libxml_rs/xml/errors/
mod.rs

1//! Error subsystem (§21, §85 Phase 1).
2//!
3//! Implements the libxml2 error reporting infrastructure:
4//!
5//! - `xmlError` struct management
6//! - Error domain/code registry
7//! - Structured error callbacks (thread-local storage)
8//! - Generic error callbacks (thread-local storage)
9//! - Last-error tracking (thread-local `xmlGetLastError`, `xmlResetLastError`, `xmlCopyError`)
10//! - Error message formatting
11//! - `xmlRaiseError()` — the central error reporting function
12//!
13//! # UPSTREAM-PARITY
14//!
15//! libxml2 has a two-tier error system:
16//!
17//! 1. **Structured errors** — `xmlStructuredErrorFunc` receives an `xmlErrorPtr`
18//!    with all structured fields (domain, code, level, line, etc.)
19//!
20//! 2. **Generic errors** — `xmlGenericErrorFunc` receives a formatted string
21//!    (printf-style). This is the older system, still widely used.
22//!
23//! Both systems coexist. When both handlers are set, both are called.
24//! The last error is stored thread-locally for retrieval via `xmlGetLastError`.
25//!
26//! # Phase 1 status
27//!
28//! Complete — all error functions are implemented.
29//! Variadic message formatting will be enhanced in Phase 2+.
30
31use core::ffi::c_void;
32use core::fmt::Write;
33use core::ptr;
34use std::os::raw::{c_char, c_int};
35
36use crate::abi::callbacks::{xmlGenericErrorFunc, xmlStructuredErrorFunc};
37use crate::abi::structs::_xmlError;
38use crate::abi::types::xmlErrorLevel::*;
39use crate::abi::types::*;
40use crate::xml::globals;
41
42// ═══════════════════════════════════════════════════════════════════════════════
43// Error Management Functions
44// ═══════════════════════════════════════════════════════════════════════════════
45
46/// Set the generic error handler.
47///
48/// # UPSTREAM-PARITY
49///
50/// ```c
51/// void xmlSetGenericErrorFunc(void *ctx, xmlGenericErrorFunc handler);
52/// ```
53///
54/// # SAFETY
55///
56/// - `handler` must be a valid function pointer or NULL (to reset to default).
57/// - If non-NULL, the handler may be called at any time with `ctx`.
58pub unsafe fn set_generic_error_func(ctx: *mut c_void, handler: Option<xmlGenericErrorFunc>) {
59    // SAFETY: Delegates to globals with same safety contract.
60    unsafe { globals::set_generic_error_func(ctx, handler) };
61}
62
63/// Set the structured error handler.
64///
65/// # UPSTREAM-PARITY
66///
67/// ```c
68/// void xmlSetStructuredErrorFunc(void *ctx, xmlStructuredErrorFunc handler);
69/// ```
70///
71/// # SAFETY
72///
73/// - `handler` must be a valid function pointer or NULL.
74pub unsafe fn set_structured_error_func(ctx: *mut c_void, handler: Option<xmlStructuredErrorFunc>) {
75    // SAFETY: Delegates to globals with same safety contract.
76    unsafe { globals::set_structured_error_func(ctx, handler) };
77}
78
79/// Get the last error for the current thread.
80///
81/// # UPSTREAM-PARITY
82///
83/// ```c
84/// xmlErrorPtr xmlGetLastError(void);
85/// ```
86///
87/// Returns a pointer to the last error, or NULL if no error occurred.
88/// The returned pointer is valid until the next libxml2 call in this thread.
89pub fn get_last_error() -> *mut _xmlError {
90    globals::get_last_error()
91}
92
93/// Copy an error from one location to another.
94///
95/// # UPSTREAM-PARITY
96///
97/// ```c
98/// xmlErrorPtr xmlCopyError(xmlErrorPtr from, xmlErrorPtr to);
99/// ```
100///
101/// Copies `from` into `to`. Returns 0 on success, -1 on error.
102///
103/// # SAFETY
104///
105/// - `from` and `to` must be valid pointers to `_xmlError` structs, or NULL.
106pub unsafe fn copy_error(from: *const _xmlError, to: *mut _xmlError) -> c_int {
107    if from.is_null() || to.is_null() {
108        return -1;
109    }
110    // SAFETY: Caller guarantees both pointers are valid.
111    unsafe {
112        ptr::copy_nonoverlapping(from, to, 1);
113    }
114    0
115}
116
117/// Reset an error structure to its default state.
118///
119/// # UPSTREAM-PARITY
120///
121/// ```c
122/// void xmlResetError(xmlErrorPtr err);
123/// ```
124///
125/// # SAFETY
126///
127/// - `err` must be a valid pointer to `_xmlError`, or NULL.
128pub unsafe fn reset_error(err: *mut _xmlError) {
129    if err.is_null() {
130        return;
131    }
132    // SAFETY: Caller guarantees pointer is valid.
133    unsafe {
134        ptr::write(
135            err,
136            _xmlError {
137                domain: XML_FROM_NONE,
138                code: XML_ERR_OK as c_int,
139                message: ptr::null_mut(),
140                level: XML_ERR_NONE as c_int,
141                file: ptr::null_mut(),
142                line: 0,
143                str1: ptr::null_mut(),
144                str2: ptr::null_mut(),
145                str3: ptr::null_mut(),
146                int1: 0,
147                int2: 0,
148                ctxt: ptr::null_mut(),
149                node: ptr::null_mut(),
150            },
151        );
152    }
153}
154
155/// Reset the last error for the current thread.
