This commit is contained in:
Leo Vasanko 2025-08-13 10:18:18 -07:00
parent 47574675a3
commit 4060a582d6
2 changed files with 456 additions and 343 deletions

View File

@ -50,12 +50,16 @@ def treeget(rootmod: list[FileEntry], path: PurePosixPath):
begin = None begin = None
ret = [] ret = []
iteration_count = 0 iteration_count = 0
for i, relpath, entry in treeiter(rootmod): for i, relpath, entry in treeiter(rootmod):
iteration_count += 1 iteration_count += 1
if iteration_count % 1000 == 0: # Log every 1000 iterations to detect infinite loops if (
logger.debug(f"DEBUG: treeget iteration {iteration_count}, i={i}, relpath={relpath}, entry.name={entry.name}") iteration_count % 1000 == 0
): # Log every 1000 iterations to detect infinite loops
logger.debug(
f"DEBUG: treeget iteration {iteration_count}, i={i}, relpath={relpath}, entry.name={entry.name}"
)
if begin is None: if begin is None:
if relpath == path: if relpath == path:
logger.debug(f"DEBUG: treeget FOUND path {path} at index {i}") logger.debug(f"DEBUG: treeget FOUND path {path} at index {i}")
@ -63,59 +67,77 @@ def treeget(rootmod: list[FileEntry], path: PurePosixPath):
ret.append(entry) ret.append(entry)
continue continue
if entry.level <= len(path.parts): if entry.level <= len(path.parts):
logger.debug(f"DEBUG: treeget BREAK: entry.level={entry.level} <= path.parts_len={len(path.parts)}") logger.debug(
f"DEBUG: treeget BREAK: entry.level={entry.level} <= path.parts_len={len(path.parts)}"
)
break break
ret.append(entry) ret.append(entry)
logger.debug(f"DEBUG: treeget EXIT: path={path}, begin={begin}, ret_len={len(ret)}, iterations={iteration_count}") logger.debug(
f"DEBUG: treeget EXIT: path={path}, begin={begin}, ret_len={len(ret)}, iterations={iteration_count}"
)
return begin, ret return begin, ret
def treeinspos(rootmod: list[FileEntry], relpath: PurePosixPath, relfile: int): def treeinspos(rootmod: list[FileEntry], relpath: PurePosixPath, relfile: int):
# Find the first entry greater than the new one # Find the first entry greater than the new one
# precondition: the new entry doesn't exist # precondition: the new entry doesn't exist
logger.debug(f"DEBUG: treeinspos ENTRY: relpath={relpath}, relfile={relfile}, rootmod_len={len(rootmod)}") logger.debug(
f"DEBUG: treeinspos ENTRY: relpath={relpath}, relfile={relfile}, rootmod_len={len(rootmod)}"
)
isfile = 0 isfile = 0
level = 0 level = 0
i = 0 i = 0
iteration_count = 0 iteration_count = 0
for i, rel, entry in treeiter(rootmod): for i, rel, entry in treeiter(rootmod):
iteration_count += 1 iteration_count += 1
# Detect potential infinite loops in treeinspos # Detect potential infinite loops in treeinspos
if iteration_count % 1000 == 0: if iteration_count % 1000 == 0:
logger.debug(f"DEBUG: treeinspos iteration {iteration_count}, i={i}, rel={rel}, entry.name={entry.name}, level={level}, entry.level={entry.level}") logger.debug(
f"DEBUG: treeinspos iteration {iteration_count}, i={i}, rel={rel}, entry.name={entry.name}, level={level}, entry.level={entry.level}"
)
if iteration_count > 10000: # Emergency brake for infinite loops if iteration_count > 10000: # Emergency brake for infinite loops
logger.error(f"ERROR: treeinspos potential infinite loop! iteration={iteration_count}, relpath={relpath}, i={i}, level={level}") logger.error(
f"ERROR: treeinspos potential infinite loop! iteration={iteration_count}, relpath={relpath}, i={i}, level={level}"
)
break break
if entry.level > level: if entry.level > level:
# We haven't found item at level, skip subdirectories # We haven't found item at level, skip subdirectories
logger.debug(f"DEBUG: treeinspos SKIP: entry.level={entry.level} > level={level}") logger.debug(
