helper

Helper utilities for GDS generation: die area, pitch, step substitution.

This module exposes utilities used by the GDS generator flows.

Functions

apply_step_substitutions(steps, substitutions)	Apply LibreLane-style step substitutions to a nested step list.
get_layer_info(config)	Read the FP_TRACKS_INFO file and return layer information.
get_offset(config)	Read the FP_TRACKS_INFO file and return track offsets for X and Y.
get_pitch(config)	Read the FP_TRACKS_INFO file and return min pitches for X and Y.
get_routing_obstructions(config)	Get the routing obstructions from the config.
merge_layered_substitutions(config_sources)	Merge meta.substituting_steps across layered config sources.
round_die_area(config)	Round the DIE_AREA to multiples of the minimum pitch.
round_die_dimension(dimension, pitch, divisions)	Round dimension up so each of its divisions equal parts is pitch-aligned.
round_up_decimal(value, pitch)	Round up value to the next multiple of pitch.

Module Contents

apply_step_substitutions(steps, substitutions) → list[type[librelane.steps.step.Step]]

Apply LibreLane-style step substitutions to a nested step list.

LibreLane’s SequentialFlow.Substitute/meta.substituting_steps only walks a flow’s top-level Steps list, so it cannot see steps nested inside a WhileStep (a FABulous-authored composite, not a LibreLane built-in). This reproduces the same matching/replace semantics for that nested list, so gds_config.yaml can target loop-internal steps with identical syntax.

Parameters:

• steps (list[type[Step]]) – The loop body to substitute into. Not mutated; a copy is returned.

• substitutions (SubstitutionsObject) – A dict or list of (id, step) tuples, using the same syntax as LibreLane’s meta.substituting_steps: a bare id replaces, +id appends after, -id prepends before, and a None value removes.

Returns:

The resulting step list.

get_layer_info(config) → dict[str, dict[str, tuple[Decimal, Decimal]]]

Read the FP_TRACKS_INFO file and return layer information.

Returns a dictionary mapping layer names to their cardinal directions and corresponding (offset, pitch) tuples.

get_offset(config) → tuple[Decimal, Decimal]

Read the FP_TRACKS_INFO file and return track offsets for X and Y.

Returns a tuple (x_offset, y_offset) where x_offset is the track offset along X-axis (IO_PIN_V_LAYER X direction) and y_offset is the track offset along Y-axis (IO_PIN_H_LAYER Y direction). The cardinal field in FP_TRACKS_INFO is expected to be ‘X’ or ‘Y’ (case-insensitive).

get_pitch(config) → tuple[Decimal, Decimal]

Read the FP_TRACKS_INFO file and return min pitches for X and Y.

Returns a tuple (x_pitch, y_pitch) where x_pitch is the minimum pitch along X-axis (IO_PIN_V_LAYER X direction) and y_pitch is minimum pitch along Y-axis (IO_PIN_H_LAYER Y direction). The cardinal field in FP_TRACKS_INFO is expected to be ‘X’ or ‘Y’ (case-insensitive).

get_routing_obstructions(config) → list[tuple[str, Decimal, Decimal, Decimal, Decimal]]

Get the routing obstructions from the config.

Returns a list of tuples (layer, x1, y1, x2, y2) representing the obstructions in the routing area.

Parameters:

config (Config) – The configuration object from liberlane.

Returns:

A list of obstruction tuples.

Raises:

ValueError – If the entry is not a valid obstruction.

merge_layered_substitutions(config_sources) → librelane.flows.sequential.SubstitutionsObject | None

Merge meta.substituting_steps across layered config sources.

Config.load() overwrites Config.meta wholesale for each source in its configs list rather than merging, so a later source with no meta: key (e.g. a tile-specific or custom override dict) silently drops an earlier source’s substituting_steps. This walks the same sources ourselves, in the same order, and concatenates each source’s own substitutions so a later source’s entries still apply after an earlier source’s.

round_die_area(config) → librelane.config.config.Config

Round the DIE_AREA to multiples of the minimum pitch.

This reads the minimum pitch from FP_TRACKS_INFO and updates the config DIE_AREA to start at (0,0) with width/height rounded up to the next multiple of that pitch.

round_die_dimension(dimension, pitch, divisions) → Decimal

Round dimension up so each of its divisions equal parts is pitch-aligned.

A super tile is split into divisions equal parts during IO placement, so dimension / divisions (not just dimension) must be a pitch multiple.

round_up_decimal(value, pitch) → Decimal

Round up value to the next multiple of pitch.