Skip to main content

fallow_api/runtime/
duplication.rs

1use std::time::Instant;
2
3use fallow_config::{DetectionMode, DuplicatesConfig};
4use fallow_engine::{project_config::ProjectConfig, session::AnalysisSession};
5use fallow_output::{
6    DUPES_PROGRAMMATIC_SCHEMA_VERSION, DupesNextStepsInput, DupesOutput, DupesOutputInput,
7    build_dupes_next_steps, build_dupes_output, dupes_meta,
8};
9use fallow_types::output_format::OutputFormat;
10use rustc_hash::FxHashSet;
11
12use crate::{
13    DupesReportPayload, DuplicationGroup, DuplicationMode, DuplicationOptions,
14    DuplicationProgrammaticOutput, ProgrammaticError,
15    analysis_context::{
16        ProgrammaticAnalysisContext, changed_files_for_run,
17        resolve_programmatic_analysis_context_deferred_workspace, workspace_roots_for_session,
18    },
19    duplication_filters::{apply_top, filter_by_diff, filter_by_workspaces},
20    next_steps::{setup_pointer_applicable, suggestions_enabled},
21};
22
23use super::{ProgrammaticResult, root_envelope_mode};
24
25/// Run duplication analysis and return typed API output before serialization.
26///
27/// # Errors
28///
29/// Returns a structured programmatic error for invalid options, config load
30/// failures, or git changed-file failures.
31pub fn run_duplication(
32    options: &DuplicationOptions,
33) -> ProgrammaticResult<DuplicationProgrammaticOutput> {
34    let resolved = resolve_programmatic_analysis_context_deferred_workspace(&options.analysis)?;
35    resolved.install(|| run_duplication_inner(options, &resolved))
36}
37
38fn run_duplication_inner(
39    options: &DuplicationOptions,
40    resolved: &ProgrammaticAnalysisContext,
41) -> ProgrammaticResult<DuplicationProgrammaticOutput> {
42    let start = Instant::now();
43    resolved.ensure_not_cancelled("config load and file discovery")?;
44    let session = load_duplication_session(options, resolved)?;
45    run_duplication_with_session(options, resolved, &session, None, start)
46}
47
48pub(super) fn run_duplication_with_session(
49    options: &DuplicationOptions,
50    resolved: &ProgrammaticAnalysisContext,
51    session: &AnalysisSession,
52    changed_files: Option<&FxHashSet<std::path::PathBuf>>,
53    start: Instant,
54) -> ProgrammaticResult<DuplicationProgrammaticOutput> {
55    resolved.ensure_not_cancelled("duplication detection")?;
56    let dupes_config = build_dupes_config(options, &session.config().duplicates);
57    let resolved_changed_files = if changed_files.is_some() {
58        None
59    } else {
60        changed_files_for_run(resolved)?
61    };
62    let cache_dir = (!resolved.no_cache).then_some(session.config().cache_dir.as_path());
63    let report = if let Some(changed_files) = changed_files.or(resolved_changed_files.as_ref()) {
64        let changed_files = changed_files.iter().cloned().collect::<Vec<_>>();
65        session
66            .find_duplicates_touching_files_with_defaults(&dupes_config, &changed_files, cache_dir)
67            .report
68    } else {
69        session
70            .find_duplicates_with_defaults(&dupes_config, cache_dir)
71            .report
72    };
73
74    // Duplication detection cannot fail, so a token set while it ran has to be
75    // reported here rather than dressed up as a complete report.