f"DEBUG: treeinspos SKIP: entry.level={entry.level} > level={level}"
)
continue continue
if entry.level < level: if entry.level < level:
# We have passed the level, so the new item is the first # We have passed the level, so the new item is the first
logger.debug(f"DEBUG: treeinspos RETURN_EARLY: entry.level={entry.level} < level={level}, returning i={i}") logger.debug(
f"DEBUG: treeinspos RETURN_EARLY: entry.level={entry.level} < level={level}, returning i={i}"
)
return i return i
if level == 0: if level == 0:
# root # root
logger.debug("DEBUG: treeinspos ROOT: incrementing level from 0 to 1") logger.debug("DEBUG: treeinspos ROOT: incrementing level from 0 to 1")
level += 1 level += 1
continue continue
ename = rel.parts[level - 1] ename = rel.parts[level - 1]
name = relpath.parts[level - 1] name = relpath.parts[level - 1]
logger.debug(f"DEBUG: treeinspos COMPARE: ename='{ename}', name='{name}', level={level}") logger.debug(
f"DEBUG: treeinspos COMPARE: ename='{ename}', name='{name}', level={level}"
)
esort = sortkey(ename) esort = sortkey(ename)
nsort = sortkey(name) nsort = sortkey(name)
# Non-leaf are always folders, only use relfile at leaf # Non-leaf are always folders, only use relfile at leaf
isfile = relfile if len(relpath.parts) == level else 0 isfile = relfile if len(relpath.parts) == level else 0
logger.debug(f"DEBUG: treeinspos SORT: esort={esort}, nsort={nsort}, isfile={isfile}, entry.isfile={entry.isfile}") logger.debug(
f"DEBUG: treeinspos SORT: esort={esort}, nsort={nsort}, isfile={isfile}, entry.isfile={entry.isfile}"
)
# First compare by isfile, then by sorting order and if that too matches then case sensitive # First compare by isfile, then by sorting order and if that too matches then case sensitive
cmp = ( cmp = (
entry.isfile - isfile entry.isfile - isfile
@ -123,24 +145,28 @@ def treeinspos(rootmod: list[FileEntry], relpath: PurePosixPath, relfile: int):
or (ename > name) - (ename < name) or (ename > name) - (ename < name)
) )
logger.debug(f"DEBUG: treeinspos CMP: cmp={cmp}") logger.debug(f"DEBUG: treeinspos CMP: cmp={cmp}")
if cmp > 0: if cmp > 0:
logger.debug(f"DEBUG: treeinspos RETURN: cmp > 0, returning i={i}") logger.debug(f"DEBUG: treeinspos RETURN: cmp > 0, returning i={i}")
return i return i
if cmp < 0: if cmp < 0:
logger.debug(f"DEBUG: treeinspos CONTINUE: cmp < 0") logger.debug(f"DEBUG: treeinspos CONTINUE: cmp < 0")
continue continue
logger.debug(f"DEBUG: treeinspos INCREMENT_LEVEL: level {level} -> {level + 1}") logger.debug(f"DEBUG: treeinspos INCREMENT_LEVEL: level {level} -> {level + 1}")
level += 1 level += 1
if level > len(relpath.parts): if level > len(relpath.parts):
logger.error(f"ERROR: insertpos level overflow: relpath={relpath}, i={i}, entry.name={entry.name}, entry.level={entry.level}, level={level}") logger.error(
f"ERROR: insertpos level overflow: relpath={relpath}, i={i}, entry.name={entry.name}, entry.level={entry.level}, level={level}"
)
break break
else: else:
logger.debug(f"DEBUG: treeinspos FOR_ELSE: incrementing i from {i} to {i + 1}") logger.debug(f"DEBUG: treeinspos FOR_ELSE: incrementing i from {i} to {i + 1}")
i += 1 i += 1
logger.debug(f"DEBUG: treeinspos EXIT: returning i={i}, iterations={iteration_count}") logger.debug(
f"DEBUG: treeinspos EXIT: returning i={i}, iterations={iteration_count}"
)
return i return i
@ -219,20 +245,26 @@ def update_root(loop):
def update_path(rootmod: list[FileEntry], relpath: PurePosixPath, loop): def update_path(rootmod: list[FileEntry], relpath: PurePosixPath, loop):
"""Called on FS updates, check the filesystem and broadcast any changes.""" """Called on FS updates, check the filesystem and broadcast any changes."""
