fhir-diff 0.1.0

//! The effective element tree of a profile: its FHIR Schema merged over every schema it is
//! based on, down to the core type. Two versions of a profile are compared on these trees, so
//! a change inherited from a base profile counts as much as one the profile makes itself.

use std::collections::{BTreeMap, BTreeSet, HashMap};

use octofhir_fhirschema::types::{FhirSchemaConstraint, FhirSchemaSliceMatch, FhirSchemaSlicing};
use octofhir_fhirschema::{FhirSchema, FhirSchemaElement, FhirVersion, StructureDefinition};
use serde::Serialize;
use serde_json::Value;

use crate::package::{Definition, Loaded};

/// Strength of a terminology binding, weakest first.
#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Serialize)]
#[serde(rename_all = "lowercase")]
pub enum Strength {
    Example,
    Preferred,
    Extensible,
    Required,
}

impl Strength {
    fn parse(text: &str) -> Self {
        match text {
            "required" => Self::Required,
            "extensible" => Self::Extensible,
            "preferred" => Self::Preferred,
            _ => Self::Example,
        }
    }

    pub fn as_str(self) -> &'static str {
        match self {
            Self::Example => "example",
            Self::Preferred => "preferred",
            Self::Extensible => "extensible",
            Self::Required => "required",
        }
    }
}

#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
pub struct Binding {
    pub strength: Strength,
    pub value_set: Option<String>,
}

#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
pub struct Constraint {
    pub severity: String,
    pub expression: String,
    pub human: String,
}

#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
pub struct Slicing {
    pub rules: String,
    pub ordered: bool,
    pub discriminators: Vec<String>,
}

/// One element of the effective tree.
#[derive(Debug, Clone, PartialEq, Serialize)]
pub struct Node {
    pub min: u32,
    /// `None` is `*`.
    pub max: Option<u32>,
    /// The element's type, or for a choice element the types it may take.
    pub types: BTreeSet<String>,
    /// Profiles a Reference (or canonical) may point to.
    pub targets: BTreeSet<String>,
    pub must_support: bool,
    pub is_modifier: bool,
    pub binding: Option<Binding>,
    pub constraints: BTreeMap<String, Constraint>,
    /// `fixed[x]` / `pattern[x]` value.
    pub pattern: Option<Value>,
    pub slicing: Option<Slicing>,
    pub slices: BTreeMap<String, Node>,
    pub children: BTreeMap<String, Node>,
    /// Canonical of the extension an extension slice holds.
    pub extension_url: Option<String>,
}

impl Node {
    /// An element nothing constrains: what a missing slice or child is compared with.
    pub fn unconstrained() -> Self {
        Self {
            min: 0,
            max: None,
            types: BTreeSet::new(),
            targets: BTreeSet::new(),
            must_support: false,
            is_modifier: false,
            binding: None,
            constraints: BTreeMap::new(),
            pattern: None,
            slicing: None,
            slices: BTreeMap::new(),
            children: BTreeMap::new(),
            extension_url: None,
        }
    }
}

/// A profile (or extension definition) with its effective tree.
#[derive(Debug, Clone, Serialize)]
pub struct Profile {
    pub url: String,
    pub version: Option<String>,
    pub name: String,
    pub title: Option<String>,
    /// The resource or data type it constrains (`Patient`, `Extension`).
    pub type_name: String,
    pub base: Option<String>,
    pub kind: String,
    /// File in the package.
    pub file: String,
    #[serde(skip)]
    pub tree: Node,
}

/// Every schema a package's profiles may be based on: the package's own, its dependencies'
/// and the core release's.
pub struct Pool {
    by_url: HashMap<String, FhirSchema>,
    core: &'static HashMap<String, FhirSchema>,
    /// The package's own constraint definitions, in file order.
    own: Vec<(String, FhirSchema, Definition)>,
    /// Definitions that did not convert, with the reason.
    pub skipped: Vec<(String, String)>,
    /// Bases that were not found (compared over the core type instead).
    pub missing: std::cell::RefCell<BTreeSet<String>>,
    /// Bases found only among the other package's dependencies (see [`Pool::borrow_bases`]).
    pub borrowed: BTreeSet<String>,
}

const CORE_PREFIX: &str = "http://hl7.org/fhir/StructureDefinition/";

