lychee_lib/types/
error.rs

1use http::StatusCode;
2use serde::{Serialize, Serializer};
3use std::error::Error;
4use std::hash::Hash;
5use std::{convert::Infallible, path::PathBuf};
6use thiserror::Error;
7use tokio::task::JoinError;
8
9use super::InputContent;
10use crate::types::StatusCodeSelectorError;
11use crate::{Uri, basic_auth::BasicAuthExtractorError, utils};
12
13/// Kinds of status errors
14/// Note: The error messages can change over time, so don't match on the output
15#[derive(Error, Debug)]
16#[non_exhaustive]
17pub enum ErrorKind {
18    /// Network error while handling request.
19    /// This does not include erroneous status codes, `RejectedStatusCode` will be used in that case.
20    #[error("Network error: {analysis} ({error})", analysis=utils::reqwest::analyze_error_chain(.0), error=.0)]
21    NetworkRequest(#[source] reqwest::Error),
22    /// Cannot read the body of the received response
23    #[error("Error reading response body: {0}")]
24    ReadResponseBody(#[source] reqwest::Error),
25    /// The network client required for making requests cannot be created
26    #[error("Error creating request client: {0}")]
27    BuildRequestClient(#[source] reqwest::Error),
28
29    /// Network error while using GitHub API
30    #[error("Network error (GitHub client)")]
31    GithubRequest(#[from] Box<octocrab::Error>),
32
33    /// Error while executing a future on the Tokio runtime
34    #[error("Task failed to execute to completion")]
35    RuntimeJoin(#[from] JoinError),
36
37    /// Error while converting a file to an input
38    #[error("Cannot read input content from file `{1}`")]
39    ReadFileInput(#[source] std::io::Error, PathBuf),
40
41    /// Error while reading stdin as input
42    #[error("Cannot read input content from stdin")]
43    ReadStdinInput(#[from] std::io::Error),
44
45    /// Errors which can occur when attempting to interpret a sequence of u8 as a string
46    #[error("Attempted to interpret an invalid sequence of bytes as a string")]
47    Utf8(#[from] std::str::Utf8Error),
48
49    /// The GitHub client required for making requests cannot be created
50    #[error("Error creating GitHub client")]
51    BuildGithubClient(#[source] Box<octocrab::Error>),
52
53    /// Invalid GitHub URL
54    #[error("GitHub URL is invalid: {0}")]
55    InvalidGithubUrl(String),
56
57    /// The input is empty and not accepted as a valid URL
58    #[error("URL cannot be empty")]
59    EmptyUrl,
60
61    /// The given string can not be parsed into a valid URL, e-mail address, or file path
62    #[error("Cannot parse '{1}' into a URL: {0}")]
63    ParseUrl(#[source] url::ParseError, String),
64
65    /// The given string is a root-relative link and cannot be parsed without a known root-dir
66    #[error("Cannot resolve root-relative link '{0}'")]
67    RootRelativeLinkWithoutRoot(String),
68
69    /// The given URI cannot be converted to a file path
70    #[error("Cannot find file")]
71    InvalidFilePath(Uri),
72
73    /// The given URI's fragment could not be found within the page content
74    #[error("Cannot find fragment")]
75    InvalidFragment(Uri),
76
77    /// Cannot resolve local directory link using the configured index files
78    #[error("Cannot find index file within directory")]
79    InvalidIndexFile(Vec<String>),
80
81    /// The given path cannot be converted to a URI
82    #[error("Invalid path to URL conversion: {0}")]
83    InvalidUrlFromPath(PathBuf),
84
85    /// The given mail address is unreachable
86    #[error("Unreachable mail address {0}")]
87    UnreachableEmailAddress(Uri, String),
88
89    /// The given header could not be parsed.
90    /// A possible error when converting a `HeaderValue` from a string or byte
91    /// slice.
92    #[error("Header could not be parsed.")]
