klamath-rs/src/library.rs

611 lines
22 KiB
Rust
Raw Normal View History

///
/// File-level read/write functionality.
///
2021-12-18 21:05:00 -08:00
use std::io::Write;
//use std::collections::HashMap;
2021-12-18 21:05:00 -08:00
2024-12-17 18:38:40 -08:00
pub use crate::record;
pub use crate::record::{RecordHeader, Record};
pub use crate::records;
pub use crate::elements;
2025-04-12 12:06:36 -07:00
pub use crate::elements::{read_elements, DListBuilder, FListBuilder};
2024-12-17 18:38:40 -08:00
pub use crate::basic::{IResult, OResult, take_bytes, fail};
2021-12-18 21:05:00 -08:00
use std::string::String;
use std::collections::HashMap;
use std::sync::Arc;
2021-12-18 21:05:00 -08:00
use arrow::datatypes::{DataType, Field, Fields};
use arrow::array::{
StructBuilder, StringBuilder, UInt64Builder, UInt32Builder, Int16Builder, Float64Builder,
2025-04-12 12:06:36 -07:00
StructArray,
};
///
/// Representation of the GDS file header.
///
/// File header records: HEADER BGNLIB LIBNAME UNITS
/// Optional records are ignored if present and never written.
///
/// Version is written as `600`.
///
#[derive(Debug, Clone)]
2021-12-18 21:05:00 -08:00
pub struct FileHeader {
/// Number of user units in one database unit
user_units_per_db_unit: f64,
/// Number of meters in one database unit
meters_per_db_unit: f64,
/// Last-modified time [y, m, d, h, m, s]
mod_time: [i16; 6],
/// Last-accessed time [y, m, d, h, m, s]
acc_time: [i16; 6],
/// Library name
name: Vec<u8>,
2021-12-18 21:05:00 -08:00
}
impl FileHeader {
pub fn new(name: &[u8], meters_per_db_unit: f64, user_units_per_db_unit: f64) -> Self {
FileHeader{
mod_time: [0, 1, 1, 0, 0, 0],
acc_time: [0, 1, 1, 0, 0, 0],
2021-12-18 21:05:00 -08:00
name: name.to_owned(),
user_units_per_db_unit,
meters_per_db_unit,
2021-12-18 21:05:00 -08:00
}
}
/// Read and construct a header from the provided input.
///
/// Args:
/// input: Seekable input to read from
///
/// Returns:
/// FileHeader object
///
pub fn read(input: &[u8]) -> IResult<Self> {
let (input, _version) = records::HEADER::read(input)?;
let (input, [mod_time, acc_time]) = records::BGNLIB::read(input)?;
let (input, name) = records::LIBNAME::skip_and_read(input)?;
let (input, (uu, dbu)) = records::UNITS::skip_and_read(input)?;
2021-12-18 21:05:00 -08:00
Ok((input, FileHeader{
mod_time,
acc_time,
name,
2021-12-18 21:05:00 -08:00
user_units_per_db_unit: uu,
meters_per_db_unit: dbu,
}))
2021-12-18 21:05:00 -08:00
}
/// Write the header to a input
///
/// Args:
/// input: input to write to
///
/// Returns:
/// number of bytes written
///
pub fn write<W: Write>(&self, ww: &mut W) -> OResult {
2021-12-18 21:05:00 -08:00
let mut size = 0;
size += records::HEADER::write(ww, &600)?;
size += records::BGNLIB::write(ww, &[self.mod_time, self.acc_time])?;
size += records::LIBNAME::write(ww, &self.name)?;
size += records::UNITS::write(ww, &(self.user_units_per_db_unit, self.meters_per_db_unit))?;
2021-12-18 21:05:00 -08:00
Ok(size)
}
}
pub fn read_library(input: &[u8]) -> IResult<StructArray> {
let input_size = input.len();
let property_t = DataType::Struct(Fields::from(vec![
Field::new("key", DataType::Int16, false),
Field::new("value", DataType::Utf8, false),
]));
let property_list_t = DataType::List(Arc::new(
Field::new_list_field(property_t, false)
));
let repetition_struct_t = DataType::Struct(Fields::from(vec![
Field::new("x0", DataType::Int32, false),
Field::new("y0", DataType::Int32, false),
Field::new("x1", DataType::Int32, false),
Field::new("y1", DataType::Int32, false),
Field::new("count0", DataType::Int16, false),
Field::new("count1", DataType::Int16, false),
]));
let ref_struct_t = DataType::Struct(Fields::from(vec![
Field::new("target", DataType::UInt32, false),
Field::new("invert_y", DataType::Boolean, true),
Field::new("mag", DataType::Float64, true),
Field::new("angle_deg", DataType::Float64, true),
Field::new("x", DataType::Int32, false),
Field::new("y", DataType::Int32, false),
Field::new("repetition", repetition_struct_t, true),
Field::new("properties", property_list_t.clone(), true),
]));
let text_struct_t = DataType::Struct(Fields::from(vec![
Field::new("layer", DataType::UInt32, false),
