rearrange options and make output cleaner

.gitignore

@@ -0,0 +1 @@
+__pycache__

README.md

@@ -0,0 +1,73 @@
+# iavlread - extract data from the state of a Cosmos SDK blockchain
+
+This is a simple tool read state data from the snapshot of a [Cosmos SDK] block chain. The state is stored in the `application.db` leveldb database, in the form of an [IAVL tree]. This tool walks the IAVL tree to get values from the state, at a desired block height.
+
+### Installation
+
+It's really just two Python files. `iavlread` is the CLI tool, and `iavltree.py` is the library which actually handles the data structure. It can also be used as a package in Python code (`store.ipynb` shows a few examples). To use `iavlread`, it's easiest to just clone this repository and make a symlink to `iavlread` from `$HOME/bin` or other directory which in `$PATH`.
+
+### Usage
+
+I'm just going to show a couple of examples on a snapshot of the [Allora] testnet, as it can be downloaded e.g. [here](https://www.imperator.co/services/chain-services/testnets/allora). That's what I've been using this for, although I'd assume it works the same for other CosmosSDK blockchains. 
+
+We assume that we're inside the snapshot, so the application db is at the path `data/application.db` from the current working directory. Otherwise, it can be specified with using `-d path_to_database`.
+
+To get the **maximum/minimum block height** contained in the snapshot
+
+    $ iavlread max_height s/k:emissions/
+    5224814
+    $ iavlread min_height s/k:emissions/
+    5221360
+
+Here `s/k:emissions/` is the prefix for a specific IAVL tree, the one corresponding to the emissions module. Other prefixes are `s/k:mint/`, `s/k:bank/`, `s/k:staking/`, `s/k:acc/`. They should generally produce the same min/max height, but that is not guaranteed.
+
+To **count** the items in a keeper, use the `count` subcommand:
+
+    $ iavlread count s/k:emissions/
+    19070982
+
+We can also count only the items with a specific key (i.e. one item of the keeper).
+
+    $ iavlread count s/k:emissions/ 62
+    9714538
+
+Here 62 corresponds to the `latestOneOutInfererInfererNetworkRegrets` field, which has type `Map[Triple[uint64, string, string], TimestampedValue]`. So the keys for the individual item in the map are quadruples consisting of 62, an integer, and two strings.
+
+To **iterate through all items** whose key starts with 62, use the `iterate` command:
+
+    $ iavlread -kQss -vpb,2=float iterate s/k:emissions/ 62 | head -n5
+    ([62, 1, 'allo1004k7wqa4spns0wlct6mvxnmfysae07w2p75xc', 'allo1004k7wqa4spns0wlct6mvxnmfysae07w2p75xc'], [(1, 1655577), (2, -2.538390795862665)])
+    ([62, 1, 'allo1004k7wqa4spns0wlct6mvxnmfysae07w2p75xc', 'allo123s56yuz7dkyh54gs7gstulrfzwj3d6ucwlwk2'], [(1, 1655577), (2, -2.9735713547935196)])
+    ([62, 1, 'allo1004k7wqa4spns0wlct6mvxnmfysae07w2p75xc', 'allo136dfvvuhazgyqyls2f68qzr5l3p07uhnk9mmk0'], [(1, 1655577), (2, -2.4482888033843784)])
+    ([62, 1, 'allo1004k7wqa4spns0wlct6mvxnmfysae07w2p75xc', 'allo15fvg8gk63u8ydf3znaaxrrukgwulezjz4azf68'], [(1, 1655577), (2, -2.4736050386063284)])
+    ([62, 1, 'allo1004k7wqa4spns0wlct6mvxnmfysae07w2p75xc', 'allo16k9af4uu7vpm5hy68t6u62jylc5frv747mz5lw'], [(1, 1655577), (2, -2.4686292734004254)])
+
+The option `-kQss` specifies the key format (a 64 bit integer `Q` followed by two strings `s`; see below). And `-vpb,2=float` specifies the value format: a protocol buffer whose field number 2 is a `float`.
+
+If we want to restrict to keys which start with 62 and 60 (i.e. get one-out regrets for topic 60 only)
+
+    $ iavlread -kQss count s/k:emissions/ 62 60 | head -n5
+    10650
+    $ iavlread -kQss -vpb,2=float iterate s/k:emissions/ 62 60 | head -n5
+    ([62, 60, 'allo107zfy4xrp5plt0jmutaj9feer02v6r30amku78', 'allo107zfy4xrp5plt0jmutaj9feer02v6r30amku78'], [(1, 5070658), (2, 0.004613475335253211)])
+    ([62, 60, 'allo107zfy4xrp5plt0jmutaj9feer02v6r30amku78', 'allo10q5h8afpwjh5x3vazxzwwkxfhpzfys9wxxw3q8'], [(1, 4785658), (2, 0.18922985130794248)])
+    ([62, 60, 'allo107zfy4xrp5plt0jmutaj9feer02v6r30amku78', 'allo10vgaxk57dkk0fd255r3gxn5quwzxaqq95m2cz2'], [(1, 4837018), (2, 0.6616848257922094)])
+    ([62, 60, 'allo107zfy4xrp5plt0jmutaj9feer02v6r30amku78', 'allo10w45atfjsh9q6vsk7mx74xh0pvuf8r42vnmt5p'], [(1, 4802038), (2, 0.010614832362991345)])
+    ([62, 60, 'allo107zfy4xrp5plt0jmutaj9feer02v6r30amku78', 'allo12hnfamvwumkfm6dnc42rt8q3yevyqpuzkdtwat'], [(1, 4471438), (2, 0.2668976650081499)])
+
+Or if we know the full key, we can use `get` instead of `iterate`. E.g. to get a balance from the bank module:
+    
+    $ iavlread -kbs -vint get s/k:bank/ 2 570DD38DC5BAF3112A7C83A420ED399A8E59C5FC uallo
+    350
+
+We can also get the value for at a different block height (by default, the max block height is used):
+
+    $ iavlread -H 5221360 -kbs -vint get s/k:bank/ 2 570DD38DC5BAF3112A7C83A420ED399A8E59C5FC uallo
+    10
+
+Or we can get all past updates to the value (that are contained in the snapshot):
+
+    $ iavlread -kbs -vint history s/k:bank/ 2 570DD38DC5BAF3112A7C83A420ED399A8E59C5FC uallo
+    5224814 150
+    5224813 30
+    5224812 60