156///
157/// # UPSTREAM-PARITY
158///
159/// ```c
160/// void xmlResetLastError(void);
161/// ```
162pub fn reset_last_error() {
163    globals::reset_last_error();
164}
165
166/// Format an error message.
167///
168/// This function creates a formatted error message from the component parts.
169/// In Phase 1, this is a basic implementation. In Phase 2+, variadic
170/// printf-style formatting will be added.
171///
172/// Returns a C string pointer (allocated with xmlMalloc) that the caller
173/// must free with xmlFreeImpl, or NULL on allocation failure.
174///
175/// # UPSTREAM-PARITY
176///
177/// Upstream libxml2 uses `vsnprintf` internally for message formatting.
178/// We use a simple formatting approach that produces compatible output
179/// for the common error patterns.
180pub fn format_error_message(
181    _domain: c_int,
182    _code: c_int,
183    msg: *const c_char,
184    str1: *const c_char,
185    str2: *const c_char,
186    str3: *const c_char,
187) -> *mut c_char {
188    // Phase 1: basic message construction.
189    // If a direct message string is provided, use it.
190    if !msg.is_null() {
191        // SAFETY: Caller guarantees msg is a valid C string.
192        let msg_str = unsafe { crate::abi::allocator::xmlMemStrdupImpl(msg) };
193        return msg_str as *mut c_char;
194    }
195
196    // Build a message from the component strings.
197    // This matches upstream behavior where domain/code are combined
198    // with str1/str2/str3 into a diagnostic message.
199    let mut buf: [u8; 1024] = [0; 1024];
200    let mut pos = 0;
201
202    // Write domain prefix
203    let domain_str = match _domain {
204        XML_FROM_PARSER => "parser",
205        XML_FROM_TREE => "tree",
206        XML_FROM_NAMESPACE => "namespace",
207        XML_FROM_DTD => "dtd",
208        XML_FROM_HTML => "html",
209        XML_FROM_MEMORY => "memory",
210        XML_FROM_OUTPUT => "output",
211        XML_FROM_IO => "io",
212        XML_FROM_XPATH => "xpath",
213        XML_FROM_XPOINTER => "xpointer",
214        XML_FROM_XINCLUDE => "xinclude",
215        XML_FROM_CATALOG => "catalog",
216        XML_FROM_C14N => "c14n",
217        XML_FROM_XSLT => "xslt",
218        XML_FROM_VALID => "valid",
219        XML_FROM_CHECK => "check",
220        XML_FROM_WRITER => "writer",
221        XML_FROM_MODULE => "module",
222        XML_FROM_I18N => "i18n",
223        XML_FROM_SCHEMATRONV => "schematron",
224        XML_FROM_BUFFER => "buffer",
225        XML_FROM_URI => "uri",
226        XML_FROM_NONE => "",
227        XML_FROM_FTP => "ftp",
228        XML_FROM_HTTP => "http",
229        XML_FROM_REGEXP => "regexp",
230        XML_FROM_DATATYPE => "datatype",
231        XML_FROM_SCHEMASP => "schema parser",
232        XML_FROM_SCHEMASV => "schema validator",
233        XML_FROM_RELAXNGP => "relaxng parser",
234        XML_FROM_RELAXNGV => "relaxng validator",
235        _ => "unknown",
236    };
237
238    if !domain_str.is_empty() {
239        let bytes = domain_str.as_bytes();
240        let len = bytes.len().min(buf.len().saturating_sub(pos + 2));
241        buf[pos..pos + len].copy_from_slice(&bytes[..len]);
242        pos += len;
243        buf[pos] = b' ';
244        pos += 1;
245    }
246
247    // Append str1 if present
248    if !str1.is_null() {
249        // SAFETY: Caller guarantees str1 is a valid C string.
250        let s = unsafe { crate::abi::versioning::c_str_to_bytes(str1).unwrap_or_default() };
251        if pos + s.len() + 3 <= buf.len() {
252            buf[pos] = b'\'';
253            pos += 1;
254            buf[pos..pos + s.len()].copy_from_slice(s);
255            pos += s.len();
256            buf[pos] = b'\'';
257            pos += 1;
258            buf[pos] = b' ';
259            pos += 1;
260        }
261    }
262
263    // Append str2 if present
264    if !str2.is_null() {
265        let s = unsafe { crate::abi::versioning::c_str_to_bytes(str2).unwrap_or_default() };
266        if pos + s.len() + 3 <= buf.len() {
267            buf[pos] = b'\'';
268            pos += 1;
269            buf[pos..pos + s.len()].copy_from_slice(s);
270            pos += s.len();
271            buf[pos] = b'\'';
272            pos += 1;
273            buf[pos] = b' ';
274            pos += 1;
275        }
276    }
277
278    // Append str3 if present
279    if !str3.is_null() {
280        let s = unsafe { crate::abi::versioning::c_str_to_bytes(str3).unwrap_or_default() };
281        if pos + s.len() + 3 <= buf.len() {
282            buf[pos] = b'\'';
283            pos += 1;
284            buf[pos..pos + s.len()].copy_from_slice(s);
285            pos += s.len();
286            buf[pos] = b'\'';
287            pos += 1;
288            buf[pos] = b' ';
289            pos += 1;
290        }
291    }
292
293    // Null-terminate
294    if pos < buf.len() {
295        buf[pos] = 0;
296    } else {
297        buf[buf.len() - 1] = 0;
298    }
299
300    // Allocate and return
301    let result = unsafe { crate::abi::allocator::xmlMallocImpl(pos + 1) };
302    if result.is_null() {
303        return ptr::null_mut();
304    }
305    unsafe {
306        ptr::copy_nonoverlapping(buf.as_ptr(), result as *mut u8, pos + 1);
307    }
308    result as *mut c_char
309}
310
311/// Raise an error — the central error reporting function.
