//! python code generation from lexicon documents

use std::collections::{HashMap, HashSet};
use std::fs;
use std::io;
use std::path::Path;

use atrium_lex::lexicon::{LexObject, LexRecord, LexUserType};
use atrium_lex::LexiconDoc;
use heck::ToSnakeCase;

use crate::builtin::builtin_lexicons;
use crate::types::{collect_external_refs, property_to_python, to_class_name, RefContext};

const HEADER: &str = r#"# auto-generated by pmgfal - do not edit

from __future__ import annotations

from typing import Any

from pydantic import BaseModel, Field
"#;

/// python keywords that need escaping as field names
const PYTHON_KEYWORDS: &[&str] = &[
    "type", "class", "import", "from", "global", "lambda", "def", "return", "yield", "raise",
    "try", "except", "finally", "with", "as", "if", "elif", "else", "for", "while", "break",
    "continue", "pass", "and", "or", "not", "in", "is", "None", "True", "False", "async", "await",
];

/// generate pydantic models for all documents
pub fn generate_models(
    docs: &[LexiconDoc],
    output_dir: &Path,
    namespace_prefix: Option<&str>,
) -> Result<Vec<String>, io::Error> {
    let filtered: Vec<_> = docs
        .iter()
        .filter(|doc| {
            namespace_prefix
                .map(|p| doc.id.starts_with(p))
                .unwrap_or(true)
        })
        .collect();

    if filtered.is_empty() {
        return Ok(vec![]);
    }

    // build lookup of all available lexicons (user + builtin)
    let mut all_docs: HashMap<&str, &LexiconDoc> = HashMap::new();
    for doc in docs {
        all_docs.insert(&doc.id, doc);
    }
    for doc in builtin_lexicons() {
        all_docs.entry(&doc.id).or_insert(doc);
    }

    // collect external refs from user documents
    let mut external_refs: HashSet<String> = HashSet::new();
    for doc in &filtered {
        external_refs.extend(collect_external_refs(doc));
    }

    // find which external refs we can resolve from builtins
    let mut resolved_externals: Vec<&LexiconDoc> = Vec::new();
    for ref_nsid in &external_refs {
        if let Some(doc) = all_docs.get(ref_nsid.as_str()) {
            // only include if not already in user docs
            if !filtered.iter().any(|d| d.id == *ref_nsid) {
                resolved_externals.push(doc);
            }
        }
    }
    resolved_externals.sort_by(|a, b| a.id.cmp(&b.id));

    fs::create_dir_all(output_dir)?;

    let mut output = String::from(HEADER);
    output.push('\n');

    // generate external deps first (so they're defined before use)
    for doc in &resolved_externals {
        output.push_str(&format!("\n# {} (builtin)\n", doc.id));
        output.push_str(&generate_document(doc));
    }

    // generate user documents
    for doc in &filtered {
        output.push_str(&format!("\n# {}\n", doc.id));
        output.push_str(&generate_document(doc));
    }

    let output_file = match namespace_prefix {
        Some(prefix) => output_dir.join(format!("{}.py", prefix.replace('.', "_"))),
        None => output_dir.join("models.py"),
    };

    fs::write(&output_file, &output)?;

    Ok(vec![output_file.to_string_lossy().to_string()])
}

/// generate python code for a single lexicon document
fn generate_document(doc: &LexiconDoc) -> String {
    let ctx = RefContext::new(&doc.id);
    let mut output = String::new();

    for (def_name, def) in &doc.defs {
        let class_name = to_class_name(&doc.id, def_name);

        match def {
            LexUserType::Record(LexRecord {
                record,
                description,
                ..
            }) => {
                let atrium_lex::lexicon::LexRecordRecord::Object(obj) = record;
                let desc = description.as_deref().unwrap_or(&doc.id);
                output.push_str(&generate_class(&class_name, obj, Some(desc), &ctx));
                output.push_str("\n\n");
            }
            LexUserType::Object(obj) => {
                output.push_str(&generate_class(
                    &class_name,
                    obj,
                    obj.description.as_deref(),
                    &ctx,
                ));
                output.push_str("\n\n");
            }
            LexUserType::Token(_) => {
                output.push_str(&format!(
                    "# token: {}\n{} = \"{}#{}\"\n\n",
                    class_name,
                    class_name.to_uppercase(),
                    doc.id,
                    def_name
                ));
            }
            _ => {}
        }
    }

    output
}

/// generate a pydantic model class
fn generate_class(
    class_name: &str,
    obj: &LexObject,
    description: Option<&str>,
    ctx: &RefContext,
) -> String {
    let mut lines = vec![format!("class {class_name}(BaseModel):")];

    if let Some(desc) = description {
        lines.push(format!("    \"\"\"{desc}\"\"\""));
    }

    if obj.properties.is_empty() {
        lines.push("    pass".into());
        return lines.join("\n");
    }

    let required: HashSet<_> = obj
        .required
        .as_ref()
        .map(|r| r.iter().map(String::as_str).collect())
        .unwrap_or_default();

    // generate required fields first, then optional
    let mut fields: Vec<_> = obj.properties.iter().collect();
    fields.sort_by_key(|(name, _)| !required.contains(name.as_str()));

    for (name, prop) in fields {
        let field_name = to_field_name(name);
        let is_required = required.contains(name.as_str());

        let mut py_type = property_to_python(prop, ctx);
        if !is_required {
            py_type = format!("{py_type} | None");
        }

        let needs_alias = field_name != *name;
        let needs_default = !is_required;

        let field_def = match (needs_alias, needs_default) {
            (false, false) => format!("    {field_name}: {py_type}"),
            (true, false) => format!("    {field_name}: {py_type} = Field(alias=\"{name}\")"),
            (false, true) => format!("    {field_name}: {py_type} = Field(default=None)"),
            (true, true) => {
                format!("    {field_name}: {py_type} = Field(default=None, alias=\"{name}\")")
            }
        };

        lines.push(field_def);
    }

    lines.join("\n")
}

/// convert property name to valid python field name
fn to_field_name(name: &str) -> String {
    let snake = name.to_snake_case();
    if PYTHON_KEYWORDS.contains(&snake.as_str()) {
        format!("{snake}_")
    } else {
        snake
    }
}