iavlread

@@ -0,0 +1,155 @@
+#!/usr/bin/env python3
+import argparse
+import plyvel
+import iavltree
+import json
+import struct
+
+def decode_protobuf(subformats: dict, format_prefix: str, data: bytes):
+    result = []
+    for (k,v) in iavltree.parse_pb(data):
+        idx = f'{format_prefix}.{k}'
+        if idx in subformats:
+            f = subformats[idx]
+            if f == 'pb':
+                decoded_value = decode_protobuf(subformats, idx, v)
+            elif f == 'pbdict':
+                decoded_value = dict(decode_protobuf(subformats, idx, v))
+            else:
+                decoded_value = decode_output(f, v)
+        else:
+            decoded_value = v
+        result.append((k, decoded_value))
+    return result
+
+def decode_output(format: str, data: bytes) -> str:
+    if format == 'str':
+        return data.decode('utf-8')
+    elif format == 'int':
+        return int(data)
+    elif format == 'float':
+        return float(data)
+    elif format == 'u64':
+        return struct.unpack('>Q', data[:8])[0]
+    elif format == 'u32':
+        return struct.unpack('>I', data[:4])[0]
+    elif format == 'u16':
+        return struct.unpack('>H', data[:2])[0]
+    elif format == 'u8':
+        return struct.unpack('>B', data[:1])[0]
+    elif format == 'i64':
+        return struct.unpack('>q', data[:8])[0]
+    elif format == 'i32':
+        return struct.unpack('>i', data[:4])[0]
+    elif format == 'i16':
+        return struct.unpack('>h', data[:2])[0]
+    elif format == 'i8':
+        return struct.unpack('>b', data[:1])[0]
+    elif format == 'i64ord':
+        return struct.unpack('>Q', data[:8])[0] - (1<<63)
+    elif format == 'i32ord':
+        return struct.unpack('>I', data[:4])[0] - (1<<31)
+    elif format == 'i16ord':
+        return struct.unpack('>H', data[:2])[0] - (1<<15)
+    elif format == 'i8ord':
+        return struct.unpack('>B', data[:1])[0] - (1<<7)
+    elif format.startswith('pbdict'):
+        subformats = {'.' + id: subformat for x in format.split(',')[1:] for id, subformat in (x.split('='),)}
+        return dict(decode_protobuf(subformats, '', data))
+    elif format.startswith('pb'):
+        subformats = {'.' + id: subformat for x in format.split(',')[1:] for id, subformat in (x.split('='),)}
+        return decode_protobuf(subformats, '', data)
+    else:
+        return data
+
+def get_args():
+    parser = argparse.ArgumentParser(description="Read the IAVL tree in a cosmos snapshot")
+
+    parser.add_argument('-d', '--database', help='Path to database (application.db folder)')
+    parser.add_argument('-H', '--height', type=int, help='Block height')
+    # parser.add_argument('-j', '--json', action='store_true', help='JSON output')
+
+    def add_key_cmd(subparsers, cmd, help, optional: bool):
+        subp = subparsers.add_parser(cmd, help = help)
+        subp.add_argument('-k', '--keyformat', help='Key format for maps (e.g. Qss)')
+        subp.add_argument('-v', '--valueformat', help='Value format')
+        subp.add_argument('prefix', help = 'Prefix (e.g. "s/k:emissions/")')
+        subp.add_argument('key', nargs='*' if optional else '+', help = 'Key parts')
+        return subp
+
+    subparsers = parser.add_subparsers(required=True, dest='cmd')
+    p_max_height = subparsers.add_parser('max_height', help = 'Get the max block height in the snapshot')
+    p_max_height.add_argument('prefix', help = 'Prefix (e.g. "s/k:emissions/")')
+    p_min_height = subparsers.add_parser('min_height', help = 'Get the min block height in the snapshot')
+    p_min_height.add_argument('prefix', help = 'Prefix (e.g. "s/k:emissions/")')