312///
313/// This is called internally when an error occurs. It:
314/// 1. Updates the thread-local last error
315/// 2. Invokes the structured error handler if one is set
316/// 3. Invokes the generic error handler if one is set (for warnings/errors)
317///
318/// # UPSTREAM-PARITY
319///
320/// ```c
321/// void xmlRaiseError(xmlErrorPtr ctxt,
322///                    xmlErrorPtr ctxt2,
323///                    xmlErrorPtr ctxt3,
324///                    xmlErrorPtr ctxt4,
325///                    xmlErrorPtr ctxt5,
326///                    int domain,
327///                    int code,
328///                    xmlErrorLevel level,
329///                    const char *file,
330///                    int line,
331///                    const char *str1,
332///                    const char *str2,
333///                    const char *str3,
334///                    int int1,
335///                    int int2,
336///                    const char *msg,
337///                    ...);
338/// ```
339///
340/// # SAFETY
341///
342/// - `ctxt` may be NULL (context of the error).
343/// - `domain`, `code`, `level`: valid error codes.
344/// - `msg` must be a valid C string or NULL.
345/// - `file` must be a valid C string or NULL.
346/// - `str1`, `str2`, `str3`: error-related strings (may be NULL).
347pub unsafe fn raise_error(
348    ctxt: *mut c_void,
349    _ctxt2: *mut c_void,
350    _ctxt3: *mut c_void,
351    _ctxt4: *mut c_void,
352    _ctxt5: *mut c_void,
353    domain: c_int,
354    code: c_int,
355    level: c_int,
356    file: *const c_char,
357    line: c_int,
358    str1: *const c_char,
359    str2: *const c_char,
360    str3: *const c_char,
361    int1: c_int,
362    int2: c_int,
363    msg: *const c_char,
364) {
365    // Format the error message
366    let formatted_msg = format_error_message(domain, code, msg, str1, str2, str3);
367
368    // UPSTREAM-PARITY (xmlVSetError): string fields are owned copies so the
369    // stored error survives transient callers.
370    let file_copy = if file.is_null() {
371        ptr::null_mut()
372    } else {
373        crate::abi::allocator::xmlMemStrdupImpl(file) as *mut c_char
374    };
375    let str1_copy = if str1.is_null() {
376        ptr::null_mut()
377    } else {
378        crate::abi::allocator::xmlMemStrdupImpl(str1) as *mut c_char
379    };
380    let str2_copy = if str2.is_null() {
381        ptr::null_mut()
382    } else {
383        crate::abi::allocator::xmlMemStrdupImpl(str2) as *mut c_char
384    };
385    let str3_copy = if str3.is_null() {
386        ptr::null_mut()
387    } else {
388        crate::abi::allocator::xmlMemStrdupImpl(str3) as *mut c_char
389    };
390
391    // Store the last error
392    let err = _xmlError {
393        domain,
394        code,
395        message: formatted_msg,
396        level,
397        file: file_copy,
398        line,
399        str1: str1_copy,
400        str2: str2_copy,
401        str3: str3_copy,
402        int1,
403        int2: 0,
404        ctxt,
405        node: ptr::null_mut(),
406    };
407
408    globals::set_last_error(err);
409
410    // UPSTREAM-PARITY (xmlCtxtVErr): the structured handler wins; the
411    // generic channel is only used when no structured handler is set.
412    if let Some(handler) = globals::get_structured_error_func() {
413        let ctx = globals::get_structured_error_ctx();
414        let err_ref = globals::get_last_error();
415        if !err_ref.is_null() {
416            handler(ctx, err_ref as *const _xmlError);
417        }
418    } else if let Some(handler) = globals::get_generic_error_func() {
419        if level != 0 {
420            let ctx = globals::get_generic_error_ctx();
421            if !formatted_msg.is_null() {
422                handler(ctx, formatted_msg as *const core::ffi::c_char);
423            } else if !msg.is_null() {
424                handler(ctx, msg);
425            }
426        }
427    }
428
429    // Free the formatted message if it was allocated
430    // Note: We keep it as the last error's message, so we don't free it here.
431    // The next call to raise_error or reset_error will free the old message.
432    // Actually, in Phase 1, we don't free because the message is the last error's.
433    // A more complete implementation would free the old message when setting a new one.
434}
435
436/// Emit a legacy-format message through the generic error channel.
437///
438/// Upstream's `xmlGenericErrorDefaultFunc` writes the message to stderr;
439/// when a custom generic handler is installed it receives the message. The
440/// `level` prefix matches upstream `xmlVFormatLegacyError` (error.c 2.15).
