Skip to main content

document_svg/document/
access.rs

1//! Bounded Microsoft Access ACE/Jet database header previews.
2//!
3//! ACCDB and MDB files are page-oriented database files whose first page
4//! identifies the storage engine (`Standard ACE DB` or `Standard Jet DB`).
5//! This adapter reads only that fixed header page and a bounded version-marker
6//! search.  Tables, queries, forms, attachments, VBA, linked databases, and
7//! encrypted payloads are never opened or executed.
8
9use std::fs::{self, File};
10use std::io::Read;
11use std::path::Path;
12
13use crate::convert::{ConvertOptions, PageConsumer};
14use crate::document::html::{HtmlBlock, render_blocks_to_pages};
15use crate::error::{Error, Result};
16use crate::table::{TableAlign, TableData};
17
18const MAX_ACCESS_BYTES: u64 = 2 * 1024 * 1024 * 1024;
19const ACCESS_HEADER_BYTES: usize = 4096;
20const ACE_MARKER: &[u8] = b"Standard ACE DB";
21const JET_MARKER: &[u8] = b"Standard Jet DB";
22const MSISAM_MARKER: &[u8] = b"MSISAM Database";
23
24pub(crate) fn looks_like_prefix(prefix: &[u8]) -> bool {
25    prefix.len() >= 4
26        && prefix[..4] == [0, 1, 0, 0]
27        && (prefix
28            .windows(ACE_MARKER.len())
29            .any(|window| window == ACE_MARKER)
30            || prefix
31                .windows(JET_MARKER.len())
32                .any(|window| window == JET_MARKER)
33            || prefix
34                .windows(MSISAM_MARKER.len())
35                .any(|window| window == MSISAM_MARKER))
36}
37
38struct AccessPageSink<'a> {
39    inner: &'a mut dyn PageConsumer,
40    warnings: &'a [String],
41}
42
43impl PageConsumer for AccessPageSink<'_> {
44    fn consume(&mut self, mut page: crate::ir::Page) -> Result<()> {
45        page.source_format = "access".into();
46        if page.title.is_empty() {
47            page.title = "Microsoft Access database header".into();
48        }
49        page.description = "ACE/Jet header metadata is rendered as bounded inert rows; database objects and executable content are not opened".into();
50        for warning in self.warnings {
51            page.warn(warning.clone());
52        }
53        self.inner.consume(page)
54    }
55}
56
57pub(crate) fn convert(
58    path: &Path,
59    options: &ConvertOptions,
60    sink: &mut dyn PageConsumer,
61) -> Result<Vec<String>> {
62    let metadata = fs::metadata(path)?;
63    let max_bytes = options.max_input_bytes.min(MAX_ACCESS_BYTES);
64    if metadata.len() > max_bytes {
65        return Err(Error::LimitExceeded(format!(
66            "Access input exceeds maximum bytes ({max_bytes})"
67        )));
68    }
69    let mut file = File::open(path)?;
70    let read_len = metadata.len().min(ACCESS_HEADER_BYTES as u64) as usize;
71    let mut header = vec![0u8; read_len];
72    file.read_exact(&mut header)?;
73    if !looks_like_prefix(&header) {
74        return Err(Error::InvalidInput(
75            "Access input has no recognized ACE/Jet database header".into(),
76        ));
77    }
78    let (engine, page_size) = if header
79        .windows(ACE_MARKER.len())
80        .any(|window| window == ACE_MARKER)
81    {
82        ("ACE / ACCDB", "4096 bytes (ACE default)")
83    } else if header
84        .windows(JET_MARKER.len())
85        .any(|window| window == JET_MARKER)
86    {
87        ("Jet / MDB", "2048 or 4096 bytes (Jet generation dependent)")
88    } else {
89        ("MSISAM", "engine-defined")
90    };
91    let rows = vec![
92        vec!["Storage engine".into(), engine.into()],
93        vec!["Header marker offset".into(), "4".into()],
94        vec!["Page size".into(), page_size.into()],
95        vec!["File bytes".into(), metadata.len().to_string()],
96        vec![
97            "AccessVersion marker".into(),
98            if header
99                .windows(b"AccessVersion".len())
100                .any(|window| window == b"AccessVersion")
101            {
102                "present"
103            } else {
104                "not in bounded header"
105            }
106            .into(),
107        ],
108        vec!["Header bytes read".into(), read_len.to_string()],
109    ];
110    let warnings = vec![
111        "Access tables, rows, columns, indexes, queries, forms, reports, relationships, attachments, and linked databases are not decoded".into(),
112        "VBA/data macros, OLE/ActiveX payloads, passwords, encryption, external paths, and database operations remain inert and are never bypassed or executed".into(),
113    ];
114    let blocks = vec![
115        HtmlBlock::Heading {
116            level: 1,
117            text: "Microsoft Access database header".into(),
118        },
119        HtmlBlock::Paragraph {
120            text: "Only the ACE/Jet engine marker and bounded first-page metadata are inspected. Database pages and executable content are never expanded.".into(),
121        },
122        HtmlBlock::Table(TableData {
123            headers: vec!["Metric".into(), "Value".into()],
124            rows,
125            alignments: vec![TableAlign::Left, TableAlign::Right],
126            raw_source: String::new(),
127        }),
128    ];
129    let mut page_sink = AccessPageSink {
130        inner: sink,
131        warnings: &warnings,
132    };
133    render_blocks_to_pages(&blocks, &mut page_sink, options)?;
134    Ok(warnings)
135}