile(from_file, to_file)
+                            # Don't try links anymore
+                            use_hardlinks = False
+                        elif e.errno == errno.EXDEV:
+                            msg = "Cross device hard link is impossible (source: %s, dest: %s). Using regular copy."
+                            logger.warn(msg, from_file, to_dir)
+                            # Copy this file
+                            shutil.copyfile(from_file, to_file)
+                            # Don't try links anymore
+                            use_hardlinks = False
+                        else:
+                            raise
+                else:
+                    shutil.copyfile(from_file, to_file)
                 end_time = datetime.datetime.now()
                 end_time = time.mktime(end_time.timetuple())
                 file_to_copy['download_time'] = end_time - start_time
-                shutil.copystat(from_file, to_file)
+                if not use_hardlinks:
+                    shutil.copystat(from_file, to_file)
 
     @staticmethod
-    def copy_files_with_regexp(from_dir, to_dir, regexps, move=False, lock=None):
+    def copy_files_with_regexp(from_dir, to_dir, regexps, move=False, lock=None,
+                               use_hardlinks=False):
         """
         Copy or move files from from_dir to to_dir matching regexps.
         Copy keeps the original file stats.
@@ -266,20 +305,28 @@ class Utils(object):
         :type move: bool
         :param lock: thread lock object for multi-threads
         :type lock: Lock
+        :param use_hardlinks: use hard links (if possible)
+        :type link: bool
         :return: list of copied files with their size
         """
         logger = logging.getLogger('biomaj')
         files_to_copy = []
-        for root, dirs, files in os.walk(from_dir, topdown=True):
+        files_list = []
+        for root, _, files in os.walk(from_dir, topdown=True):
             for name in files:
                 for reg in regexps:
                     file_relative_path = os.path.join(root, name).replace(from_dir, '')
                     if file_relative_path.startswith('/'):
                         file_relative_path = file_relative_path.replace('/', '', 1)
+                    # sometimes files appear twice.... check not already managed
+                    if file_relative_path in files_list:
+                        continue
                     if reg == "**/*":
                         files_to_copy.append({'name': file_relative_path})
+                        files_list.append(file_relative_path)
                         continue
                     if re.match(reg, file_relative_path):
+                        files_list.append(file_relative_path)
                         files_to_copy.append({'name': file_relative_path})
                         continue
 
@@ -305,8 +352,43 @@ class Utils(object):
             if move:
                 shutil.move(from_file, to_file)
             else:
-                shutil.copyfile(from_file, to_file)
-                shutil.copystat(from_file, to_file)
+                if use_hardlinks:
+                    try:
+                        os.link(from_file, to_file)
+                        logger.debug("Using hardlinks to copy %s",
+                                     file_to_copy['name'])
+                    except OSError as e:
+                        if e.errno in (errno.ENOSYS, errno.ENOTSUP):
+                            msg = "Your system doesn't support hard links. Using regular copy."
+                            logger.warn(msg)
+                            # Copy this file (the stats are copied at the end
+                            # of the function)
+                            shutil.copyfile(from_file, to_file)
+                            # Don't try links anymore
+                            use_hardlinks = False
+                        elif e.errno == errno.EPERM:
+                            msg = "The FS at %s doesn't support hard links. Using regular copy."
+                            logger.warn(msg, to_dir)
+                            # Copy this file (we copy the stats here because
+                            # it's not done at the end of the function)
+                            shutil.copyfile(from_file, to_file)
+                            shutil.copystat(from_file, to_file)
+                            # Don't try links anymore
+                            use_hardlinks = False
+                        elif e.errno == errno.EXDEV:
+                            msg = "Cross device hard link is impossible (source: %s, dest: %s). Using regular copy."
+                            logger.warn(msg, from_file, to_dir)
+                            # Copy this file (we copy the stats here because
+                            # it's not done at the end of the function)
+                            shutil.copyfile(from_file, to_file)
+                            shutil.copystat(from_file, to_file)
+                            # Don't try links anymore
+                            use_hardlinks = False
+                        else:
+                            raise
+                else:
+                    shutil.copyfile(from_file, to_file)
+                    shutil.copystat(from_file, to_file)
             file_to_copy['size'] = os.path.getsize(to_file)
             f_stat = datetime.datetime.fromtimestamp(os.path.getmtime(to_file))
             file_to_copy['year'] = str(f_stat.year)
@@ -384,3 +466,28 @@ class Utils(object):
             os.remove(archivefile)
 
