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    // Call the structured error handler if set
411    if let Some(handler) = globals::get_structured_error_func() {
412        let ctx = globals::get_structured_error_ctx();
413        let err_ref = globals::get_last_error();
414        if !err_ref.is_null() {
415            handler(ctx, err_ref as *const _xmlError);
416        }
417    }
418
419    // Call the generic error handler if set (for warnings/errors)
420    if let Some(handler) = globals::get_generic_error_func() {
421        if level != 0 {
422            let ctx = globals::get_generic_error_ctx();
423            if !formatted_msg.is_null() {
424                handler(ctx, formatted_msg as *const core::ffi::c_char);
425            } else if !msg.is_null() {
426                handler(ctx, msg);
427            }
428        }
429    }
430
431    // Free the formatted message if it was allocated
432    // Note: We keep it as the last error's message, so we don't free it here.
433    // The next call to raise_error or reset_error will free the old message.
434    // Actually, in Phase 1, we don't free because the message is the last error's.
435    // A more complete implementation would free the old message when setting a new one.
436}
437
438/// Emit a legacy-format message through the generic error channel.
439///
440/// Upstream's `xmlGenericErrorDefaultFunc` writes the message to stderr;
441/// when a custom generic handler is installed it receives the message. The
442/// `level` prefix matches upstream `xmlVFormatLegacyError` (error.c 2.15).
443unsafe fn emit_legacy_message(level: &str, msg: *const c_char) {
444    if msg.is_null() {
445        return;
446    }
447    let len = libc::strlen(msg) as usize;
448    let text = core::slice::from_raw_parts(msg as *const u8, len);
449    let mut full = Vec::with_capacity(level.len() + 2 + len);
450    full.extend_from_slice(level.as_bytes());
451    full.push(b':');
452    full.push(b' ');
453    full.extend_from_slice(text);
454    if let Some(handler) = globals::get_generic_error_func() {
455        let ctx = globals::get_generic_error_ctx();
456        let mut cmsg = full.clone();
457        cmsg.push(0);
458        handler(ctx, cmsg.as_ptr() as *const c_char);
459    } else {
460        // Upstream default (xmlGenericErrorDefaultFunc): stderr.
461        let _ = libc::write(2, full.as_ptr() as *const libc::c_void, full.len());
462    }
463}
464
465// ═══════════════════════════════════════════════════════════════════════════════
466// Generic-channel fragment streaming (upstream error.c `xmlFormatError`)
467// ═══════════════════════════════════════════════════════════════════════════════
468//
469// Upstream streams each error through the generic channel as a sequence of
470// variadic calls (e.g. `channel(data, "%s:%d: ", file, line)` followed by the
471// domain, level, message and source-context fragments). Custom handlers and the
472// built-in default (an x86_64 SysV va_list shim, see data_globals.rs) both
473// observe the same per-fragment calls. Stable Rust cannot express a variadic
474// call, so each fragment goes through a tiny x86_64 trampoline that places the
475// fixed arguments in the ABI registers and does an indirect call.
476
477/// `channel(data, fmt)` — no variadic arguments.
478#[cfg(target_arch = "x86_64")]
479#[inline]
480unsafe fn ch_call0(handler: xmlGenericErrorFunc, data: *mut c_void, fmt: *const c_char) {
481    // SAFETY: `handler` is a C-compatible generic error callback; per the
482    // SysV ABI the callee sees (data, fmt) with no additional registers
483    // consumed (rdx/rcx zeroed so a va_list-reading callee finds nothing).
484    // The compiler guarantees 16-byte stack alignment at the asm block, so
485    // the `call` is correctly aligned.
486    unsafe {
487        core::arch::asm!(
488            "xor edx, edx",
489            "xor ecx, ecx",
490            "call {h}",
491            h = in(reg) handler as usize,
492            in("rdi") data,
493            in("rsi") fmt,
494            out("rdx") _, out("rcx") _,
495            lateout("rax") _, lateout("r8") _, lateout("r9") _, lateout("r10") _, lateout("r11") _,
496        );
497    }
498}
499
500/// `channel(data, fmt, a1)` — one pointer-sized variadic argument.
501#[cfg(target_arch = "x86_64")]
502#[inline]
503unsafe fn ch_call1(handler: xmlGenericErrorFunc, data: *mut c_void, fmt: *const c_char, a1: usize) {
504    // SAFETY: as ch_call0; `a1` lands in the va_list slot after the two
505    // fixed args (rdx).
