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    let session = load_duplication_session(options, resolved)?;
44    run_duplication_with_session(options, resolved, &session, None, start)
45}
46
47pub(super) fn run_duplication_with_session(
48    options: &DuplicationOptions,
49    resolved: &ProgrammaticAnalysisContext,
50    session: &AnalysisSession,
51    changed_files: Option<&FxHashSet<std::path::PathBuf>>,
52    start: Instant,
53) -> ProgrammaticResult<DuplicationProgrammaticOutput> {
54    let dupes_config = build_dupes_config(options, &session.config().duplicates);
55    let resolved_changed_files = if changed_files.is_some() {
56        None
57    } else {
58        changed_files_for_run(resolved)?
59    };
60    let cache_dir = (!resolved.no_cache).then_some(session.config().cache_dir.as_path());
61    let report = if let Some(changed_files) = changed_files.or(resolved_changed_files.as_ref()) {
62        let changed_files = changed_files.iter().cloned().collect::<Vec<_>>();
63        session
64            .find_duplicates_touching_files_with_defaults(&dupes_config, &changed_files, cache_dir)
65            .report
66    } else {
67        session
68            .find_duplicates_with_defaults(&dupes_config, cache_dir)
69            .report
70    };
71
72    run_duplication_report_with_session(options, resolved, session, report, start)
73}
74
75pub(super) fn run_duplication_report_with_session(
76    options: &DuplicationOptions,
77    resolved: &ProgrammaticAnalysisContext,
78    session: &AnalysisSession,
79    mut report: fallow_engine::duplicates::DuplicationReport,
80    start: Instant,
81) -> ProgrammaticResult<DuplicationProgrammaticOutput> {
82    let dupes_config = build_dupes_config(options, &session.config().duplicates);
83    if let Some(diff) = resolved.diff.as_ref() {
84        filter_by_diff(&mut report, diff, session.root());
85    }
86    let workspace_roots = workspace_roots_for_session(resolved, session.workspaces())?;
87    if let Some(workspace_roots) = workspace_roots.as_ref() {
88        filter_by_workspaces(&mut report, workspace_roots, session.root());
89    }
90    if let Some(top) = options.top {
91        apply_top(&mut report, top, session.root());
92    }
93
94    let root = session.root();
95    let payload = DupesReportPayload::from_report(&report);
96    let clone_fingerprints = payload
97        .clone_groups
98        .iter()
99        .map(|group| group.fingerprint.as_str())
100        .collect::<Vec<_>>();
101    let next_steps = build_dupes_next_steps(DupesNextStepsInput {
102        suggestions_enabled: suggestions_enabled(),
103        clone_fingerprints: &clone_fingerprints,
104        offer_setup: setup_pointer_applicable(root),
105        impact_digest: None,
106        audit_changed: fallow_engine::churn::is_git_repo(root),
107    });
108    let output: DupesOutput<DupesReportPayload, DuplicationGroup> =
109        build_dupes_output(DupesOutputInput {
110            schema_version: DUPES_PROGRAMMATIC_SCHEMA_VERSION,
111            version: env!("CARGO_PKG_VERSION").to_string(),
112            elapsed: start.elapsed(),
113            report: payload,
114            grouped_by: None,
115            total_issues: None,
116            groups: None,
117            meta: resolved.explain_enabled().then(dupes_meta),
118            workspace_diagnostics: session.current_workspace_diagnostics(),
119            next_steps,
120        });
121    Ok(DuplicationProgrammaticOutput {
122        output,
123        root: session.root().to_path_buf(),
124        threshold: dupes_config.threshold,
125        envelope_mode: root_envelope_mode(),
126        telemetry_analysis_run_id: None,
127    })
128}
129
130pub(super) fn load_duplication_session(
131    options: &DuplicationOptions,
132    resolved: &ProgrammaticAnalysisContext,
133) -> ProgrammaticResult<AnalysisSession> {
134    let project_config = fallow_engine::project_config::config_for_project_with_load_options(
135        &resolved.root,
136        resolved.config_path.as_deref(),
137        fallow_config::ConfigLoadOptions {
138            allow_remote_extends: resolved.allow_remote_extends(),
139        },
140    )
141    .map_err(|err| {
142        ProgrammaticError::new(format!("failed to load config: {err}"), 2)
143            .with_code("FALLOW_CONFIG_LOAD_FAILED")
144            .with_context("analysis.configPath")
145    })?;
146    let project_config = configure_project_for_duplication(project_config, options, resolved);
147    Ok(AnalysisSession::from_config(project_config))
148}
149
150fn configure_project_for_duplication(
151    mut project_config: ProjectConfig,
152    options: &DuplicationOptions,
153    resolved: &ProgrammaticAnalysisContext,
154) -> ProjectConfig {
155    let production = resolved
156        .production_override
157        .unwrap_or(project_config.config.production);
158    project_config.config.production = production;
159    project_config.config.output = OutputFormat::Json;
160    project_config.config.threads = resolved.threads;
161    project_config.config.no_cache = resolved.no_cache;
162    project_config.config.duplicates =
163        build_dupes_config(options, &project_config.config.duplicates);
164    project_config
165}
166
167pub(super) fn build_dupes_config(
168    options: &DuplicationOptions,
169    config: &DuplicatesConfig,
170) -> DuplicatesConfig {
171    DuplicatesConfig {
172        enabled: true,
173        mode: options.mode.map_or(config.mode, duplication_mode_to_config),
174        near: options.near.unwrap_or(config.near),
175        min_tokens: options.min_tokens.unwrap_or(config.min_tokens),
176        min_lines: options.min_lines.unwrap_or(config.min_lines),
177        min_occurrences: options.min_occurrences.unwrap_or(config.min_occurrences),
178        threshold: options.threshold.unwrap_or(config.threshold),
179        ignore: config.ignore.clone(),
180        ignored_clones: config.ignored_clones.clone(),
181        ignore_defaults: config.ignore_defaults,
182        skip_local: options.skip_local.unwrap_or(config.skip_local),
183        cross_language: options.cross_language.unwrap_or(config.cross_language),
184        ignore_imports: options.ignore_imports.unwrap_or(config.ignore_imports),
185        normalization: config.normalization.clone(),
186        min_corpus_size_for_shingle_filter: config.min_corpus_size_for_shingle_filter,
187        min_corpus_size_for_token_cache: config.min_corpus_size_for_token_cache,
188    }
189}
190
191const fn duplication_mode_to_config(mode: DuplicationMode) -> DetectionMode {
192    match mode {
193        DuplicationMode::Strict => DetectionMode::Strict,
194        DuplicationMode::Mild => DetectionMode::Mild,
195        DuplicationMode::Weak => DetectionMode::Weak,
196        DuplicationMode::Semantic => DetectionMode::Semantic,
197    }
198}
199
200#[cfg(test)]
201mod tests {
202    use super::*;
203
204    #[test]
205    fn duplication_options_override_near_and_preserve_reviewed_clones() {
206        let config = DuplicatesConfig {
207            near: true,
208            ignored_clones: vec!["dup:12345678:2".to_string()],
209            ..DuplicatesConfig::default()
210        };
211        let options = DuplicationOptions {
212            near: Some(false),
213            ..DuplicationOptions::default()
214        };
215
216        let merged = build_dupes_config(&options, &config);
217
218        assert!(!merged.near);
219        assert_eq!(merged.ignored_clones, config.ignored_clones);
220    }
221}