logger.debug(f"DEBUG: update_path ENTRY: path={relpath}, rootmod_len={len(rootmod)}") logger.debug(
f"DEBUG: update_path ENTRY: path={relpath}, rootmod_len={len(rootmod)}"
)
# Add timing for walk operation # Add timing for walk operation
walk_start = time.perf_counter() walk_start = time.perf_counter()
new = walk(relpath) new = walk(relpath)
walk_end = time.perf_counter() walk_end = time.perf_counter()
logger.debug(f"DEBUG: walk({relpath}) took {walk_end - walk_start:.4f}s, returned {len(new)} entries") logger.debug(
f"DEBUG: walk({relpath}) took {walk_end - walk_start:.4f}s, returned {len(new)} entries"
)
# Add timing for treeget operation # Add timing for treeget operation
treeget_start = time.perf_counter() treeget_start = time.perf_counter()
obegin, old = treeget(rootmod, relpath) obegin, old = treeget(rootmod, relpath)
treeget_end = time.perf_counter() treeget_end = time.perf_counter()
logger.debug(f"DEBUG: treeget({relpath}) took {treeget_end - treeget_start:.4f}s, obegin={obegin}, old_len={len(old) if old else 0}") logger.debug(
f"DEBUG: treeget({relpath}) took {treeget_end - treeget_start:.4f}s, obegin={obegin}, old_len={len(old) if old else 0}"
)
if old == new: if old == new:
logger.debug( logger.debug(
f"Watch: Event without changes needed {relpath}" f"Watch: Event without changes needed {relpath}"
@ -241,29 +273,37 @@ def update_path(rootmod: list[FileEntry], relpath: PurePosixPath, loop):
) )
logger.debug(f"DEBUG: update_path EARLY_EXIT: no changes for {relpath}") logger.debug(f"DEBUG: update_path EARLY_EXIT: no changes for {relpath}")
return return
# Debug the deletion operation # Debug the deletion operation
if obegin is not None: if obegin is not None:
logger.debug(f"DEBUG: DELETING entries from rootmod[{obegin}:{obegin + len(old)}] for path {relpath}") logger.debug(
f"DEBUG: DELETING entries from rootmod[{obegin}:{obegin + len(old)}] for path {relpath}"
)
del rootmod[obegin : obegin + len(old)] del rootmod[obegin : obegin + len(old)]
logger.debug(f"DEBUG: DELETED entries, rootmod_len now {len(rootmod)}") logger.debug(f"DEBUG: DELETED entries, rootmod_len now {len(rootmod)}")
if new: if new:
logger.debug(f"Watch: Update {relpath}" if old else f"Watch: Created {relpath}") logger.debug(f"Watch: Update {relpath}" if old else f"Watch: Created {relpath}")
# Add timing for treeinspos operation - this is where hangs might occur # Add timing for treeinspos operation - this is where hangs might occur
inspos_start = time.perf_counter() inspos_start = time.perf_counter()
i = treeinspos(rootmod, relpath, new[0].isfile) i = treeinspos(rootmod, relpath, new[0].isfile)
inspos_end = time.perf_counter() inspos_end = time.perf_counter()
logger.debug(f"DEBUG: treeinspos({relpath}) took {inspos_end - inspos_start:.4f}s, returned index={i}") logger.debug(
f"DEBUG: treeinspos({relpath}) took {inspos_end - inspos_start:.4f}s, returned index={i}"
logger.debug(f"DEBUG: INSERTING {len(new)} entries at position {i} for path {relpath}") )
logger.debug(
f"DEBUG: INSERTING {len(new)} entries at position {i} for path {relpath}"
)
rootmod[i:i] = new rootmod[i:i] = new
logger.debug(f"DEBUG: INSERTED entries, rootmod_len now {len(rootmod)}") logger.debug(f"DEBUG: INSERTED entries, rootmod_len now {len(rootmod)}")
else: else:
logger.debug(f"Watch: Removed {relpath}") logger.debug(f"Watch: Removed {relpath}")
logger.debug(f"DEBUG: update_path EXIT: path={relpath}, final_rootmod_len={len(rootmod)}") logger.debug(
f"DEBUG: update_path EXIT: path={relpath}, final_rootmod_len={len(rootmod)}"
)
def update_space(loop): def update_space(loop):
@ -284,38 +324,46 @@ def update_space(loop):
def format_update(old, new): def format_update(old, new):
logger.debug(f"DEBUG: format_update ENTRY: old_len={len(old)}, new_len={len(new)}") logger.debug(f"DEBUG: format_update ENTRY: old_len={len(old)}, new_len={len(new)}")
# Make keep/del/insert diff until one of the lists ends # Make keep/del/insert diff until one of the lists ends
oidx, nidx = 0, 0 oidx, nidx = 0, 0
