organic/src/compare/compare_field.rs

632 lines
22 KiB
Rust
Raw Normal View History

2023-10-10 00:05:34 -04:00
use std::collections::BTreeSet;
use std::fmt::Debug;
2023-10-09 21:04:41 -04:00
use std::str::FromStr;
use super::diff::artificial_diff_scope;
2023-10-11 14:03:42 -04:00
use super::diff::artificial_owned_diff_scope;
use super::diff::compare_ast_node;
use super::diff::DiffEntry;
use super::diff::DiffStatus;
2023-10-08 14:40:01 -04:00
use super::sexp::unquote;
use super::sexp::Token;
use super::util::get_property;
2023-10-09 21:04:41 -04:00
use super::util::get_property_numeric;
use super::util::get_property_quoted_string;
use super::util::get_property_unquoted_atom;
use crate::types::AstNode;
2023-10-10 00:43:02 -04:00
use crate::types::CharOffsetInLine;
use crate::types::LineNumber;
2023-10-11 13:55:43 -04:00
use crate::types::Object;
2023-10-10 00:43:02 -04:00
use crate::types::RetainLabels;
use crate::types::SwitchNumberLines;
#[derive(Debug)]
pub(crate) enum EmacsField<'s> {
Required(&'s str),
Optional(&'s str),
}
#[derive(Debug)]
pub(crate) enum ComparePropertiesResult<'b, 's> {
NoChange,
/// Return when you want the status for "this" node to change (as opposed to collecting child status).
SelfChange(DiffStatus, Option<String>),
DiffEntry(DiffEntry<'b, 's>),
}
impl<'b, 's> ComparePropertiesResult<'b, 's> {
pub(crate) fn apply(
self,
child_status: &mut Vec<DiffEntry<'b, 's>>,
this_status: &mut DiffStatus,
message: &mut Option<String>,
) {
match self {
ComparePropertiesResult::NoChange => {}
ComparePropertiesResult::SelfChange(new_status, new_message) => {
*this_status = new_status;
*message = new_message
}
ComparePropertiesResult::DiffEntry(diff_entry) => child_status.push(diff_entry),
}
// foo
}
}
/// Do no comparison.
///
/// This is for when you want to acknowledge that a field exists in the emacs token, but you do not have any validation for it when using the compare_properties!() macro. Ideally, this should be kept to a minimum since this represents untested values.
2023-10-06 16:32:49 -04:00
pub(crate) fn compare_noop<'b, 's, 'x, R, RG>(
_source: &'s str,
_emacs: &'b Token<'s>,
_rust_node: R,
_emacs_field: &'x str,
_rust_value_getter: RG,
) -> Result<ComparePropertiesResult<'b, 's>, Box<dyn std::error::Error>> {
Ok(ComparePropertiesResult::NoChange)
}
/// Do no comparison.
///
/// This is for when you want to acknowledge that a field exists in the emacs token, but you do not have any validation for it when using the compare_properties!() macro. Ideally, this should be kept to a minimum since this represents untested values.
pub(crate) fn compare_identity() -> () {
()
}
/// Assert that the emacs value is always nil or absent.
///
/// This is usually used for fields which, in my testing, are always nil. Using this compare function instead of simply doing a compare_noop will enable us to be alerted when we finally come across an org-mode document that has a value other than nil for the property.
