Merge pull request #3749 from ComputeCanada/checksums_external

add support for checksums specified in external `checksums.json` file

Author: boegel

easybuild/framework/easyblock.py

@@ -42,6 +42,7 @@
 import copy
 import glob
 import inspect
+import json
 import os
 import re
 import stat
@@ -67,7 +68,7 @@
 from easybuild.tools.build_details import get_build_stats
 from easybuild.tools.build_log import EasyBuildError, dry_run_msg, dry_run_warning, dry_run_set_dirs
 from easybuild.tools.build_log import print_error, print_msg, print_warning
-from easybuild.tools.config import DEFAULT_ENVVAR_USERS_MODULES
+from easybuild.tools.config import CHECKSUM_PRIORITY_JSON, DEFAULT_ENVVAR_USERS_MODULES
 from easybuild.tools.config import FORCE_DOWNLOAD_ALL, FORCE_DOWNLOAD_PATCHES, FORCE_DOWNLOAD_SOURCES
 from easybuild.tools.config import build_option, build_path, get_log_filename, get_repository, get_repositorypath
 from easybuild.tools.config import install_path, log_path, package_path, source_paths
@@ -156,6 +157,7 @@ def __init__(self, ec):
         self.patches = []
         self.src = []
         self.checksums = []
+        self.json_checksums = None
 
         # build/install directories
         self.builddir = None
@@ -347,23 +349,55 @@ def get_checksum_for(self, checksums, filename=None, index=None):
         Obtain checksum for given filename.
 
         :param checksums: a list or tuple of checksums (or None)
-        :param filename: name of the file to obtain checksum for (Deprecated)
+        :param filename: name of the file to obtain checksum for
         :param index: index of file in list
         """
-        # Filename has never been used; flag it as deprecated
-        if filename:
-            self.log.deprecated("Filename argument to get_checksum_for() is deprecated", '5.0')
+        checksum = None
+
+        # sometimes, filename are specified as a dict
+        if isinstance(filename, dict):
+            filename = filename['filename']
 
         # if checksums are provided as a dict, lookup by source filename as key
-        if isinstance(checksums, (list, tuple)):
+        if isinstance(checksums, dict):
+            if filename is not None and filename in checksums:
+                checksum = checksums[filename]
+            else:
+                checksum = None
+        elif isinstance(checksums, (list, tuple)):
             if index is not None and index < len(checksums) and (index >= 0 or abs(index) <= len(checksums)):
-                return checksums[index]
+                checksum = checksums[index]
             else:
-                return None
+                checksum = None
         elif checksums is None:
-            return None
+            checksum = None
+        else:
+            raise EasyBuildError("Invalid type for checksums (%s), should be dict, list, tuple or None.",
+                                 type(checksums))
+
+        if checksum is None or build_option("checksum_priority") == CHECKSUM_PRIORITY_JSON:
+            json_checksums = self.get_checksums_from_json()
+            return json_checksums.get(filename, None)
         else:
-            raise EasyBuildError("Invalid type for checksums (%s), should be list, tuple or None.", type(checksums))
+            return checksum
+
+    def get_checksums_from_json(self, always_read=False):
+        """
+        Get checksums for this software that are provided in a checksums.json file
+
+        :param: always_read: always read the checksums.json file, even if it has been read before
+        """
+        if always_read or self.json_checksums is None:
+            try:
+                path = self.obtain_file("checksums.json", no_download=True)
+                self.log.info("Loading checksums from file %s", path)
+                json_txt = read_file(path)
+                self.json_checksums = json.loads(json_txt)
+            # if the file can't be found, return an empty dict
+            except EasyBuildError:
+                self.json_checksums = {}
+
+        return self.json_checksums
 
     def fetch_source(self, source, checksum=None, extension=False, download_instructions=None):
         """
@@ -445,7 +479,8 @@ def fetch_sources(self, sources=None, checksums=None):
             if source is None:
                 raise EasyBuildError("Empty source in sources list at index %d", index)
 
