Skip to main content

provenant/parsers/
hex_lock.rs

1// SPDX-FileCopyrightText: Provenant contributors
2// SPDX-License-Identifier: Apache-2.0
3
4use std::collections::HashMap;
5use std::path::Path;
6
7use crate::parser_warn as warn;
8use crate::parsers::utils::{
9    MAX_ITERATION_COUNT, RecursionGuard, capped_iteration_limit, read_file_to_string,
10    truncate_field,
11};
12use packageurl::PackageUrl;
13use serde_json::Value as JsonValue;
14
15use crate::models::{
16    DatasourceId, Dependency, PackageData, PackageType, ResolvedPackage, Sha256Digest,
17};
18
19use super::PackageParser;
20use super::metadata::ParserMetadata;
21
22pub struct HexLockParser;
23
24#[derive(Clone, Debug)]
25enum Term {
26    Map(Vec<(Term, Term)>),
27    Tuple(Vec<Term>),
28    List(Vec<Term>),
29    KeywordList(Vec<(String, Term)>),
30    String(String),
31    Atom(String),
32    Bool(bool),
33    Integer(i64),
34}
35
36struct Parser<'a> {
37    chars: Vec<char>,
38    pos: usize,
39    source: &'a str,
40    guard: RecursionGuard<()>,
41}
42
43impl PackageParser for HexLockParser {
44    const PACKAGE_TYPE: PackageType = PackageType::Hex;
45
46    fn metadata() -> Vec<ParserMetadata> {
47        vec![ParserMetadata {
48            description: "Hex mix.lock lockfile",
49            file_patterns: &["**/mix.lock"],
50            package_type: "hex",
51            primary_language: "Elixir",
52            documentation_url: Some("https://hexdocs.pm/mix/Mix.Tasks.Deps.html"),
53        }]
54    }
55
56    fn is_match(path: &Path) -> bool {
57        path.file_name().and_then(|name| name.to_str()) == Some("mix.lock")
58    }
59
60    fn extract_packages(path: &Path) -> Vec<PackageData> {
61        let content = match read_file_to_string(path, None) {
62            Ok(content) => content,
63            Err(e) => {
64                warn!("Failed to read mix.lock at {:?}: {}", path, e);
65                return vec![default_package_data()];
66            }
67        };
68
69        match parse_mix_lock(&content) {
70            Ok(package_data) => vec![package_data],
71            Err(e) => {
72                warn!("Failed to parse mix.lock at {:?}: {}", path, e);
73                vec![default_package_data()]
74            }
75        }
76    }
77}
78
79fn default_package_data() -> PackageData {
80    PackageData {
81        package_type: Some(PackageType::Hex),
82        primary_language: Some("Elixir".to_string()),
83        datasource_id: Some(DatasourceId::HexMixLock),
84        ..Default::default()
85    }
86}
87
88fn parse_mix_lock(content: &str) -> Result<PackageData, String> {
89    let mut parser = Parser::new(content);
90    let term = parser.parse_term()?;
91    parser.skip_ws();
92    if !parser.is_eof() {
93        return Err("Unexpected trailing content in mix.lock".to_string());
94    }
95
96    let entries = match term {
97        Term::Map(entries) => entries,
98        _ => return Err("mix.lock root must be a map".to_string()),
99    };
100
101    let mut dependencies = Vec::new();
102    let limit = capped_iteration_limit(entries.len(), "mix.lock entries");
103    for (key, value) in entries.into_iter().take(limit) {
104        if let Some(dep) = build_dependency_from_lock_entry(&key, &value)? {
105            dependencies.push(dep);
106        }
107    }
108
109    let mut package = default_package_data();
110    package.dependencies = dependencies;
111    Ok(package)
112}
113
114fn build_dependency_from_lock_entry(
115    key: &Term,
116    value: &Term,
117) -> Result<Option<Dependency>, String> {
118    let app_name = truncate_field(term_to_string(key)?);
119
120    let tuple = match value {
121        Term::Tuple(items) => items,
122        _ => return Ok(None),
123    };
124
125    // hex dependency tuples have grown over time; accept all historical shapes rather than
126    // dropping older lockfiles:
127    //   6 elements (legacy): {:hex, name, version, inner_checksum, managers, deps}
128    //   7 elements: + repo
129    //   8 elements (modern, Hex >= 0.20 / Elixir >= 1.7): + outer_checksum
130    if tuple.len() < 6 {
131        return Ok(None);
132    }
133
134    let kind = term_to_atom(&tuple[0])?;
135    if kind != "hex" {
136        return Ok(None);
137    }
138
139    let package_name = truncate_field(term_to_atom(&tuple[1])?);
140    let version = truncate_field(term_to_string(&tuple[2])?);
141    let inner_checksum = truncate_field(term_to_string(&tuple[3])?);
142    let managers = term_to_atom_list(&tuple[4])?;
143    let nested_dependencies = term_to_dependency_tuples(&tuple[5])?;
144    // repo (element 7) was added later; default to the public "hexpm" repository when absent.