+    add_key_cmd(subparsers, 'get', 'Retrieve a single item', False)
+    add_key_cmd(subparsers, 'history', 'Get all stored past values of the item', False)
+    add_key_cmd(subparsers, 'count', 'Count number of items with a prefix', True)
+    add_key_cmd(subparsers, 'iterate', 'Iterate over items with some prefix', True)
+    add_key_cmd(subparsers, 'iterate_keys', 'Iterate over items with some prefix, output keys only', True)
+    add_key_cmd(subparsers, 'iterate_values', 'Iterate over items with some prefix, output values only', True)
+
+    return parser.parse_args()
+
+def run(args):
+    dbpath = args.database if args.database is not None else 'data/application.db'
+    keyformat = args.keyformat if hasattr(args, 'keyformat') and args.keyformat is not None else ''
+    valueformat = args.valueformat if hasattr(args, 'valueformat') and args.valueformat is not None else 'b'
+
+    if args.cmd == 'max_height' or args.cmd == 'min_height' or args.key is None or len(args.key) == 0:
+        key = None
+    else:
+        if len(args.key) > len(keyformat) + 1:
+            raise Exception('Too many key elements for keyformat')
+        key = [int(args.key[0])]
+        for f, k in zip(keyformat, args.key[1:]):
+            if f in ['i', 'I', 'q', 'Q']:
+                key.append(int(k))
+            else:
+                key.append(k)
+
+    with plyvel.DB(dbpath) as db:
+        if args.height is None or args.cmd == 'max_height':
+            height = iavltree.max_height(db, args.prefix.encode('utf-8'))
+        else:
+            height = args.height
+
+        if args.cmd == 'max_height':
+            print(height)
+        elif args.cmd == 'min_height':
+            hmin, _ = iavltree.min_max_height(db, args.prefix.encode('utf-8'))
+            print(hmin)
+        elif args.cmd == 'get':
+            result = iavltree.get(db, args.prefix, height, keyformat, key)
+
+            if result is not None:
+                print(decode_output(valueformat, result))
+        elif args.cmd == 'history':
+            it = iavltree.history(db, args.prefix, keyformat, key, height)
+
+            try:
+                for h, v in it:
+                    print(f'{h} {decode_output(valueformat, v)}')
+            except BrokenPipeError:
+                pass
+        elif args.cmd == 'count':
+            result = iavltree.count(db, args.prefix, height, keyformat, key = key)
+
+            print(result)
+        elif args.cmd == 'iterate' or args.cmd == 'iterate_keys' or args.cmd == 'iterate_values':
+            it = iavltree.iterate(db, args.prefix, height, keyformat, key = key)
+
+            try:
+                for k, v in it:
+                    if args.cmd == 'iterate_keys':
+                        print(' '.join([str(x) for x in k]))
+                    elif args.cmd == 'iterate_values':
+                        print(decode_output(valueformat,v))
+                    else:
+                        print(' '.join([str(x) for x in k]), decode_output(valueformat, v))
+            except BrokenPipeError:
+                pass
+
+if __name__ == '__main__':
+    args = get_args()
+    run(args)

iavltree.py

@@ -1,9 +1,8 @@
 import plyvel
 import struct
-import numpy as np
 
 # functions for reading IAVL tree
-def read_varint(x: bytes, offset: int = 0) -> int:
+def read_varint(x: bytes, offset: int = 0) -> tuple[int, int]:
     result = 0
     factor = 1
 
@@ -15,7 +14,7 @@ def read_varint(x: bytes, offset: int = 0) -> int:
             return result // 2, offset+i+1
         factor *= 128
 
-def read_uvarint(x: bytes, offset: int = 0) -> int:
+def read_uvarint(x: bytes, offset: int = 0) -> tuple[int, int]:
     result = 0
     factor = 1
 