506    unsafe {
507        core::arch::asm!(
508            "xor ecx, ecx",
509            "call {h}",
510            h = in(reg) handler as usize,
511            in("rdi") data,
512            in("rsi") fmt,
513            in("rdx") a1,
514            out("rcx") _,
515            lateout("rax") _, lateout("r8") _, lateout("r9") _, lateout("r10") _, lateout("r11") _,
516        );
517    }
518}
519
520/// `channel(data, fmt, a1, a2)` — two pointer-sized variadic arguments.
521#[cfg(target_arch = "x86_64")]
522#[inline]
523unsafe fn ch_call2(
524    handler: xmlGenericErrorFunc,
525    data: *mut c_void,
526    fmt: *const c_char,
527    a1: usize,
528    a2: usize,
529) {
530    // SAFETY: as ch_call0; a1/a2 land in the va_list slots (rdx, rcx).
531    unsafe {
532        core::arch::asm!(
533            "call {h}",
534            h = in(reg) handler as usize,
535            in("rdi") data,
536            in("rsi") fmt,
537            in("rdx") a1,
538            in("rcx") a2,
539            lateout("rax") _, lateout("r8") _, lateout("r9") _, lateout("r10") _, lateout("r11") _,
540        );
541    }
542}
543
544/// Emit one raise through the generic channel with upstream's
545/// `xmlFormatError` fragment sequence (error.c 2.15): file/line prefix,
546/// domain, level, message, then the source window and caret line.
547///
548/// `file`/`line` come from the raising site's input; `source_window` is the
549/// current input line text plus the 0-based caret column (upstream
550/// `xmlParserInputGetWindow`).
551///
552/// # SAFETY
553///
554/// - `file` and `message` must be valid C strings or NULL.
555/// - `source_window` bytes must be valid for the duration of the call.
556#[cfg(target_arch = "x86_64")]
557unsafe fn format_error_streamed(
558    domain: c_int,
559    code: c_int,
560    level: c_int,
561    message: *const c_char,
562    file: *const c_char,
563    line: c_int,
564    source_window: Option<(&[u8], usize)>,
565    enc_bytes: Option<[u8; 4]>,
566) {
567    // SAFETY: reads the exported C globals (upstream reads the same).
568    let Some(handler) = globals::get_generic_error_func() else {
569        return;
570    };
571    let data = globals::get_generic_error_ctx();
572
573    // 1. File/line prefix (xmlFormatError).
574    if !file.is_null() {
575        ch_call2(
576            handler,
577            data,
578            b"%s:%d: \0".as_ptr() as *const c_char,
579            file as usize,
580            line as usize,
581        );
582    } else if line != 0
583        && (domain == XML_FROM_PARSER
584            || domain == XML_FROM_SCHEMASV
585            || domain == XML_FROM_SCHEMASP
586            || domain == XML_FROM_DTD
587            || domain == XML_FROM_RELAXNGP
588            || domain == XML_FROM_RELAXNGV)
589    {
590        ch_call1(
591            handler,
592            data,
593            b"Entity: line %d: \0".as_ptr() as *const c_char,
594            line as usize,
595        );
596    }
597
598    // 2. Domain fragment (xmlFormatError switch).
599    let dom: &[u8] = match domain {
600        XML_FROM_PARSER => b"parser \0",
601        XML_FROM_NAMESPACE => b"namespace \0",
602        XML_FROM_DTD | XML_FROM_VALID => b"validity \0",
603        XML_FROM_HTML => b"HTML parser \0",
604        XML_FROM_MEMORY => b"memory \0",
605        XML_FROM_OUTPUT => b"output \0",
606        XML_FROM_IO => b"I/O \0",
607        XML_FROM_XINCLUDE => b"XInclude \0",
608        XML_FROM_XPATH => b"XPath \0",
609        XML_FROM_XPOINTER => b"parser \0",
610        XML_FROM_REGEXP => b"regexp \0",
611        XML_FROM_MODULE => b"module \0",
612        XML_FROM_SCHEMASV => b"Schemas validity \0",
613        XML_FROM_SCHEMASP => b"Schemas parser \0",
614        XML_FROM_RELAXNGP => b"Relax-NG parser \0",
615        XML_FROM_RELAXNGV => b"Relax-NG validity \0",
616        XML_FROM_CATALOG => b"Catalog \0",
617        XML_FROM_C14N => b"C14N \0",
618        XML_FROM_XSLT => b"XSLT \0",
619        XML_FROM_I18N => b"encoding \0",
620        XML_FROM_SCHEMATRONV => b"schematron \0",
621        XML_FROM_BUFFER => b"internal buffer \0",
622        XML_FROM_URI => b"URI \0",
623        _ => b"\0",
624    };
625    if !dom.is_empty() && dom[0] != 0 {
626        ch_call0(handler, data, dom.as_ptr() as *const c_char);
627    }
628
629    // 3. Level fragment (xmlFormatError switch).