145    let repo = match tuple.get(6) {
146        Some(term) => truncate_field(term_to_string(term)?),
147        None => "hexpm".to_string(),
148    };
149    // outer_checksum (element 8) is only present in modern lockfiles.
150    let outer_checksum = match tuple.get(7) {
151        Some(term) => Some(truncate_field(term_to_string(term)?)),
152        None => None,
153    };
154
155    let purl = build_hex_purl(&package_name, Some(&version), Some(&repo));
156    let resolved_package = ResolvedPackage {
157        primary_language: Some("Elixir".to_string()),
158        download_url: None,
159        sha1: None,
160        sha256: Sha256Digest::from_hex(&inner_checksum).ok(),
161        sha512: None,
162        md5: None,
163        is_virtual: true,
164        extra_data: Some({
165            let mut extra = HashMap::from([
166                (
167                    "repo".to_string(),
168                    JsonValue::String(truncate_field(repo.clone())),
169                ),
170                (
171                    "managers".to_string(),
172                    JsonValue::Array(
173                        managers
174                            .into_iter()
175                            .map(|m| JsonValue::String(truncate_field(m)))
176                            .collect(),
177                    ),
178                ),
179            ]);
180            if let Some(ref outer) = outer_checksum {
181                extra.insert(
182                    "outer_checksum".to_string(),
183                    JsonValue::String(truncate_field(outer.clone())),
184                );
185            }
186            extra
187        }),
188        dependencies: nested_dependencies
189            .into_iter()
190            .map(build_nested_dependency)
191            .collect::<Result<Vec<_>, _>>()?,
192        repository_homepage_url: Some(truncate_field(build_hexdocs_homepage(&package_name, &repo))),
193        repository_download_url: None,
194        api_data_url: Some(truncate_field(build_hex_api_url(&package_name, &repo))),
195        datasource_id: Some(DatasourceId::HexMixLock),
196        purl: build_hex_purl(&package_name, Some(&version), Some(&repo)).map(truncate_field),
197        ..ResolvedPackage::new(
198            PackageType::Hex,
199            if repo == "hexpm" {
200                String::new()
201            } else {
202                repo.clone()
203            },
204            package_name.clone(),
205            version.clone(),
206        )
207    };
208
209    Ok(Some(Dependency {
210        purl: purl.map(truncate_field),
211        extracted_requirement: Some(truncate_field(version)),
212        scope: Some("dependencies".to_string()),
213        is_runtime: None,
214        is_optional: None,
215        is_pinned: Some(true),
216        is_direct: None,
217        resolved_package: Some(Box::new(resolved_package)),
218        extra_data: Some(HashMap::from([(
219            "app".to_string(),
220            JsonValue::String(truncate_field(app_name)),
221        )])),
222    }))
223}
224
225fn build_nested_dependency(tuple: DependencyTuple) -> Result<Dependency, String> {
226    let package_name = truncate_field(
227        tuple
228            .hex_name
229            .clone()
230            .unwrap_or_else(|| tuple.app_name.clone()),
231    );
232    Ok(Dependency {
233        purl: build_hex_purl(&package_name, None, tuple.repo.as_deref()).map(truncate_field),
234        extracted_requirement: Some(truncate_field(tuple.requirement)),
235        scope: Some("dependencies".to_string()),
236        is_runtime: Some(!tuple.optional),
237        is_optional: Some(tuple.optional),
238        is_pinned: Some(false),
239        is_direct: Some(true),
240        resolved_package: None,
241        extra_data: None,
242    })
243}
244
245#[derive(Debug)]
246struct DependencyTuple {
247    app_name: String,
248    requirement: String,
249    hex_name: Option<String>,
250    repo: Option<String>,
251    optional: bool,
252}
253
254fn term_to_dependency_tuples(term: &Term) -> Result<Vec<DependencyTuple>, String> {
255    let items = match term {
256        Term::List(items) => items,
257        _ => return Ok(Vec::new()),
258    };
259
260    let mut result = Vec::new();
261    let limit = capped_iteration_limit(items.len(), "mix.lock dependency tuples");
262    for item in items.iter().take(limit) {
263        let tuple = match item {
264            Term::Tuple(items) if items.len() == 3 => items,
265            _ => continue,
266        };
267
268        let app_name = truncate_field(term_to_atom(&tuple[0])?);
269        let requirement = truncate_field(term_to_string(&tuple[1])?);
270        let opts = term_to_keyword_map(&tuple[2])?;
271        let hex_name = opts
272            .get("hex")
273            .map(term_to_atom)
274            .transpose()?