441unsafe fn emit_legacy_message(level: &str, msg: *const c_char) {
442    if msg.is_null() {
443        return;
444    }
445    let len = libc::strlen(msg) as usize;
446    let text = core::slice::from_raw_parts(msg as *const u8, len);
447    let mut full = Vec::with_capacity(level.len() + 2 + len);
448    full.extend_from_slice(level.as_bytes());
449    full.push(b':');
450    full.push(b' ');
451    full.extend_from_slice(text);
452    if let Some(handler) = globals::get_generic_error_func() {
453        let ctx = globals::get_generic_error_ctx();
454        let mut cmsg = full.clone();
455        cmsg.push(0);
456        handler(ctx, cmsg.as_ptr() as *const c_char);
457    } else {
458        // Upstream default (xmlGenericErrorDefaultFunc): stderr.
459        let _ = libc::write(2, full.as_ptr() as *const libc::c_void, full.len());
460    }
461}
462
463// ═══════════════════════════════════════════════════════════════════════════════
464// Generic-channel fragment streaming (upstream error.c `xmlFormatError`)
465// ═══════════════════════════════════════════════════════════════════════════════
466//
467// Upstream streams each error through the generic channel as a sequence of
468// variadic calls (e.g. `channel(data, "%s:%d: ", file, line)` followed by the
469// domain, level, message and source-context fragments). Custom handlers and the
470// built-in default (an x86_64 SysV va_list shim, see data_globals.rs) both
471// observe the same per-fragment calls. Stable Rust cannot express a variadic
472// call, so each fragment goes through a tiny x86_64 trampoline that places the
473// fixed arguments in the ABI registers and does an indirect call.
474
475/// `channel(data, fmt)` — no variadic arguments.
476#[cfg(target_arch = "x86_64")]
477#[inline]
478unsafe fn ch_call0(handler: xmlGenericErrorFunc, data: *mut c_void, fmt: *const c_char) {
479    // SAFETY: `handler` is a C-compatible generic error callback; per the
480    // SysV ABI the callee sees (data, fmt) with no additional registers
481    // consumed (rdx/rcx zeroed so a va_list-reading callee finds nothing).
482    // The compiler guarantees 16-byte stack alignment at the asm block, so
483    // the `call` is correctly aligned.
484    unsafe {
485        core::arch::asm!(
486            "xor edx, edx",
487            "xor ecx, ecx",
488            "call {h}",
489            h = in(reg) handler as usize,
490            in("rdi") data,
491            in("rsi") fmt,
492            out("rdx") _, out("rcx") _,
493            lateout("rax") _, lateout("r8") _, lateout("r9") _, lateout("r10") _, lateout("r11") _,
494        );
495    }
496}
497
498/// `channel(data, fmt, a1)` — one pointer-sized variadic argument.
499#[cfg(target_arch = "x86_64")]
500#[inline]
501unsafe fn ch_call1(handler: xmlGenericErrorFunc, data: *mut c_void, fmt: *const c_char, a1: usize) {
502    // SAFETY: as ch_call0; `a1` lands in the va_list slot after the two
503    // fixed args (rdx).
504    unsafe {
505        core::arch::asm!(
506            "xor ecx, ecx",
507            "call {h}",
508            h = in(reg) handler as usize,
509            in("rdi") data,
510            in("rsi") fmt,
511            in("rdx") a1,
512            out("rcx") _,
513            lateout("rax") _, lateout("r8") _, lateout("r9") _, lateout("r10") _, lateout("r11") _,
514        );
515    }
516}
517
518/// `channel(data, fmt, a1, a2)` — two pointer-sized variadic arguments.
519#[cfg(target_arch = "x86_64")]
520#[inline]
521unsafe fn ch_call2(
522    handler: xmlGenericErrorFunc,
523    data: *mut c_void,
524    fmt: *const c_char,
525    a1: usize,
526    a2: usize,
527) {
528    // SAFETY: as ch_call0; a1/a2 land in the va_list slots (rdx, rcx).
529    unsafe {
530        core::arch::asm!(
531            "call {h}",
532            h = in(reg) handler as usize,
533            in("rdi") data,
534            in("rsi") fmt,
535            in("rdx") a1,
536            in("rcx") a2,
537            lateout("rax") _, lateout("r8") _, lateout("r9") _, lateout("r10") _, lateout("r11") _,
538        );
539    }
540}
541
542/// Emit one raise through the generic channel with upstream's
543/// `xmlFormatError` fragment sequence (error.c 2.15): file/line prefix,
544/// domain, level, message, then the source window and caret line.
545///
546/// `file`/`line` come from the raising site's input; `source_window` is the
547/// current input line text plus the 0-based caret column (upstream
548/// `xmlParserInputGetWindow`).
549///
550/// # SAFETY
551///
552/// - `file` and `message` must be valid C strings or NULL.
553/// - `source_window` bytes must be valid for the duration of the call.
554#[cfg(target_arch = "x86_64")]
555unsafe fn format_error_streamed(
556    domain: c_int,
557    code: c_int,
558    level: c_int,
559    message: *const c_char,
560    file: *const c_char,
561    line: c_int,
562    source_window: Option<(&[u8], usize)>,
563    enc_bytes: Option<[u8; 4]>,
564) {
565    // SAFETY: reads the exported C globals (upstream reads the same).