         return True
+
+    @staticmethod
+    def to_bool(value):
+        if isinstance(value, bool):
+            return value
+        if not value:
+            return False
+        try:
+            if value.lower() == 'true' or value == '1':
+                return True
+            else:
+                return False
+        except Exception:
+            return False
+
+    @staticmethod
+    def to_int(value):
+        if isinstance(value, int):
+            return value
+        if not value:
+            return 0
+        try:
+            return int(value)
+        except Exception:
+            return 0

debian/changelog

+biomaj3-core (3.0.19-1) UNRELEASED; urgency=medium
+
+  * New upstream release 
+
+ -- Olivier Sallou <osallou@debian.org>  Wed, 16 Oct 2019 13:14:38 +0000
+
 biomaj3-core (3.0.16-1) unstable; urgency=medium
 
   * New upstream release 

setup.py

@@ -21,7 +21,7 @@ config = {
     'url': 'http://biomaj.genouest.org',
     'download_url': 'http://biomaj.genouest.org',
     'author_email': 'olivier.sallou@irisa.fr',
-    'version': '3.0.16',
+    'version': '3.0.19',
      'classifiers': [
         'Development Status :: 5 - Production/Stable',
         'Environment :: Console',

tests/biomaj_tests.py

+from __future__ import print_function
+
 from nose.tools import *
 from nose.plugins.attrib import attr
 
 import shutil
 import os
+import sys
 import tempfile
 import stat
 
@@ -24,6 +27,7 @@ class UtilsForTest:
         Setup the temp dirs and files.
         """
         self.global_properties = None
+        self.global_properties_hl = None
         self.bank_properties = None
 
         self.test_dir = tempfile.mkdtemp('biomaj')
@@ -50,6 +54,9 @@ class UtilsForTest:
         if self.global_properties is None:
             self.__copy_global_properties()
 
+        if self.global_properties_hl is None:
+            self.__copy_global_properties_hl()
+
         if self.bank_properties is None:
             self.__copy_test_bank_properties()
 
@@ -86,7 +93,8 @@ class UtilsForTest:
                       'error.properties', 'local.properties',
                       'localprocess.properties', 'testhttp.properties',
                       'computed.properties', 'computed2.properties',
-                      'sub1.properties', 'sub2.properties']
+                      'sub1.properties', 'sub2.properties',
+                      'hardlinks.properties']
         for prop in properties:
             from_file = os.path.join(curdir, prop)
             to_file = os.path.join(self.conf_dir, prop)
@@ -126,6 +134,29 @@ class UtilsForTest:
                     fout.write(line)
         fout.close()
 
+    def __copy_global_properties_hl(self):
+        if self.global_properties_hl is not None:
+            return
+        self.global_properties_hl = os.path.join(self.conf_dir, 'global_hardlinks.properties')
+        curdir = os.path.dirname(os.path.realpath(__file__))
+        global_template = os.path.join(curdir, 'global_hardlinks.properties')
+        fout = open(self.global_properties_hl, 'w')
+        with open(global_template,'r') as fin:
+            for line in fin:
+                if line.startswith('conf.dir'):
+                    fout.write("conf.dir="+self.conf_dir+"\n")
+                elif line.startswith('log.dir'):
+                    fout.write("log.dir="+self.log_dir+"\n")
+                elif line.startswith('data.dir'):
+                    fout.write("data.dir="+self.data_dir+"\n")
+                elif line.startswith('process.dir'):
+                    fout.write("process.dir="+self.process_dir+"\n")
+                elif line.startswith('lock.dir'):
+                    fout.write("lock.dir="+self.lock_dir+"\n")
+                else:
+                    fout.write(line)
+        fout.close()
+
 
 class TestBiomajUtils(unittest.TestCase):
 
@@ -163,6 +194,24 @@ class TestBiomajUtils(unittest.TestCase):
         endpoint = Utils.get_service_endpoint(config, 'process')
         self.assertTrue(endpoint == 'http://localhost')
 
