portindex.reef - zygaena/coral

root / src / core / portindex.reef
portindex.reef Reef 577 lines 18.8 KB
/******************************************************************************
               __               ____                __
              / /   ___  ____ _/ __/_____________ _/ /__
             / /   / _ \/ __ `/ /_/ ___/ ___/ __ `/ / _ \
            / /___/  __/ /_/ / __(__  ) /__/ /_/ / /  __/
           /_____/\___/\__,_/_/ /____/\___/\__,_/_/\___/

   (C)opyright 2025, Leafscale, LLC -  https://www.leafscale.com

   Project: Zygaena
  Filename: portindex.reef
   Authors: Chris Tusa <chris.tusa@leafscale.com>
   License: <see LICENSE file included with this source code>
Description: Ports index for fast port lookups

    The ports index provides fast indexed access to port metadata
    without parsing individual package.toml files from disk.

    Index file: /var/lib/coral/db/ports.db
    Format: Base64-encoded MsgPack (same as packages.db, files.db)

    Rebuild triggers: coral sync, coral db rebuild, coral db init
    Manual: coral ports cache rebuild

******************************************************************************/

module core.portindex

import types
import io.file
import io.dir
import io.path
import core.str
import core.port
import core.config
import encoding.msgpack
import encoding.base64
import core.result as res

export
    type PortIndexEntry
    type PortsIndex

    // Rebuild the ports index from the ports tree
    fn rebuild_port_index(): bool

    // Check if ports.db exists
    fn port_index_exists(): bool

    // Load the ports index from disk
    fn load_port_index(): PortsIndex

    // Find a single port by name or category/name
    fn find_port(name: string): PortIndexEntry

    // Search ports by substring match on name + description
    fn search_ports(term: string, results: [PortIndexEntry], max_count: int): int

    // List all indexed ports
    fn list_all_ports(entries: [PortIndexEntry], max_count: int): int

    // Get count of indexed ports
    fn get_port_count(): int

    // Find ports that depend on a given package name
    fn find_reverse_deps(name: string, results: [PortIndexEntry], max_count: int): int
end export

// ============================================================================
// Types
// ============================================================================

type PortIndexEntry = struct
    name: string
    category: string
    version: string
    release: int
    description: string
    port_dir: string
    runtime_deps: [string]
    runtime_dep_count: int
    found: bool
end PortIndexEntry

type PortsIndex = struct
    version: int
    generated: string
    port_count: int
    entries: [PortIndexEntry]
    loaded: bool
end PortsIndex

// ============================================================================
// Constants
// ============================================================================

const PORT_INDEX_VERSION: int = 1
const MAX_PORTS: int = 4096

// ============================================================================
// Path helpers
// ============================================================================

fn ports_db_path(): string
    return types.get_index_dir() + "/ports.db"
end ports_db_path

// Get current timestamp as string
fn get_timestamp(): string
    return "2026-02-19"
end get_timestamp

// ============================================================================
// Empty constructors
// ============================================================================

fn empty_port_index(): PortsIndex
    return PortsIndex{
        version: 0,
        generated: "",
        port_count: 0,
        entries: new [PortIndexEntry](MAX_PORTS),
        loaded: false
    }
end empty_port_index

fn empty_port_entry(): PortIndexEntry
    return PortIndexEntry{
        name: "",
        category: "",
        version: "",
        release: 0,
        description: "",
        port_dir: "",
        runtime_deps: new [string](64),
        runtime_dep_count: 0,
        found: false
    }
end empty_port_entry

// ============================================================================
// Buffer size estimation
// ============================================================================

fn estimate_buffer_size(port_count: int): int
    // Header: ~100 bytes
    // Per port: ~300 bytes (name, category, version, description, port_dir, deps)
    return 100 + (port_count * 300)
end estimate_buffer_size

// ============================================================================
// MsgPack header size helpers (same as pkgdb.reef)
// ============================================================================

fn map_header_size(buf: [int], offset: int): int
    let b = buf[offset] & 255
    if (b & 0xF0) == 0x80
        return 1
    elif b == 0xDE
        return 3
    elif b == 0xDF
        return 5
    end if
    return 1
end map_header_size

fn array_header_size(buf: [int], offset: int): int
    let b = buf[offset] & 255
    if (b & 0xF0) == 0x90
        return 1
    elif b == 0xDC
        return 3
    elif b == 0xDD
        return 5
    end if
    return 1
end array_header_size