566    let Some(handler) = globals::get_generic_error_func() else {
567        return;
568    };
569    let data = globals::get_generic_error_ctx();
570
571    // 1. File/line prefix (xmlFormatError).
572    if !file.is_null() {
573        ch_call2(
574            handler,
575            data,
576            b"%s:%d: \0".as_ptr() as *const c_char,
577            file as usize,
578            line as usize,
579        );
580    } else if line != 0
581        && (domain == XML_FROM_PARSER
582            || domain == XML_FROM_SCHEMASV
583            || domain == XML_FROM_SCHEMASP
584            || domain == XML_FROM_DTD
585            || domain == XML_FROM_RELAXNGP
586            || domain == XML_FROM_RELAXNGV)
587    {
588        ch_call1(
589            handler,
590            data,
591            b"Entity: line %d: \0".as_ptr() as *const c_char,
592            line as usize,
593        );
594    }
595
596    // 2. Domain fragment (xmlFormatError switch).
597    let dom: &[u8] = match domain {
598        XML_FROM_PARSER => b"parser \0",
599        XML_FROM_NAMESPACE => b"namespace \0",
600        XML_FROM_DTD | XML_FROM_VALID => b"validity \0",
601        XML_FROM_HTML => b"HTML parser \0",
602        XML_FROM_MEMORY => b"memory \0",
603        XML_FROM_OUTPUT => b"output \0",
604        XML_FROM_IO => b"I/O \0",
605        XML_FROM_XINCLUDE => b"XInclude \0",
606        XML_FROM_XPATH => b"XPath \0",
607        XML_FROM_XPOINTER => b"parser \0",
608        XML_FROM_REGEXP => b"regexp \0",
609        XML_FROM_MODULE => b"module \0",
610        XML_FROM_SCHEMASV => b"Schemas validity \0",
611        XML_FROM_SCHEMASP => b"Schemas parser \0",
612        XML_FROM_RELAXNGP => b"Relax-NG parser \0",
613        XML_FROM_RELAXNGV => b"Relax-NG validity \0",
614        XML_FROM_CATALOG => b"Catalog \0",
615        XML_FROM_C14N => b"C14N \0",
616        XML_FROM_XSLT => b"XSLT \0",
617        XML_FROM_I18N => b"encoding \0",
618        XML_FROM_SCHEMATRONV => b"schematron \0",
619        XML_FROM_BUFFER => b"internal buffer \0",
620        XML_FROM_URI => b"URI \0",
621        _ => b"\0",
622    };
623    if !dom.is_empty() && dom[0] != 0 {
624        ch_call0(handler, data, dom.as_ptr() as *const c_char);
625    }
626
627    // 3. Level fragment (xmlFormatError switch).
628    let lvl: &[u8] = if level == XML_ERR_NONE as c_int {
629        b": \0"
630    } else if level == XML_ERR_WARNING as c_int {
631        b"warning : \0"
632    } else if level == XML_ERR_ERROR as c_int || level == XML_ERR_FATAL as c_int {
633        b"error : \0"
634    } else {
635        b"\0"
636    };
637    if !lvl.is_empty() && lvl[0] != 0 {
638        ch_call0(handler, data, lvl.as_ptr() as *const c_char);
639    }
640
641    // 4. Message fragment.
642    if !message.is_null() {
643        let msg = message as *const u8;
644        let mut len = 0usize;
645        while unsafe { *msg.add(len) } != 0 {
646            len += 1;
647        }
648        let ends_nl = len > 0 && unsafe { *msg.add(len - 1) } == b'\n';
649        let fmt: &[u8] = if ends_nl { b"%s\0" } else { b"%s\n\0" };
650        ch_call1(handler, data, fmt.as_ptr() as *const c_char, msg as usize);
651    }
652
653    // 4b. Invalid-encoding byte dump (upstream xmlFormatError: the first 4
654    // bytes at the error position, only for XML_ERR_INVALID_ENCODING).
655    if code == XML_ERR_INVALID_ENCODING {
656        if let Some(bytes) = enc_bytes {
657            ch_call0(handler, data, b"Bytes:\0".as_ptr() as *const c_char);
658            for b in bytes {
659                // " 0x%02X"
660                let hex = format!(" 0x{:02X}\0", b);
661                ch_call0(handler, data, hex.as_ptr() as *const c_char);
662            }
663            ch_call0(handler, data, b"\n\0".as_ptr() as *const c_char);
664        }
665    }
666
667    // 5. Source window + caret (xmlParserPrintFileContextInternal).
668    if let Some((window, caret)) = source_window {
669        let mut win = window.to_vec();
670        win.push(0);
671        ch_call1(
672            handler,
673            data,
674            b"%s\n\0".as_ptr() as *const c_char,
675            win.as_ptr() as usize,
676        );
677        let mut caret_line = Vec::with_capacity(caret + 2);
678        for &b in window.iter().take(caret) {
679            caret_line.push(if b == b'\t' { b'\t' } else { b' ' });
680        }
681        caret_line.push(b'^');
682        caret_line.push(0);
683        ch_call1(
684            handler,
685            data,
686            b"%s\n\0".as_ptr() as *const c_char,
687            caret_line.as_ptr() as usize,
688        );
689    }
690}
691
692/// How a raise delivers to the generic side of the error system (upstream
693/// `xmlVRaiseError` channel selection, error.c 2.15).