Field::new("presentation_horiz", DataType::UInt8, true),
Field::new("presentation_vert", DataType::UInt8, true),
Field::new("presentation_font", DataType::UInt8, true),
Field::new("path_type", DataType::Int16, true),
Field::new("width", DataType::Int32, true),
Field::new("invert_y", DataType::Boolean, true),
Field::new("mag", DataType::Float64, true),
Field::new("angle_deg", DataType::Float64, true),
Field::new("x", DataType::Int32, false),
Field::new("y", DataType::Int32, false),
Field::new("string", DataType::Utf8, false),
Field::new("properties", property_list_t.clone(), true),
]));
let coords_t = DataType::List(Arc::new(
Field::new_list_field(DataType::Int32, false)
));
let boundary_struct_t = DataType::Struct(Fields::from(vec![
Field::new("layer", DataType::UInt32, false),
Field::new("xy", coords_t.clone(), false),
Field::new("properties", property_list_t.clone(), true),
]));
let path_struct_t = DataType::Struct(Fields::from(vec![
Field::new("layer", DataType::UInt32, false),
2025-04-12 12:06:36 -07:00
Field::new("path_type", DataType::Int16, true),
Field::new("extension_start", DataType::Int32, true),
Field::new("extension_end", DataType::Int32, true),
Field::new("width", DataType::Int32, false),
Field::new("xy", coords_t.clone(), false),
Field::new("properties", property_list_t.clone(), true),
]));
let boxnode_struct_t = DataType::Struct(Fields::from(vec![
Field::new("layer", DataType::UInt32, false),
Field::new("xy", coords_t.clone(), false),
Field::new("properties", property_list_t.clone(), true),
]));
let ref_list_t = DataType::List(Arc::new(
Field::new_list_field(ref_struct_t, false)
));
let text_list_t = DataType::List(Arc::new(
Field::new_list_field(text_struct_t, false)
));
let boundary_list_t = DataType::List(Arc::new(
Field::new_list_field(boundary_struct_t, false)
));
let path_list_t = DataType::List(Arc::new(
Field::new_list_field(path_struct_t, false)
));
let boxnode_list_t = DataType::List(Arc::new(
Field::new_list_field(boxnode_struct_t, false)
));
let name_list_t = DataType::List(Arc::new(
Field::new_list_field(DataType::Utf8, false)
));
let layer_list_t = DataType::List(Arc::new(
Field::new_list_field(DataType::UInt32, false)
));
let time_t = DataType::FixedSizeList(Arc::new(
Field::new_list_field(DataType::Int16, false),
),
6,
);
let cell_struct_t = DataType::Struct(Fields::from(vec![
Field::new("id", DataType::UInt32, false),
Field::new("file_offset", DataType::UInt64, false),
Field::new("refs", ref_list_t, false),
Field::new("boundaries", boundary_list_t, false),
Field::new("paths", path_list_t, false),
Field::new("nodes", boxnode_list_t.clone(), true),
Field::new("boxes", boxnode_list_t.clone(), true),
Field::new("texts", text_list_t, false),
]));
2025-04-12 12:06:36 -07:00
let cells_list_t = DataType::List(Arc::new(
Field::new_list_field(cell_struct_t, false)
));
let mut lib_builder = StructBuilder::from_fields(vec![
Field::new("meters_per_db_unit", DataType::Float64, false),
Field::new("user_units_per_db_unit", DataType::Float64, false),
Field::new("lib_name", DataType::Utf8, false),
Field::new("mod_time", time_t.clone(), false),
Field::new("acc_time", time_t.clone(), false),
2025-04-12 12:06:36 -07:00
Field::new("cells", cells_list_t, false),
Field::new("cell_names", name_list_t, false),
Field::new("layers", layer_list_t, false),
],
0,
);
let (input, header) = FileHeader::read(input)?;
let dbu_builder = lib_builder.field_builder::<Float64Builder>(0).unwrap();
dbu_builder.append_value(header.meters_per_db_unit);
let uu_builder = lib_builder.field_builder::<Float64Builder>(1).unwrap();
uu_builder.append_value(header.user_units_per_db_unit);
let libname_builder = lib_builder.field_builder::<StringBuilder>(2).unwrap();
libname_builder.append_value(String::from_utf8(header.name).unwrap());
2025-04-12 12:06:36 -07:00
let mtl_builder = lib_builder.field_builder::<FListBuilder>(3).unwrap();
let mt_builder = mtl_builder.values().as_any_mut().downcast_mut::<Int16Builder>().unwrap();
mt_builder.append_values(&header.mod_time, &[true; 6]);
mtl_builder.append(true);
let atl_builder = lib_builder.field_builder::<FListBuilder>(4).unwrap();
