ensure that 'pkgbuilds' and 'pkgentries' are loaded before 'pkgfiles'wip-2022-02-15

1 files changed, 48 insertions, 26 deletions

diff --git a/parabola_repolint/repocache.py b/parabola_repolint/repocache.py
index 09893da..911c1b4 100644
--- a/parabola_repolint/repocache.py
+++ b/parabola_repolint/repocache.py
@@ -678,51 +678,39 @@ class Repo():
 
     def __init__(self, name, have_abs_tree=False):
         ''' constructor '''
-        # NOTE: the order of Repo instance var initialization is critical
-        #       _load_pkgbuilds and _load_pkgentries must be called before _load_pkgfiles;
-        #         because PkgFile initialization (in _load_pkgfiles) assumes this object's
-        #         _pkgbuild_cache and _pkgentries_cache to be populated prior.
         self._name = name
         self._pkgbuild_dir = os.path.join(RepoCache.ABSLIBRE_DIR, name) if have_abs_tree else None
         self._pkgentries_dir = os.path.join(RepoCache.PKGENTRIES_DIR, name)
         self._pkgfiles_dir = os.path.join(RepoCache.PKGFILES_DIR, name)
         self._mirror_dir = os.path.join(RepoCache.MIRROR_DIR, name)
 
+        # initialize pkgbuilds cache (per abslibre)
         self._pkgbuilds = []
         self._pkgbuild_cache = {}
-        if self._pkgbuild_dir is not None:
-            self._load_pkgbuilds()
-            logging.info('%s pkgbuilds: %i', name, len(self._pkgbuilds))
-            with open(os.path.join(self._pkgbuild_dir, '.pkgbuilds'), 'w') as out:
-                out.write(json.dumps(self._pkgbuilds, indent=4, sort_keys=True, default=str))
-            with open(os.path.join(self._pkgbuild_dir, '.pkgbuild_cache'), 'w') as out:
-                out.write(json.dumps(self._pkgbuild_cache, indent=4, sort_keys=True, default=str))
+        self._load_pkgbuilds()
 
+        # initialize pkgentries cache (per found repo.db files)
         self._pkgentries = []
         self._pkgentries_cache = {}
         self._provides_cache = {}
         self._load_pkgentries()
 
-        logging.info('%s pkgentries: %i', name, len(self._pkgentries))
-        with open(os.path.join(self._pkgentries_dir, '.pkgentries'), 'w') as out:
-            out.write(json.dumps(self._pkgentries, indent=4, sort_keys=True, default=str))
-        with open(os.path.join(self._pkgentries_dir, '.pkgentries_cache'), 'w') as out:
-            out.write(json.dumps(self._pkgentries_cache, indent=4, sort_keys=True, default=str))
-        with open(os.path.join(self._pkgentries_dir, '.provides_cache'), 'w') as out:
-            out.write(json.dumps(self._provides_cache, indent=4, sort_keys=True, default=str))
-
+        # initialize pkgfile cache (per found pool files)
+        # TODO: this does not actually check the pool, but the symlinks forest
+        #       it is not capable of distinguishing missing symlinks from missing files,
+        #         nor identifying pool files with no corresponding symlinks
+        #       some files in the pool could be usedless cruft,
+        #         so the latter is not the same case as a missing symlink,
+        #         because the implication only holds from symlinks to files in the pool
+        #       actually, the semantics of this class is more like 'PkgLinks',
+        #         because it represents the symlinks forest, and not any pool files
+        #       sorting/correlating them may require a new 'PoolFiles' class
+        #         (and perhaps to rename this class to 'PkgLinks')
         self._pkgfiles = []
         self._pkgstatus = {}
         self._litterfiles = []
         self._load_pkgfiles()
 