93    InvalidHeader(#[from] http::header::InvalidHeaderValue),
94
95    /// The given string can not be parsed into a valid base URL or base directory
96    #[error("Error with base dir '{0}': {1}")]
97    InvalidBase(String, String),
98
99    /// Invalid root directory given
100    #[error("Invalid root directory '{0}': {1}")]
101    InvalidRootDir(PathBuf, #[source] std::io::Error),
102
103    /// The given URI type is not supported
104    #[error("Unsupported URI type: '{0}'")]
105    UnsupportedUriType(String),
106
107    /// The given input can not be parsed into a valid URI remapping
108    #[error("Error remapping URL: `{0}`")]
109    InvalidUrlRemap(String),
110
111    /// The given path does not resolve to a valid file
112    #[error("Invalid file path: {0}")]
113    InvalidFile(PathBuf),
114
115    /// Error while traversing an input directory
116    #[error("Cannot traverse input directory: {0}")]
117    DirTraversal(#[from] ignore::Error),
118
119    /// The given glob pattern is not valid
120    #[error("UNIX glob pattern is invalid")]
121    InvalidGlobPattern(#[from] glob::PatternError),
122
123    /// The GitHub API could not be called because of a missing GitHub token.
124    #[error(
125        "GitHub token not specified. To check GitHub links reliably, use `--github-token` flag / `GITHUB_TOKEN` env var."
126    )]
127    MissingGitHubToken,
128
129    /// Used an insecure URI where a secure variant was reachable
130    #[error("This URI is available in HTTPS protocol, but HTTP is provided. Use '{0}' instead")]
131    InsecureURL(Uri),
132
133    /// Error while sending/receiving messages from MPSC channel
134    #[error("Cannot send/receive message from channel")]
135    Channel(#[from] tokio::sync::mpsc::error::SendError<InputContent>),
136
137    /// A URL without a host was found
138    #[error("URL is missing a host")]
139    InvalidUrlHost,
140
141    /// Cannot parse the given URI
142    #[error("The given URI is invalid: {0}")]
143    InvalidURI(Uri),
144
145    /// The given status code is invalid (not in the range 100-1000)
146    #[error("Invalid status code: {0}")]
147    InvalidStatusCode(u16),
148
149    /// The given status code was not accepted (this depends on the `accept` configuration)
150    #[error(
151        r#"Rejected status code: {code} {reason} (configurable with "accept" option)"#,
152        code = .0.as_str(),
153        reason = .0.canonical_reason().unwrap_or("Unknown status code")
154    )]
155    RejectedStatusCode(StatusCode),
156
157    /// Regex error
158    #[error("Error when using regex engine: {0}")]
159    Regex(#[from] regex::Error),
160
161    /// Basic auth extractor error
162    #[error("Basic auth extractor error")]
163    BasicAuthExtractorError(#[from] BasicAuthExtractorError),
164
165    /// Cannot load cookies
166    #[error("Cannot load cookies")]
167    Cookies(String),
168
169    /// Status code selector parse error
170    #[error("Status code range error")]
171    StatusCodeSelectorError(#[from] StatusCodeSelectorError),
172
173    /// Preprocessor command error
174    #[error("Preprocessor command '{command}' failed: {reason}")]
175    PreprocessorError {
176        /// The command which did not execute successfully
177        command: String,
178        /// The reason the command failed
179        reason: String,
180    },
181
182    /// The extracted `WikiLink` could not be found by searching the directory
183    #[error("Wikilink {0} not found at {1}")]
184    WikilinkNotFound(Uri, PathBuf),
185
186    /// Error on creation of the `WikilinkResolver`
187    #[error("Failed to initialize wikilink checker: {0}")]
188    WikilinkInvalidBase(String),
189}
190
191impl ErrorKind {
192    /// Return more details about the given [`ErrorKind`]
193    ///
194    /// Which additional information we can extract depends on the underlying
195    /// request type. The output is purely meant for humans (e.g. for status
196    /// messages) and future changes are expected.