/// The core release of a package's `fhirVersions`.
pub fn release(fhir_versions: &[String]) -> FhirVersion {
    let version = fhir_versions.first().map(String::as_str).unwrap_or("4.0.1");
    if version.starts_with("4.3") {
        FhirVersion::R4B
    } else if version.starts_with("5.") {
        FhirVersion::R5
    } else if version.starts_with("6.") {
        FhirVersion::R6
    } else {
        FhirVersion::R4
    }
}

impl Pool {
    pub fn new(loaded: &Loaded) -> Self {
        let core = octofhir_fhirschema::get_schemas(release(&loaded.package.fhir_versions));
        let mut pool = Self {
            by_url: HashMap::new(),
            core,
            own: Vec::new(),
            skipped: Vec::new(),
            missing: Default::default(),
            borrowed: BTreeSet::new(),
        };
        for package in &loaded.dependencies {
            for definition in &package.definitions {
                if let Ok(schema) = convert(&definition.json) {
                    pool.by_url.insert(schema.url.clone(), schema);
                }
            }
        }
        for definition in &loaded.package.definitions {
            match convert(&definition.json) {
                Ok(schema) => {
                    pool.by_url.insert(schema.url.clone(), schema.clone());
                    if definition.json["derivation"] == "constraint" {
                        pool.own
                            .push((schema.url.clone(), schema, definition.clone()));
                    }
                }
                Err(error) => pool.skipped.push((definition.file.clone(), error)),
            }
        }
        pool
    }

    /// The pools of two packages compared with each other, each lent the bases it lacks by the
    /// other ([`Pool::borrow_bases`]).
    pub fn pair(old: &Loaded, new: &Loaded) -> (Self, Self) {
        let (mut old_pool, mut new_pool) = (Self::new(old), Self::new(new));
        old_pool.borrow_bases(&new_pool);
        new_pool.borrow_bases(&old_pool);
        (old_pool, new_pool)
    }

    /// Take the bases this package's profiles need but its manifest does not provide from
    /// `other`, the package compared with. US Core 6.1.0 bases a profile on SDC without listing
    /// SDC; comparing that profile over the core type would report SDC's own constraints as
    /// changes. A base with the same canonical URL from the other side's dependencies isolates
    /// the package's own changes.
    pub fn borrow_bases(&mut self, other: &Pool) {
        let mut pending: Vec<String> = self
            .own
            .iter()
            .filter_map(|(_, schema, _)| schema.base.clone())
            .collect();
        let mut seen = BTreeSet::new();
        while let Some(base) = pending.pop() {
            let plain = base.split('|').next().unwrap_or(&base).to_owned();
            if !seen.insert(plain.clone()) {
                continue;
            }
            let schema = match self.schema(&plain) {
                Some(schema) => schema.clone(),
                None => match other.by_url.get(&plain) {
                    Some(schema) => {
                        self.by_url.insert(plain.clone(), schema.clone());
                        self.borrowed.insert(plain);
                        schema.clone()
                    }
                    None => continue,
                },
            };
            pending.extend(schema.base);
        }
    }

    fn schema(&self, url: &str) -> Option<&FhirSchema> {
        let plain = url.split('|').next().unwrap_or(url);
        self.by_url.get(plain).or_else(|| {
            plain
                .strip_prefix(CORE_PREFIX)
                .and_then(|name| self.core.get(name))
        })
    }

    /// The package's profiles and extension definitions with their effective trees.
    pub fn profiles(&self) -> Vec<Profile> {
        self.own
            .iter()
            .map(|(url, schema, definition)| Profile {
                url: url.clone(),
                version: schema.version.clone(),
                name: schema.name.clone(),
                title: definition.json["title"].as_str().map(str::to_owned),
                type_name: schema.type_name.clone(),
                base: schema.base.clone(),
                kind: schema.kind.clone(),
                file: definition.file.clone(),
                tree: self.effective(schema),
            })
            .collect()
    }

    /// The type a profile constrains (`Patient` for US Core Patient), or the type itself for a
    /// core URL.
    pub fn type_of(&self, url: &str) -> Option<String> {
        let plain = url.split('|').next().unwrap_or(url);
        match self.by_url.get(plain) {
            Some(schema) => Some(schema.type_name.clone()),
            None => plain.strip_prefix(CORE_PREFIX).map(str::to_owned),
        }
    }

