From f3bcd3c9a91f6212c30b9c778c11f3c8a9f6f1d4 Mon Sep 17 00:00:00 2001
From: Richard Purdie <richard.purdie@linuxfoundation.org>
Date: Sun, 27 Nov 2022 21:16:16 +0000
Subject: bitbake: ast/data/codeparser: Add dependencies from python module
 functions

Moving code into python modules is a very effective way to reduce parsing
time and overhead in recipes. The downside has always been that any
dependency information on which variables those functions access is lost
and the hashes can therefore become less reliable.

This patch adds parsing of the imported module functions and that dependency
information is them injected back into the hash dependency information.

Intermodule function references are resolved to the full function
call names in our module namespace to ensure interfunction dependencies
are correctly handled too.

(Bitbake rev: 605c478ce14cdc3c02d6ef6d57146a76d436a83c)

(Bitbake rev: 91441e157e495b02db44e19e836afad366ee8924)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
---
 bitbake/lib/bb/codeparser.py | 43 ++++++++++++++++++++++++++++++++++++++-----
 1 file changed, 38 insertions(+), 5 deletions(-)

(limited to 'bitbake/lib/bb/codeparser.py')

diff --git a/bitbake/lib/bb/codeparser.py b/bitbake/lib/bb/codeparser.py
index 9d66d3ae41..c0b1d7966f 100644
--- a/bitbake/lib/bb/codeparser.py
+++ b/bitbake/lib/bb/codeparser.py
@@ -27,6 +27,7 @@ import ast
 import sys
 import codegen
 import logging
+import inspect
 import bb.pysh as pysh
 import bb.utils, bb.data
 import hashlib
@@ -58,10 +59,33 @@ def check_indent(codestr):
 
     return codestr
 
-# A custom getstate/setstate using tuples is actually worth 15% cachesize by
-# avoiding duplication of the attribute names!
+modulecode_deps = {}
 
+def add_module_functions(fn, functions, namespace):
+    fstat = os.stat(fn)
+    fixedhash = fn + ":" + str(fstat.st_size) +  ":" + str(fstat.st_mtime)
+    for f in functions:
+        name = "%s.%s" % (namespace, f)
+        parser = PythonParser(name, logger)
+        try:
+            parser.parse_python(None, filename=fn, lineno=1, fixedhash=fixedhash+f)
+            #bb.warn("Cached %s" % f)
+        except KeyError:
+            lines, lineno = inspect.getsourcelines(functions[f])
+            src = "".join(lines)
+            parser.parse_python(src, filename=fn, lineno=lineno, fixedhash=fixedhash+f)
+            #bb.warn("Not cached %s" % f)
+        execs = parser.execs.copy()
+        # Expand internal module exec references
+        for e in parser.execs:
+            if e in functions:
+                execs.remove(e)
+                execs.add(namespace + "." + e)
+        modulecode_deps[name] = [parser.references.copy(), execs, parser.var_execs.copy(), parser.contains.copy()]
+        #bb.warn("%s: %s\nRefs:%s Execs: %s %s %s" % (name, src, parser.references, parser.execs, parser.var_execs, parser.contains))
 
+# A custom getstate/setstate using tuples is actually worth 15% cachesize by
+# avoiding duplication of the attribute names!
 class SetCache(object):
     def __init__(self):
         self.setcache = {}
@@ -289,11 +313,17 @@ class PythonParser():
         self.unhandled_message = "in call of %s, argument '%s' is not a string literal"
         self.unhandled_message = "while parsing %s, %s" % (name, self.unhandled_message)
 
-    def parse_python(self, node, lineno=0, filename="<string>"):
-        if not node or not node.strip():
+    # For the python module code it is expensive to have the function text so it is
+    # uses a different fixedhash to cache against. We can take the hit on obtaining the
+    # text if it isn't in the cache.
+    def parse_python(self, node, lineno=0, filename="<string>", fixedhash=None):
+        if not fixedhash and (not node or not node.strip()):
             return
 
-        h = bbhash(str(node))
+        if fixedhash:
+            h = fixedhash
+        else:
+            h = bbhash(str(node))
 
         if h in codeparsercache.pythoncache:
             self.references = set(codeparsercache.pythoncache[h].refs)
@@ -311,6 +341,9 @@ class PythonParser():
                 self.contains[i] = set(codeparsercache.pythoncacheextras[h].contains[i])
             return
 
+        if fixedhash and not node:
+            raise KeyError
+
         # Need to parse so take the hit on the real log buffer
         self.log = BufferedLogger('BitBake.Data.PythonParser', logging.DEBUG, self._log)
 
-- 
cgit v1.2.3-54-g00ecf