197    #[must_use]
198    #[allow(clippy::too_many_lines)]
199    pub fn details(&self) -> Option<String> {
200        match self {
201            ErrorKind::NetworkRequest(e) => {
202                // Get detailed, actionable error analysis
203                Some(utils::reqwest::analyze_error_chain(e))
204            }
205            ErrorKind::RejectedStatusCode(status) => status
206                .is_redirection()
207                .then_some(r#"Redirects may have been limited by "max-redirects"."#.to_string()),
208            ErrorKind::GithubRequest(e) => {
209                if let octocrab::Error::GitHub { source, .. } = &**e {
210                    Some(source.message.clone())
211                } else {
212                    // Fall back to generic error analysis
213                    Some(e.to_string())
214                }
215            }
216            ErrorKind::InvalidFilePath(_uri) => {
217                Some("File not found. Check if file exists and path is correct".to_string())
218            }
219            ErrorKind::ReadFileInput(e, path) => match e.kind() {
220                std::io::ErrorKind::NotFound => Some("Check if file path is correct".to_string()),
221                std::io::ErrorKind::PermissionDenied => Some(format!(
222                    "Permission denied: '{}'. Check file permissions",
223                    path.display()
224                )),
225                std::io::ErrorKind::IsADirectory => Some(format!(
226                    "Path is a directory, not a file: '{}'. Check file path",
227                    path.display()
228                )),
229                _ => Some(format!("File read error for '{}': {}", path.display(), e)),
230            },
231            ErrorKind::ReadStdinInput(e) => match e.kind() {
232                std::io::ErrorKind::UnexpectedEof => {
233                    Some("Stdin input ended unexpectedly. Check input data".to_string())
234                }
235                std::io::ErrorKind::InvalidData => {
236                    Some("Invalid data from stdin. Check input format".to_string())
237                }
238                _ => Some(format!("Stdin read error: {e}")),
239            },
240            ErrorKind::ParseUrl(e, _url) => match e {
241                url::ParseError::RelativeUrlWithoutBase => Some(
242                    "This relative link was found inside an input source that has no base location"
243                        .to_string(),
244                ),
245                _ => None,
246            },
247            ErrorKind::RootRelativeLinkWithoutRoot(_) => Some(
248                "To resolve root-relative links in local files, provide a root dir".to_string(),
249            ),
250            ErrorKind::EmptyUrl => {
251                Some("Empty URL found. Check for missing links or malformed markdown".to_string())
252            }
253            ErrorKind::InvalidFile(path) => Some(format!(
254                "Invalid file path: '{}'. Check if file exists and is readable",
255                path.display()
256            )),
257            ErrorKind::ReadResponseBody(error) => Some(format!(
258                "Failed to read response body: {error}. Server may have sent invalid data",
259            )),
260            ErrorKind::BuildRequestClient(error) => Some(format!(
261                "Failed to create HTTP client: {error}. Check system configuration",
262            )),
263            ErrorKind::RuntimeJoin(join_error) => Some(format!(
264                "Task execution failed: {join_error}. Internal processing error"
265            )),
266            ErrorKind::Utf8(_utf8_error) => {
267                Some("Invalid UTF-8 sequence found. File contains non-UTF-8 characters".to_string())
268            }
269            ErrorKind::BuildGithubClient(error) => Some(format!(
270                "Failed to create GitHub client: {error}. Check token and network connectivity",
271            )),
272            ErrorKind::InvalidGithubUrl(url) => Some(format!(
273                "Invalid GitHub URL format: '{url}'. Check URL syntax",
274            )),
275            ErrorKind::InvalidFragment(_uri) => Some(
276                "Fragment not found in document. Check if fragment exists or page structure"
277                    .to_string(),
278            ),
279            ErrorKind::InvalidUrlFromPath(path_buf) => Some(format!(
280                "Cannot convert path to URL: '{}'. Check path format",
281                path_buf.display()
282            )),
283            ErrorKind::UnreachableEmailAddress(_uri, reason) => Some(reason.clone()),
284            ErrorKind::InvalidHeader(invalid_header_value) => Some(format!(
285                "Invalid HTTP header: {invalid_header_value}. Check header format",
