mig-types 0.6.0

use serde::{Deserialize, Serialize};

use super::common::{Cardinality, CodeDefinition};

/// Complete MIG schema for a message type.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct MigSchema {
    /// The EDIFACT message type (e.g., "UTILMD", "ORDERS").
    pub message_type: String,
    /// Optional variant (e.g., "Strom", "Gas").
    pub variant: Option<String>,
    /// Version number from the MIG (e.g., "S2.1", "1.4a").
    pub version: String,
    /// Publication date string.
    pub publication_date: String,
    /// Author (typically "BDEW").
    pub author: String,
    /// Format version directory (e.g., "FV2504").
    pub format_version: String,
    /// Path to the source XML file.
    pub source_file: String,
    /// Top-level segment definitions (not in groups).
    pub segments: Vec<MigSegment>,
    /// Segment group definitions (contain more segments).
    pub segment_groups: Vec<MigSegmentGroup>,
}

impl MigSchema {
    /// Whether this MIG includes the interchange envelope (UNA/UNB) as
    /// top-level segments. UTILMD/MSCONS/INVOIC/REMADV do; UTILTS/PRICAT/
    /// ORDERS/COMDIS start at UNH.
    ///
    /// Callers that assemble from raw EDIFACT need this to decide whether
    /// to feed `MessageChunk::all_segments()` (envelope + UNH + body + UNT)
    /// or `MessageChunk::message_segments()` (UNH + body + UNT) to the
    /// assembler — feeding envelope segments to a UNH-start MIG aborts
    /// assembly at the first segment.
    pub fn includes_envelope(&self) -> bool {
        self.segments.iter().any(|s| s.id == "UNA" || s.id == "UNB")
    }
}

/// A segment (S_*) definition from the MIG.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct MigSegment {
    /// Segment identifier (e.g., "UNH", "BGM", "NAD").
    pub id: String,
    /// Human-readable name.
    pub name: String,
    /// Description of the segment.
    pub description: Option<String>,
    /// Position counter (e.g., "0010", "0020").
    pub counter: Option<String>,
    /// Nesting level (0=root, 1=first level, etc.).
    pub level: i32,
    /// Sequence number within the message.
    pub number: Option<String>,
    /// Standard maximum repetitions.
    pub max_rep_std: i32,
    /// Specification maximum repetitions.
    pub max_rep_spec: i32,
    /// Standard status (M=Mandatory, C=Conditional, etc.).
    pub status_std: Option<String>,
    /// Specification status (M, R, D, O, N).
    pub status_spec: Option<String>,
    /// Example EDIFACT string.
    pub example: Option<String>,
    /// Direct child data elements.
    pub data_elements: Vec<MigDataElement>,
    /// Child composite elements.
    pub composites: Vec<MigComposite>,
}

impl MigSegment {
    /// Returns the effective cardinality based on spec or std status.
    pub fn cardinality(&self) -> Cardinality {
        let status = self
            .status_spec
            .as_deref()
            .or(self.status_std.as_deref())
            .unwrap_or("C");
        Cardinality::from_status(status)
    }

    /// Returns the effective max repetitions (spec overrides std).
    pub fn max_rep(&self) -> i32 {
        self.max_rep_spec.max(self.max_rep_std)
    }
}

/// A segment group (G_SG*) definition from the MIG.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct MigSegmentGroup {
    /// Group identifier (e.g., "SG1", "SG2", "SG10").
    pub id: String,
    /// Human-readable name.
    pub name: String,
    /// Description of the segment group.
    pub description: Option<String>,
    /// Position counter (e.g., "0070", "0500").
    pub counter: Option<String>,
    /// Nesting level.
    pub level: i32,
    /// Standard maximum repetitions.
    pub max_rep_std: i32,
    /// Specification maximum repetitions.
    pub max_rep_spec: i32,
    /// Standard status.
    pub status_std: Option<String>,
    /// Specification status.
    pub status_spec: Option<String>,
    /// Segments directly in this group.
    pub segments: Vec<MigSegment>,
    /// Nested segment groups.
    pub nested_groups: Vec<MigSegmentGroup>,
    /// Optional variant qualifier code for the entry segment.
    /// When set, the assembler only matches segments whose entry qualifier
    /// equals this code (e.g., "Z98" for SEQ+Z98, "ZD5" for SEQ+ZD5).
    #[serde(default, skip_serializing_if = "Option::is_none")]
    pub variant_code: Option<String>,
    /// Position of the variant qualifier in the entry segment:
    /// (element_index, component_index). Defaults to (0, 0) when absent.
    /// Some segments have the qualifier in a composite at a non-zero position
    /// (e.g., CCI with qualifier in C240/D7037 at element index 2).
    #[serde(default, skip_serializing_if = "Option::is_none")]
    pub variant_qualifier_position: Option<(usize, usize)>,
    /// All allowed qualifier codes for this variant (assembler matches ANY).
    #[serde(default)]
    pub variant_codes: Vec<String>,
    /// Number of MIG XML variants that were merged into this group definition.
    /// When multiple SG2 definitions (MS, MR, DP, etc.) are merged into one,
    /// this holds the count of merged variants. Used to compute the correct
    /// max_reps for PID schema generation (variant count, not max of individual max_reps).
    #[serde(default, skip_serializing_if = "Option::is_none")]
    pub merged_variant_count: Option<u32>,
    /// Entry-segment qualifiers of the MIG variants merged into this group
    /// definition, in MIG order (see `merge_group_variants` in
    /// `mig-assembly::pid_filter`). Empty for a group that is not a merge — its
    /// own [`entry_qualifier`](Self::entry_qualifier) applies.
    ///
    /// Merging unions the variants' segments, which loses which repetition
    /// belongs to which variant. The reverse mapping needs exactly that to
    /// emit repetitions in MIG variant order (`NAD+MS` before `NAD+MR`) rather
    /// than in the order the BO4E JSON happens to list them.
    #[serde(default, skip_serializing_if = "Vec::is_empty")]
    pub variant_entry_qualifiers: Vec<EntryQualifier>,
}