+    def test_use_hardlinks_config(self):
+        """
+        Test that hardlinks are disabled by default and can be overridden.
+        """
+        BiomajConfig.load_config(self.utils.global_properties,
+                                 allow_user_config=False)
+        # Must be disabled in local.properties
+        config = BiomajConfig('local')
+        self.assertFalse(config.get_bool("use_hardlinks"))
+        # Must be enabled for hardlinks.properties (override)
+        config = BiomajConfig('hardlinks')
+        self.assertTrue(config.get_bool("use_hardlinks"))
+        # Reload file with use_hardlinks=1
+        BiomajConfig.load_config(self.utils.global_properties_hl,
+                                 allow_user_config=False)
+        config = BiomajConfig('local')
+        self.assertTrue(config.get_bool("use_hardlinks"))
+
     def test_mimes(self):
         fasta_file = os.path.join(os.path.dirname(os.path.realpath(__file__)),
                                   'bank/test2.fasta')
@@ -186,6 +235,39 @@ class TestBiomajUtils(unittest.TestCase):
         Utils.copy_files_with_regexp(from_dir, to_dir, ['.*\.py'])
         self.assertTrue(os.path.exists(to_dir+'/biomaj_tests.py'))
 
+    def test_copy_with_regexp_hardlink(self):
+        """
+        Test copy with hardlinks: we create files in data_dir and try to
+        link them. This should work unless /tmp don't accept hardlinks.
+        """
+        # Create 5 files and a directory in data_dir. We don't destroy them
+        # since they are in /tmp.
+        suffix = ".dat"
+        regexp = ".*\\" + suffix
+        orig_file_full = [
+            tempfile.mkstemp(dir=self.utils.data_dir, suffix=suffix)[1]
+            for i in range(5)
+        ]
+        to_dir = tempfile.mkdtemp(dir=self.utils.data_dir)
+        new_file_full = [os.path.join(to_dir, os.path.basename(f))
+                         for f in orig_file_full]
+        # Copy
+        from_dir = self.utils.data_dir
+        Utils.copy_files_with_regexp(from_dir, to_dir, [regexp],
+                                     use_hardlinks=True)
+        # Check if files was copied
+        for orig, new in zip(orig_file_full, new_file_full):
+            self.assertTrue(os.path.exists(new))
+            # Check if it's really a hardlink. This may fail so we catch
+            # any exceptions.
+            orig_file_stat = os.stat(orig)
+            new_file_stat = os.stat(new)
+            try:
+                self.assertTrue(orig_file_stat.st_ino == new_file_stat.st_ino)
+            except Exception:
+                msg = "In %s: copy worked but hardlinks were not used." % self.id()
+                print(msg, file=sys.stderr)
+
     def test_copy(self):
         from_dir = os.path.dirname(os.path.realpath(__file__))
         local_file = 'biomaj_tests.py'
@@ -194,6 +276,33 @@ class TestBiomajUtils(unittest.TestCase):
         Utils.copy_files(files_to_copy, to_dir)
         self.assertTrue(os.path.exists(to_dir+'/biomaj_tests.py'))
 
+    def test_copy_hardlink(self):
+        """
+        Test copy with hardlinks: we create a file in data_dir and try to
+        link it. This should work unless /tmp don't accept hardlinks.
+        """
+        # Create a file and a directory in data_dir. We don't destroy them
+        # since they are in /tmp.
+        _, orig_file_full = tempfile.mkstemp(dir=self.utils.data_dir)
+        orig_file = os.path.basename(orig_file_full)
+        to_dir = tempfile.mkdtemp(dir=self.utils.data_dir)
+        new_file_full = os.path.join(to_dir, orig_file)
+        # Copy
+        from_dir = self.utils.data_dir
+        files_to_copy = [{'root': from_dir, 'name': orig_file}]
+        Utils.copy_files(files_to_copy, to_dir, use_hardlinks=True)
+        # Check if file was copied
+        self.assertTrue(os.path.exists(new_file_full))
+        # Check if it's really a hardlink. This may fail so we catch
+        # any exceptions.
+        orig_file_stat = os.stat(orig_file_full)
+        new_file_stat = os.stat(new_file_full)
+        try:
+            self.assertTrue(orig_file_stat.st_ino == new_file_stat.st_ino)
+        except Exception:
+            msg = "In %s: copy worked but hardlinks were not used." % self.id()
+            print(msg, file=sys.stderr)
+
     @attr('check')
     def test_check_method(self):
         """Check .name, .exe and .args are well check during bank configuration