286            )),
287            ErrorKind::InvalidBase(base, reason) => {
288                Some(format!("Invalid base URL or directory: '{base}'. {reason}",))
289            }
290            ErrorKind::InvalidRootDir(_, _) => {
291                Some("Check the root dir exists and is accessible".to_string())
292            }
293            ErrorKind::UnsupportedUriType(uri_type) => Some(format!(
294                "Unsupported URI type: '{uri_type}'. {}",
295                "Only http, https, file, and mailto are supported",
296            )),
297            ErrorKind::InvalidUrlRemap(remap) => Some(format!(
298                "Invalid URL remapping: '{remap}'. Check remapping syntax",
299            )),
300            ErrorKind::DirTraversal(error) => Some(format!(
301                "Directory traversal failed: {error}. Check directory permissions",
302            )),
303            ErrorKind::InvalidGlobPattern(pattern_error) => Some(format!(
304                "Invalid glob pattern: {pattern_error}. Check pattern syntax",
305            )),
306            ErrorKind::MissingGitHubToken => Some(format!(
307                "GitHub token required. {}",
308                "Use --github-token flag or GITHUB_TOKEN environment variable",
309            )),
310            ErrorKind::InsecureURL(uri) => Some(format!(
311                "Insecure HTTP URL detected: use '{}' instead of HTTP",
312                uri.as_str().replace("http://", "https://")
313            )),
314            ErrorKind::Channel(_send_error) => {
315                Some("Internal communication error. Processing thread failed".to_string())
316            }
317            ErrorKind::InvalidUrlHost => Some("URL missing hostname. Check URL format".to_string()),
318            ErrorKind::InvalidURI(uri) => {
319                Some(format!("Invalid URI format: '{uri}'. Check URI syntax",))
320            }
321            ErrorKind::InvalidStatusCode(code) => Some(format!(
322                "Invalid HTTP status code: {code}. Must be between 100-999",
323            )),
324            ErrorKind::Regex(error) => Some(format!(
325                "Regular expression error: {error}. Check regex syntax",
326            )),
327            ErrorKind::BasicAuthExtractorError(basic_auth_extractor_error) => Some(format!(
328                "Basic authentication error: {basic_auth_extractor_error}. {}",
329                "Check credentials format",
330            )),
331            ErrorKind::Cookies(reason) => Some(format!(
332                "Cookie handling error: {reason}. Check cookie file format",
333            )),
334            ErrorKind::StatusCodeSelectorError(status_code_selector_error) => Some(format!(
335                "Status code selector error: {status_code_selector_error}. {}",
336                "Check accept configuration",
337            )),
338            ErrorKind::InvalidIndexFile(index_files) => match &index_files[..] {
339                [] => "No directory links are allowed because index_files is defined and empty"
340                    .to_string(),
341                [name] => format!("An index file ({name}) is required"),
342                [init @ .., tail] => format!(
343                    "An index file ({}, or {}) is required",
344                    init.join(", "),
345                    tail
346                ),
347            }
348            .into(),
349            ErrorKind::PreprocessorError { command, reason } => Some(format!(
350                "Command '{command}' failed {reason}. Check value of the pre option"
351            )),
352            ErrorKind::WikilinkNotFound(uri, pathbuf) => Some(format!(
353                "WikiLink {uri} could not be found at {:}",
354                pathbuf.display()
355            )),
356            ErrorKind::WikilinkInvalidBase(reason) => {
357                Some(format!("WikiLink Resolver could not be created: {reason} ",))
358            }
359        }
360    }
361
362    /// Return the underlying source of the given [`ErrorKind`]
363    /// if it is a `reqwest::Error`.
364    /// This is useful for extracting the status code of a failed request.
365    /// If the error is not a `reqwest::Error`, `None` is returned.
366    #[must_use]
367    #[allow(clippy::redundant_closure_for_method_calls)]
368    pub(crate) fn reqwest_error(&self) -> Option<&reqwest::Error> {
369        self.source()
370            .and_then(|e| e.downcast_ref::<reqwest::Error>())
371    }
372
373    /// Return the underlying source of the given [`ErrorKind`]
374    /// if it is a `octocrab::Error`.
375    /// This is useful for extracting the status code of a failed request.
376    /// If the error is not a `octocrab::Error`, `None` is returned.
