swift_mt_message/parser/

generated.rs

//! Generated parser functions
//!
//! This module contains parser functions that are generated by the swift-mt-message-macros crate.
//! These functions replace the manual implementations that were previously in parser_impl.rs.

use crate::errors::{ParseError, Result};
use std::collections::HashMap;

// Include the generated parser functions from the macro crate
// These will be generated at compile time based on the macro definitions

/// Parse block 4 fields into a field map with enhanced position tracking
///
/// This function is generated by the macro system to ensure consistent parsing
/// across all field types and messages.
pub fn parse_block4_fields(block4: &str) -> Result<HashMap<String, Vec<(String, usize)>>> {
    // Pre-allocate HashMap with estimated capacity based on typical field count
    // Most messages have between 10-60 fields
    let estimated_fields = block4.matches("\n:").count().max(10);
    let mut field_map: HashMap<String, Vec<(String, usize)>> =
        HashMap::with_capacity(estimated_fields);

    // Remove leading/trailing whitespace and newlines
    let content = block4.trim();

    // Split by field markers (:XX:)
    let mut current_pos = 0;
    let mut field_position = 0; // Track sequential position for consumption ordering
    let mut line_number = 1;

    while current_pos < content.len() {
        // Track line numbers for better error reporting
        if current_pos > 0 && content.chars().nth(current_pos - 1) == Some('\n') {
            line_number += 1;
        }

        // Find next field marker
        if let Some(field_start) = content[current_pos..].find(':') {
            let field_start = current_pos + field_start;

            // Extract field tag (characters after : until next :)
            if let Some(tag_end) = content[field_start + 1..].find(':') {
                let tag_end = field_start + 1 + tag_end;
                let raw_field_tag = &content[field_start + 1..tag_end];

                // Normalize field tag by removing option letters (A, F, K, etc.)
                let field_tag = normalize_field_tag(raw_field_tag);

                // Find the end of field value (next field marker or end of content)
                let value_start = tag_end + 1;
                let value_end = if let Some(next_field) = content[value_start..].find("\n:") {
                    value_start + next_field
                } else {
                    content.len()
                };

                // Avoid unnecessary string allocation - trim inline during push
                let field_value_slice = &content[value_start..value_end];
                let trimmed_value = field_value_slice.trim();

                // Store field value with enhanced position info (line number encoded with field position)
                // High 16 bits: line number, Low 16 bits: field position
                let position_info = (line_number << 16) | (field_position & 0xFFFF);

                // Add to existing Vec or create new Vec for this field tag
                field_map
                    .entry(field_tag.into_owned())
                    .or_default()
                    .push((trimmed_value.to_string(), position_info));

                field_position += 1; // Increment position for next field
                current_pos = value_end;
            } else {
                // Last field or malformed - provide detailed error
                return Err(ParseError::InvalidBlockStructure {
                    block: "4".to_string(),
                    message: format!(
                        "Malformed field tag at line {line_number}, position {current_pos}"
                    ),
                });
            }
        } else {
            break;
        }
    }

    Ok(field_map)
}

/// Normalize field tag by removing option letters (A, F, K, etc.)
/// Example: "50K" -> "50", "59A" -> "59", "20" -> "20"
/// But preserve option letters for fields that have multiple variants like 23B/23E, 71A/71F/71G
/// Also preserve numbered field tags like "50#1", "50#2"
pub fn normalize_field_tag(raw_tag: &str) -> std::borrow::Cow<'_, str> {
    use std::borrow::Cow;

    // Special handling for numbered field tags (e.g., "50#1", "50#2")
    if raw_tag.contains('#') {
        // For numbered fields, we need to keep the full tag including the # and number
        // This is used in MT101 and other messages to distinguish multiple occurrences
        return Cow::Borrowed(raw_tag);
    }

    // Find where the numeric part ends
    let numeric_end = raw_tag
        .find(|c: char| !c.is_ascii_digit())
        .unwrap_or(raw_tag.len());

    // If no suffix, return the tag as-is
    if numeric_end == raw_tag.len() {
        return Cow::Borrowed(raw_tag);
    }

    let numeric_part = &raw_tag[..numeric_end];
    let suffix = &raw_tag[numeric_end..];

    // For certain field numbers, preserve the option letter to avoid conflicts
    match numeric_part {
        "11" | "13" | "21" | "23" | "25" | "26" | "28" | "32" | "33" | "34" | "37" | "50"
        | "52" | "53" | "54" | "55" | "56" | "57" | "58" | "59" | "60" | "62" | "71" | "77" => {
            // Keep option letters for fields that have multiple variants or specific formats
            Cow::Borrowed(raw_tag)
        }
        _ => {
            // For other fields, check if suffix is just uppercase letters
            if suffix.chars().all(|c| c.is_ascii_uppercase()) {
                // It's an option letter, return just the numeric part
                Cow::Owned(numeric_part.to_string())
            } else {
                // Not a simple option letter, keep the full tag
                Cow::Borrowed(raw_tag)
            }
        }
    }
}

/// Extract base tag by removing index suffix (e.g., "50#1" -> "50")
pub fn extract_base_tag(tag: &str) -> &str {
    if let Some(index_pos) = tag.find('#') {
        &tag[..index_pos]
    } else {
        tag
    }
}