@@ -27,6 +26,20 @@ def read_uvarint(x: bytes, offset: int = 0) -> int:
             return result, offset+i+1
         factor *= 128
 
+def write_uvarint(x: int) -> list[int]:
+    if x < 0:
+        raise Exception('write_uvarint only supports positive integers')
+    elif x == 0:
+        return [0]
+
+    result = []
+    while x > 0:
+        result.append(128 + x % 128)
+        x //= 128
+    result[-1] -= 128
+    return result
+
+
 def read_key(key: bytes) -> tuple[int, int] | None:
     if not key.startswith(b's'):
         return None
@@ -43,7 +56,6 @@ def write_key(key: tuple[int, int]) -> bytes:
     return b's' + version + nonce
 
 def read_node(node: bytes) -> tuple[int, int, bytes, tuple[int, int], tuple[int, int]] | tuple[int, int, list[int], bytes] | tuple[int, int]:
-
     if node.startswith(b's'):
         return read_key(node)
 
@@ -74,53 +86,36 @@ def read_node(node: bytes) -> tuple[int, int, bytes, tuple[int, int], tuple[int,
 
         return (height, length, key, (left_version, left_nonce), (right_version, right_nonce))
 
-def walk(tree, version, searchkey):
-    if (version, 1) not in tree:
+def get_raw(db: plyvel.DB, prefix: bytes, version: int, searchkey: bytes) -> None | tuple[int, int, int, int, bytes, bytes]:
+    key = db.get(prefix + write_key((version, 1)))
+    if key is None:
         return None
 
-    node = tree[(version, 1)]
-    if len(node) == 2: # root copy?
-        node = tree[node]
-
-    while node[0] > 0:
-        nodekey = node[2]
-        if searchkey < nodekey:
-            next = node[3]
-        else:
-            next = node[4]
-
-        node = tree[next]
-
-    return node[3]
-
-def walk_disk_raw(db, prefix: bytes, version: int, searchkey: bytes) -> None | bytes:
-
-    root = db.get(prefix + write_key((version, 1)))
-    if root is None:
-        return None
-
-    node = read_node(root)
+    node = read_node(key)
 
     if len(node) == 2: # root copy?
-        node = read_node(db.get(prefix + write_key(node)))
+        key = node
+        node = read_node(db.get(prefix + write_key(key)))
 
     while node[0] > 0:
         # print(node)
 
         nodekey = node[2]
         if searchkey < nodekey:
-            next = node[3]
+            key = node[3]
         else:
-            next = node[4]
+            key = node[4]
 
-        node = read_node(db.get(prefix + write_key(next)))
+        node = read_node(db.get(prefix + write_key(key)))
 
     if node[2] == searchkey:
-        return node[3]
+        (version, nonce) = key
+        (height, length, itemkey, value) = node
+        return (version, nonce, height, length, itemkey, value)
     else:
         return None
 
-def walk_disk_next_key_raw(db, prefix: bytes, version: int, searchkey: bytes) -> None | bytes:
+def get_next_key_raw(db: plyvel.DB, prefix: bytes, version: int, searchkey: bytes) -> None | bytes:
     root = db.get(prefix + write_key((version, 1)))
     if root is None:
         return None
@@ -146,10 +141,11 @@ def walk_disk_next_key_raw(db, prefix: bytes, version: int, searchkey: bytes) ->
 
     return lowest_geq_key
 
-def walk_disk(db, prefix: str, version: int, format: str, searchkey: list) -> None | bytes:
-    return walk_disk_raw(db, prefix.encode('utf-8'), version, encode_key(format, searchkey))
+def get(db: plyvel.DB, prefix: str, version: int, format: str, searchkey: list) -> None | bytes:
+    x = get_raw(db, prefix.encode('utf-8'), version, encode_key(format, searchkey))
+    return x[5] if x is not None else None
 
-def parse_struct(data):
+def parse_pb(data):
     n = 0
     results = []
 
@@ -180,26 +176,51 @@ def next_key(db, k: bytes) -> bytes | None:
     finally:
         it.close()
 
-def max_height(db) -> int:
+def max_height(db: plyvel.DB, prefix: bytes | str) -> int:
+    if isinstance(prefix, str):
+        prefix = prefix.encode('utf-8')
+
     testnr = 1<<63
 
     for i in range(62, -1, -1):
-        prefix = b's/k:emissions/s'
-        n = next_key(db, prefix + struct.pack('>Q', testnr))
+        n = next_key(db, prefix + b's' + struct.pack('>Q', testnr) + struct.pack('>I', 1))
 
         if n is not None and n.startswith(prefix):
-            # print(f'{testnr:16x} is low')
+            # print(f'{testnr} is low')
             testnr += 1 << i
         else:
-            # print(f'{testnr:16x} is high')
+            # print(f'{testnr} is high')
             testnr -= 1 << i
 