let at_builder = atl_builder.values().as_any_mut().downcast_mut::<Int16Builder>().unwrap();
at_builder.append_values(&header.acc_time, &[true; 6]);
atl_builder.append(true);
let mut names = HashMap::<String, u32>::new();
let mut layers = HashMap::<u32, u32>::new();
let cells_builder = lib_builder.field_builder::<DListBuilder>(5).unwrap();
let (mut input, mut header) = RecordHeader::read(input)?;
while header.tag != records::RTAG_ENDLIB {
(input, _) = take_bytes(input, header.data_size)?;
if header.tag == records::RTAG_BGNSTR {
let name_bytes;
(input, name_bytes) = records::STRNAME::read(input)?;
let name = String::from_utf8(name_bytes).unwrap();
2025-04-12 12:06:36 -07:00
println!("{name}");
let next_id = names.len();
let id = names.entry(name).or_insert(next_id.try_into().unwrap());
let position = input_size - input.len();
let cell_builder = cells_builder.values().as_any_mut().downcast_mut::<StructBuilder>().unwrap();
let id_builder = cell_builder.field_builder::<UInt32Builder>(0).unwrap();
id_builder.append_value(*id);
let offset_builder = cell_builder.field_builder::<UInt64Builder>(1).unwrap();
offset_builder.append_value(position.try_into().unwrap());
(input, _) = read_elements(input, cell_builder, &mut names, &mut layers)?;
2025-04-12 12:06:36 -07:00
cell_builder.append(true);
}
(input, header) = RecordHeader::read(input)?;
}
2025-04-12 12:06:36 -07:00
cells_builder.append(true);
let mut ids: HashMap<u32, String> = names.into_iter().map(|(kk, vv)| (vv, kk)).collect();
let names_builder = lib_builder.field_builder::<DListBuilder>(6).unwrap();
2025-04-12 12:06:36 -07:00
let name_builder = names_builder.values().as_any_mut().downcast_mut::<StringBuilder>().unwrap();
for id in 0..ids.len() {
2025-04-12 12:06:36 -07:00
name_builder.append_value(ids.remove(&id.try_into().unwrap()).unwrap());
}
2025-04-12 12:06:36 -07:00
names_builder.append(true);
let mut layer_ids: HashMap<u32, u32> = layers.into_iter().map(|(kk, vv)| (vv, kk)).collect();
let layers_builder = lib_builder.field_builder::<DListBuilder>(7).unwrap();
let layer_builder = layers_builder.values().as_any_mut().downcast_mut::<UInt32Builder>().unwrap();
for layer_id in 0..layer_ids.len() {
layer_builder.append_value(layer_ids.remove(&layer_id.try_into().unwrap()).unwrap());
}
layers_builder.append(true);
lib_builder.append(true);
let lib = lib_builder.finish();
Ok((input, lib))
}
/*
///
/// Scan through a GDS file, building a table of
/// {b'structure_name': byte_offset}.
/// The intent of this function is to enable random access
/// and/or partial (structure-by-structure) reads.
///
/// Args:
/// input: Seekable input to read from. Should be positioned
/// before the first structure record, but possibly
/// already past the file header.
///
pub fn scan_structs(input: &[u8]) -> IResult<HashMap::<Vec<u8>, usize>> {
2021-12-18 21:05:00 -08:00
let input_size = input.len();
let mut positions = HashMap::new();
2021-12-18 21:05:00 -08:00
let (mut input, mut header) = RecordHeader::read(input)?;
2021-12-18 21:05:00 -08:00
while header.tag != records::RTAG_ENDLIB {
(input, _) = take_bytes(input, header.data_size)?;
if header.tag == records::RTAG_BGNSTR {
let name;
(input, name) = records::STRNAME::read(input)?;
if positions.contains_key(&name) {
return fail(input, format!("Duplicate structure name: {:?}", name));
2021-12-18 21:05:00 -08:00
}
let position = input_size - input.len();
positions.insert(name, position);
}
(input, header) = RecordHeader::read(input)?;
2021-12-18 21:05:00 -08:00
}
Ok((input, positions))
2021-12-18 21:05:00 -08:00
}
*/
2021-12-18 21:05:00 -08:00
/*
#[derive(Debug, Clone)]
2021-12-18 21:05:00 -08:00
pub struct Cell {
name: Vec<u8>,
2021-12-18 21:05:00 -08:00
boundaries: Vec<elements::Boundary>,
paths: Vec<elements::Path>,
nodes: Vec<elements::Node>,
boxes: Vec<elements::GDSBox>,
texts: Vec<elements::Text>,
refs: Vec<elements::Reference>,
2021-12-18 21:05:00 -08:00
}
impl Cell {
/// Build an empty cell
pub fn new(name: Vec<u8>) -> Self {
Cell{
name: name,
boundaries: Vec::new(),
paths: Vec::new(),
nodes: Vec::new(),
boxes: Vec::new(),
texts: Vec::new(),
refs: Vec::new(),
}
}
/// Skip to the next structure and attempt to read it.