-            src_spec = self.fetch_source(source, self.get_checksum_for(checksums=checksums, index=index))
+            checksum = self.get_checksum_for(checksums=checksums, filename=source, index=index)
+            src_spec = self.fetch_source(source, checksum=checksum)
             if src_spec:
                 self.src.append(src_spec)
             else:
@@ -477,7 +512,7 @@ def fetch_patches(self, patch_specs=None, extension=False, checksums=None):
             if path:
                 self.log.debug('File %s found for patch %s', path, patch_spec)
                 patch_info['path'] = path
-                patch_info['checksum'] = self.get_checksum_for(checksums, index=index)
+                patch_info['checksum'] = self.get_checksum_for(checksums, filename=patch_info['name'], index=index)
 
                 if extension:
                     patches.append(patch_info)
@@ -638,7 +673,7 @@ def collect_exts_file_info(self, fetch_files=True, verify_checksums=True):
 
                         # verify checksum (if provided)
                         self.log.debug('Verifying checksums for extension source...')
-                        fn_checksum = self.get_checksum_for(checksums, index=0)
+                        fn_checksum = self.get_checksum_for(checksums, filename=src_fn, index=0)
                         if verify_checksum(src_path, fn_checksum):
                             self.log.info('Checksum for extension source %s verified', src_fn)
                         elif build_option('ignore_checksums'):
@@ -672,7 +707,7 @@ def collect_exts_file_info(self, fetch_files=True, verify_checksums=True):
                                 patch = patch['path']
                                 patch_fn = os.path.basename(patch)
 
-                                checksum = self.get_checksum_for(checksums[1:], index=idx)
+                                checksum = self.get_checksum_for(checksums, filename=patch_fn, index=idx+1)
                                 if verify_checksum(patch, checksum):
                                     self.log.info('Checksum for extension patch %s verified', patch_fn)
                                 elif build_option('ignore_checksums'):
@@ -694,7 +729,7 @@ def collect_exts_file_info(self, fetch_files=True, verify_checksums=True):
         return exts_sources
 
     def obtain_file(self, filename, extension=False, urls=None, download_filename=None, force_download=False,
-                    git_config=None, download_instructions=None, alt_location=None):
+                    git_config=None, no_download=False, download_instructions=None, alt_location=None):
         """
         Locate the file with the given name
         - searches in different subdirectories of source path
@@ -705,6 +740,7 @@ def obtain_file(self, filename, extension=False, urls=None, download_filename=No
         :param download_filename: filename with which the file should be downloaded, and then renamed to <filename>
         :param force_download: always try to download file, even if it's already available in source path
         :param git_config: dictionary to define how to download a git repository
+        :param no_download: do not try to download the file
         :param download_instructions: instructions to manually add source (used for complex cases)
         :param alt_location: alternative location to use instead of self.name
         """
@@ -818,6 +854,13 @@ def obtain_file(self, filename, extension=False, urls=None, download_filename=No
                 if self.dry_run:
                     self.dry_run_msg("  * %s found at %s", filename, foundfile)
                 return foundfile
+            elif no_download:
+                if self.dry_run:
+                    self.dry_run_msg("  * %s (MISSING)", filename)
+                    return filename
+                else:
+                    raise EasyBuildError("Couldn't find file %s anywhere, and downloading it is disabled... "
+                                         "Paths attempted (in order): %s ", filename, ', '.join(failedpaths))
             elif git_config:
                 return get_source_tarball_from_git(filename, targetdir, git_config)
             else:
@@ -2280,7 +2323,7 @@ def fetch_step(self, skip_checksums=False):
 
         # fetch patches
         if self.cfg['patches'] + self.cfg['postinstallpatches']:
-            if isinstance(self.cfg['checksums'], (list, tuple)):
+            if self.cfg['checksums'] and isinstance(self.cfg['checksums'], (list, tuple)):
                 # if checksums are provided as a list, first entries are assumed to be for sources
                 patches_checksums = self.cfg['checksums'][len(self.cfg['sources']):]
             else:
@@ -2367,6 +2410,20 @@ def check_checksums_for(self, ent, sub='', source_cnt=None):
         patches = ent.get('patches', [])
         checksums = ent.get('checksums', [])
 
