summaryrefslogtreecommitdiff
path: root/examples/program_analysis/program_understanding.py
diff options
context:
space:
mode:
Diffstat (limited to 'examples/program_analysis/program_understanding.py')
-rw-r--r--examples/program_analysis/program_understanding.py72
1 files changed, 72 insertions, 0 deletions
diff --git a/examples/program_analysis/program_understanding.py b/examples/program_analysis/program_understanding.py
new file mode 100644
index 0000000..2b1e141
--- /dev/null
+++ b/examples/program_analysis/program_understanding.py
@@ -0,0 +1,72 @@
+"""Comparative program understanding between shell implementation snippets."""
+import random
+import os
+from timeit import default_timer as timer
+from tactic_utils import Fix, ApplyRulesMatching
+from ui import serve
+from lazy_structure import LazyStructure
+from analyzelib import LoadDocument, AnalyzeCodelets
+from analogy_utils import Analogy
+
+def Main():
+ """Runs the analogy-maker and displays the output."""
+ start = timer()
+ print("Setting up the structure...")
+ random.seed(24)
+ demo_path = os.environ.get("BUILD_WORKSPACE_DIRECTORY", ".")
+ demo_path += "/examples/program_analysis/paper_demos"
+
+ extra_special = [
+ "static_shell_builtins", "shell_builtins", "builtin_address_internal",
+ "builtin_datas", "builtin_lookup", "_builtin", "builtin_"]
+ bash = LoadDocument(f"{demo_path}/bash.txt", extra_special=extra_special)
+ fish = LoadDocument(f"{demo_path}/fish.txt", extra_special=extra_special)
+
+ # These could be provided, eg., by an AST, or we could even have Sifter
+ # rewrite rules which identify them. We can also make each the membor of
+ # its sub-struct then make the sub-struct a member of the parent struct.
+ bash.AnnotateChunks(dict({
+ bash.FindChunks("shell_builtins")[0]: "/:Semantics:Collection",
+ bash.FindChunks("cd")[2]: "/:Semantics:CollectionMember",
+ }))
+ fish.AnnotateChunks(dict({
+ fish.FindChunks("builtin_datas")[0]: "/:Semantics:Collection",
+ fish.FindChunks("cd")[2]: "/:Semantics:CollectionMember",
+ }))
+ bash.AnnotateChunks(dict({
+ bash.FindChunks("shell_builtins")[-1]: "/:Semantics:FunctionBody",
+ bash.FindChunks("builtin_address_internal")[-1]: "/:Semantics:Function",
+ }))
+ fish.AnnotateChunks(dict({
+ fish.FindChunks("builtin_datas")[-1]: "/:Semantics:FunctionBody",
+ fish.FindChunks("builtin_lookup")[-1]: "/:Semantics:Function",
+ }))
+
+ structure = LazyStructure([bash, fish], AnalyzeCodelets)
+ for document in structure.documents:
+ for chunk in document.chunks:
+ structure.ChunkToNode(document, chunk)
+
+ print("Identifying word pairs...")
+ for word in ["cd", "shell_builtins", "builtin_datas"]:
+ ApplyRulesMatching(structure.rt, "SameWord", dict({
+ "/:Rules:SameWord:MustMap:Word": structure.dictionary[word].full_name,
+ }))
+
+ print("Finding the analogy...")
+ analogy = Analogy.Begin(structure.rt, dict({
+ "/:Mapper:NoSlipRules:A":
+ structure.NodeOfChunk(bash, bash.FindChunk("cd")),
+ "/:Mapper:NoSlipRules:B":
+ structure.NodeOfChunk(fish, fish.FindChunk("cd")),
+ }), extend_here=False)
+
+ Fix(analogy.ExtendMap,
+ ["/:Semantics:CollectionMember", "/:Semantics:Function",
+ "/:Semantics:FunctionBody", "/:SameWord"])
+
+ print(timer() - start)
+ serve.start_server(structure)
+
+if __name__ == "__main__":
+ Main()
generated by cgit on debian on lair
contact matthew@masot.net with questions or feedback