694#[derive(Clone, Copy)]
695pub enum GenericDelivery {
696    /// Custom SAX channel: single call `channel(ctx, msg)`.
697    Custom(xmlGenericErrorFunc, *mut c_void),
698    /// Legacy/default channel: stream the `xmlFormatError` fragments through
699    /// the global generic handler.
700    Stream,
701    /// No channel (SAX slot NULL): no generic delivery.
702    None,
703}
704
705/// Raise an error with upstream's full routing (error.c 2.15
706/// `xmlVRaiseError`): update the last error, then deliver to the structured
707/// handler **or** the selected generic channel — never both.
708///
709/// `file`/`line`/`source_window` feed the generic fragment stream (the
710/// structured handler receives the complete `xmlError` instead). `col` is
711/// the 1-based byte column (upstream `input->col` → `err->int2`); `str1`..
712/// `str3`/`int1` are the upstream extra fields; `enc_bytes` feeds the
713/// `XML_ERR_INVALID_ENCODING` "Bytes:" fragment.
714///
715/// # UPSTREAM-PARITY (ownership)
716///
717/// Like upstream `xmlVSetError`, every string field of the stored error is
718/// owned (`xmlStrdup`): `file`/`str1`/`str2`/`str3` are heap copies, so the
719/// caller may pass transient C strings.
720///
721/// # SAFETY
722///
723/// - `ctxt` may be NULL.
724/// - `msg`, `file`, `str1`, `str2`, `str3` must be valid C strings or NULL.
725/// - `source_window` bytes must be valid for the duration of the call.
726pub unsafe fn raise_error_streamed(
727    ctxt: *mut c_void,
728    domain: c_int,
729    code: c_int,
730    level: c_int,
731    file: *const c_char,
732    line: c_int,
733    col: c_int,
734    str1: *const c_char,
735    str2: *const c_char,
736    str3: *const c_char,
737    int1: c_int,
738    msg: *const c_char,
739    source_window: Option<(&[u8], usize)>,
740    enc_bytes: Option<[u8; 4]>,
741    delivery: GenericDelivery,
742) {
743    // The streamed generic-error channel below uses an x86_64 SysV va_list
744    // trampoline (ch_call0/1/2 — register-based). Other ABIs (i686 cdecl,
745    // ARM/aarch64 AAPCS, ...) fall back to the plain raise path; full
746    // streamed-fragment parity there is an unexecuted platform obligation
747    // (atlas/PLATFORM_SURFACE_ATLAS.md, OBLIG-WORDSIZE-32 / compiler-ABI).
748    #[cfg(not(target_arch = "x86_64"))]
749    {
750        raise_error(
751            ctxt,
752            ptr::null_mut(),
753            ptr::null_mut(),
754            ptr::null_mut(),
755            ptr::null_mut(),
756            domain,
757            code,
758            level,
759            file,
760            line,
761            str1,
762            str2,
763            str3,
764            int1,
765            col,
766            msg,
767        );
768        return;
769    }
770
771    #[cfg(target_arch = "x86_64")]
772    {
773        // UPSTREAM-PARITY (xmlVRaiseError): warnings are suppressed when the
774        // global xmlGetWarningsDefaultValue is zero.
775        if level == xmlErrorLevel::XML_ERR_WARNING as c_int
776            && unsafe { crate::abi::data_globals::xmlGetWarningsDefaultValue } == 0
777        {
778            return;
779        }
780
781        raise_error_streamed_x86_64(
782            ctxt,
783            domain,
784            code,
785            level,
786            file,
787            line,
788            col,
789            str1,
790            str2,
791            str3,
792            int1,
793            msg,
794            source_window,
795            enc_bytes,
796            delivery,
797        );
798    }
799}
800
801/// x86-64 streamed raise (SysV va_list channel). See `raise_error_streamed`.
802#[cfg(target_arch = "x86_64")]
803unsafe fn raise_error_streamed_x86_64(
804    ctxt: *mut c_void,
805    domain: c_int,
806    code: c_int,
807    level: c_int,
808    file: *const c_char,
809    line: c_int,
810    col: c_int,
811    str1: *const c_char,
812    str2: *const c_char,
813    str3: *const c_char,
814    int1: c_int,
815    msg: *const c_char,
816    source_window: Option<(&[u8], usize)>,
817    enc_bytes: Option<[u8; 4]>,
818    delivery: GenericDelivery,
819) {
820    // UPSTREAM-PARITY (xmlVRaiseError): warnings are suppressed when the
821    // global xmlGetWarningsDefaultValue is zero.
822    if level == xmlErrorLevel::XML_ERR_WARNING as c_int
823        && unsafe { crate::abi::data_globals::xmlGetWarningsDefaultValue } == 0
824    {
825        return;
826    }
827
828    // Format the error message (same as raise_error).