// ============================================================================
// Index existence check
// ============================================================================

fn port_index_exists(): bool
    return file.fileExists(ports_db_path())
end port_index_exists

// ============================================================================
// Rebuild: scan all ports, serialize to ports.db
// ============================================================================

fn rebuild_port_index(): bool
    // Ensure db directory exists
    let db_dir = types.get_index_dir()
    if not dir.dir_exists(db_dir)
        dir.create_dir_all(db_dir)
    end if

    // Get all categories dynamically
    mut cats: [string] = new [string](32)
    let cat_count = port.get_categories(cats, 32)

    // Collect all port entries
    mut entries: [PortIndexEntry] = new [PortIndexEntry](MAX_PORTS)
    mut total = 0

    mut i = 0
    while i < cat_count and total < MAX_PORTS
        let category = cats[i]
        mut port_names: [string] = new [string](256)
        let port_count = port.list_category(category, port_names, 256)

        mut j = 0
        while j < port_count and total < MAX_PORTS
            let result = port.load(get_port_path(category, port_names[j]))
            if result.success
                let p = result.port
                mut entry = empty_port_entry()
                entry.name = p.info.name
                entry.category = p.category
                entry.version = p.info.version
                entry.release = p.info.release
                entry.description = p.info.description
                entry.port_dir = p.port_dir
                entry.found = true

                // Copy runtime deps
                mut k = 0
                while k < p.deps.runtime_count and k < 64
                    entry.runtime_deps[k] = p.deps.runtime[k]
                    k = k + 1
                end while
                entry.runtime_dep_count = p.deps.runtime_count

                entries[total] = entry
                total = total + 1
            end if
            j = j + 1
        end while

        i = i + 1
    end while

    // Serialize to MsgPack
    let buf_size = estimate_buffer_size(total)
    mut buf: [int] = new [int](buf_size)
    let buf_len = serialize_port_index(entries, total, buf, buf_size)

    if buf_len == 0
        return false
    end if

    // Encode to base64 and write
    let encoded = base64.base64_encode_bytes(buf, buf_len)
    return res.is_ok(file.writeFile(ports_db_path(), encoded))
end rebuild_port_index

// Build a port path from category and name
fn get_port_path(category: string, name: string): string
    let cfg = config.load()
    let ports_dir = config.get_ports_dir(cfg)
    return path.join_path(ports_dir, path.join_path(category, name))
end get_port_path

// ============================================================================
// Serialization
// ============================================================================

fn serialize_port_index(entries: [PortIndexEntry], count: int, buf: [int], max_size: int): int
    mut pos = 0

    // Pack map header (4 keys: version, generated, port_count, ports)
    pos = pos + msgpack.msgpack_pack_map_header(4, buf, pos)

    // version
    pos = pos + msgpack.msgpack_pack_string("version", buf, pos)
    pos = pos + msgpack.msgpack_pack_int(PORT_INDEX_VERSION, buf, pos)

    // generated
    pos = pos + msgpack.msgpack_pack_string("generated", buf, pos)
    pos = pos + msgpack.msgpack_pack_string(get_timestamp(), buf, pos)

    // port_count
    pos = pos + msgpack.msgpack_pack_string("port_count", buf, pos)
    pos = pos + msgpack.msgpack_pack_int(count, buf, pos)

    // ports array
    pos = pos + msgpack.msgpack_pack_string("ports", buf, pos)
    pos = pos + msgpack.msgpack_pack_array_header(count, buf, pos)

    mut i = 0
    while i < count and pos < max_size - 2000
        let entry = entries[i]

        // Each port is a map with 7 keys
        pos = pos + msgpack.msgpack_pack_map_header(7, buf, pos)

        // name
        pos = pos + msgpack.msgpack_pack_string("name", buf, pos)
        pos = pos + msgpack.msgpack_pack_string(entry.name, buf, pos)

        // category
        pos = pos + msgpack.msgpack_pack_string("category", buf, pos)
        pos = pos + msgpack.msgpack_pack_string(entry.category, buf, pos)

        // version
        pos = pos + msgpack.msgpack_pack_string("version", buf, pos)
        pos = pos + msgpack.msgpack_pack_string(entry.version, buf, pos)