+        if not checksums:
+            checksums_from_json = self.get_checksums_from_json()
+            # recreate a list of checksums. If each filename is found, the generated list of checksums should match
+            # what is expected in list format
+            for fn in sources + patches:
+                # if the filename is a tuple, the actual source file name is the first element
+                if isinstance(fn, tuple):
+                    fn = fn[0]
+                # if the filename is a dict, the actual source file name is the "filename" element
+                if isinstance(fn, dict):
+                    fn = fn["filename"]
+                if fn in checksums_from_json.keys():
+                    checksums += [checksums_from_json[fn]]
+
         if source_cnt is None:
             source_cnt = len(sources)
         patch_cnt, checksum_cnt = len(patches), len(checksums)
@@ -4406,6 +4463,67 @@ class StopException(Exception):
     pass
 
 
+def inject_checksums_to_json(ecs, checksum_type):
+    """
+    Inject checksums of given type in corresponding json files
+
+    :param ecs: list of EasyConfig instances to calculate checksums and inject them into checksums.json
+    :param checksum_type: type of checksum to use
+    """
+    for ec in ecs:
+        ec_fn = os.path.basename(ec['spec'])
+        ec_dir = os.path.dirname(ec['spec'])
+        print_msg("injecting %s checksums for %s in checksums.json" % (checksum_type, ec['spec']), log=_log)
+
+        # get easyblock instance and make sure all sources/patches are available by running fetch_step
+        print_msg("fetching sources & patches for %s..." % ec_fn, log=_log)
+        app = get_easyblock_instance(ec)
+        app.update_config_template_run_step()
+        app.fetch_step(skip_checksums=True)
+
+        # compute & inject checksums for sources/patches
+        print_msg("computing %s checksums for sources & patches for %s..." % (checksum_type, ec_fn), log=_log)
+        checksums = {}
+        for entry in app.src + app.patches:
+            checksum = compute_checksum(entry['path'], checksum_type)
+            print_msg("* %s: %s" % (os.path.basename(entry['path']), checksum), log=_log)
+            checksums[os.path.basename(entry['path'])] = checksum
+
+        # compute & inject checksums for extension sources/patches
+        if app.exts:
+            print_msg("computing %s checksums for extensions for %s..." % (checksum_type, ec_fn), log=_log)
+
+            for ext in app.exts:
+                # compute checksums for extension sources & patches
+                if 'src' in ext:
+                    src_fn = os.path.basename(ext['src'])
+                    checksum = compute_checksum(ext['src'], checksum_type)
+                    print_msg(" * %s: %s" % (src_fn, checksum), log=_log)
+                    checksums[src_fn] = checksum
+                for ext_patch in ext.get('patches', []):
+                    patch_fn = os.path.basename(ext_patch['path'])
+                    checksum = compute_checksum(ext_patch['path'], checksum_type)
+                    print_msg(" * %s: %s" % (patch_fn, checksum), log=_log)
+                    checksums[patch_fn] = checksum
+
+        # actually inject new checksums or overwrite existing ones (if --force)
+        existing_checksums = app.get_checksums_from_json(always_read=True)
+        for filename in checksums:
+            if filename not in existing_checksums:
+                existing_checksums[filename] = checksums[filename]
+            # don't do anything if the checksum already exist and is the same
+            elif checksums[filename] != existing_checksums[filename]:
+                if build_option('force'):
+                    print_warning("Found existing checksums for %s, overwriting them (due to --force)..." % ec_fn)
+                    existing_checksums[filename] = checksums[filename]
+                else:
+                    raise EasyBuildError("Found existing checksum for %s, use --force to overwrite them" % filename)
+
+        # actually write the checksums
+        with open(os.path.join(ec_dir, 'checksums.json'), 'w') as outfile:
+            json.dump(existing_checksums, outfile, indent=2, sort_keys=True)
+
+
 def inject_checksums(ecs, checksum_type):
     """
     Inject checksums of given type in specified easyconfig files

easybuild/main.py

@@ -34,6 +34,7 @@
 :author: Toon Willems (Ghent University)
 :author: Ward Poelmans (Ghent University)
 :author: Fotis Georgatos (Uni.Lu, NTUA)
+:author: Maxime Boissonneault (Compute Canada)
 """
 import copy
 import os
@@ -45,7 +46,7 @@
 #  expect missing log output when this not the case!
 from easybuild.tools.build_log import EasyBuildError, print_error, print_msg, print_warning, stop_logging
 