    /// The core definition of element `child` of type `type_name`: what a child that only one
    /// version constrains is compared with.
    pub fn base_node(&self, type_name: &str, child: &str) -> Option<Node> {
        let schema = self.core.get(type_name)?;
        let parent = level(schema);
        let element = parent.elements.as_ref()?.get(child)?;
        Some(node(self.core, element, Some(&parent), child))
    }

    /// `schema` merged over its base chain, as one tree rooted at the constrained type.
    pub fn effective(&self, schema: &FhirSchema) -> Node {
        let mut chain = vec![level(schema)];
        let mut seen = BTreeSet::from([schema.url.clone()]);
        let mut current = schema.base.clone();
        while let Some(base) = current {
            if !seen.insert(base.clone()) {
                break;
            }
            let Some(base_schema) = self.schema(&base) else {
                // A base from a package the manifest does not list: carry on from the core
                // type, so the comparison still sees the core definitions.
                if let Some(core) = self.core.get(&schema.type_name)
                    && seen.insert(core.url.clone())
                {
                    self.missing.borrow_mut().insert(base.clone());
                    chain.push(level(core));
                }
                break;
            };
            chain.push(level(base_schema));
            current = base_schema.base.clone();
        }
        let merged = chain
            .into_iter()
            .rev()
            .reduce(|base, overlay| merge(&base, &overlay))
            .unwrap_or_default();
        let mut root = node(self.core, &merged, None, "");
        root.min = 0;
        root.max = None;
        root.types = BTreeSet::from([schema.type_name.clone()]);
        root
    }
}

fn convert(json: &Value) -> Result<FhirSchema, String> {
    let definition: StructureDefinition =
        serde_json::from_value(json.clone()).map_err(|e| e.to_string())?;
    octofhir_fhirschema::translate(definition, None).map_err(|e| e.to_string())
}

/// A schema's top level as an element, so every level merges the same way.
fn level(schema: &FhirSchema) -> FhirSchemaElement {
    FhirSchemaElement {
        type_name: Some(schema.type_name.clone()),
        elements: schema.elements.clone(),
        required: schema.required.clone(),
        excluded: schema.excluded.clone(),
        constraint: schema.constraint.clone(),
        extensions: schema.extensions.clone(),
        ..Default::default()
    }
}

/// `overlay` over `base`: the overlay's constraints win, collections are united.
fn merge(base: &FhirSchemaElement, overlay: &FhirSchemaElement) -> FhirSchemaElement {
    let mut result = base.clone();
    macro_rules! take {
        ($($field:ident),*) => {$(
            if overlay.$field.is_some() {
                result.$field = overlay.$field.clone();
            }
        )*};
    }
    take!(
        type_name,
        array,
        min,
        max,
        refers,
        binding,
        pattern,
        choices,
        choice_of,
        url,
        must_support,
        is_modifier,
        short
    );
    if let Some(constraints) = &overlay.constraint {
        let merged = result.constraint.get_or_insert_with(HashMap::new);
        for (key, constraint) in constraints {
            merged.insert(key.clone(), constraint.clone());
        }
    }
    for (target, source) in [
        (&mut result.required, &overlay.required),
        (&mut result.excluded, &overlay.excluded),
    ] {
        if let Some(names) = source {
            let list = target.get_or_insert_with(Vec::new);
            for name in names {
                if !list.contains(name) {
                    list.push(name.clone());
                }
            }
        }
    }
    if let Some(elements) = &overlay.elements {
        let merged = result.elements.get_or_insert_with(HashMap::new);
        for (name, element) in elements {
            let next = match merged.get(name) {
                Some(existing) => merge(existing, element),
                None => element.clone(),
            };
            merged.insert(name.clone(), next);
        }
    }
    if let Some(slicing) = &overlay.slicing {
        result.slicing = Some(merge_slicing(result.slicing.as_ref(), slicing));
    }
    result
}

fn merge_slicing(
    base: Option<&FhirSchemaSlicing>,
    overlay: &FhirSchemaSlicing,
) -> FhirSchemaSlicing {
    let Some(base) = base else {
        return overlay.clone();
    };
    let mut result = base.clone();
    if overlay.discriminator.is_some() {
        result.discriminator = overlay.discriminator.clone();
    }
    if overlay.rules.is_some() {
        result.rules = overlay.rules.clone();
    }
    if overlay.ordered.is_some() {
        result.ordered = overlay.ordered;
    }
    if let Some(slices) = &overlay.slices {
        let merged = result.slices.get_or_insert_with(HashMap::new);
        for (name, slice) in slices {
            let next = match merged.get(name) {
                Some(existing) => FhirSchemaSliceMatch {
                    match_value: slice.match_value.clone().or(existing.match_value.clone()),
                    schema: match (&existing.schema, &slice.schema) {
                        (Some(a), Some(b)) => Some(merge(a, b)),
                        (a, b) => b.clone().or(a.clone()),
                    },
                    min: slice.min.or(existing.min),
                    max: slice.max.or(existing.max),
                },
                None => slice.clone(),
            };
            merged.insert(name.clone(), next);
        }
    }
    result
}