        // release
        pos = pos + msgpack.msgpack_pack_string("release", buf, pos)
        pos = pos + msgpack.msgpack_pack_int(entry.release, buf, pos)

        // description
        pos = pos + msgpack.msgpack_pack_string("description", buf, pos)
        pos = pos + msgpack.msgpack_pack_string(entry.description, buf, pos)

        // port_dir
        pos = pos + msgpack.msgpack_pack_string("port_dir", buf, pos)
        pos = pos + msgpack.msgpack_pack_string(entry.port_dir, buf, pos)

        // runtime_deps (array of strings)
        pos = pos + msgpack.msgpack_pack_string("runtime_deps", buf, pos)
        pos = pos + msgpack.msgpack_pack_array_header(entry.runtime_dep_count, buf, pos)

        mut d = 0
        while d < entry.runtime_dep_count
            pos = pos + msgpack.msgpack_pack_string(entry.runtime_deps[d], buf, pos)
            d = d + 1
        end while

        i = i + 1
    end while

    return pos
end serialize_port_index

// ============================================================================
// Deserialization
// ============================================================================

fn load_port_index(): PortsIndex
    mut index = empty_port_index()

    let db_path = ports_db_path()
    if not file.fileExists(db_path)
        return index
    end if

    let encoded = res.unwrap_or(file.readFile(db_path), "")
    if str.length(encoded) == 0
        return index
    end if

    let buf_size = str.length(encoded)
    mut buf: [int] = new [int](buf_size)
    let buf_len = res.unwrap_or(base64.base64_decode_bytes(encoded, buf, buf_size), 0)

    if buf_len == 0
        return index
    end if

    mut offset = 0

    // Expect map header
    if not msgpack.msgpack_is_map(buf, offset)
        return index
    end if

    let map_len = res.unwrap_or(msgpack.msgpack_unpack_map_len(buf, offset), 0)
    offset = offset + map_header_size(buf, offset)

    mut entry_idx = 0
    while entry_idx < map_len and offset < buf_len
        let key = res.unwrap_or(msgpack.msgpack_unpack_string(buf, offset), "")
        offset = offset + msgpack.msgpack_value_size(buf, offset)

        if str.equals(key, "version")
            index.version = res.unwrap_or(msgpack.msgpack_unpack_int(buf, offset), 0)
            offset = offset + msgpack.msgpack_value_size(buf, offset)
        elif str.equals(key, "generated")
            index.generated = res.unwrap_or(msgpack.msgpack_unpack_string(buf, offset), "")
            offset = offset + msgpack.msgpack_value_size(buf, offset)
        elif str.equals(key, "port_count")
            index.port_count = res.unwrap_or(msgpack.msgpack_unpack_int(buf, offset), 0)
            offset = offset + msgpack.msgpack_value_size(buf, offset)
        elif str.equals(key, "ports")
            // Parse ports array
            let arr_len = res.unwrap_or(msgpack.msgpack_unpack_array_len(buf, offset), 0)
            offset = offset + array_header_size(buf, offset)

            mut port_idx = 0
            while port_idx < arr_len and port_idx < MAX_PORTS and offset < buf_len
                let port_map_len = res.unwrap_or(msgpack.msgpack_unpack_map_len(buf, offset), 0)
                offset = offset + map_header_size(buf, offset)

                mut entry = empty_port_entry()
                entry.found = true

                mut port_field = 0
                while port_field < port_map_len and offset < buf_len
                    let field_key = res.unwrap_or(msgpack.msgpack_unpack_string(buf, offset), "")
                    offset = offset + msgpack.msgpack_value_size(buf, offset)