-    n = next_key(db, prefix + struct.pack('>Q', testnr))
+    n = db.get(prefix + struct.pack('>Q', testnr))
     if n is not None and n.startswith(prefix):
         return testnr
     else:
         return testnr - 1
 
+def min_max_height(db: plyvel.DB, prefix: bytes) -> tuple[int, int]:
+    if isinstance(prefix, str):
+        prefix = prefix.encode('utf-8')
+
+    hmax = max_height(db, prefix)
+
+    h = 1<<hmax.bit_length()
+    inc = h>>1
+
+    for _ in range(25):
+        if h > hmax:
+            highenough = True
+        else:
+            root = db.get(prefix + write_key((h, 1)))
+            highenough = root is not None
+        # print(h, highenough, inc)
+        (h, inc) = (h + (1 - 2*highenough) * inc, inc >> 1)
+    if not highenough:
+        h += 1
+
+    return (h, hmax)
+
+
 # encode and decode keys
 def encode_key(format: str, key: list) -> bytes:
     result_bytes = []
@@ -217,6 +238,10 @@ def encode_key(format: str, key: list) -> bytes:
             result_bytes += list(struct.pack('>Q', key[i+1]))
         elif f == 'q':
             result_bytes += list(struct.pack('>Q', key[i+1] + (1<<63)))
+        elif f == 'b':
+            data = list(bytes.fromhex(key[i+1]))
+            result_bytes += write_uvarint(len(data))
+            result_bytes += data
 
     return bytes(result_bytes)
 
@@ -244,6 +269,11 @@ def decode_key(format: str, key: bytes) -> list:
             v = struct.unpack('>Q', key[idx:idx+8])[0]
             result.append(v - (1<<63))
             idx += 8
+        elif f == 'b':
+            length, offset = read_uvarint(key[idx:])
+            data = key[idx+offset:idx+offset+length]
+            result.append(data.hex().upper())
+            idx += offset + length
 
     if idx < len(key):
         result.append(key[idx:])
@@ -252,26 +282,32 @@ def decode_key(format: str, key: bytes) -> list:
 
 # iteration
 class IAVLTreeIteratorRaw:
-    def __init__(self, db, prefix: bytes, version: int, start: bytes | None = None, end: bytes | None = None):
+    def __init__(self, db: plyvel.DB, prefix: bytes, version: int, start: bytes | None = None, end: bytes | None = None):
         self.db = db
         self.prefix = prefix
         self.version = version
         self.start = start
         self.end = end
         self.stack = []
+        self.lookups = []
 
     def __iter__(self):
         return self
 
+    def get_node(self, key):
+        key_enc = self.prefix + write_key(key)
+        self.lookups.append(key_enc)
+        return self.db.get(key_enc)
+
     def __next__(self):
         if len(self.stack) == 0:
             # get root node
-            root = db.get(self.prefix + write_key((self.version, 1)))
+            root = self.get_node((self.version, 1))
             if root is None:
                 raise StopIteration
             node = read_node(root)
             if len(node) == 2: # link to other root node
-                node = read_node(db.get(self.prefix + write_key(node)))
+                node = read_node(self.get_node(node))
             self.stack.append(((self.version, 1), node))
 
             # walk tree to either last before start or first after start
@@ -282,7 +318,7 @@ class IAVLTreeIteratorRaw:
                     next = node[3]
                 else:
                     next = node[4]
-                node = read_node(db.get(self.prefix + write_key(next)))
+                node = read_node(self.get_node(next))
                 self.stack.append((next, node))
 
             # return early if we ended up at first item after start
@@ -308,13 +344,13 @@ class IAVLTreeIteratorRaw:
             raise StopIteration
 
         # go right
-        node = read_node(db.get(self.prefix + write_key(key)))
+        node = read_node(self.get_node(key))
         self.stack.append((key, node))
 
         # go left until at a leaf
         while node[0] > 0:
             key = node[3]
-            node = read_node(db.get(self.prefix + write_key(key)))
+            node = read_node(self.get_node(key))
             self.stack.append((key, node))
 
         if self.end is not None and node[2] >= self.end:
@@ -323,11 +359,9 @@ class IAVLTreeIteratorRaw:
         return (node[2], node[3])
 
 class IAVLTreeIterator:
-    def __init__(self, db, prefix: str, version: int, format: str, start: list | None = None, end: list | None = None):
+    def __init__(self, db: plyvel.DB, prefix: bytes, version: int, format: str, start: bytes | None = None, end: bytes | None = None):
         self.format = format
-        start_enc = encode_key(format, start) if start is not None else None
-        end_enc = encode_key(format, end) if end is not None else None
-        self.inner = IAVLTreeIteratorRaw(db, prefix.encode('utf-8'), version, start = start_enc, end = end_enc)
+        self.inner = IAVLTreeIteratorRaw(db, prefix, version, start, end)
 
     def __iter__(self):
         return self
@@ -336,15 +370,58 @@ class IAVLTreeIterator:
         (k, v) = next(self.inner)
         return (decode_key(self.format, k), v)
 
