Commit: 8b60dc0d6200a93008d69c118732d46bdc7a72d3 - cammy (git)

A categorical programming language

Commit MetaInfo

• Revision: 8b60dc0d6200a93008d69c118732d46bdc7a72d3 (tree)
• Time: 2024-04-25 10:34:53
• Author: Corbin <cds@corb...>
• Commiter: Corbin

Log Message

Get 2to3.py to build a v3 hive.

I haven't verified that the contents of the hive are correct, and
git-fsck is (unsurprisingly) very unhappy with me. I used a monkey patch
on dulwich to allow creation of refs with question marks in their names,
because I know it's legal on Linux and I don't care about other kernels.

Change Summary

• modified: 2to3.py

Incremental Difference

--- a/2to3.py
+++ b/2to3.py
@@ -1,21 +1,45 @@
 import json
+import os
 import sys
+import time
 
-import dulwich
+from dulwich.repo import Repo
+from dulwich.objects import Blob, Commit, Tree
 
-with open(sys.argv[-1]) as handle: d = json.load(handle)
+# Monkey patch: Allow ? in ref names.
+from dulwich import refs
+refs.BAD_REF_CHARS = set(b"\177 ~^:*[")
+
+with open(sys.argv[-2]) as handle: d = json.load(handle)
+repoPath = sys.argv[-1]
+if os.path.exists(repoPath): repo = Repo(repoPath)
+else: repo = Repo.init(repoPath, mkdir=True)
+
+# XXX implement following four functions, all should return object hash :str
 
 def insertBlob(s):
-    return "blob"
+    blob = Blob.from_string(s.encode("utf-8"))
+    repo.object_store.add_object(blob)
+    return blob.id
 
 def insertTree(d):
-    return "tree"
+    tree = Tree()
+    for k, v in d.items():
+        tree.add(k.encode("utf-8"), 0o100644, v)
+    repo.object_store.add_object(tree)
+    return tree.id
 
+author = b"Corbin <[e-mail]>"
 def insertCommit(tree, message):
-    return "commit"
-
-def addTag(t, obj):
-    return "tag"
+    commit = Commit()
+    commit.author = commit.committer = author
+    commit.author_time = commit.commit_time = int(time.time())
+    commit.author_timezone = commit.commit_timezone = 0
+    commit.encoding = b"UTF-8"
+    commit.message = message.encode("utf-8")
+    commit.tree = tree
+    repo.object_store.add_object(commit)
+    return commit.id
 
 heap = []
 for row in d["heap"]:
@@ -25,10 +49,14 @@ for row in d["heap"]:
         tree = dict((("_%d" % i, heap[j]) for i, j in enumerate(row[1:])))
         tree["_con"] = insertBlob(row[0])
         heap.append(insertTree(tree))
+    if not (len(heap) % 100): print("h", end="", flush=True)
+print()
 
-print("Heap is processed! Total expressions:", len(heap))
-
+exprCount = 0
 def insertExpr(expr):
+    global exprCount
+    exprCount += 1
+    if not (exprCount % 100): print("e", end="", flush=True)
     if isinstance(expr, str): return insertBlob(expr)
     elif isinstance(expr, int): return insertBlob("_%d" % expr)
     else:
@@ -36,18 +64,21 @@ def insertExpr(expr):
         tree = dict(("_%d" % i, j) for i, j in enumerate(exprs))
         return insertTree(tree)
 
-for name, expr in d["jets"].items(): addTag("jets/" + name, insertExpr(expr))
+def refName(ns, s): return (("refs/%s/%s" % (ns, s)).encode("utf-8"))
 
-print("Jets are processed! Total jets:", len(d["jets"]))
+for name, expr in d["jets"].items():
+    repo.refs[refName("jets", name)] = insertExpr(expr)
 
 for name, row in d["symbols"].items():
     pointer, trail = row
-    addTag("local/" + name, insertCommit(heap[pointer], trail))
-
-print("Symbols are processed! Total names:", len(d["symbols"]))
+    repo.refs[refName("local", name)] = insertCommit(heap[pointer], trail)
 
 for name, row in d["templates"].items():
     expr, trail = row
-    addTag("local/" + name, insertCommit(insertExpr(expr), trail))
+    repo.refs[refName("local", name)] = insertCommit(insertExpr(expr), trail)
 
+print()
+print("Heap is processed! Total expressions:", len(heap))
 print("Templates are processed! Total names:", len(d["templates"]))
+print("Jets are processed! Total jets:", len(d["jets"]))
+print("Symbols are processed! Total names:", len(d["symbols"]))