                    if str.equals(field_key, "name")
                        entry.name = res.unwrap_or(msgpack.msgpack_unpack_string(buf, offset), "")
                        offset = offset + msgpack.msgpack_value_size(buf, offset)
                    elif str.equals(field_key, "category")
                        entry.category = res.unwrap_or(msgpack.msgpack_unpack_string(buf, offset), "")
                        offset = offset + msgpack.msgpack_value_size(buf, offset)
                    elif str.equals(field_key, "version")
                        entry.version = res.unwrap_or(msgpack.msgpack_unpack_string(buf, offset), "")
                        offset = offset + msgpack.msgpack_value_size(buf, offset)
                    elif str.equals(field_key, "release")
                        entry.release = res.unwrap_or(msgpack.msgpack_unpack_int(buf, offset), 0)
                        offset = offset + msgpack.msgpack_value_size(buf, offset)
                    elif str.equals(field_key, "description")
                        entry.description = res.unwrap_or(msgpack.msgpack_unpack_string(buf, offset), "")
                        offset = offset + msgpack.msgpack_value_size(buf, offset)
                    elif str.equals(field_key, "port_dir")
                        entry.port_dir = res.unwrap_or(msgpack.msgpack_unpack_string(buf, offset), "")
                        offset = offset + msgpack.msgpack_value_size(buf, offset)
                    elif str.equals(field_key, "runtime_deps")
                        // Parse deps array
                        let deps_len = res.unwrap_or(msgpack.msgpack_unpack_array_len(buf, offset), 0)
                        offset = offset + array_header_size(buf, offset)

                        mut dep_idx = 0
                        while dep_idx < deps_len and dep_idx < 64 and offset < buf_len
                            entry.runtime_deps[dep_idx] = res.unwrap_or(msgpack.msgpack_unpack_string(buf, offset), "")
                            offset = offset + msgpack.msgpack_value_size(buf, offset)
                            dep_idx = dep_idx + 1
                        end while
                        entry.runtime_dep_count = dep_idx
                    else
                        offset = offset + msgpack.msgpack_value_size(buf, offset)
                    end if

                    port_field = port_field + 1
                end while

                index.entries[port_idx] = entry
                port_idx = port_idx + 1
            end while
        else
            offset = offset + msgpack.msgpack_value_size(buf, offset)
        end if

        entry_idx = entry_idx + 1
    end while

    index.loaded = true
    return index
end load_port_index

// ============================================================================
// Query functions
// ============================================================================

// Find a single port by name or category/name
fn find_port(name: string): PortIndexEntry
    let index = load_port_index()
    if not index.loaded
        return empty_port_entry()
    end if

    // Check if name contains a slash (category/name format)
    let slash_idx = str.index_of(name, "/")

    mut i = 0
    while i < index.port_count
        let entry = index.entries[i]

        if slash_idx >= 0
            // Match category/name format
            let full_name = entry.category + "/" + entry.name
            // Also check directory-based name (category/dirname)
            let dir_name = path.basename(entry.port_dir)
            let full_dir_name = entry.category + "/" + dir_name
            if str.equals(full_name, name) or str.equals(full_dir_name, name)
                return entry
            end if
        else
            // Match just the name
            if str.equals(entry.name, name)
                return entry
            end if
        end if

        i = i + 1
    end while

    return empty_port_entry()
end find_port

// Search ports by substring match on name + description
fn search_ports(term: string, results: [PortIndexEntry], max_count: int): int
    let index = load_port_index()
    if not index.loaded
        return 0
    end if

    let lower_term = str.to_lower(term)
    mut count = 0

    mut i = 0
    while i < index.port_count and count < max_count
        let entry = index.entries[i]
        let lower_name = str.to_lower(entry.name)
        let lower_desc = str.to_lower(entry.description)

        if str.contains(lower_name, lower_term) or str.contains(lower_desc, lower_term)
            results[count] = entry
            count = count + 1
        end if

        i = i + 1
    end while

    return count
end search_ports

// List all indexed ports
fn list_all_ports(entries: [PortIndexEntry], max_count: int): int
    let index = load_port_index()
    if not index.loaded
        return 0
    end if

    mut count = 0
    while count < index.port_count and count < max_count
        entries[count] = index.entries[count]
        count = count + 1
    end while

    return count
end list_all_ports

// Get count of indexed ports
fn get_port_count(): int
    let index = load_port_index()
    if index.loaded
        return index.port_count
    end if
    return 0
end get_port_count

// Find ports that have `name` in their runtime dependencies
fn find_reverse_deps(name: string, results: [PortIndexEntry], max_count: int): int
    let index = load_port_index()
    if not index.loaded
        return 0
    end if

    mut count = 0

    mut i = 0
    while i < index.port_count and count < max_count
        let entry = index.entries[i]

        mut j = 0
        while j < entry.runtime_dep_count
            if str.equals(entry.runtime_deps[j], name)
                results[count] = entry
                count = count + 1
                break
            end if
            j = j + 1
        end while

        i = i + 1
    end while

    return count
end find_reverse_deps

end module