275            .map(truncate_field);
276        let repo = opts
277            .get("repo")
278            .map(term_to_string)
279            .transpose()?
280            .map(truncate_field);
281        let optional = opts
282            .get("optional")
283            .and_then(|term| match term {
284                Term::Bool(value) => Some(*value),
285                _ => None,
286            })
287            .unwrap_or(false);
288
289        result.push(DependencyTuple {
290            app_name,
291            requirement,
292            hex_name,
293            repo,
294            optional,
295        });
296    }
297
298    Ok(result)
299}
300
301fn term_to_keyword_map(term: &Term) -> Result<HashMap<String, Term>, String> {
302    match term {
303        Term::KeywordList(entries) => Ok(entries.iter().cloned().collect()),
304        Term::List(entries) => {
305            let mut map = HashMap::new();
306            for entry in entries {
307                if let Term::Tuple(items) = entry
308                    && items.len() == 2
309                {
310                    map.insert(term_to_atom(&items[0])?, items[1].clone());
311                }
312            }
313            Ok(map)
314        }
315        _ => Ok(HashMap::new()),
316    }
317}
318
319fn build_hex_purl(name: &str, version: Option<&str>, repo: Option<&str>) -> Option<String> {
320    let mut purl = PackageUrl::new("hex", name).ok()?;
321    if let Some(repo) = repo
322        && repo != "hexpm"
323    {
324        purl.with_namespace(repo).ok()?;
325    }
326    if let Some(version) = version {
327        purl.with_version(version).ok()?;
328    }
329    Some(purl.to_string())
330}
331
332fn build_hexdocs_homepage(name: &str, repo: &str) -> String {
333    if repo == "hexpm" {
334        format!("https://hex.pm/packages/{}", name)
335    } else {
336        format!("https://hex.pm/packages/{}?repo={}", name, repo)
337    }
338}
339
340fn build_hex_api_url(name: &str, repo: &str) -> String {
341    if repo == "hexpm" {
342        format!("https://hex.pm/api/packages/{}", name)
343    } else {
344        format!("https://hex.pm/api/repos/{}/packages/{}", repo, name)
345    }
346}
347
348fn term_to_string(term: &Term) -> Result<String, String> {
349    match term {
350        Term::String(value) => Ok(value.clone()),
351        Term::Atom(value) => Ok(value.clone()),
352        Term::Integer(value) => Ok(value.to_string()),
353        _ => Err("Expected string-like term".to_string()),
354    }
355}
356
357fn term_to_atom(term: &Term) -> Result<String, String> {
358    match term {
359        Term::Atom(value) => Ok(value.clone()),
360        _ => Err("Expected atom".to_string()),
361    }
362}
363
364fn term_to_atom_list(term: &Term) -> Result<Vec<String>, String> {
365    let items = match term {
366        Term::List(items) => items,
367        _ => return Ok(Vec::new()),
368    };
369    items.iter().map(term_to_atom).collect()
370}
371
372impl<'a> Parser<'a> {
373    fn new(source: &'a str) -> Self {
374        Self {
375            chars: source.chars().collect(),
376            pos: 0,
377            source,
378            guard: RecursionGuard::depth_only(),
379        }
380    }
381
382    fn parse_term(&mut self) -> Result<Term, String> {
383        if self.guard.descend() {
384            return Err("recursion depth exceeded".to_string());
385        }
386        self.skip_ws();
387        let result = match self.peek() {
388            Some('%') => self.parse_map(),
389            Some('{') => self.parse_tuple(),
390            Some('[') => self.parse_list(),
391            Some('"') => self.parse_string().map(Term::String),
392            Some(':') => self.parse_atom().map(Term::Atom),
393            Some(c) if c.is_ascii_digit() || c == '-' => self.parse_integer().map(Term::Integer),
394            Some('t') | Some('f') => self.parse_bool().map(Term::Bool),
395            Some(other) => Err(format!("Unexpected character '{}' at {}", other, self.pos)),
396            None => Err("Unexpected end of mix.lock".to_string()),
397        };
