Skip to main content

file_parser/
module.rs

1use std::sync::{Arc, OnceLock};
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    service: OnceLock<Arc<FileParserService>>,
22}
23
24impl Default for FileParserModule {
25    fn default() -> Self {
26        Self {
27            service: OnceLock::new(),
28        }
29    }
30}
31
32#[async_trait]
33impl Module for FileParserModule {
34    #[allow(clippy::cast_possible_truncation)]
35    async fn init(&self, ctx: &ModuleCtx) -> anyhow::Result<()> {
36        const BYTES_IN_MB: u64 = 1024_u64 * 1024;
37
38        // Load module configuration
39        let cfg: FileParserConfig = ctx.config()?;
40        debug!(
41            "Loaded file-parser config: max_file_size_mb={}",
42            cfg.max_file_size_mb
43        );
44
45        // Build parser backends
46        let parsers: Vec<Arc<dyn crate::domain::parser::FileParserBackend>> = vec![
47            Arc::new(PlainTextParser::new()),
48            Arc::new(HtmlParser::new()),
49            Arc::new(PdfParser::new()),
50            Arc::new(DocxParser::new()),
51            Arc::new(XlsxParser::new()),
52            Arc::new(PptxParser::new()),
53            Arc::new(ImageParser::new()),
54            Arc::new(StubParser::new()),
55        ];
56
57        info!("Registered {} parser backends", parsers.len());
58
59        // Canonicalize at startup so we only do it once.
60        let allowed_local_base_dir = cfg.allowed_local_base_dir.canonicalize().map_err(|e| {
61            anyhow::anyhow!(
62                "allowed_local_base_dir '{}' cannot be resolved: {e}",
63                cfg.allowed_local_base_dir.display()
64            )
65        })?;
66        if !allowed_local_base_dir.is_dir() {
67            return Err(anyhow::anyhow!(
68                "allowed_local_base_dir '{}' is not a directory",
69                allowed_local_base_dir.display()
70            ));
71        }
72        info!(
73            allowed_local_base_dir = %allowed_local_base_dir.display(),
74            "Local file parsing restricted to base directory"
75        );
76
77        // Create service config from module config
78        let service_config = ServiceConfig {
79            max_file_size_bytes: usize::try_from(cfg.max_file_size_mb * BYTES_IN_MB)
80                .unwrap_or(usize::MAX),
81            allowed_local_base_dir,
82        };
83
84        // Create file parser service
85        let file_parser_service = Arc::new(FileParserService::new(parsers, service_config));
86
87        // Store service for REST usage
88        self.service
89            .set(file_parser_service)
90            .map_err(|_| anyhow::anyhow!("{} module already initialized", Self::MODULE_NAME))?;
91
92        Ok(())
93    }
94}
95
96impl RestApiCapability for FileParserModule {
97    fn register_rest(
98        &self,
99        _ctx: &ModuleCtx,
100        router: axum::Router,
101        openapi: &dyn OpenApiRegistry,
102    ) -> anyhow::Result<axum::Router> {
103        info!("Registering file-parser REST routes");
104
105        let service = self
106            .service
107            .get()
108            .ok_or_else(|| anyhow::anyhow!("Service not initialized"))?
109            .clone();
110
111        let router = crate::api::rest::routes::register_routes(router, openapi, service);
112
113        info!("File parser REST routes registered successfully");
114        Ok(router)
115    }
116}