oremain, nremain = set(old), set(new) oremain, nremain = set(old), set(new)
update = [] update = []
keep_count = 0 keep_count = 0
iteration_count = 0 iteration_count = 0
while oidx < len(old) and nidx < len(new): while oidx < len(old) and nidx < len(new):
iteration_count += 1 iteration_count += 1
# Log every 1000 iterations to detect infinite loops # Log every 1000 iterations to detect infinite loops
if iteration_count % 1000 == 0: if iteration_count % 1000 == 0:
logger.debug(f"DEBUG: format_update iteration {iteration_count}, oidx={oidx}/{len(old)}, nidx={nidx}/{len(new)}") logger.debug(
f"DEBUG: format_update iteration {iteration_count}, oidx={oidx}/{len(old)}, nidx={nidx}/{len(new)}"
)
# Emergency brake for potential infinite loops # Emergency brake for potential infinite loops
if iteration_count > 50000: if iteration_count > 50000:
logger.error(f"ERROR: format_update potential infinite loop! iteration={iteration_count}, oidx={oidx}, nidx={nidx}") logger.error(
raise Exception(f"format_update infinite loop detected at iteration {iteration_count}") f"ERROR: format_update potential infinite loop! iteration={iteration_count}, oidx={oidx}, nidx={nidx}"
)
raise Exception(
f"format_update infinite loop detected at iteration {iteration_count}"
)
modified = False modified = False
# Matching entries are kept # Matching entries are kept
if old[oidx] == new[nidx]: if old[oidx] == new[nidx]:
entry = old[oidx] entry = old[oidx]
logger.debug(f"DEBUG: format_update MATCH: entry={entry.name}, oidx={oidx}, nidx={nidx}") logger.debug(
f"DEBUG: format_update MATCH: entry={entry.name}, oidx={oidx}, nidx={nidx}"
)
oremain.remove(entry) oremain.remove(entry)
nremain.remove(entry) nremain.remove(entry)
keep_count += 1 keep_count += 1
oidx += 1 oidx += 1
nidx += 1 nidx += 1
continue continue
if keep_count > 0: if keep_count > 0:
logger.debug(f"DEBUG: format_update KEEP: adding UpdKeep({keep_count})") logger.debug(f"DEBUG: format_update KEEP: adding UpdKeep({keep_count})")
modified = True modified = True
@ -326,12 +374,16 @@ def format_update(old, new):
del_count = 0 del_count = 0
del_start_oidx = oidx del_start_oidx = oidx
while oidx < len(old) and old[oidx] not in nremain: while oidx < len(old) and old[oidx] not in nremain:
logger.debug(f"DEBUG: format_update DELETE: removing old[{oidx}]={old[oidx].name}") logger.debug(
f"DEBUG: format_update DELETE: removing old[{oidx}]={old[oidx].name}"
)
oremain.remove(old[oidx]) oremain.remove(old[oidx])
del_count += 1 del_count += 1
oidx += 1 oidx += 1
if del_count: if del_count:
logger.debug(f"DEBUG: format_update DEL: adding UpdDel({del_count}), oidx {del_start_oidx}->{oidx}") logger.debug(
f"DEBUG: format_update DEL: adding UpdDel({del_count}), oidx {del_start_oidx}->{oidx}"
)
update.append(UpdDel(del_count)) update.append(UpdDel(del_count))
continue continue
@ -340,19 +392,29 @@ def format_update(old, new):
ins_start_nidx = nidx ins_start_nidx = nidx
while nidx < len(new) and new[nidx] not in oremain: while nidx < len(new) and new[nidx] not in oremain:
entry = new[nidx] entry = new[nidx]
logger.debug(f"DEBUG: format_update INSERT: adding new[{nidx}]={entry.name}") logger.debug(
f"DEBUG: format_update INSERT: adding new[{nidx}]={entry.name}"
)
nremain.remove(entry) nremain.remove(entry)
insert_items.append(entry) insert_items.append(entry)
nidx += 1 nidx += 1
if insert_items: if insert_items:
logger.debug(f"DEBUG: format_update INS: adding UpdIns({len(insert_items)} items), nidx {ins_start_nidx}->{nidx}") logger.debug(
f"DEBUG: format_update INS: adding UpdIns({len(insert_items)} items), nidx {ins_start_nidx}->{nidx}"
)
modified = True modified = True
update.append(UpdIns(insert_items)) update.append(UpdIns(insert_items))
if not modified: if not modified:
logger.error(f"ERROR: format_update INFINITE_LOOP: nidx={nidx}, oidx={oidx}, old_len={len(old)}, new_len={len(new)}") logger.error(