///
/// Args:
/// input: Seekable input to read from.
///
/// Returns:
/// (name, elements) if a structure was found.
/// None if no structure was found before the end of the library.
///
pub fn read(input: &[u8]) -> IResult<Option<Cell>> {
let (input, success) = records::BGNSTR::skip_past(input)?;
if !success {
return Ok((input, None))
}
let (input, name) = records::STRNAME::read(input)?;
let mut cell = Cell::new(name);
let (input, _) = cell.read_elements(input)?;
Ok((input, Some(cell)))
}
/// Read elements from the input until an ENDSTR
/// record is encountered. The ENDSTR record is also
/// consumed.
///
/// Args:
/// input: Seekable input to read from.
///
/// Returns:
/// List of element objects.
///
pub fn read_elements<'a>(&mut self, input: &'a [u8]) -> IResult<'a, ()> {
let (mut input, mut header) = RecordHeader::read(input)?;
while header.tag != records::RTAG_ENDSTR {
match header.tag {
records::RTAG_BOUNDARY => {
let boundary;
(input, _) = records::BOUNDARY::read(input)?;
(input, boundary) = elements::Boundary::read(input)?;
self.boundaries.push(boundary);
},
records::RTAG_PATH => {
let path;
(input, _) = records::PATH::read(input)?;
(input, path) = elements::Path::read(input)?;
self.paths.push(path);
},
records::RTAG_NODE => {
let node;
(input, _) = records::NODE::read(input)?;
(input, node) = elements::Node::read(input)?;
self.nodes.push(node);
},
records::RTAG_BOX => {
let gds_box;
(input, _) = records::BOX::read(input)?;
(input, gds_box) = elements::GDSBox::read(input)?;
self.boxes.push(gds_box);
},
records::RTAG_TEXT => {
let txt;
(input, _) = records::TEXT::read(input)?;
(input, txt) = elements::Text::read(input)?;
self.texts.push(txt);
},
records::RTAG_SREF => {
let sref;
(input, _) = records::SREF::read(input)?;
(input, sref) = elements::Reference::read(input)?;
self.refs.push(sref);
},
records::RTAG_AREF => {
let aref;
(input, _) = records::AREF::read(input)?;
(input, aref) = elements::Reference::read(input)?;
self.refs.push(aref);
},
_ => {
// don't care, skip
(input, _) = take_bytes(input, header.data_size)?;
}
}
(input, header) = RecordHeader::read(input)?;
}
Ok((input, ()))
}
///
/// Write a structure to the provided input.
///
/// Args:
/// name: Structure name (ascii-encoded).
/// elements: List of Elements containing the geometry and text in this struct.
/// cre_time: Creation time (optional).
/// mod_time: Modification time (optional).
///
/// Return:
/// Number of bytes written
///
2021-12-18 21:05:00 -08:00
pub fn write<W: Write>(
&self,
ww: &mut W,
2021-12-18 21:05:00 -08:00
cre_time: Option<[i16; 6]>,
mod_time: Option<[i16; 6]>,
) -> OResult {
2021-12-18 21:05:00 -08:00
let mut size = 0;
size += records::BGNSTR::write(ww, &[cre_time.unwrap_or(DEFAULT_DATE),
mod_time.unwrap_or(DEFAULT_DATE)])?;
size += records::STRNAME::write(ww, &self.name)?;
size += self.write_elements(ww)?;
size += records::ENDSTR::write(ww, &())?;
2021-12-18 21:05:00 -08:00
Ok(size)
}
pub fn write_elements<W: Write>(&self, ww: &mut W) -> OResult {
let mut size = 0;
for boundary in &self.boundaries {
size += boundary.write(ww)?;
}
for path in &self.paths {
size += path.write(ww)?;
}
for node in &self.nodes {
size += node.write(ww)?;
}
for gds_box in &self.boxes {
size += gds_box.write(ww)?;
}
for text in &self.texts {
size += text.write(ww)?;
}
for reference in &self.refs {
size += reference.write(ww)?;
2021-12-18 21:05:00 -08:00
}
Ok(size)
2021-12-18 21:05:00 -08:00
}
}
*/
2021-12-18 21:05:00 -08:00
2022-05-08 16:41:43 -07:00
/*
///
/// Scan through a GDS file, building a table of instance counts
/// `{b'structure_name': {b'ref_name': count}}`.