-def iterate(db, prefix, version, format = '', field = None, start = None, end = None):
-    if field is not None:
-        return IAVLTreeIterator(db, prefix, version, format, start = [field], end = [field+1] if field < 255 else None)
-    else:
-        return IAVLTreeIterator(db, prefix, version, format, start = start, end = end)
+def next_bs(x: bytes) -> bytes | None:
+    if len(x) == 0:
+        return None
 
-def indexof_raw(db, prefix: bytes, version: int, key: bytes) -> int:
+    x_enc = None
+    for i in range(len(x),0,-1):
+        if x[i-1] != 255:
+            x_enc = x[:i-1] + bytes([x[i-1] + 1]) + bytes([0 for _ in range(len(x)-i)])
+            break
+
+    return x_enc
+
+def iterate(db: plyvel.DB, prefix, version, format = '', key = None, start = None, end = None):
+    prefix_enc = prefix.encode('utf-8')
+
+    if key is not None:
+        start_enc = encode_key(format, key)
+        end_enc = next_bs(start_enc)
+    else:
+        start_enc = encode_key(format, start) if start is not None else None
+        end_enc = encode_key(format, end) if end is not None else None
+
+    return IAVLTreeIterator(db, prefix_enc, version, format, start = start_enc, end = end_enc)
+
+def count(db: plyvel.DB, prefix, version, format = '', key = None, start = None, end = None):
+    prefix_enc = prefix.encode('utf-8')
+
+    if key is not None:
+        start_enc = encode_key(format, key)
+        end_enc = next_bs(start_enc)
+    else:
+        start_enc = encode_key(format, start) if start is not None else None
+        end_enc = encode_key(format, end) if end is not None else None
+
+    startidx = indexof_raw(db, prefix_enc, version, start_enc) if start_enc is not None else 0
+
+    if end_enc is not None:
+        endidx = indexof_raw(db, prefix_enc, version, end_enc)
+    else:
+        # get full count
+        it = IAVLTreeIteratorRaw(db, prefix_enc, version)
+        try:
+            next(it)
+            endidx = it.stack[0][1][1] # just read the length field of the root element
+        except StopIteration:
+            endidx = 0
+
+    return endidx - startidx
+
+def indexof_raw(db: plyvel.DB, prefix: bytes, version: int, key: bytes) -> int:
     """
-    Find how many items come before `key` in the tree. If `key` doesn't exist, how many 
+    Find how many items come before `key` in the tree. If `key` doesn't exist, how many
     items come before the slot it would get inserted at
     """
     it = IAVLTreeIteratorRaw(db, prefix, version, start=key)
@@ -352,13 +429,39 @@ def indexof_raw(db, prefix: bytes, version: int, key: bytes) -> int:
         next(it)
     except StopIteration:
         # get root count
-        return read_node(db.get(prefix + write_key(it.stack[0][0])))[1]
-    
+        return it.stack[0][1][1]
+
     keys = [p[1][3] for p, c in zip(it.stack, it.stack[1:]) if c[0] == p[1][4]]
     keys_encoded = [prefix + write_key(k) for k in keys]
     count = sum([read_node(db.get(k))[1] for k in keys_encoded])
 
     return count
 
-def indexof(db, prefix: str, version: int, format: str, key: list) -> int:
+def indexof(db: plyvel.DB, prefix: str, version: int, format: str, key: list) -> int:
     return indexof_raw(db, prefix.encode('utf-8'), version, encode_key(format, key))
+
+class IAVLTreeHistoryIterator:
+    def __init__(self, db: plyvel.DB, prefix: bytes, key: bytes, max_height: int, min_height: int = 0):
+        self.db = db
+        self.prefix = prefix
+        self.key = key
+        self.height = max_height
+        self.min_height = min_height
+
+    def __iter__(self):
+        return self
+
+    def __next__(self) -> tuple[int, bytes]:
+        if self.height < self.min_height:
+            raise StopIteration
+        result = get_raw(self.db, self.prefix, self.height, self.key)
+        if result is None:
+            raise StopIteration
+        (h, _, _, _, _, v) = result
+        self.height = h - 1
+        return (h, v)
+
+def history(db: plyvel.DB, prefix: str, format: str, key: list, max_height: int, min_height: int = 0) -> IAVLTreeHistoryIterator:
+    prefix_enc = prefix.encode('utf-8')
+    key_enc = encode_key(format, key)
+    return IAVLTreeHistoryIterator(db, prefix_enc, key_enc, max_height, min_height)