398        self.guard.ascend();
399        result
400    }
401
402    fn parse_map(&mut self) -> Result<Term, String> {
403        self.expect('%')?;
404        self.expect('{')?;
405        let mut entries = Vec::new();
406        let mut count = 0usize;
407        loop {
408            self.skip_ws();
409            if self.peek() == Some('}') {
410                self.pos += 1;
411                break;
412            }
413            if count >= MAX_ITERATION_COUNT {
414                warn!("map entry count exceeded MAX_ITERATION_COUNT in mix.lock");
415                break;
416            }
417            let key = self.parse_term()?;
418            self.skip_ws();
419            if self.starts_with("=>") {
420                self.expect_sequence("=>")?;
421            } else {
422                self.expect(':')?;
423            }
424            let value = self.parse_term()?;
425            entries.push((key, value));
426            count += 1;
427            self.skip_ws();
428            if self.peek() == Some(',') {
429                self.pos += 1;
430            }
431        }
432        Ok(Term::Map(entries))
433    }
434
435    fn parse_tuple(&mut self) -> Result<Term, String> {
436        self.expect('{')?;
437        let mut items = Vec::new();
438        let mut count = 0usize;
439        loop {
440            self.skip_ws();
441            if self.peek() == Some('}') {
442                self.pos += 1;
443                break;
444            }
445            if count >= MAX_ITERATION_COUNT {
446                warn!("tuple item count exceeded MAX_ITERATION_COUNT in mix.lock");
447                break;
448            }
449            items.push(self.parse_term()?);
450            count += 1;
451            self.skip_ws();
452            if self.peek() == Some(',') {
453                self.pos += 1;
454            }
455        }
456        Ok(Term::Tuple(items))
457    }
458
459    fn parse_list(&mut self) -> Result<Term, String> {
460        self.expect('[')?;
461        let mut keyword_entries = Vec::new();
462        let mut items = Vec::new();
463        let mut saw_keyword = false;
464        let mut count = 0usize;
465
466        loop {
467            self.skip_ws();
468            if self.peek() == Some(']') {
469                self.pos += 1;
470                break;
471            }
472            if count >= MAX_ITERATION_COUNT {
473                warn!("list item count exceeded MAX_ITERATION_COUNT in mix.lock");
474                break;
475            }
476
477            if let Some(keyword) = self.try_parse_keyword_key() {
478                saw_keyword = true;
479                let value = self.parse_term()?;
480                keyword_entries.push((keyword, value));
481            } else {
482                items.push(self.parse_term()?);
483            }
484
485            count += 1;
486            self.skip_ws();
487            if self.peek() == Some(',') {
488                self.pos += 1;
489            }
490        }
491
492        if saw_keyword && items.is_empty() {
493            Ok(Term::KeywordList(keyword_entries))
494        } else if saw_keyword {
495            let mut merged = items;
496            merged.extend(
497                keyword_entries
498                    .into_iter()
499                    .map(|(k, v)| Term::Tuple(vec![Term::Atom(k), v])),
500            );
501            Ok(Term::List(merged))
502        } else {
503            Ok(Term::List(items))
504        }
505    }
506
507    fn try_parse_keyword_key(&mut self) -> Option<String> {
508        let saved = self.pos;
509        self.skip_ws();
510        let start = self.pos;
511        while let Some(c) = self.peek() {
512            if c.is_ascii_alphanumeric() || c == '_' || c == '?' || c == '!' {
513                self.pos += 1;
514            } else {
515                break;
516            }
517        }
518        if self.pos == start || self.peek() != Some(':') || self.peek_n(1) == Some(':') {
519            self.pos = saved;
520            return None;
521        }
522        let key: String = self.chars[start..self.pos].iter().collect();
