Triforce v0.7.0 "Sage Of Courage"

The 0.6 crashes after warps and area changes fixed (no write-back into memory the capture leaves out), invisible objects fixed (occlusion gate), DLSS frame generation flicker fixed, achievements, the renderer built with a play profile, faster translated game code, helper threads judged on settled scenes, memory, shader preparation, audio effects, launcher, controls and Linux fixes.
This commit is contained in:
ruckus7382sfgsgdf
2026-11-22 11:59:06 -06:00
parent 70c210cee1
commit e6f2f248f2
529 changed files with 55958 additions and 1898 deletions

View File

@@ -7,10 +7,60 @@ HEAD_BYTES HEAD, for bytes (build.py's memory estimate reads the files
The lift writes its C with \n line ends; a file with \r\n ones (written in text mode on Windows) is read the same: none of the
patterns needs the end of the definition's line, and a note's line may end in \r\n.
(Triforce, 11-22, the performance audit's IO-40, after Source A's) The scans themselves, for the callers:
definitions(text) HEAD's matches (HEAD_BYTES's for bytes), one after the other, as HEAD.finditer gives them
noted_starts(text) where HEAD_NOTED's matches start, as [m.start() for m in HEAD_NOTED.finditer(text)]
A pattern that begins with ^ is tried by finditer at every line of 1.5 GB of C. These jump from one "\nvoid f_" to the
next with str.find and try the pattern only there: the same matches, far fewer tries (Source A: two scans over 1.56 GB in
1.49 s where they took 11.49 s). HOOKSHOT_LIFTED_C_REGEX=1 (read once): both are the patterns' own finditer, as before.
tests/test_lifted_c_scan.py holds them to the patterns.
"""
import os
import re
_DEF = r"void (f_[0-9a-f]+)\(Ctx \*__restrict A64_NOESC c\) \{"
HEAD = re.compile(r"^" + _DEF, re.M)
HEAD_NOTED = re.compile(r"^(?:/\* .* \*/\r?\n)?" + _DEF, re.M)
HEAD_BYTES = re.compile((r"^" + _DEF).encode(), re.M)
REGEX_SCAN = os.environ.get("HOOKSHOT_LIFTED_C_REGEX", "0") not in ("", "0")
_NOTE = re.compile(r"/\* .* \*/\r?\n") # (HEAD_NOTED's optional first line, by itself)
def definitions(text):
"""Each definition in text (str or bytes), in order: the match HEAD (or HEAD_BYTES) has for it"""
head, mark = (HEAD_BYTES, b"\nvoid f_") if isinstance(text, (bytes, bytearray, memoryview)) else (HEAD, "\nvoid f_")
if REGEX_SCAN:
yield from head.finditer(text)
return
# (^ is the start of the text or the place after a \n: the first line, then every "\nvoid f_". A definition is one
# line and holds no "\n", so the next one is looked for from where this one's line began)
at = 0
while at >= 0:
m = head.match(text, at)
if m:
yield m
at = text.find(mark, at)
if at >= 0:
at += 1
def noted_starts(text):
"""Where each definition in text (str) starts, its note's line counted in when the line just above is one: the starts
of HEAD_NOTED's matches, in order"""
if REGEX_SCAN:
for m in HEAD_NOTED.finditer(text):
yield m.start()
return
# (HEAD_NOTED matches from the note's line when the line above the definition, whole, is "/* ... */" and its line
# end; a match never reaches into the line after a definition's, so that line is free to be the next one's note)
for m in definitions(text):
start = m.start()
if start:
above = text.rfind("\n", 0, start - 1) + 1
if _NOTE.fullmatch(text, above, start):
start = above
yield start