/// The node of `element` (named `name` in a parent whose `required` / `excluded` lists
/// apply to it).
fn node(
    core: &HashMap<String, FhirSchema>,
    element: &FhirSchemaElement,
    parent: Option<&FhirSchemaElement>,
    name: &str,
) -> Node {
    let listed =
        |list: Option<&Vec<String>>| list.is_some_and(|names| names.iter().any(|n| n == name));
    let required = parent.is_some_and(|p| listed(p.required.as_ref()));
    let excluded = parent.is_some_and(|p| listed(p.excluded.as_ref()));
    let min = element.min.unwrap_or(0).max(0) as u32;
    let min = if required { min.max(1) } else { min };
    let max = if excluded {
        Some(0)
    } else if element.array == Some(true) {
        element.max.map(|max| max.max(0) as u32)
    } else {
        Some(element.max.map_or(1, |max| max.max(0) as u32))
    };
    let types = match &element.choices {
        Some(choices) => choices.iter().cloned().collect(),
        None => element.type_name.iter().cloned().collect(),
    };
    // What the element's type itself requires or forbids of its children (Narrative.status).
    let type_schema = element.type_name.as_deref().and_then(|t| core.get(t));
    let mut scope = element.clone();
    if let Some(type_schema) = type_schema {
        for (target, source) in [
            (&mut scope.required, &type_schema.required),
            (&mut scope.excluded, &type_schema.excluded),
        ] {
            for name in source.iter().flatten() {
                let list = target.get_or_insert_with(Vec::new);
                if !list.contains(name) {
                    list.push(name.clone());
                }
            }
        }
    }
    let mut children: BTreeMap<String, Node> = element
        .elements
        .iter()
        .flatten()
        .map(|(child, schema)| {
            // A nested constraint (`type.coding` min 1) says nothing of what the child
            // inherits from its type (`CodeableConcept.coding` repeats): merge it over that.
            let inherited = element
                .type_name
                .as_deref()
                .and_then(|type_name| core.get(type_name))
                .and_then(|type_schema| type_schema.elements.as_ref()?.get(child));
            let schema = match inherited {
                Some(base) => merge(base, schema),
                None => schema.clone(),
            };
            (child.clone(), node(core, &schema, Some(&scope), child))
        })
        .collect();
    normalize_choices(&mut children);
    let (slicing, slices) = match &element.slicing {
        Some(slicing) => (
            Some(Slicing {
                rules: slicing.rules.clone().unwrap_or_else(|| "open".to_owned()),
                ordered: slicing.ordered.unwrap_or(false),
                discriminators: slicing
                    .discriminator
                    .iter()
                    .flatten()
                    .map(|d| format!("{}:{}", d.type_name, d.path))
                    .collect(),
            }),
            slicing
                .slices
                .iter()
                .flatten()
                .map(|(slice_name, slice)| {
                    // An extension slice's cardinality is in the parent's `extensions`.
                    let extension = matches!(name, "extension" | "modifierExtension")
                        .then(|| parent.and_then(|p| p.extensions.as_ref()?.get(slice_name)))
                        .flatten();
                    (
                        slice_name.clone(),
                        slice_node(core, element, slice, extension),
                    )
                })
                .collect(),
        ),
        None => (None, BTreeMap::new()),
    };
    Node {
        min,
        max,
        types,
        targets: element.refers.iter().flatten().cloned().collect(),
        must_support: element.must_support.unwrap_or(false),
        is_modifier: element.is_modifier.unwrap_or(false),
        binding: element.binding.as_ref().map(|binding| Binding {
            strength: Strength::parse(&binding.strength),
            value_set: binding.value_set.clone(),
        }),
        constraints: element
            .constraint
            .iter()
            .flatten()
            .map(|(key, constraint)| (key.clone(), constraint_of(constraint)))
            .collect(),
        pattern: element
            .pattern
            .as_ref()
            .map(|pattern| pattern.value.clone()),
        slicing,
        slices,
        children,
        extension_url: element.url.clone(),
    }
}