-from easybuild.framework.easyblock import build_and_install_one, inject_checksums
+from easybuild.framework.easyblock import build_and_install_one, inject_checksums, inject_checksums_to_json
 from easybuild.framework.easyconfig import EASYCONFIGS_PKG_SUBDIR
 from easybuild.framework.easystack import parse_easystack
 from easybuild.framework.easyconfig.easyconfig import clean_up_easyconfigs
@@ -425,7 +426,8 @@ def main(args=None, logfile=None, do_build=None, testing=False, modtool=None):
             sys.exit(31)  # exit -> 3x1t -> 31
 
     # read easyconfig files
-    easyconfigs, generated_ecs = parse_easyconfigs(paths, validate=not options.inject_checksums)
+    validate = not options.inject_checksums and not options.inject_checksums_to_json
+    easyconfigs, generated_ecs = parse_easyconfigs(paths, validate=validate)
 
     # handle --check-contrib & --check-style options
     if run_contrib_style_checks([ec['ec'] for ec in easyconfigs], options.check_contrib, options.check_style):
@@ -453,6 +455,7 @@ def main(args=None, logfile=None, do_build=None, testing=False, modtool=None):
 
     keep_available_modules = forced or dry_run_mode or options.extended_dry_run or pr_options or options.copy_ec
     keep_available_modules = keep_available_modules or options.inject_checksums or options.sanity_check_only
+    keep_available_modules = keep_available_modules or options.inject_checksums_to_json
 
     # skip modules that are already installed unless forced, or unless an option is used that warrants not skipping
     if not keep_available_modules:
@@ -538,8 +541,12 @@ def main(args=None, logfile=None, do_build=None, testing=False, modtool=None):
         with rich_live_cm():
             inject_checksums(ordered_ecs, options.inject_checksums)
 
+    elif options.inject_checksums_to_json:
+        inject_checksums_to_json(ordered_ecs, options.inject_checksums_to_json)
+
     # cleanup and exit after dry run, searching easyconfigs or submitting regression test
     stop_options = [options.check_conflicts, dry_run_mode, options.dump_env_script, options.inject_checksums]
+    stop_options += [options.inject_checksums_to_json]
     if any(no_ec_opts) or any(stop_options):
         clean_exit(logfile, eb_tmpdir, testing)
 

easybuild/tools/config.py

@@ -34,6 +34,7 @@
 :author: Ward Poelmans (Ghent University)
 :author: Damian Alvarez (Forschungszentrum Juelich GmbH)
 :author: Andy Georges (Ghent University)
+:author: Maxime Boissonneault (Compute Canada)
 """
 import copy
 import glob
@@ -126,6 +127,11 @@
 FORCE_DOWNLOAD_CHOICES = [FORCE_DOWNLOAD_ALL, FORCE_DOWNLOAD_PATCHES, FORCE_DOWNLOAD_SOURCES]
 DEFAULT_FORCE_DOWNLOAD = FORCE_DOWNLOAD_SOURCES
 
+CHECKSUM_PRIORITY_JSON = "json"
+CHECKSUM_PRIORITY_EASYCONFIG = "easyconfig"
+CHECKSUM_PRIORITY_CHOICES = [CHECKSUM_PRIORITY_JSON, CHECKSUM_PRIORITY_EASYCONFIG]
+DEFAULT_CHECKSUM_PRIORITY = CHECKSUM_PRIORITY_EASYCONFIG
+
 # package name for generic easyblocks
 GENERIC_EASYBLOCK_PKG = 'generic'
 
@@ -180,6 +186,7 @@ def mk_full_default_path(name, prefix=DEFAULT_PREFIX):
         'accept_eula_for',
         'aggregate_regtest',
         'backup_modules',
+        'checksum_priority',
         'container_config',
         'container_image_format',
         'container_image_name',