/// The qualifier that identifies a group variant: the codes the MIG allows in
/// the first coded data element of the group's entry segment.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct EntryQualifier {
    /// Entry segment tag, e.g. `"NAD"`.
    pub tag: String,
    /// Element index within the segment (0-based).
    pub element: usize,
    /// Component index within the element (0 for a simple data element).
    pub component: usize,
    /// Allowed codes, in MIG order.
    pub codes: Vec<String>,
    /// IDs of the groups nested in this variant (e.g. `["SG3"]` for the SG2
    /// `NAD+MS` variant that holds the sender's contact).
    #[serde(default, skip_serializing_if = "Vec::is_empty")]
    pub nested_group_ids: Vec<String>,
}

impl EntryQualifier {
    /// Whether `other` identifies the same variant (nested groups aside).
    pub fn same_qualifier(&self, other: &EntryQualifier) -> bool {
        self.tag == other.tag
            && self.element == other.element
            && self.component == other.component
            && self.codes == other.codes
    }

    /// Whether a segment's elements carry one of this qualifier's codes.
    pub fn matches(&self, elements: &[Vec<String>]) -> bool {
        elements
            .get(self.element)
            .and_then(|e| e.get(self.component))
            .is_some_and(|v| self.codes.iter().any(|c| c.eq_ignore_ascii_case(v)))
    }
}

impl MigSegmentGroup {
    /// The qualifier of this group's entry segment: its first coded data
    /// element by (element, component) position. `None` when the entry segment
    /// has no coded element (the variant cannot be told apart by qualifier).
    pub fn entry_qualifier(&self) -> Option<EntryQualifier> {
        let entry = self.segments.first()?;
        let simple = entry.data_elements.iter().map(|d| (d.position, 0, d));
        let composite = entry.composites.iter().flat_map(|c| {
            c.data_elements
                .iter()
                .map(move |d| (c.position, d.position, d))
        });
        simple
            .chain(composite)
            .filter(|(_, _, d)| d.codes.iter().any(|c| !c.value.is_empty()))
            .min_by_key(|(e, c, _)| (*e, *c))
            .map(|(element, component, d)| EntryQualifier {
                tag: entry.id.clone(),
                element,
                component,
                codes: d
                    .codes
                    .iter()
                    .filter(|c| !c.value.is_empty())
                    .map(|c| c.value.clone())
                    .collect(),
                nested_group_ids: self.nested_group_ids(),
            })
    }

    /// IDs of the directly nested groups, deduplicated, in MIG order.
    pub fn nested_group_ids(&self) -> Vec<String> {
        let mut ids: Vec<String> = Vec::new();
        for g in &self.nested_groups {
            if !ids.contains(&g.id) {
                ids.push(g.id.clone());
            }
        }
        ids
    }

    /// Returns the effective cardinality.
    pub fn cardinality(&self) -> Cardinality {
        let status = self
            .status_spec
            .as_deref()
            .or(self.status_std.as_deref())
            .unwrap_or("C");
        Cardinality::from_status(status)
    }
}

/// A composite element (C_*) definition from the MIG.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct MigComposite {
    /// Composite identifier (e.g., "S009", "C002").
    pub id: String,
    /// Human-readable name.
    pub name: String,
    /// Description.
    pub description: Option<String>,
    /// Standard status.
    pub status_std: Option<String>,
    /// Specification status.
    pub status_spec: Option<String>,
    /// Child data elements within this composite.
    pub data_elements: Vec<MigDataElement>,
    /// Position of this composite within its parent segment (0-based).
    pub position: usize,
}

/// A data element (D_*) definition from the MIG.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct MigDataElement {
    /// Element identifier (e.g., "0062", "3035").
    pub id: String,
    /// Human-readable name.
    pub name: String,
    /// Description.
    pub description: Option<String>,
    /// Standard status.
    pub status_std: Option<String>,
    /// Specification status.
    pub status_spec: Option<String>,
    /// Standard format (e.g., "an..14", "n13").
    pub format_std: Option<String>,
    /// Specification format.
    pub format_spec: Option<String>,
    /// Allowed code values, if restricted.
    pub codes: Vec<CodeDefinition>,
    /// Position within parent (0-based).
    pub position: usize,
}