Skip to main content

file_parser/
module.rs

1use std::sync::Arc;
2
3use async_trait::async_trait;
4use modkit::api::OpenApiRegistry;
5use modkit::{Module, ModuleCtx, RestApiCapability};
6use tracing::{debug, info};
7
8use crate::config::FileParserConfig;
9use crate::domain::service::{FileParserService, ServiceConfig};
10use crate::infra::parsers::{
11    DocxParser, HtmlParser, ImageParser, PdfParser, PlainTextParser, PptxParser, StubParser,
12    XlsxParser,
13};
14
15/// Main module struct for file parsing
16#[modkit::module(
17    name = "file_parser",
18    capabilities = [rest]
19)]
20pub struct FileParserModule {
21    // Keep the service behind ArcSwap for cheap read-mostly access.
22    service: arc_swap::ArcSwapOption<FileParserService>,
23}
24
25impl Default for FileParserModule {
26    fn default() -> Self {
27        Self {
28            service: arc_swap::ArcSwapOption::from(None),
29        }
30    }
31}
32
33impl Clone for FileParserModule {
34    fn clone(&self) -> Self {
35        Self {
36            service: arc_swap::ArcSwapOption::new(self.service.load().as_ref().map(Clone::clone)),
37        }
38    }
39}
40
41#[async_trait]
42impl Module for FileParserModule {
43    #[allow(clippy::cast_possible_truncation)]
44    async fn init(&self, ctx: &ModuleCtx) -> anyhow::Result<()> {
45        const BYTES_IN_MB: u64 = 1024_u64 * 1024;
46
47        info!("Initializing file_parser module");
48
49        // Load module configuration
50        let cfg: FileParserConfig = ctx.config()?;
51        debug!(
52            "Loaded file_parser config: max_file_size_mb={}, download_timeout_secs={}",
53            cfg.max_file_size_mb, cfg.download_timeout_secs
54        );
55
56        // Build parser backends
57        let parsers: Vec<Arc<dyn crate::domain::parser::FileParserBackend>> = vec![
58            Arc::new(PlainTextParser::new()),
59            Arc::new(HtmlParser::new()),
60            Arc::new(PdfParser::new()),
61            Arc::new(DocxParser::new()),
62            Arc::new(XlsxParser::new()),
63            Arc::new(PptxParser::new()),
64            Arc::new(ImageParser::new()),
65            Arc::new(StubParser::new()),
66        ];
67
68        info!("Registered {} parser backends", parsers.len());
69
70        // Create service config from module config
71        let service_config = ServiceConfig {
72            max_file_size_bytes: usize::try_from(cfg.max_file_size_mb * BYTES_IN_MB)
73                .unwrap_or(usize::MAX),
74            download_timeout_secs: cfg.download_timeout_secs,
75        };
76
77        // Create file parser service
78        let file_parser_service = Arc::new(FileParserService::new(parsers, service_config));
79
80        // Store service for REST usage
81        self.service.store(Some(file_parser_service));
82
83        info!("FileParserService initialized successfully");
84        Ok(())
85    }
86}
87
88impl RestApiCapability for FileParserModule {
89    fn register_rest(
90        &self,
91        _ctx: &ModuleCtx,
92        router: axum::Router,
93        openapi: &dyn OpenApiRegistry,
94    ) -> anyhow::Result<axum::Router> {
95        info!("Registering file_parser REST routes");
96
97        let service = self
98            .service
99            .load()
100            .as_ref()
101            .ok_or_else(|| anyhow::anyhow!("Service not initialized"))?
102            .clone();
103
104        let router = crate::api::rest::routes::register_routes(router, openapi, service);
105
106        info!("File parser REST routes registered successfully");
107        Ok(router)
108    }
109}