pub(crate) fn compare_property_always_nil<'b, 's, 'x, R, RG>(
_source: &'s str,
emacs: &'b Token<'s>,
_rust_node: R,
emacs_field: &'x str,
_rust_value_getter: RG,
) -> Result<ComparePropertiesResult<'b, 's>, Box<dyn std::error::Error>> {
let value = get_property(emacs, emacs_field)?;
if value.is_some() {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} was expected to always be nil: {:?}",
emacs_field, value
));
Ok(ComparePropertiesResult::SelfChange(this_status, message))
} else {
Ok(ComparePropertiesResult::NoChange)
}
}
pub(crate) fn compare_property_quoted_string<
'b,
's,
'x,
R,
RV: AsRef<str> + std::fmt::Debug,
RG: Fn(R) -> Option<RV>,
>(
_source: &'s str,
emacs: &'b Token<'s>,
rust_node: R,
2023-10-06 16:03:41 -04:00
emacs_field: &'x str,
rust_value_getter: RG,
) -> Result<ComparePropertiesResult<'b, 's>, Box<dyn std::error::Error>> {
let value = get_property_quoted_string(emacs, emacs_field)?;
let rust_value = rust_value_getter(rust_node);
if rust_value.as_ref().map(|s| s.as_ref()) != value.as_ref().map(String::as_str) {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, value, rust_value
));
Ok(ComparePropertiesResult::SelfChange(this_status, message))
} else {
Ok(ComparePropertiesResult::NoChange)
}
}
pub(crate) fn compare_property_unquoted_atom<'b, 's, 'x, R, RG: Fn(R) -> Option<&'s str>>(
_source: &'s str,
emacs: &'b Token<'s>,
rust_node: R,
emacs_field: &'x str,
rust_value_getter: RG,
) -> Result<ComparePropertiesResult<'b, 's>, Box<dyn std::error::Error>> {
let value = get_property_unquoted_atom(emacs, emacs_field)?;
let rust_value = rust_value_getter(rust_node);
2023-10-09 21:04:41 -04:00
if rust_value != value {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, value, rust_value
));
Ok(ComparePropertiesResult::SelfChange(this_status, message))
} else {
Ok(ComparePropertiesResult::NoChange)
}
}
pub(crate) fn compare_property_numeric<
'b,
's,
'x,
R,
RV: FromStr + PartialEq + Debug,
RG: Fn(R) -> Option<RV>,
>(
_source: &'s str,
emacs: &'b Token<'s>,
rust_node: R,
emacs_field: &'x str,
rust_value_getter: RG,
) -> Result<ComparePropertiesResult<'b, 's>, Box<dyn std::error::Error + 's>>
where
<RV as FromStr>::Err: std::error::Error,
<RV as FromStr>::Err: 's,
{
let value = get_property_numeric::<RV>(emacs, emacs_field)?;
let rust_value = rust_value_getter(rust_node);
if rust_value != value {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, value, rust_value
));
Ok(ComparePropertiesResult::SelfChange(this_status, message))
} else {
Ok(ComparePropertiesResult::NoChange)
}
}
2023-10-08 14:40:01 -04:00
pub(crate) fn compare_property_list_of_quoted_string<
'b,
's,
'x,
R,
2023-10-08 15:08:21 -04:00
RV: AsRef<str> + std::fmt::Debug,
RI: Iterator<Item = RV>,
RG: Fn(R) -> Option<RI>,
2023-10-08 14:40:01 -04:00
>(
_source: &'s str,
2023-10-08 14:40:01 -04:00
emacs: &'b Token<'s>,
rust_node: R,
emacs_field: &'x str,
rust_value_getter: RG,
) -> Result<ComparePropertiesResult<'b, 's>, Box<dyn std::error::Error>> {
2023-10-08 14:40:01 -04:00
let value = get_property(emacs, emacs_field)?
.map(Token::as_list)
.map_or(Ok(None), |r| r.map(Some))?;
let rust_value = rust_value_getter(rust_node);
2023-10-08 15:08:21 -04:00
// TODO: Seems we are needlessly coverting to a vec here.
let rust_value: Option<Vec<RV>> = rust_value.map(|it| it.collect());
match (value, &rust_value) {
2023-10-08 14:40:01 -04:00
(None, None) => {}
(None, Some(_)) | (Some(_), None) => {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, value, rust_value
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
2023-10-08 14:40:01 -04:00
}
(Some(el), Some(rl)) if el.len() != rl.len() => {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, value, rust_value
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
2023-10-08 14:40:01 -04:00
}
(Some(el), Some(rl)) => {
for (e, r) in el.iter().zip(rl) {
let e = unquote(e.as_atom()?)?;
let r = r.as_ref();
if e != r {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}. Full list: {:?} != {:?}",
emacs_field, e, r, value, rust_value
2023-10-08 14:40:01 -04:00
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
2023-10-08 14:40:01 -04:00
}
}
}
}
Ok(ComparePropertiesResult::NoChange)
2023-10-08 14:40:01 -04:00
}
2023-10-08 18:14:48 -04:00
2023-10-10 00:05:34 -04:00
pub(crate) fn compare_property_set_of_quoted_string<
'a,
'b,
's,
'x,
R,
RV: AsRef<str> + std::fmt::Debug + Ord + 'a + ?Sized,
RI: Iterator<Item = &'a RV>,
RG: Fn(R) -> Option<RI>,
>(
_source: &'s str,
emacs: &'b Token<'s>,
rust_node: R,
emacs_field: &'x str,
rust_value_getter: RG,
) -> Result<ComparePropertiesResult<'b, 's>, Box<dyn std::error::Error>> {
let value = get_property(emacs, emacs_field)?
