provenant/parsers/
microsoft_update_manifest.rs

1// SPDX-FileCopyrightText: Provenant contributors
2// SPDX-License-Identifier: Apache-2.0
3
4//! Parser for Microsoft Update Manifest (.mum) files.
5//!
6//! Extracts Windows Update package metadata from .mum XML manifest files.
7//!
8//! # Supported Formats
9//! - `*.mum` - Microsoft Update Manifest XML files
10//!
11//! # Implementation Notes
12//! - Format: XML with assembly and package metadata
13//! - Spec: Windows Update manifests
14
15use crate::models::{DatasourceId, PackageType};
16use std::path::Path;
17
18use crate::parser_warn as warn;
19use quick_xml::events::Event;
20use quick_xml::reader::Reader;
21
22use crate::models::PackageData;
23use crate::parsers::utils::{MAX_ITERATION_COUNT, read_file_to_string, truncate_field};
24
25use super::PackageParser;
26
27const PACKAGE_TYPE: PackageType = PackageType::WindowsUpdate;
28
29pub struct MicrosoftUpdateManifestParser;
30
31impl PackageParser for MicrosoftUpdateManifestParser {
32    const PACKAGE_TYPE: PackageType = PACKAGE_TYPE;
33
34    fn is_match(path: &Path) -> bool {
35        path.extension().is_some_and(|ext| ext == "mum")
36    }
37
38    fn extract_packages(path: &Path) -> Vec<PackageData> {
39        let content = match read_file_to_string(path, None) {
40            Ok(c) => c,
41            Err(e) => {
42                warn!("Failed to read .mum file {:?}: {}", path, e);
43                return vec![PackageData {
44                    package_type: Some(PACKAGE_TYPE),
45                    datasource_id: Some(DatasourceId::MicrosoftUpdateManifestMum),
46                    ..Default::default()
47                }];
48            }
49        };
50
51        vec![parse_mum_xml(&content)]
52    }
53}
54
55pub(crate) fn parse_mum_xml(content: &str) -> PackageData {
56    let mut reader = Reader::from_str(content);
57    reader.config_mut().trim_text(true);
58
59    let mut name = None;
60    let mut version = None;
61    let mut description = None;
62    let mut copyright = None;
63    let mut homepage_url = None;
64
65    let mut buf = Vec::new();
66    let mut iteration_count: usize = 0;
67
68    loop {
69        iteration_count += 1;
70        if iteration_count > MAX_ITERATION_COUNT {
71            warn!(
72                "Exceeded MAX_ITERATION_COUNT ({}) parsing .mum XML, stopping",
73                MAX_ITERATION_COUNT
74            );
75            break;
76        }
77        match reader.read_event_into(&mut buf) {
78            Ok(Event::Empty(e)) => {
79                if e.name().as_ref() == b"assemblyIdentity" {
80                    for attr in e.attributes().filter_map(|a| a.ok()) {
81                        match attr.key.as_ref() {
82                            b"name" => {
83                                let raw = attr.value.to_vec();
84                                let has_invalid = String::from_utf8(raw.clone()).is_err();
85                                let val = String::from_utf8_lossy(&raw).into_owned();
86                                if has_invalid {
87                                    warn!(
88                                        "Invalid UTF-8 in 'name' attribute, using lossy conversion"
89                                    );
90                                }
91                                name = Some(truncate_field(val));
92                            }
93                            b"version" => {
94                                let raw = attr.value.to_vec();
95                                let has_invalid = String::from_utf8(raw.clone()).is_err();
96                                let val = String::from_utf8_lossy(&raw).into_owned();
97                                if has_invalid {
98                                    warn!(
99                                        "Invalid UTF-8 in 'version' attribute, using lossy conversion"
100                                    );
101                                }
102                                version = Some(truncate_field(val));
103                            }
104                            _ => {}
105                        }
106                    }
107                }
108            }
109            Ok(Event::Start(e)) => {
110                if e.name().as_ref() == b"assembly" {
111                    for attr in e.attributes().filter_map(|a| a.ok()) {
112                        match attr.key.as_ref() {
113                            b"description" => {
114                                let raw = attr.value.to_vec();
115                                let has_invalid = String::from_utf8(raw.clone()).is_err();
116                                let val = String::from_utf8_lossy(&raw).into_owned();
117                                if has_invalid {
118                                    warn!(
119                                        "Invalid UTF-8 in 'description' attribute, using lossy conversion"
120                                    );
121                                }
122                                description = Some(truncate_field(val));
123                            }
124                            b"copyright" => {
125                                let raw = attr.value.to_vec();
126                                let has_invalid = String::from_utf8(raw.clone()).is_err();
127                                let val = String::from_utf8_lossy(&raw).into_owned();
128                                if has_invalid {
129                                    warn!(
130                                        "Invalid UTF-8 in 'copyright' attribute, using lossy conversion"
131                                    );
132                                }
133                                copyright = Some(truncate_field(val));
134                            }
135                            b"supportInformation" => {
136                                let raw = attr.value.to_vec();
137                                let has_invalid = String::from_utf8(raw.clone()).is_err();
138                                let val = String::from_utf8_lossy(&raw).into_owned();
139                                if has_invalid {
140                                    warn!(
141                                        "Invalid UTF-8 in 'supportInformation' attribute, using lossy conversion"
142                                    );
143                                }
144                                homepage_url = Some(truncate_field(val));
145                            }
146                            _ => {}
147                        }
148                    }
149                }
150            }
151            Ok(Event::Eof) => break,
152            Err(e) => {
153                warn!(
154                    "Error parsing XML at position {}: {}",
155                    reader.buffer_position(),
156                    e
157                );
158                break;
159            }
160            _ => {}
161        }
162        buf.clear();
163    }
164
165    PackageData {
166        package_type: Some(PACKAGE_TYPE),
167        name,
168        version,
169        description,
170        homepage_url,
171        copyright,
172        datasource_id: Some(DatasourceId::MicrosoftUpdateManifestMum),
173        ..Default::default()
174    }
175}
176
177crate::register_parser!(
178    "Microsoft Update Manifest .mum file",
179    &["*.mum"],
180    "windows-update",
181    "",
182    None,
183);
provenant/parsers/microsoft_update_manifest.rs

provenant/parsers/
microsoft_update_manifest.rs