377    #[must_use]
378    #[allow(clippy::redundant_closure_for_method_calls)]
379    pub(crate) fn github_error(&self) -> Option<&octocrab::Error> {
380        self.source()
381            .and_then(|e| e.downcast_ref::<octocrab::Error>())
382    }
383}
384
385#[allow(clippy::match_same_arms)]
386impl PartialEq for ErrorKind {
387    fn eq(&self, other: &Self) -> bool {
388        match (self, other) {
389            (Self::NetworkRequest(e1), Self::NetworkRequest(e2)) => {
390                e1.to_string() == e2.to_string()
391            }
392            (Self::ReadResponseBody(e1), Self::ReadResponseBody(e2)) => {
393                e1.to_string() == e2.to_string()
394            }
395            (Self::BuildRequestClient(e1), Self::BuildRequestClient(e2)) => {
396                e1.to_string() == e2.to_string()
397            }
398            (Self::RuntimeJoin(e1), Self::RuntimeJoin(e2)) => e1.to_string() == e2.to_string(),
399            (Self::ReadFileInput(e1, s1), Self::ReadFileInput(e2, s2)) => {
400                e1.kind() == e2.kind() && s1 == s2
401            }
402            (Self::ReadStdinInput(e1), Self::ReadStdinInput(e2)) => e1.kind() == e2.kind(),
403            (Self::GithubRequest(e1), Self::GithubRequest(e2)) => e1.to_string() == e2.to_string(),
404            (Self::InvalidGithubUrl(s1), Self::InvalidGithubUrl(s2)) => s1 == s2,
405            (Self::ParseUrl(s1, e1), Self::ParseUrl(s2, e2)) => s1 == s2 && e1 == e2,
406            (Self::UnreachableEmailAddress(u1, ..), Self::UnreachableEmailAddress(u2, ..)) => {
407                u1 == u2
408            }
409            (Self::InsecureURL(u1), Self::InsecureURL(u2)) => u1 == u2,
410            (Self::InvalidGlobPattern(e1), Self::InvalidGlobPattern(e2)) => {
411                e1.msg == e2.msg && e1.pos == e2.pos
412            }
413            (Self::InvalidHeader(_), Self::InvalidHeader(_))
414            | (Self::MissingGitHubToken, Self::MissingGitHubToken) => true,
415            (Self::InvalidStatusCode(c1), Self::InvalidStatusCode(c2)) => c1 == c2,
416            (Self::InvalidUrlHost, Self::InvalidUrlHost) => true,
417            (Self::InvalidURI(u1), Self::InvalidURI(u2)) => u1 == u2,
418            (Self::Regex(e1), Self::Regex(e2)) => e1.to_string() == e2.to_string(),
419            (Self::DirTraversal(e1), Self::DirTraversal(e2)) => e1.to_string() == e2.to_string(),
420            (Self::Channel(_), Self::Channel(_)) => true,
421            (Self::BasicAuthExtractorError(e1), Self::BasicAuthExtractorError(e2)) => {
422                e1.to_string() == e2.to_string()
423            }
424            (Self::Cookies(e1), Self::Cookies(e2)) => e1 == e2,
425            (Self::InvalidFile(p1), Self::InvalidFile(p2)) => p1 == p2,
426            (Self::InvalidFilePath(u1), Self::InvalidFilePath(u2)) => u1 == u2,
427            (Self::InvalidFragment(u1), Self::InvalidFragment(u2)) => u1 == u2,
428            (Self::InvalidIndexFile(p1), Self::InvalidIndexFile(p2)) => p1 == p2,
429            (Self::InvalidUrlFromPath(p1), Self::InvalidUrlFromPath(p2)) => p1 == p2,
430            (Self::InvalidBase(b1, e1), Self::InvalidBase(b2, e2)) => b1 == b2 && e1 == e2,
431            (Self::InvalidUrlRemap(r1), Self::InvalidUrlRemap(r2)) => r1 == r2,
432            (Self::EmptyUrl, Self::EmptyUrl) => true,
433            (Self::RejectedStatusCode(c1), Self::RejectedStatusCode(c2)) => c1 == c2,
434
435            _ => false,
436        }
437    }
438}
439
440impl Eq for ErrorKind {}
441
442#[allow(clippy::match_same_arms)]
443impl Hash for ErrorKind {
444    fn hash<H>(&self, state: &mut H)
445    where
446        H: std::hash::Hasher,
447    {
448        match self {