829    let formatted_msg = format_error_message(domain, code, msg, str1, str2, str3);
830    let file_copy = if file.is_null() {
831        ptr::null_mut()
832    } else {
833        crate::abi::allocator::xmlMemStrdupImpl(file) as *mut c_char
834    };
835    let str1_copy = if str1.is_null() {
836        ptr::null_mut()
837    } else {
838        crate::abi::allocator::xmlMemStrdupImpl(str1) as *mut c_char
839    };
840    let str2_copy = if str2.is_null() {
841        ptr::null_mut()
842    } else {
843        crate::abi::allocator::xmlMemStrdupImpl(str2) as *mut c_char
844    };
845    let str3_copy = if str3.is_null() {
846        ptr::null_mut()
847    } else {
848        crate::abi::allocator::xmlMemStrdupImpl(str3) as *mut c_char
849    };
850    let err = _xmlError {
851        domain,
852        code,
853        message: formatted_msg,
854        level,
855        file: file_copy,
856        line,
857        str1: str1_copy,
858        str2: str2_copy,
859        str3: str3_copy,
860        int1,
861        int2: col,
862        ctxt,
863        node: ptr::null_mut(),
864    };
865
866    globals::set_last_error(err);
867
868    // Structured handler wins (upstream `else if` chain).
869    if let Some(handler) = globals::get_structured_error_func() {
870        let ctx = globals::get_structured_error_ctx();
871        let err_ref = globals::get_last_error();
872        if !err_ref.is_null() {
873            handler(ctx, err_ref as *const _xmlError);
874        }
875        return;
876    }
877
878    match delivery {
879        GenericDelivery::Custom(channel, ctx) => {
880            if !msg.is_null() {
881                // SAFETY: the caller provided a valid C callback.
882                unsafe { channel(ctx, msg) };
883            }
884        }
885        GenericDelivery::Stream => {
886            if globals::get_generic_error_func().is_some() {
887                unsafe {
888                    format_error_streamed(
889                        domain,
890                        code,
891                        level,
892                        formatted_msg,
893                        file,
894                        line,
895                        source_window,
896                        enc_bytes,
897                    )
898                };
899            }
900        }
901        GenericDelivery::None => {}
902    }
903}
904
905/// Default SAX v1 error handler — `void xmlParserError(void *ctx, const char *msg, ...)`.
906///
907/// # SAFETY
908///
909/// - `ctx` may be NULL (unused by the candidate's legacy path).
910/// - `msg` must be a valid NUL-terminated C string or NULL.
911#[no_mangle]
912pub unsafe extern "C" fn xmlParserError(ctx: *mut c_void, msg: *const c_char) {
913    let _ = ctx;
914    unsafe { emit_legacy_message("error", msg) };
915}
916
917/// Default SAX v1 warning handler — `void xmlParserWarning(void *ctx, const char *msg, ...)`.
918#[no_mangle]
919pub unsafe extern "C" fn xmlParserWarning(ctx: *mut c_void, msg: *const c_char) {
920    let _ = ctx;
921    unsafe { emit_legacy_message("warning", msg) };
922}
923
924/// Default validity error handler — `void xmlParserValidityError(void *ctx, const char *msg, ...)`.
925#[no_mangle]
926pub unsafe extern "C" fn xmlParserValidityError(ctx: *mut c_void, msg: *const c_char) {
927    let _ = ctx;
928    unsafe { emit_legacy_message("validity error", msg) };
929}
930
931/// Default validity warning handler — `void xmlParserValidityWarning(void *ctx, const char *msg, ...)`.
932#[no_mangle]
933pub unsafe extern "C" fn xmlParserValidityWarning(ctx: *mut c_void, msg: *const c_char) {
934    let _ = ctx;
935    unsafe { emit_legacy_message("validity warning", msg) };
936}
937
938// ═══════════════════════════════════════════════════════════════════════════════
939// Tests
940// ═══════════════════════════════════════════════════════════════════════════════
941
942#[cfg(test)]
943mod tests {
944    use super::*;
945    use crate::abi::allocator;
946    use core::ffi::c_void;
947
948    #[test]
949    fn test_error_default_reset() {
950        unsafe {
951            let mut err = _xmlError {
952                domain: XML_FROM_PARSER,
953                code: XML_ERR_NO_MEMORY,
954                message: ptr::null_mut(),
955                level: XML_ERR_ERROR as c_int,
956                file: ptr::null_mut(),
957                line: 42,
958                str1: ptr::null_mut(),
959                str2: ptr::null_mut(),
960                str3: ptr::null_mut(),
961                int1: 0,
962                int2: 0,
963                ctxt: ptr::null_mut(),
964                node: ptr::null_mut(),
965            };
966
967            reset_error(&mut err);
968            assert_eq!(err.domain, XML_FROM_NONE);
969            assert_eq!(err.code, XML_ERR_OK as c_int);
970            assert_eq!(err.level, XML_ERR_NONE as c_int);
971            assert_eq!(err.line, 0);
972        }
973    }
974
975    #[test]
976    fn test_copy_error() {
977        unsafe {
978            let from = _xmlError {
979                domain: XML_FROM_PARSER,
980                code: XML_ERR_NO_MEMORY,
981                message: ptr::null_mut(),
982                level: XML_ERR_FATAL as c_int,
983                file: ptr::null_mut(),
984                line: 100,
985                str1: ptr::null_mut(),
986                str2: ptr::null_mut(),
987                str3: ptr::null_mut(),
988                int1: 1,
989                int2: 2,
990                ctxt: ptr::null_mut(),
991                node: ptr::null_mut(),
992            };