perftest.py

@@ -0,0 +1,29 @@
+#!/usr/bin/env python
+import plyvel
+import iavltree
+from tqdm import tqdm
+
+with plyvel.DB('../node/nodedir/data/application.db') as db:
+    height = iavltree.max_height(db)
+
+    total = iavltree.count(db, 's/k:emissions/', height, 'Qss', key = [62])
+    progress = tqdm(total = total)
+
+    it = iavltree.iterate(db, 's/k:emissions/', height, 'Qss', key = [62])
+
+    for k, v in it:
+        progress.update(1)
+
+    progress.close()
+
+    keys = it.inner.lookups
+
+print(f'Number of items: {total}')
+print(f'Lookups needed: {len(keys)}')
+
+with plyvel.DB('../node/nodedir/data/application.db') as db:
+    progress = tqdm(total = len(keys))
+    for k in keys:
+        db.get(k)
+        progress.update(1)
+    progress.close()

store.ipynb

@@ -2,50 +2,54 @@
  "cells": [
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 1,
    "metadata": {},
    "outputs": [],
    "source": [
     "import plyvel\n",
+    "from itertools import islice\n",
+    "import iavltree"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# db = plyvel.DB('../node/nodedir/data/application.db')\n",
+    "height = iavltree.max_height(db)\n",
+    "height"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "[k for k, v in iavltree.iterate(db, 's/k:mint/', height)]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "dict(iavltree.parse_pb(next(iavltree.iterate(db, 's/k:mint/', height, key = [138]))[1]))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "it = iavltree.iterate(db, 's/k:emissions/', height, key = [62, 64], format = 'Qss')\n",
+    "ooiiregrets = [(k[2],k[3],value[1],float(value[2])) for k,v in it for value in (dict(iavltree.parse_pb(v)),)]\n",
     "\n",
-    "%run -i read_tree.py"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# db = plyvel.DB('../testnode/nodedir/data/application.db')\n",
-    "max_height(db)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "[k for k, v in iterate(db, 's/k:mint/', 5224815)]"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "parse_struct(next(iterate(db, 's/k:mint/', 5224815, field = 138))[1])"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "[k for k,v in iterate(db, 's/k:emissions/', 5224815, start = [62, 60], end = [62, 61], format = 'Qss')]"
+    "len(ooiiregrets), len(it.inner.lookups)"
    ]
   },
   {
@@ -54,16 +58,20 @@
    "metadata": {},
    "outputs": [],
    "source": [
+    "import numpy as np\n",
+    "keynames = {0: \"Params\", 1: \"TotalStake\", 2: \"TopicStake\", 3: \"Rewards\", 4: \"NextTopicId\", 5: \"Topics\", 6: \"TopicWorkers\", 7: \"TopicReputers\", 8: \"DelegatorStake\", 9: \"DelegateStakePlacement\", 10: \"TargetStake\", 11: \"Inferences\", 12: \"Forecasts\", 13: \"WorkerNodes\", 14: \"ReputerNodes\", 15: \"LatestInferencesTs\", 16: \"ActiveTopics\", 17: \"AllInferences\", 18: \"AllForecasts\", 19: \"AllLossBundles\", 20: \"StakeRemoval\", 21: \"StakeByReputerAndTopicId\", 22: \"DelegateStakeRemoval\", 23: \"AllTopicStakeSum\", 24: \"AddressTopics\", 24: \"WhitelistAdmins\", 25: \"ChurnableTopics\", 26: \"RewardableTopics\", 27: \"NetworkLossBundles\", 28: \"NetworkRegrets\", 29: \"StakeByReputerAndTopicId\", 30: \"ReputerScores\", 31: \"InferenceScores\", 32: \"ForecastScores\", 33: \"ReputerListeningCoefficient\", 34: \"InfererNetworkRegrets\", 35: \"ForecasterNetworkRegrets\", 36: \"OneInForecasterNetworkRegrets\", 37: \"OneInForecasterSelfNetworkRegrets\", 38: \"UnfulfilledWorkerNonces\", 39: \"UnfulfilledReputerNonces\", 40: \"FeeRevenueEpoch\", 41: \"TopicFeeRevenue\", 