///
/// This is intended to provide a fast overview of the file's
/// contents without performing a full read of all elements.
///
/// Args:
/// input: Seekable input to read from. Should be positioned
/// before the first structure record, but possibly
/// already past the file header.
///
pub fn scan_hierarchy(input: &[u8]) -> IResult<HashMap::<Vec<u8>, HashMap::<Vec<u8>, u32>>> {
let mut structures = HashMap::new();
2021-12-18 21:05:00 -08:00
2022-05-08 16:41:43 -07:00
let mut ref_name = None;
let mut ref_count = None;
let (mut input, mut header) = RecordHeader::read(input)?;
2022-05-08 16:41:43 -07:00
let mut cur_structure = HashMap::new();
2021-12-18 21:05:00 -08:00
while header.tag != records::RTAG_ENDLIB {
match header.tag {
records::RTAG_BGNSTR => {
(input, _) = take_bytes(input, header.data_size)?;
let result = records::STRNAME::read(input)?;
input = result.0;
let name = result.1;
if structures.contains_key(&name) {
return fail(input, format!("Duplicate structure name: {:?}", name));
2021-12-18 21:05:00 -08:00
}
let mut cur_structure = HashMap::new();
2021-12-18 21:05:00 -08:00
structures.insert(name, cur_structure);
ref_name = None;
ref_count = None;
},
records::RTAG_SNAME => {
let result = records::SNAME::read_data(input, header.data_size)?;
input = result.0;
ref_name = Some(result.1);
2021-12-18 21:05:00 -08:00
},
records::RTAG_COLROW => {
let result = records::COLROW::read_data(input, header.data_size)?;
input = result.0;
let (col, row) = (result.1[0], result.1[1]);
ref_count = Some((col * row) as u32);
2021-12-18 21:05:00 -08:00
},
records::RTAG_ENDEL => {
*cur_structure.entry(ref_name.unwrap()).or_insert(0) += ref_count.unwrap_or(1);
},
_ => {
(input, _) = take_bytes(input, header.data_size)?;
2021-12-18 21:05:00 -08:00
},
}
(input, header) = RecordHeader::read(input)?;
}
Ok((input, structures))
}
2022-05-08 16:41:43 -07:00
*/
2022-05-08 16:41:43 -07:00
/*
pub fn count_ref(input: &[u8]) -> IResult<Option((Vec<u8>, u32))> {
let (input, found_struc) = records::BGNSTR.skip_past(input)?;
2022-05-08 16:41:43 -07:00
if !found_struc {
return Ok((input, None))
}
let mut cur_structure = HashMap::new();
let (input, name) = records::STRNAME::read(input)?;
if structures.contains_key(&name) {
return fail(input, format!("Duplicate structure name: {:?}", name));
}
let (mut input, mut header) = RecordHeader::read(input)?;
while header.tag != records::RTAG_ENDSTR {
let mut ref_name = None;
let mut ref_count = None;
while header.tag != records::RTAG_ENDEL {
match header.tag {
records::RTAG_SNAME => {
let result = records::SNAME::read_data(input1, header.data_size)?;
input1 = result.0;
ref_name = Some(result.1);
},
records::RTAG_COLROW => {
let result = records::COLROW::read_data(input1, header.data_size)?;
input1 = result.0;
let (col, row) = (result.1[0], result.1[1]);
ref_count = Some((col * row) as u32);
},
_ => {
(input1, _) = take_bytes(input1, header.data_size)?;
},
}
(input1, header) = RecordHeader::read(input1)?;
}
// got ENDEL, update count for this reference
*cur_structure.entry(ref_name.unwrap()).or_insert(0) += ref_count.unwrap_or(1);
(input1, header) = RecordHeader::read(input1)?;
2021-12-18 21:05:00 -08:00
}
structures.insert(name, cur_structure);
(input, header) = RecordHeader::read(input1)?;
2021-12-18 21:05:00 -08:00
}
2022-05-08 16:41:43 -07:00
*/