993            let mut to = _xmlError {
994                domain: XML_FROM_NONE,
995                code: XML_ERR_OK as c_int,
996                message: ptr::null_mut(),
997                level: XML_ERR_NONE as c_int,
998                file: ptr::null_mut(),
999                line: 0,
1000                str1: ptr::null_mut(),
1001                str2: ptr::null_mut(),
1002                str3: ptr::null_mut(),
1003                int1: 0,
1004                int2: 0,
1005                ctxt: ptr::null_mut(),
1006                node: ptr::null_mut(),
1007            };
1008
1009            let result = copy_error(&from, &mut to);
1010            assert_eq!(result, 0);
1011            assert_eq!(to.domain, XML_FROM_PARSER);
1012            assert_eq!(to.code, XML_ERR_NO_MEMORY);
1013            assert_eq!(to.level, XML_ERR_FATAL as c_int);
1014            assert_eq!(to.line, 100);
1015            assert_eq!(to.int1, 1);
1016            assert_eq!(to.int2, 2);
1017        }
1018    }
1019
1020    #[test]
1021    fn test_raise_and_get_last_error() {
1022        unsafe {
1023            reset_last_error();
1024            assert!(get_last_error().is_null());
1025
1026            let file = b"test.xml\0" as *const u8 as *const c_char;
1027            let str1 = b"element\0" as *const u8 as *const c_char;
1028
1029            raise_error(
1030                ptr::null_mut(),
1031                ptr::null_mut(),
1032                ptr::null_mut(),
1033                ptr::null_mut(),
1034                ptr::null_mut(),
1035                XML_FROM_PARSER,
1036                XML_ERR_TAG_NAME_MISMATCH,
1037                XML_ERR_ERROR as c_int,
1038                file,
1039                10,
1040                str1,
1041                ptr::null(),
1042                ptr::null(),
1043                0,
1044                0,
1045                ptr::null(),
1046            );
1047
1048            let last = get_last_error();
1049            assert!(!last.is_null());
1050            assert_eq!((*last).domain, XML_FROM_PARSER);
1051            assert_eq!((*last).code, XML_ERR_TAG_NAME_MISMATCH);
1052            assert_eq!((*last).level, XML_ERR_ERROR as c_int);
1053            assert_eq!((*last).line, 10);
1054
1055            // Check file was stored
1056            let last_file = (*last).file;
1057            assert!(!last_file.is_null());
1058
1059            reset_last_error();
1060            assert!(get_last_error().is_null());
1061        }
1062    }
1063
1064    #[test]
1065    fn test_structured_error_callback() {
1066        unsafe {
1067            reset_last_error();
1068
1069            // Set up a structured error handler that captures the error
1070            let mut captured_domain: c_int = 0;
1071            let captured_ptr = &mut captured_domain as *mut c_int as *mut c_void;
1072
1073            // SAFETY: The callback writes to captured_ptr which lives on the stack
1074            // for the duration of this test.
1075            extern "C" fn test_handler(ctx: *mut c_void, _err: *const _xmlError) {
1076                // SAFETY: ctx is valid for the test duration.
1077                unsafe {
1078                    let captured = &mut *(ctx as *mut c_int);
1079                    *captured = 42;
1080                }
1081            }
1082
1083            set_structured_error_func(captured_ptr, Some(test_handler as xmlStructuredErrorFunc));
1084
1085            raise_error(
1086                ptr::null_mut(),
1087                ptr::null_mut(),
1088                ptr::null_mut(),
1089                ptr::null_mut(),
1090                ptr::null_mut(),
1091                XML_FROM_PARSER,
1092                XML_ERR_OK as c_int,
1093                XML_ERR_WARNING as c_int,
1094                ptr::null(),
1095                0,
1096                ptr::null(),
1097                ptr::null(),
1098                ptr::null(),
1099                0,
1100                0,
1101                ptr::null(),
1102            );
1103
1104            assert_eq!(captured_domain, 42);
1105
1106            // Reset
1107            set_structured_error_func(ptr::null_mut(), None);
1108            reset_last_error();
1109        }
1110    }
1111
1112    #[test]
1113    fn test_format_error_message() {
1114        unsafe {
1115            // Test with direct message
1116            let msg = b"test error\0" as *const u8 as *const c_char;
1117            let formatted = format_error_message(
1118                XML_FROM_NONE,
1119                XML_ERR_OK as c_int,
1120                msg,
1121                ptr::null(),
1122                ptr::null(),
1123                ptr::null(),
1124            );
1125            assert!(!formatted.is_null());
1126            let formatted_str = std::ffi::CStr::from_ptr(formatted);
1127            assert_eq!(formatted_str.to_bytes(), b"test error");
1128
1129            // Free the allocated message
1130            allocator::xmlFreeImpl(formatted as *mut c_void);
1131
1132            // Test with domain and str1
1133            let str1 = b"foo\0" as *const u8 as *const c_char;
1134            let formatted2 = format_error_message(
1135                XML_FROM_PARSER,
1136                XML_ERR_OK as c_int,
1137                ptr::null(),
1138                str1,
1139                ptr::null(),
1140                ptr::null(),
1141            );
1142            assert!(!formatted2.is_null());
1143            allocator::xmlFreeImpl(formatted2 as *mut c_void);
1144        }
1145    }
1146}
libxml_rs/xml/errors/mod.rs

libxml_rs/xml/errors/
mod.rs