.map(Token::as_list)
.map_or(Ok(None), |r| r.map(Some))?;
let empty = Vec::new();
let value = value.unwrap_or(&empty);
let rust_value = rust_value_getter(rust_node);
let rust_value = if let Some(rust_value) = rust_value {
let slices: BTreeSet<&str> = rust_value.map(|rv| rv.as_ref()).collect();
slices
} else {
BTreeSet::new()
};
let value: Vec<&str> = value
.iter()
.map(|e| e.as_atom())
.collect::<Result<Vec<_>, _>>()?;
let value: Vec<String> = value
.into_iter()
.map(unquote)
.collect::<Result<Vec<_>, _>>()?;
let value: BTreeSet<&str> = value.iter().map(|e| e.as_str()).collect();
let mismatched: Vec<_> = value
.symmetric_difference(&rust_value)
.map(|val| *val)
.collect();
if !mismatched.is_empty() {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch. Mismatched values: {}",
emacs_field,
mismatched.join(", ")
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
}
Ok(ComparePropertiesResult::NoChange)
}
2023-10-08 18:14:48 -04:00
pub(crate) fn compare_property_boolean<'b, 's, 'x, R, RG: Fn(R) -> bool>(
_source: &'s str,
2023-10-08 18:14:48 -04:00
emacs: &'b Token<'s>,
rust_node: R,
emacs_field: &'x str,
rust_value_getter: RG,
) -> Result<ComparePropertiesResult<'b, 's>, Box<dyn std::error::Error>> {
2023-10-08 18:14:48 -04:00
// get_property already converts nil to None.
let value = get_property(emacs, emacs_field)?.is_some();
let rust_value = rust_value_getter(rust_node);
if rust_value != value {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, value, rust_value
));
Ok(ComparePropertiesResult::SelfChange(this_status, message))
2023-10-08 18:14:48 -04:00
} else {
Ok(ComparePropertiesResult::NoChange)
}
}
pub(crate) fn compare_property_single_ast_node<
'b,
's,
'x: 'b + 's,
R,
RV: std::fmt::Debug,
RG: Fn(R) -> Option<RV>,
>(
source: &'s str,
emacs: &'b Token<'s>,
rust_node: R,
emacs_field: &'x str,
rust_value_getter: RG,
) -> Result<ComparePropertiesResult<'b, 's>, Box<dyn std::error::Error>>
where
AstNode<'b, 's>: From<RV>,
{
let value = get_property(emacs, emacs_field)?;
let rust_value = rust_value_getter(rust_node);
match (value, rust_value) {
(None, None) => {}
(None, rv @ Some(_)) | (Some(_), rv @ None) => {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, value, rv
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
}
(Some(ev), Some(rv)) => {
let child_status: Vec<DiffEntry<'b, 's>> =
vec![compare_ast_node(source, ev, rv.into())?];
let diff_scope = artificial_diff_scope(emacs_field, child_status)?;
return Ok(ComparePropertiesResult::DiffEntry(diff_scope));
}
}
Ok(ComparePropertiesResult::NoChange)
}
pub(crate) fn compare_property_list_of_ast_nodes<
'b,
's,
'x: 'b + 's,
R,
RV: std::fmt::Debug,
RI: Iterator<Item = RV>,
RG: Fn(R) -> Option<RI>,
>(
source: &'s str,
emacs: &'b Token<'s>,
rust_node: R,
emacs_field: &'x str,
rust_value_getter: RG,
) -> Result<ComparePropertiesResult<'b, 's>, Box<dyn std::error::Error>>
where
AstNode<'b, 's>: From<RV>,
{
let value = get_property(emacs, emacs_field)?