630    let lvl: &[u8] = if level == XML_ERR_NONE as c_int {
631        b": \0"
632    } else if level == XML_ERR_WARNING as c_int {
633        b"warning : \0"
634    } else if level == XML_ERR_ERROR as c_int || level == XML_ERR_FATAL as c_int {
635        b"error : \0"
636    } else {
637        b"\0"
638    };
639    if !lvl.is_empty() && lvl[0] != 0 {
640        ch_call0(handler, data, lvl.as_ptr() as *const c_char);
641    }
642
643    // 4. Message fragment.
644    if !message.is_null() {
645        let msg = message as *const u8;
646        let mut len = 0usize;
647        while unsafe { *msg.add(len) } != 0 {
648            len += 1;
649        }
650        let ends_nl = len > 0 && unsafe { *msg.add(len - 1) } == b'\n';
651        let fmt: &[u8] = if ends_nl { b"%s\0" } else { b"%s\n\0" };
652        ch_call1(handler, data, fmt.as_ptr() as *const c_char, msg as usize);
653    }
654
655    // 4b. Invalid-encoding byte dump (upstream xmlFormatError: the first 4
656    // bytes at the error position, only for XML_ERR_INVALID_ENCODING).
657    if code == XML_ERR_INVALID_ENCODING {
658        if let Some(bytes) = enc_bytes {
659            ch_call0(handler, data, b"Bytes:\0".as_ptr() as *const c_char);
660            for b in bytes {
661                // " 0x%02X"
662                let hex = format!(" 0x{:02X}\0", b);
663                ch_call0(handler, data, hex.as_ptr() as *const c_char);
664            }
665            ch_call0(handler, data, b"\n\0".as_ptr() as *const c_char);
666        }
667    }
668
669    // 5. Source window + caret (xmlParserPrintFileContextInternal).
670    if let Some((window, caret)) = source_window {
671        let mut win = window.to_vec();
672        win.push(0);
673        ch_call1(
674            handler,
675            data,
676            b"%s\n\0".as_ptr() as *const c_char,
677            win.as_ptr() as usize,
678        );
679        let mut caret_line = Vec::with_capacity(caret + 2);
680        for &b in window.iter().take(caret) {
681            caret_line.push(if b == b'\t' { b'\t' } else { b' ' });
682        }
683        caret_line.push(b'^');
684        caret_line.push(0);
685        ch_call1(
686            handler,
687            data,
688            b"%s\n\0".as_ptr() as *const c_char,
689            caret_line.as_ptr() as usize,
690        );
691    }
692}
693
694/// How a raise delivers to the generic side of the error system (upstream
695/// `xmlVRaiseError` channel selection, error.c 2.15).
696#[derive(Clone, Copy)]
697pub enum GenericDelivery {
698    /// Custom SAX channel: single call `channel(ctx, msg)`.
699    Custom(xmlGenericErrorFunc, *mut c_void),
700    /// Legacy/default channel: stream the `xmlFormatError` fragments through
701    /// the global generic handler.
702    Stream,
703    /// No channel (SAX slot NULL): no generic delivery.
704    None,
705}
706
707/// Raise an error with upstream's full routing (error.c 2.15
708/// `xmlVRaiseError`): update the last error, then deliver to the structured
709/// handler **or** the selected generic channel — never both.
710///
711/// `file`/`line`/`source_window` feed the generic fragment stream (the
712/// structured handler receives the complete `xmlError` instead). `col` is
713/// the 1-based byte column (upstream `input->col` → `err->int2`); `str1`..
714/// `str3`/`int1` are the upstream extra fields; `enc_bytes` feeds the
715/// `XML_ERR_INVALID_ENCODING` "Bytes:" fragment.