logger.error(f"ERROR: old[oidx]={old[oidx].name if oidx < len(old) else 'OUT_OF_BOUNDS'}") f"ERROR: format_update INFINITE_LOOP: nidx={nidx}, oidx={oidx}, old_len={len(old)}, new_len={len(new)}"
logger.error(f"ERROR: new[nidx]={new[nidx].name if nidx < len(new) else 'OUT_OF_BOUNDS'}") )
logger.error(
f"ERROR: old[oidx]={old[oidx].name if oidx < len(old) else 'OUT_OF_BOUNDS'}"
)
logger.error(
f"ERROR: new[nidx]={new[nidx].name if nidx < len(new) else 'OUT_OF_BOUNDS'}"
)
raise Exception( raise Exception(
f"Infinite loop in diff {nidx=} {oidx=} {len(old)=} {len(new)=}" f"Infinite loop in diff {nidx=} {oidx=} {len(old)=} {len(new)=}"
) )
@ -362,13 +424,19 @@ def format_update(old, new):
logger.debug(f"DEBUG: format_update FINAL_KEEP: adding UpdKeep({keep_count})") logger.debug(f"DEBUG: format_update FINAL_KEEP: adding UpdKeep({keep_count})")
update.append(UpdKeep(keep_count)) update.append(UpdKeep(keep_count))
if oremain: if oremain:
logger.debug(f"DEBUG: format_update FINAL_DEL: adding UpdDel({len(oremain)}) for remaining old items") logger.debug(
f"DEBUG: format_update FINAL_DEL: adding UpdDel({len(oremain)}) for remaining old items"
)
update.append(UpdDel(len(oremain))) update.append(UpdDel(len(oremain)))
elif nremain: elif nremain:
logger.debug(f"DEBUG: format_update FINAL_INS: adding UpdIns({len(new[nidx:])}) for remaining new items") logger.debug(
f"DEBUG: format_update FINAL_INS: adding UpdIns({len(new[nidx:])}) for remaining new items"
)
update.append(UpdIns(new[nidx:])) update.append(UpdIns(new[nidx:]))
logger.debug(f"DEBUG: format_update EXIT: generated {len(update)} operations, iterations={iteration_count}") logger.debug(
f"DEBUG: format_update EXIT: generated {len(update)} operations, iterations={iteration_count}"
)
return msgspec.json.encode({"update": update}).decode() return msgspec.json.encode({"update": update}).decode()
@ -440,16 +508,24 @@ def watcher_inotify(loop):
logger.debug(f"Watch: {interesting=} {event=}") logger.debug(f"Watch: {interesting=} {event=}")
if interesting: if interesting:
# Update modified path # Update modified path
logger.debug(f"DEBUG: inotify PROCESSING: event={event}, path={event[2]}/{event[3]}") logger.debug(
f"DEBUG: inotify PROCESSING: event={event}, path={event[2]}/{event[3]}"
)
t0 = time.perf_counter() t0 = time.perf_counter()
path = PurePosixPath(event[2]) / event[3] path = PurePosixPath(event[2]) / event[3]
try: try:
rel_path = path.relative_to(rootpath) rel_path = path.relative_to(rootpath)
logger.debug(f"DEBUG: inotify CALLING update_path: rel_path={rel_path}") logger.debug(
f"DEBUG: inotify CALLING update_path: rel_path={rel_path}"
)
update_path(rootmod, rel_path, loop) update_path(rootmod, rel_path, loop)
logger.debug(f"DEBUG: inotify update_path COMPLETED: rel_path={rel_path}") logger.debug(
f"DEBUG: inotify update_path COMPLETED: rel_path={rel_path}"
)
except Exception as e: except Exception as e:
logger.error(f"ERROR: inotify update_path FAILED: path={path}, error={e}") logger.error(
f"ERROR: inotify update_path FAILED: path={path}, error={e}"
)
raise raise
t1 = time.perf_counter() t1 = time.perf_counter()
logger.debug(f"Watch: Update {event[3]} took {t1 - t0:.1f}s") logger.debug(f"Watch: Update {event[3]} took {t1 - t0:.1f}s")
@ -461,7 +537,9 @@ def watcher_inotify(loop):
logger.debug("DEBUG: inotify TIMEOUT: breaking due to 0.5s timeout") logger.debug("DEBUG: inotify TIMEOUT: breaking due to 0.5s timeout")
break break
if dirty and state.root != rootmod: if dirty and state.root != rootmod:
logger.debug(f"DEBUG: inotify BATCH_UPDATE: state.root_len={len(state.root)}, rootmod_len={len(rootmod)}") logger.debug(
f"DEBUG: inotify BATCH_UPDATE: state.root_len={len(state.root)}, rootmod_len={len(rootmod)}"
)
t0 = time.perf_counter() t0 = time.perf_counter()
logger.debug("DEBUG: inotify CALLING format_update") logger.debug("DEBUG: inotify CALLING format_update")
update = format_update(state.root, rootmod) update = format_update(state.root, rootmod)

File diff suppressed because it is too large Load Diff