.map(Token::as_list)
.map_or(Ok(None), |r| r.map(Some))?;
let rust_value = rust_value_getter(rust_node);
// TODO: Seems we are needlessly coverting to a vec here.
let rust_value: Option<Vec<RV>> = rust_value.map(|it| it.collect());
match (value, rust_value) {
(None, None) => {}
(Some(el), None)
if el.len() == 1
&& el.into_iter().all(|t| {
if let Ok(r#""""#) = t.as_atom() {
true
} else {
false
}
}) => {}
(None, rv @ Some(_)) | (Some(_), rv @ None) => {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, value, rv
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
}
(Some(el), Some(rl)) if el.len() != rl.len() => {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, el, rl
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
}
(Some(el), Some(rl)) => {
let mut child_status: Vec<DiffEntry<'b, 's>> = Vec::with_capacity(rl.len());
for (e, r) in el.iter().zip(rl) {
child_status.push(compare_ast_node(source, e, r.into())?);
}
let diff_scope = artificial_diff_scope(emacs_field, child_status)?;
return Ok(ComparePropertiesResult::DiffEntry(diff_scope));
}
2023-10-08 18:14:48 -04:00
}
Ok(ComparePropertiesResult::NoChange)
2023-10-08 18:14:48 -04:00
}
2023-10-10 00:43:02 -04:00
2023-10-11 13:55:43 -04:00
/// Special compare used for affiliate keywords that are parsed as objects.
///
/// Org-mode seems to store these as a 3-deep list:
/// - Outer list with 1 element per #+caption keyword (or other parsed keyword).
2023-10-11 17:53:32 -04:00
/// - Middle list which has:
/// - first element is a list of objects representing the value after the colon.
/// - every additional element is a list of objects from inside the square brackets (the optional value).
2023-10-11 13:55:43 -04:00
pub(crate) fn compare_property_list_of_list_of_list_of_ast_nodes<
'b,
's,
2023-10-11 14:03:42 -04:00
'x,
2023-10-11 13:55:43 -04:00
R,
2023-10-11 18:44:21 -04:00
RG: Fn(R) -> Option<&'b Vec<(Option<Vec<Object<'s>>>, Vec<Object<'s>>)>>,
2023-10-11 13:55:43 -04:00
>(
source: &'s str,
emacs: &'b Token<'s>,
rust_node: R,
emacs_field: &'x str,
rust_value_getter: RG,
) -> Result<ComparePropertiesResult<'b, 's>, Box<dyn std::error::Error>> {
// TODO: Replace Object<'s> with generics. I hard-coded Object in to make lifetimes easier.
2023-10-11 13:55:43 -04:00
let rust_value = rust_value_getter(rust_node);
let value = get_property(emacs, emacs_field)?
.map(Token::as_list)
.map_or(Ok(None), |r| r.map(Some))?;
let (value, rust_value) = match (value, rust_value) {
(None, None) => {
return Ok(ComparePropertiesResult::NoChange);
}
(None, Some(_)) | (Some(_), None) => {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, value, rust_value
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
}
(Some(value), Some(rust_value)) if value.len() != rust_value.len() => {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, value, rust_value
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
}
(Some(value), Some(rust_value)) => (value, rust_value),
};
2023-10-11 18:59:13 -04:00
let mut child_status: Vec<DiffEntry<'b, 's>> = Vec::with_capacity(rust_value.len());
2023-10-11 13:55:43 -04:00
// Iterate the outer lists
2023-10-11 18:44:21 -04:00
for (value, (rust_optional, rust_value)) in value.iter().zip(rust_value.iter()) {
let mut middle_value = value.as_list()?.iter();
// First element of middle list is the mandatory value (the value past the colon).