/// A profile may narrow a choice through one of its variants (`Observation.effectiveDateTime`
/// with `min` 1 restricts `effective[x]` to dateTime and requires it) rather than through the
/// choice element's own types: carry that onto the choice element, so both ways of saying it
/// compare equal.
fn normalize_choices(children: &mut BTreeMap<String, Node>) {
    let choices: Vec<(String, Vec<String>)> = children
        .iter()
        .filter(|(_, node)| {
            node.types.len() > 1 || node.types.iter().any(|t| looks_like_variant(t))
        })
        .map(|(name, node)| (name.clone(), node.types.iter().cloned().collect()))
        .collect();
    for (name, variants) in choices {
        let present: Vec<String> = variants
            .iter()
            .filter(|variant| children.contains_key(*variant))
            .cloned()
            .collect();
        if present.is_empty() {
            continue;
        }
        let required: Vec<&String> = present
            .iter()
            .filter(|variant| children[*variant].min >= 1)
            .collect();
        let excluded: BTreeSet<String> = present
            .iter()
            .filter(|variant| children[*variant].max == Some(0))
            .cloned()
            .collect();
        let only = (required.len() == 1).then(|| required[0].clone());
        // A variant named by a type slice keeps its own mustSupport; one named by the
        // renamed path (`effectiveDateTime`) is the choice element itself.
        let sliced = !children[&name].slices.is_empty();
        let variant_must_support = !sliced
            && only
                .as_ref()
                .is_some_and(|variant| children[variant].must_support);
        if let Some(choice) = children.get_mut(&name) {
            if let Some(only) = only {
                choice.min = choice.min.max(1);
                choice.types = BTreeSet::from([only]);
                choice.must_support |= variant_must_support;
            }
            choice.types.retain(|variant| !excluded.contains(variant));
        }
    }
}

/// Whether a type name is a choice variant (`valueQuantity`): a lowercase prefix then an
/// uppercase type name.
fn looks_like_variant(name: &str) -> bool {
    let mut chars = name.chars();
    chars.next().is_some_and(char::is_lowercase) && name.chars().any(char::is_uppercase)
}

/// Whether `name` is one of the variants a sibling choice element lists.
pub fn is_variant(siblings: &BTreeMap<String, Node>, name: &str) -> bool {
    siblings
        .values()
        .any(|node| node.types.contains(name) && node.types.iter().all(|t| looks_like_variant(t)))
}

/// A slice: its schema over the sliced element (a slice of `Coverage.class` keeps what
/// `class` requires of its children), with the slice's own cardinality.
fn slice_node(
    core: &HashMap<String, FhirSchema>,
    sliced: &FhirSchemaElement,
    slice: &FhirSchemaSliceMatch,
    extension: Option<&Value>,
) -> Node {
    let mut base = sliced.clone();
    base.slicing = None;
    base.min = None;
    base.max = None;
    let schema = match &slice.schema {
        Some(schema) => merge(&base, schema),
        None => base,
    };
    let mut slice_node = node(core, &schema, None, "");
    let number = |value: Option<&Value>| match value {
        Some(Value::Number(n)) => n.as_u64().map(|n| n as u32),
        Some(Value::String(s)) => s.parse().ok(),
        _ => None,
    };
    let unbounded = |value: Option<&Value>| matches!(value, Some(Value::String(s)) if s == "*");
    slice_node.min = slice
        .min
        .map(|min| min.max(0) as u32)
        .or_else(|| number(extension.and_then(|e| e.get("min"))))
        .unwrap_or(0);
    slice_node.max = match slice.max {
        Some(max) => Some(max.max(0) as u32),
        None => match extension.and_then(|e| e.get("max")) {
            max if unbounded(max) => None,
            Some(max) => number(Some(max)),
            None => slice
                .schema
                .as_ref()
                .and_then(|s| s.max)
                .map(|max| max.max(0) as u32),
        },
    };
    slice_node
}

fn constraint_of(constraint: &FhirSchemaConstraint) -> Constraint {
    Constraint {
        severity: constraint.severity.clone(),
        expression: constraint.expression.clone(),
        human: constraint.human.clone(),
    }
}