523        self.pos += 1;
524        Some(key)
525    }
526
527    fn parse_string(&mut self) -> Result<String, String> {
528        self.expect('"')?;
529        let mut out = String::new();
530        while let Some(c) = self.peek() {
531            self.pos += 1;
532            match c {
533                '"' => return Ok(out),
534                '\\' => {
535                    let escaped = self
536                        .peek()
537                        .ok_or_else(|| "Unterminated string escape".to_string())?;
538                    self.pos += 1;
539                    out.push(match escaped {
540                        'n' => '\n',
541                        'r' => '\r',
542                        't' => '\t',
543                        '"' => '"',
544                        '\\' => '\\',
545                        other => other,
546                    });
547                }
548                other => out.push(other),
549            }
550        }
551        Err("Unterminated string literal".to_string())
552    }
553
554    fn parse_atom(&mut self) -> Result<String, String> {
555        self.expect(':')?;
556        let start = self.pos;
557        while let Some(c) = self.peek() {
558            if c.is_ascii_alphanumeric() || c == '_' || c == '?' || c == '!' || c == '@' {
559                self.pos += 1;
560            } else {
561                break;
562            }
563        }
564        if self.pos == start {
565            return Err("Expected atom after ':'".to_string());
566        }
567        Ok(self.chars[start..self.pos].iter().collect())
568    }
569
570    fn parse_integer(&mut self) -> Result<i64, String> {
571        let start = self.pos;
572        if self.peek() == Some('-') {
573            self.pos += 1;
574        }
575        while let Some(c) = self.peek() {
576            if c.is_ascii_digit() {
577                self.pos += 1;
578            } else {
579                break;
580            }
581        }
582        self.source[start..self.byte_index(self.pos)]
583            .parse::<i64>()
584            .map_err(|e| format!("Invalid integer: {}", e))
585    }
586
587    fn parse_bool(&mut self) -> Result<bool, String> {
588        if self.starts_with("true") {
589            self.pos += 4;
590            Ok(true)
591        } else if self.starts_with("false") {
592            self.pos += 5;
593            Ok(false)
594        } else {
595            Err("Invalid boolean".to_string())
596        }
597    }
598
599    fn skip_ws(&mut self) {
600        while let Some(c) = self.peek() {
601            if c.is_whitespace() {
602                self.pos += 1;
603            } else {
604                break;
605            }
606        }
607    }
608
609    fn expect(&mut self, expected: char) -> Result<(), String> {
610        match self.peek() {
611            Some(c) if c == expected => {
612                self.pos += 1;
613                Ok(())
614            }
615            Some(c) => Err(format!("Expected '{}' but found '{}'", expected, c)),
616            None => Err(format!("Expected '{}' but reached end of input", expected)),
617        }
618    }
619
620    fn expect_sequence(&mut self, expected: &str) -> Result<(), String> {
621        if self.starts_with(expected) {
622            self.pos += expected.chars().count();
623            Ok(())
624        } else {
625            Err(format!("Expected '{}' at {}", expected, self.pos))
626        }
627    }
628
629    fn starts_with(&self, s: &str) -> bool {
630        self.chars[self.pos..]
631            .iter()
632            .collect::<String>()
633            .starts_with(s)
634    }
635
636    fn peek(&self) -> Option<char> {
637        self.chars.get(self.pos).copied()
638    }
639
640    fn peek_n(&self, n: usize) -> Option<char> {
641        self.chars.get(self.pos + n).copied()
642    }
643
644    fn is_eof(&self) -> bool {
645        self.pos >= self.chars.len()
646    }
647
648    fn byte_index(&self, char_pos: usize) -> usize {
649        self.chars.iter().take(char_pos).map(|c| c.len_utf8()).sum()
650    }
651}