-        logging.info('%s pkgfiles: %i', name, len(self._pkgfiles))
-        os.makedirs(self._pkgfiles_dir, exist_ok=True)
-        with open(os.path.join(self._pkgfiles_dir, '.pkgfiles'), 'w') as out:
-            out.write(json.dumps(self._pkgfiles, indent=4, sort_keys=True, default=str))
-        with open(os.path.join(self._pkgfiles_dir, '.litterfiles'), 'w') as out:
-            out.write(json.dumps(self._litterfiles, indent=4, sort_keys=True, default=str))
-
     @property
     def name(self):
         ''' produce the name of the repo '''
@@ -765,6 +753,9 @@ class Repo():
 
     def _load_pkgbuilds(self):
         ''' load the pkgbuilds from abslibre '''
+        if self._pkgbuild_dir is None:
+            return
+
         i = 0
         for root, _, files in os.walk(self._pkgbuild_dir):
             if 'PKGBUILD' in files:
@@ -788,6 +779,13 @@ class Repo():
                             self._pkgbuild_cache[arch][pkgname] = []
                         self._pkgbuild_cache[arch][pkgname].append(pkgbuild)
 
+        # cache results
+        logging.info('%s pkgbuilds: %i', name, len(self._pkgbuilds))
+        with open(os.path.join(self._pkgbuild_dir, '.pkgbuilds'), 'w') as out:
+            out.write(json.dumps(self._pkgbuilds, indent=4, sort_keys=True, default=str))
+        with open(os.path.join(self._pkgbuild_dir, '.pkgbuild_cache'), 'w') as out:
+            out.write(json.dumps(self._pkgbuild_cache, indent=4, sort_keys=True, default=str))
+
     def _load_pkgentries(self):
         ''' extract and then load the entries in the db.tar.xz '''
         arches_dir = os.path.join(self._mirror_dir, 'os')
@@ -847,8 +845,24 @@ class Repo():
                         self._provides_cache[pkgentry.arch][provides] = []
                     self._provides_cache[pkgentry.arch][provides].append(pkgentry)
 
+        # cache results
+        logging.info('%s pkgentries: %i', name, len(self._pkgentries))
+        with open(os.path.join(self._pkgentries_dir, '.pkgentries'), 'w') as out:
+            out.write(json.dumps(self._pkgentries, indent=4, sort_keys=True, default=str))
+        with open(os.path.join(self._pkgentries_dir, '.pkgentries_cache'), 'w') as out:
+            out.write(json.dumps(self._pkgentries_cache, indent=4, sort_keys=True, default=str))
+        with open(os.path.join(self._pkgentries_dir, '.provides_cache'), 'w') as out:
+            out.write(json.dumps(self._provides_cache, indent=4, sort_keys=True, default=str))
+
     def _load_pkgfiles(self):
         ''' load the pkg.tar.xz files from the repo '''
+        # NOTE: the order of Repo instance var initialization is critical
+        #       _load_pkgbuilds and _load_pkgentries must be called before _load_pkgfiles;
+        #         because PkgFile initialization (in _load_pkgfiles) assumes this object's
+        #         _pkgbuild_cache and _pkgentries_cache to be populated prior.
+        if not self._pkgbuilds: _load_pkgbuilds(self)
+        if not self._pkgentries: _load_pkgentries(self)
+
         i = 0
         arches_dir = os.path.join(self._mirror_dir, 'os')
         arch_dirs = sorted(os.scandir(arches_dir), key=lambda arch: arch.name)
@@ -884,6 +898,14 @@ class Repo():
                         sys.stdout.write(' %s pkgfiles: %i\r' % (self._name, i))
                         sys.stdout.flush()
 
+        # cache results
+        logging.info('%s pkgfiles: %i', name, len(self._pkgfiles))
+        os.makedirs(self._pkgfiles_dir, exist_ok=True)
+        with open(os.path.join(self._pkgfiles_dir, '.pkgfiles'), 'w') as out:
+            out.write(json.dumps(self._pkgfiles, indent=4, sort_keys=True, default=str))
+        with open(os.path.join(self._pkgfiles_dir, '.litterfiles'), 'w') as out:
+            out.write(json.dumps(self._litterfiles, indent=4, sort_keys=True, default=str))
+
 
     def __repr__(self):
         ''' produce a string representation of the repo '''