dumpobj 0.1.1

No project description provided

dumpobj

A tiny utility to introspect and pretty-print Python objects as a structured tree.

It walks an object, builds a neutral Node tree (key/props/attrs/value/children), and lets you render it with pluggable formatters (plain, color, json).

• Zero dependencies, Python 3.10+
• Handles built-in containers, primitives, classes, instances, exceptions, and ellipsis
• Controls for depth, item count limits, and recursion handling
• Hook to register custom type handlers

Installation

pip install dumpobj

Requires Python 3.10 or above.

Quick start

from dumpobj import Dump
from dumpobj.formatter.plain_formatter import PlainFormatter

obj = {
    "a": [1, 2, 3, 4, 5],
    "b": "ABCDEFG",
    "c": {"x": 1, "y": 2},
}

# Configure
d = Dump()
d.set_inline(True)      # detailed tree with children; False -> compact summaries
d.set_head_count(3)     # print only the first 3 items in containers/strings
d.set_depth(5)          # traverse at most 5 levels
d.set_str_if_recur("<recur>")  # or Ellipsis for default

# Choose a formatter (optional; default is PlainFormatter)
d.set_formatter(PlainFormatter())

# Print directly — dump() returns a generator of lines/items
for line in d.dump(obj):
    print(line)

# If you need the raw Node tree for advanced processing:
node = d.dump_raw(obj)

Use JSON output

from dumpobj import Dump
from dumpobj.formatter.json_formatter import JSONFormatter

obj = {"a": 1, "b": [1, 2, 3]}

d = Dump()
d.set_formatter(JSONFormatter())

for chunk in d.dump(obj):
    print(chunk)  # prints a JSON string of the object structure

Notes:

• The plain formatter prints a lightweight tree. You can implement your own formatter by subclassing Formatter.
• "More N items..." indicates truncated content when head_count is reached.
• dump_raw(obj) returns the intermediate Node tree without rendering.

Concepts and data model

The dumper constructs an intermediate tree of Node objects before rendering.

Node fields:

• key: str — the field/index name (e.g., a dict key or list index like "[0]")
• props: dict — core descriptors
  • title: e.g., the class name of an object
  • type: e.g., "dict", "list", "str", "int", or a fully-qualified class name
• attrs: dict[str, Any] — extra attributes collected per type, e.g.:
  • "@": object id in hex
  • "len": container length
  • "sizeof": Python sizeof
  • exceptions include a "msg" field
• value: Any — leaf value as a string (e.g., numbers, truncated strings, compact container preview when not in detail mode)
• children: list[Node] — nested nodes

API

Dump

Location: dumpobj._dumpobj.Dump

Purpose: Traverse an object and produce a Node tree that represents its structure.

Handled types (by default):

• dict, list, tuple, set
• str, bool, int, float, complex, None, Ellipsis
• BaseException, type, object (generic fallback)

Key methods and properties:

• set_inline(in_detail: bool):
  • True: build a full tree with children for containers/objects
  • False: store a compact summary string in value for many types
• set_head_count(n: int): limit items/characters per container/string; default 100
• get_head_count() -> int | None
• set_depth(n: int | None): maximum recursion depth; default 5; None means unlimited
• get_depth() -> int | None
• set_str_if_recur(mark: str | Ellipsis | None): customize how recursion is marked
  • Ellipsis: sets the key to "..."
  • str: sets the key to that string
  • None: adds a Ref@ attr pointing to the already-seen object id
• get_str_if_recur() -> str | Ellipsis | None
• register_handle(t: type, handle: Callable[[Node, object, int], Node]): add/override type handlers
• dump_raw(obj: object) -> Node: start a new dump pass and return the root node
• dump(obj: object) -> Generator[Any, Any, None]: render via the configured formatter, yielding printable lines/items

Attributes collected per type are controlled by an internal attr_config. For example, containers and strings include @, __len__, and __sizeof__.

Recursion: The dumper tracks seen object ids. For value types (numbers/strings/bytes/bool), it does not track ids; for others it detects cycles and marks them according to str_if_recur.

Depth and truncation:

• Depth limit applies to child traversal when in_detail is True.
• Item limit applies to containers; strings are shortened to the first head_count characters.

Caveat: head_count must be a positive integer in the current version.

Node

Location: dumpobj.node.Node

Accessor methods:

• set_key/get_key
• set_prop/get_prop; PropKeys are "title" | "type"
• set_attr/set_attrs/get_attr/get_attrs
• set_value/get_value
• append_node(iter_children): tree building helpers
• set_parent/get_parent

Formatter, PlainFormatter, JSONFormatter

• Base class: dumpobj.formatter.base_formatter.Formatter
  • Override _format_key/_format_props/_format_attrs/_format_value/_format_header
  • Lifecycle hooks: _pre_render/_post_render/_pre_node/_post_node
  • render(node) yields lines (or items) lazily
• Plain implementation: dumpobj.formatter.plain_formatter.PlainFormatter
  • Config: attr_key_rename dict can rename attribute keys in output
  • Indentation: 4 spaces per level, with +-- tree markers
• JSON implementation: dumpobj.formatter.json_formatter.JSONFormatter
  • Config: compact, indent, ensure_ascii
  • Produces a JSON string representing the object structure

Example: Rename the "@" attribute key to "@":

pf = PlainFormatter()
pf.config["attr_key_rename"] = {"@": "@"}
for line in pf.render(Dump().dump_raw(obj)):
    print(line)

Custom handlers and error handling

You can override how a type is dumped by registering a handler. Handlers receive the current Node, the object, and the current depth, and should mutate/return the node.

If a user-defined handler raises an exception, it’s captured and represented as an ErrorNode in the output.

from datetime import datetime
from dumpobj import Dump
from dumpobj.node import Node

d = Dump()

def dump_datetime(node: Node, obj: datetime, depth: int) -> Node:
    node.set_prop("type", "datetime")
    node.set_value(obj.isoformat())
    return node

d.register_handle(datetime, dump_datetime)

License

MIT License. See LICENSE for details.

Links

• Repository: https://github.com/Ruilx/dumpobj
• Author: Ruilx