let mandatory_value = middle_value.next();
let mandatory_value = match mandatory_value {
Some(mandatory_value) => mandatory_value,
None => {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, value, rust_value
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
}
};
// Compare optional value
if let Some(rust_optional) = rust_optional {
if rust_optional.len() != middle_value.len() {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, value, rust_value
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
}
for (e, r) in middle_value.zip(rust_optional) {
2023-10-11 18:59:13 -04:00
eprintln!("Comparing optval! {:?} {:?}", e, r);
2023-10-11 18:44:21 -04:00
child_status.push(compare_ast_node(source, e, r.into())?);
}
}
// Compare mandatory value
let mandatory_value = mandatory_value.as_list()?;
if rust_value.len() != mandatory_value.len() {
2023-10-11 13:55:43 -04:00
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, value, rust_value
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
}
2023-10-11 18:44:21 -04:00
for (e, r) in mandatory_value.iter().zip(rust_value) {
2023-10-11 18:59:13 -04:00
eprintln!("Comparing mandatory! {:?} {:?}", e, r);
2023-10-11 13:55:43 -04:00
child_status.push(compare_ast_node(source, e, r.into())?);
}
2023-10-11 18:59:13 -04:00
}
if child_status.is_empty() {
Ok(ComparePropertiesResult::NoChange)
} else {
2023-10-11 14:03:42 -04:00
let diff_scope = artificial_owned_diff_scope(emacs_field, child_status)?;
2023-10-11 18:59:13 -04:00
Ok(ComparePropertiesResult::DiffEntry(diff_scope))
2023-10-11 13:55:43 -04:00
}
}
2023-10-10 00:43:02 -04:00
pub(crate) fn compare_property_number_lines<
'b,
's,
'x,
'y,
R,
RG: Fn(R) -> Option<&'y SwitchNumberLines>,
>(
_source: &'s str,
emacs: &'b Token<'s>,
rust_node: R,
emacs_field: &'x str,
rust_value_getter: RG,
) -> Result<ComparePropertiesResult<'b, 's>, Box<dyn std::error::Error>> {
let number_lines = get_property(emacs, emacs_field)?;
let rust_value = rust_value_getter(rust_node);
match (number_lines, &rust_value) {
(None, None) => {}
(Some(number_lines), Some(rust_number_lines)) => {
let token_list = number_lines.as_list()?;
let number_type = token_list
.get(0)
.map(Token::as_atom)
.map_or(Ok(None), |r| r.map(Some))?
.ok_or(":number-lines should have a type.")?;
let number_value = token_list
.get(2)
.map(Token::as_atom)
.map_or(Ok(None), |r| r.map(Some))?
.map(|val| val.parse::<LineNumber>())
.map_or(Ok(None), |r| r.map(Some))?
.ok_or(":number-lines should have a value.")?;
match (number_type, number_value, rust_number_lines) {
("new", emacs_val, SwitchNumberLines::New(rust_val)) if emacs_val == *rust_val => {}
("continued", emacs_val, SwitchNumberLines::Continued(rust_val))
if emacs_val == *rust_val => {}
_ => {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, number_lines, rust_value
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
}
}
}
_ => {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, number_lines, rust_value
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
}
};
Ok(ComparePropertiesResult::NoChange)
}
pub(crate) fn compare_property_retain_labels<'b, 's, 'x, 'y, R, RG: Fn(R) -> &'y RetainLabels>(
_source: &'s str,
emacs: &'b Token<'s>,
rust_node: R,
emacs_field: &'x str,
rust_value_getter: RG,
) -> Result<ComparePropertiesResult<'b, 's>, Box<dyn std::error::Error + 's>> {
let rust_value = rust_value_getter(rust_node);
let retain_labels = get_property_unquoted_atom(emacs, ":retain-labels")?;
if let Some(retain_labels) = retain_labels {
if retain_labels == "t" {
match rust_value {
RetainLabels::Yes => {}
_ => {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, retain_labels, rust_value
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
}
}
} else {
let retain_labels: CharOffsetInLine = get_property_numeric(emacs, ":retain-labels")?.expect("Cannot be None or else the earlier get_property_unquoted_atom would have been None.");
match (retain_labels, rust_value) {
(e, RetainLabels::Keep(r)) if e == *r => {}
_ => {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, retain_labels, rust_value
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
}
}
}
} else {
match rust_value {
RetainLabels::No => {}
_ => {
let this_status = DiffStatus::Bad;
let message = Some(format!(
"{} mismatch (emacs != rust) {:?} != {:?}",
emacs_field, retain_labels, rust_value
));
return Ok(ComparePropertiesResult::SelfChange(this_status, message));
}
}
}
Ok(ComparePropertiesResult::NoChange)
}