76    resolved.ensure_not_cancelled("the duplication report")?;
77    run_duplication_report_with_session(options, resolved, session, report, start)
78}
79
80pub(super) fn run_duplication_report_with_session(
81    options: &DuplicationOptions,
82    resolved: &ProgrammaticAnalysisContext,
83    session: &AnalysisSession,
84    mut report: fallow_engine::duplicates::DuplicationReport,
85    start: Instant,
86) -> ProgrammaticResult<DuplicationProgrammaticOutput> {
87    let dupes_config = build_dupes_config(options, &session.config().duplicates);
88    if let Some(diff) = resolved.diff.as_ref() {
89        filter_by_diff(&mut report, diff, session.root());
90    }
91    let workspace_roots = workspace_roots_for_session(resolved, session.workspaces())?;
92    if let Some(workspace_roots) = workspace_roots.as_ref() {
93        filter_by_workspaces(&mut report, workspace_roots, session.root());
94    }
95    if let Some(top) = options.top {
96        apply_top(&mut report, top, session.root());
97    }
98
99    let root = session.root();
100    let payload = DupesReportPayload::from_report(&report);
101    let clone_fingerprints = payload
102        .clone_groups
103        .iter()
104        .map(|group| group.fingerprint.as_str())
105        .collect::<Vec<_>>();
106    let next_steps = build_dupes_next_steps(DupesNextStepsInput {
107        suggestions_enabled: suggestions_enabled(),
108        clone_fingerprints: &clone_fingerprints,
109        offer_setup: setup_pointer_applicable(root),
110        impact_digest: None,
111        audit_changed: fallow_engine::churn::is_git_repo(root),
112    });
113    let output: DupesOutput<DupesReportPayload, DuplicationGroup> =
114        build_dupes_output(DupesOutputInput {
115            schema_version: DUPES_PROGRAMMATIC_SCHEMA_VERSION,
116            version: env!("CARGO_PKG_VERSION").to_string(),
117            elapsed: start.elapsed(),
118            report: payload,
119            grouped_by: None,
120            total_issues: None,
121            groups: None,
122            meta: resolved.explain_enabled().then(dupes_meta),
123            workspace_diagnostics: session.workspace_diagnostics().to_vec(),
124            next_steps,
125        });
126    Ok(DuplicationProgrammaticOutput {
127        output,
128        root: session.root().to_path_buf(),
129        threshold: dupes_config.threshold,
130        envelope_mode: root_envelope_mode(),
131        telemetry_analysis_run_id: None,
132    })
133}
134
135pub(super) fn load_duplication_session(
136    options: &DuplicationOptions,
137    resolved: &ProgrammaticAnalysisContext,
138) -> ProgrammaticResult<AnalysisSession> {
139    let project_config = fallow_engine::project_config::config_for_project_with_load_options(
140        &resolved.root,
141        resolved.config_path.as_deref(),
142        fallow_config::ConfigLoadOptions {
143            allow_remote_extends: resolved.allow_remote_extends(),
144        },
145    )
146    .map_err(|err| {
147        ProgrammaticError::new(format!("failed to load config: {err}"), 2)
148            .with_code("FALLOW_CONFIG_LOAD_FAILED")
149            .with_context("analysis.configPath")
150    })?;
151    let project_config = configure_project_for_duplication(project_config, options, resolved);
152    Ok(super::dead_code::attach_cancellation(
153        AnalysisSession::from_config(project_config),
154        resolved,
155    ))
156}
157
158fn configure_project_for_duplication(
159    mut project_config: ProjectConfig,
160    options: &DuplicationOptions,
161    resolved: &ProgrammaticAnalysisContext,
162) -> ProjectConfig {
163    let production = resolved
164        .production_override
165        .unwrap_or(project_config.config.production);
166    project_config.config.production = production;
167    project_config.config.output = OutputFormat::Json;
168    project_config.config.threads = resolved.threads;
169    project_config.config.no_cache = resolved.no_cache;
170    project_config.config.duplicates =
171        build_dupes_config(options, &project_config.config.duplicates);
172    project_config
173}
174
175pub(super) fn build_dupes_config(
176    options: &DuplicationOptions,
177    config: &DuplicatesConfig,
178) -> DuplicatesConfig {
179    DuplicatesConfig {
180        enabled: true,
181        mode: options.mode.map_or(config.mode, duplication_mode_to_config),
182        near: options.near.unwrap_or(config.near),
183        min_tokens: options.min_tokens.unwrap_or(config.min_tokens),
184        min_lines: options.min_lines.unwrap_or(config.min_lines),
185        min_occurrences: options.min_occurrences.unwrap_or(config.min_occurrences),
186        threshold: options.threshold.unwrap_or(config.threshold),
187        ignore: config.ignore.clone(),
188        ignored_clones: config.ignored_clones.clone(),
189        ignore_defaults: config.ignore_defaults,
190        skip_local: options.skip_local.unwrap_or(config.skip_local),
191        cross_language: options.cross_language.unwrap_or(config.cross_language),
192        ignore_imports: options.ignore_imports.unwrap_or(config.ignore_imports),
193        normalization: config.normalization.clone(),
194        min_corpus_size_for_shingle_filter: config.min_corpus_size_for_shingle_filter,
195        min_corpus_size_for_token_cache: config.min_corpus_size_for_token_cache,
196    }
197}
198
199const fn duplication_mode_to_config(mode: DuplicationMode) -> DetectionMode {
200    match mode {
201        DuplicationMode::Strict => DetectionMode::Strict,
202        DuplicationMode::Mild => DetectionMode::Mild,
203        DuplicationMode::Weak => DetectionMode::Weak,
204        DuplicationMode::Semantic => DetectionMode::Semantic,
205    }
206}
207
208#[cfg(test)]
209mod tests {
210    use super::*;
211
212    #[test]
213    fn duplication_options_override_near_and_preserve_reviewed_clones() {
214        let config = DuplicatesConfig {
215            near: true,
216            ignored_clones: vec!["dup:12345678:2".to_string()],
217            ..DuplicatesConfig::default()
218        };
219        let options = DuplicationOptions {
220            near: Some(false),
221            ..DuplicationOptions::default()
222        };
223
224        let merged = build_dupes_config(&options, &config);
225
226        assert!(!merged.near);
227        assert_eq!(merged.ignored_clones, config.ignored_clones);
228    }
229}