716///
717/// # UPSTREAM-PARITY (ownership)
718///
719/// Like upstream `xmlVSetError`, every string field of the stored error is
720/// owned (`xmlStrdup`): `file`/`str1`/`str2`/`str3` are heap copies, so the
721/// caller may pass transient C strings.
722///
723/// # SAFETY
724///
725/// - `ctxt` may be NULL.
726/// - `msg`, `file`, `str1`, `str2`, `str3` must be valid C strings or NULL.
727/// - `source_window` bytes must be valid for the duration of the call.
728pub unsafe fn raise_error_streamed(
729    ctxt: *mut c_void,
730    domain: c_int,
731    code: c_int,
732    level: c_int,
733    file: *const c_char,
734    line: c_int,
735    col: c_int,
736    str1: *const c_char,
737    str2: *const c_char,
738    str3: *const c_char,
739    int1: c_int,
740    msg: *const c_char,
741    source_window: Option<(&[u8], usize)>,
742    enc_bytes: Option<[u8; 4]>,
743    delivery: GenericDelivery,
744) {
745    // The streamed generic-error channel below uses an x86_64 SysV va_list
746    // trampoline (ch_call0/1/2 — register-based). Other ABIs (i686 cdecl,
747    // ARM/aarch64 AAPCS, ...) fall back to the plain raise path; full
748    // streamed-fragment parity there is an unexecuted platform obligation
749    // (atlas/PLATFORM_SURFACE_ATLAS.md, OBLIG-WORDSIZE-32 / compiler-ABI).
750    #[cfg(not(target_arch = "x86_64"))]
751    {
752        raise_error(
753            ctxt,
754            ptr::null_mut(),
755            ptr::null_mut(),
756            ptr::null_mut(),
757            ptr::null_mut(),
758            domain,
759            code,
760            level,
761            file,
762            line,
763            str1,
764            str2,
765            str3,
766            int1,
767            col,
768            msg,
769        );
770        return;
771    }
772
773    #[cfg(target_arch = "x86_64")]
774    {
775        // UPSTREAM-PARITY (xmlVRaiseError): warnings are suppressed when the
776        // global xmlGetWarningsDefaultValue is zero.
777        if level == xmlErrorLevel::XML_ERR_WARNING as c_int
778            && unsafe { crate::abi::data_globals::xmlGetWarningsDefaultValue } == 0
779        {
780            return;
781        }
782
783        raise_error_streamed_x86_64(
784            ctxt,
785            domain,
786            code,
787            level,
788            file,
789            line,
790            col,
791            str1,
792            str2,
793            str3,
794            int1,
795            msg,
796            source_window,
797            enc_bytes,
798            delivery,
799        );
800    }
801}
802
803/// x86-64 streamed raise (SysV va_list channel). See `raise_error_streamed`.
804#[cfg(target_arch = "x86_64")]
805unsafe fn raise_error_streamed_x86_64(
806    ctxt: *mut c_void,
807    domain: c_int,
808    code: c_int,
809    level: c_int,
810    file: *const c_char,
811    line: c_int,
812    col: c_int,
813    str1: *const c_char,
814    str2: *const c_char,
815    str3: *const c_char,
816    int1: c_int,
817    msg: *const c_char,
818    source_window: Option<(&[u8], usize)>,
819    enc_bytes: Option<[u8; 4]>,
820    delivery: GenericDelivery,
821) {
822    // UPSTREAM-PARITY (xmlVRaiseError): warnings are suppressed when the
823    // global xmlGetWarningsDefaultValue is zero.
824    if level == xmlErrorLevel::XML_ERR_WARNING as c_int
825        && unsafe { crate::abi::data_globals::xmlGetWarningsDefaultValue } == 0
826    {
827        return;
828    }
829
830    // Format the error message (same as raise_error).