449            Self::RuntimeJoin(e) => e.to_string().hash(state),
450            Self::ReadFileInput(e, s) => (e.kind(), s).hash(state),
451            Self::ReadStdinInput(e) => e.kind().hash(state),
452            Self::NetworkRequest(e) => e.to_string().hash(state),
453            Self::ReadResponseBody(e) => e.to_string().hash(state),
454            Self::BuildRequestClient(e) => e.to_string().hash(state),
455            Self::BuildGithubClient(e) => e.to_string().hash(state),
456            Self::GithubRequest(e) => e.to_string().hash(state),
457            Self::InvalidGithubUrl(s) => s.hash(state),
458            Self::DirTraversal(e) => e.to_string().hash(state),
459            Self::InvalidFile(e) => e.to_string_lossy().hash(state),
460            Self::EmptyUrl => "Empty URL".hash(state),
461            Self::ParseUrl(e, s) => (e.to_string(), s).hash(state),
462            Self::RootRelativeLinkWithoutRoot(s) => s.hash(state),
463            Self::InvalidURI(u) => u.hash(state),
464            Self::InvalidUrlFromPath(p) => p.hash(state),
465            Self::Utf8(e) => e.to_string().hash(state),
466            Self::InvalidFilePath(u) => u.hash(state),
467            Self::InvalidFragment(u) => u.hash(state),
468            Self::InvalidIndexFile(p) => p.hash(state),
469            Self::UnreachableEmailAddress(u, ..) => u.hash(state),
470            Self::InsecureURL(u, ..) => u.hash(state),
471            Self::InvalidBase(base, e) => (base, e).hash(state),
472            Self::InvalidRootDir(s, _) => s.hash(state),
473            Self::UnsupportedUriType(s) => s.hash(state),
474            Self::InvalidUrlRemap(remap) => (remap).hash(state),
475            Self::InvalidHeader(e) => e.to_string().hash(state),
476            Self::InvalidGlobPattern(e) => e.to_string().hash(state),
477            Self::InvalidStatusCode(c) => c.hash(state),
478            Self::RejectedStatusCode(c) => c.hash(state),
479            Self::Channel(e) => e.to_string().hash(state),
480            Self::MissingGitHubToken | Self::InvalidUrlHost => {
481                std::mem::discriminant(self).hash(state);
482            }
483            Self::Regex(e) => e.to_string().hash(state),
484            Self::BasicAuthExtractorError(e) => e.to_string().hash(state),
485            Self::Cookies(e) => e.hash(state),
486            Self::StatusCodeSelectorError(e) => e.to_string().hash(state),
487            Self::PreprocessorError { command, reason } => (command, reason).hash(state),
488            Self::WikilinkNotFound(uri, pathbuf) => (uri, pathbuf).hash(state),
489            Self::WikilinkInvalidBase(e) => e.hash(state),
490        }
491    }
492}
493
494impl Serialize for ErrorKind {
495    fn serialize<S>(&self, serializer: S) -> std::result::Result<S::Ok, S::Error>
496    where
497        S: Serializer,
498    {
499        serializer.collect_str(self)
500    }
501}
502
503impl From<Infallible> for ErrorKind {
504    fn from(_: Infallible) -> Self {
505        // tautological
506        unreachable!()
507    }
508}
509
510#[cfg(test)]
511mod tests {
512    use crate::ErrorKind;
513    #[test]
514    fn test_error_kind_details() {
515        // Test rejected status code
516        let status_error = ErrorKind::RejectedStatusCode(http::StatusCode::NOT_FOUND);
517        assert!(status_error.to_string().contains("Not Found"));
518
519        // Test redirected status code
520        let redir_error = ErrorKind::RejectedStatusCode(http::StatusCode::MOVED_PERMANENTLY);
521        assert!(redir_error.details().is_some_and(|x| x.contains(
522            "Redirects may have been limited by \"max-redirects\""
523        )));
524    }
525}
lychee_lib/types/error.rs

lychee_lib/types/
error.rs