42: \"PreviousTopicWeight\", 43: \"PreviousReputerRewardFraction\", 44: \"PreviousInferenceRewardFraction\", 45: \"PreviousForecastRewardFraction\", 46: \"InfererScoreEmas\", 47: \"ForecasterScoreEmas\", 48: \"ReputerScoreEmas\", 49: \"TopicRewardNonce\", 50: \"DelegateRewardPerShare\", 51: \"PreviousPercentageRewardToStakedReputers\", 52: \"StakeRemovalsByBlock\", 53: \"DelegateStakeRemovalsByBlock\", 54: \"StakeRemovalsByActor\", 55: \"DelegateStakeRemovalsByActor\", 56: \"TopicLastWorkerCommit\", 57: \"TopicLastReputerCommit\", 58: \"TopicLastWorkerPayload\", 59: \"TopicLastReputerPayload\", 60: \"OpenWorkerWindows\", 61: \"LatestNaiveInfererNetworkRegrets\", 62: \"LatestOneOutInfererInfererNetworkRegrets\", 63: \"LatestOneOutInfererForecasterNetworkRegrets\", 64: \"LatestOneOutForecasterInfererNetworkRegrets\", 65: \"LatestOneOutForecasterForecasterNetworkRegrets\", 66: \"PreviousForecasterScoreRatio\", 67: \"LastDripBlock\", 68: \"TopicToNextPossibleChurningBlock\", 69: \"BlockToActiveTopics\", 70: \"BlockToLowestActiveTopicWeight\", 71: \"PreviousTopicQuantileInfererScoreEma\", 72: \"PreviousTopicQuantileForecasterScoreEma\", 73: \"PreviousTopicQuantileReputerScoreEma\", 74: \"CountInfererInclusionsInTopic\", 75: \"CountForecasterInclusionsInTopic\", 76: \"ActiveInferers\", 77: \"ActiveForecasters\", 78: \"ActiveReputers\", 79: \"LowestInfererScoreEma\", 80: \"LowestForecasterScoreEma\", 81: \"LowestReputerScoreEma\", 82: \"LossBundles\", 83: \"TotalSumPreviousTopicWeights\", 84: \"RewardCurrentBlockEmission\", 85: \"GlobalWhitelist\", 86: \"TopicCreatorWhitelist\", 87: \"TopicWorkerWhitelist\", 88: \"TopicReputerWhitelist\", 89: \"TopicWorkerWhitelistEnabled\", 90: \"TopicReputerWhitelistEnabled\", 91: \"LastMedianInferences\", 92: \"MadInferences\", 93: \"InitialInfererEmaScore\", 94: \"InitialForecasterEmaScore\", 95: \"InitialReputerEmaScore\", 96: \"GlobalWorkerWhitelist\", 97: \"GlobalReputerWhitelist\", 98: \"GlobalAdminWhitelist\", 99: \"LatestRegretStdNorm\", 100: \"LatestInfererWeights\", 101: \"LatestForecasterWeights\", 102: \"NetworkInferences\", 103: \"OutlierResistantNetworkInferences\", 104: \"MonthlyReputerRewards\", 105: \"MonthlyTopicRewards\",}\n",
     "lens = np.zeros(256, dtype = int)\n",
     "\n",
-    "with plyvel.DB('../testnode/nodedir/data/application.db') as db:\n",
-    "    height = max_height(db)\n",
-    "    for field in range(255):\n",
-    "        count1 = indexof(db, 's/k:emissions/', height, '', [field])\n",
-    "        count2 = indexof(db, 's/k:emissions/', height, '', [field+1])\n",
-    "        lens[field] = count2 - count1\n",
+    "for field in range(255):\n",
+    "    lens[field] = iavltree.count(db, 's/k:emissions/', height, key = [field])\n",
     "\n",
-    "np.argsort(lens)"
+    "order = np.lexsort((np.arange(256)[::-1], lens))[::-1]\n",
+    "\n",
+    "for i in range(len(order)):\n",
+    "    if lens[order[i]] == 0 and order[i] not in keynames:\n",
+    "        break\n",
+    "\n",
+    "    print(f'{keynames[order[i]]:50} {lens[order[i]]:9d}')"
    ]
   },
   {
@@ -102,6 +110,19 @@
     "\n",
     "# found"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# allora testnet module addresses\n",
+    "# mod allorapendingrewards 54C6D62FF29ECFEE9A5F0366DEC0F9CB44C10BB4\n",
+    "# mod allorarewards F3CA54C42E5B7DC7CB2A347B21E77AC248D914D2\n",
+    "# mod allorastaking 3C19B4642DA1C2DBB7E44679FA48F72FD9A97E5E\n",
+    "# mod ecosystem 570DD38DC5BAF3112A7C83A420ED399A8E59C5FC"
+   ]
   }
  ],
  "metadata": {