tests/global_hardlinks.properties

+[GENERAL]
+test=1
+conf.dir=/tmp/biomaj/config
+log.dir=/tmp/biomaj/log
+process.dir=/tmp/biomaj/process
+#The root directory where all databases are stored.
+#If your data is not stored under one directory hirearchy
+#you can override this value in the database properties file.
+data.dir=/tmp/biomaj/
+lock.dir=/tmp/biomaj/lock
+cache.dir=/tmp/biomaj/cache
+use_hardlinks=1
+
+db.url=mongodb://localhost:27017
+db.name=biomaj_test
+
+use_ldap=1
+ldap.host=localhost
+ldap.port=389
+ldap.dn=nodomain
+
+# Use ElasticSearch for index/search capabilities
+use_elastic=0
+#Comma separated list of elasticsearch nodes  host1,host2:port2
+elastic_nodes=localhost
+elastic_index=biomaj_test
+
+celery.queue=biomaj
+celery.broker=mongodb://localhost:27017/biomaj_celery
+
+# Get directory stats (can be time consuming depending on number of files etc...)
+data.stats=1
+
+# List of user admin (linux user id, comma separated)
+admin=
+
+# Auto publish on updates (do not need publish flag, can be ovveriden in bank property file)
+auto_publish=0
+
+########################
+# Global properties file
+
+
+#To override these settings for a specific database go to its
+#properties file and uncomment or add the specific line you want
+#to override.
+
+#----------------
+# Mail Configuration
+#---------------
+#Uncomment thes lines if you want receive mail when the workflow is finished
+
+mail.smtp.host=
+mail.admin=
+mail.from=
+
+#---------------------
+#Proxy authentification
+#---------------------
+#proxyHost=
+#proxyPort=
+#proxyUser=
+#proxyPassword=
+
+#Number of thread for processes
+bank.num.threads=2
+
+#Number of threads to use for downloading
+files.num.threads=4
+
+#to keep more than one release increase this value
+keep.old.version=0
+
+#----------------------
+# Release configuration
+#----------------------
+release.separator=_
+
+#The historic log file is generated in log/
+#define level information for output : DEBUG,INFO,WARN,ERR
+historic.logfile.level=DEBUG
+
+#http.parse.dir.line=<a[\s]+href="([\S]+)/".*alt="\[DIR\]">.*([\d]{2}-[\w\d]{2,5}-[\d]{4}\s[\d]{2}:[\d]{2})
+http.parse.dir.line=<img[\s]+src="[\S]+"[\s]+alt="\[DIR\]"[\s]*/?>[\s]*<a[\s]+href="([\S]+)/"[\s]*>.*([\d]{2}-[\w\d]{2,5}-[\d]{4}\s[\d]{2}:[\d]{2})
+http.parse.file.line=<img[\s]+src="[\S]+"[\s]+alt="\[[\s]+\]"[\s]*/?>[\s]<a[\s]+href="([\S]+)".*([\d]{2}-[\w\d]{2,5}-[\d]{4}\s[\d]{2}:[\d]{2})[\s]+([\d\.]+[MKG]{0,1})
+
+http.group.dir.name=1
+http.group.dir.date=2
+http.group.file.name=1
+http.group.file.date=2
+http.group.file.size=3
+
+
+# Bank default access
+visibility.default=public
+
+
+[loggers]
+keys = root, biomaj
+
+[handlers]
+keys = console
+
+[formatters]
+keys = generic
+
+[logger_root]
+level = INFO
+handlers = console
+
+[logger_biomaj]
+level = DEBUG
+handlers = console
+qualname = biomaj
+propagate=0
+
+[handler_console]
+class = StreamHandler
+args = (sys.stderr,)
+level = DEBUG
+formatter = generic
+
+[formatter_generic]
+format = %(asctime)s %(levelname)-5.5s [%(name)s][%(threadName)s] %(message)s

tests/hardlinks.properties

+[GENERAL]
+######################
+### Initialization ###
+
+db.fullname="local system bank test with hardlinks"
+db.name=local
+db.type=nucleic_protein
+
+offline.dir.name=offline/test/local_tmp
+dir.version=test/local
+
+frequency.update=0
+
+use_hardlinks=1
+
+### Synchronization ###
+
+files.num.threads=1
+
+# NCBI (download fasta)
+protocol=local
+server=
+
+release.file=
+release.regexp=
+release.file.compressed=
+
+remote.dir=/tmp/
+remote.files=^test.*
+
+#Uncomment if you don't want to extract the data files.
+#no.extract=true
+
+local.files=^test.*
+
+## Post Process  ##  The files should be located in the projectfiles/process directory
+
+db.post.process=
+
+
+### Deployment ###
+
+keep.old.version=1