831    let formatted_msg = format_error_message(domain, code, msg, str1, str2, str3);
832    let file_copy = if file.is_null() {
833        ptr::null_mut()
834    } else {
835        crate::abi::allocator::xmlMemStrdupImpl(file) as *mut c_char
836    };
837    let str1_copy = if str1.is_null() {
838        ptr::null_mut()
839    } else {
840        crate::abi::allocator::xmlMemStrdupImpl(str1) as *mut c_char
841    };
842    let str2_copy = if str2.is_null() {
843        ptr::null_mut()
844    } else {
845        crate::abi::allocator::xmlMemStrdupImpl(str2) as *mut c_char
846    };
847    let str3_copy = if str3.is_null() {
848        ptr::null_mut()
849    } else {
850        crate::abi::allocator::xmlMemStrdupImpl(str3) as *mut c_char
851    };
852    let err = _xmlError {
853        domain,
854        code,
855        message: formatted_msg,
856        level,
857        file: file_copy,
858        line,
859        str1: str1_copy,
860        str2: str2_copy,
861        str3: str3_copy,
862        int1,
863        int2: col,
864        ctxt,
865        node: ptr::null_mut(),
866    };
867
868    globals::set_last_error(err);
869
870    // Structured handler wins (upstream `else if` chain).
871    if let Some(handler) = globals::get_structured_error_func() {
872        let ctx = globals::get_structured_error_ctx();
873        let err_ref = globals::get_last_error();
874        if !err_ref.is_null() {
875            handler(ctx, err_ref as *const _xmlError);
876        }
877        return;
878    }
879
880    match delivery {
881        GenericDelivery::Custom(channel, ctx) => {
882            if !msg.is_null() {
883                // SAFETY: the caller provided a valid C callback.
884                unsafe { channel(ctx, msg) };
885            }
886        }
887        GenericDelivery::Stream => {
888            if globals::get_generic_error_func().is_some() {
889                unsafe {
890                    format_error_streamed(
891                        domain,
892                        code,
893                        level,
894                        formatted_msg,
895                        file,
896                        line,
897                        source_window,
898                        enc_bytes,
899                    )
900                };
901            }
902        }
903        GenericDelivery::None => {}
904    }
905}
906
907/// Default SAX v1 error handler — `void xmlParserError(void *ctx, const char *msg, ...)`.
908///
909/// # SAFETY
910///
911/// - `ctx` may be NULL (unused by the candidate's legacy path).
912/// - `msg` must be a valid NUL-terminated C string or NULL.
913#[no_mangle]
914pub unsafe extern "C" fn xmlParserError(ctx: *mut c_void, msg: *const c_char) {
915    let _ = ctx;
916    unsafe { emit_legacy_message("error", msg) };
917}
918
919/// Default SAX v1 warning handler — `void xmlParserWarning(void *ctx, const char *msg, ...)`.
920#[no_mangle]
921pub unsafe extern "C" fn xmlParserWarning(ctx: *mut c_void, msg: *const c_char) {
922    let _ = ctx;
923    unsafe { emit_legacy_message("warning", msg) };
924}
925
926/// Default validity error handler — `void xmlParserValidityError(void *ctx, const char *msg, ...)`.
927#[no_mangle]
928pub unsafe extern "C" fn xmlParserValidityError(ctx: *mut c_void, msg: *const c_char) {
929    let _ = ctx;
930    unsafe { emit_legacy_message("validity error", msg) };
931}
932
933/// Default validity warning handler — `void xmlParserValidityWarning(void *ctx, const char *msg, ...)`.
934#[no_mangle]
935pub unsafe extern "C" fn xmlParserValidityWarning(ctx: *mut c_void, msg: *const c_char) {
936    let _ = ctx;
937    unsafe { emit_legacy_message("validity warning", msg) };
938}
939
940// ═══════════════════════════════════════════════════════════════════════════════
941// Tests
942// ═══════════════════════════════════════════════════════════════════════════════
943
944#[cfg(test)]
945mod tests {
946    use super::*;
947    use crate::abi::allocator;
948    use core::ffi::c_void;
949
950    #[test]
951    fn test_error_default_reset() {
952        unsafe {
953            let mut err = _xmlError {
954                domain: XML_FROM_PARSER,
955                code: XML_ERR_NO_MEMORY,
956                message: ptr::null_mut(),
957                level: XML_ERR_ERROR as c_int,
958                file: ptr::null_mut(),
959                line: 42,
960                str1: ptr::null_mut(),
961                str2: ptr::null_mut(),
962                str3: ptr::null_mut(),
963                int1: 0,
964                int2: 0,
965                ctxt: ptr::null_mut(),
966                node: ptr::null_mut(),
967            };
968
969            reset_error(&mut err);
970            assert_eq!(err.domain, XML_FROM_NONE);
971            assert_eq!(err.code, XML_ERR_OK as c_int);
972            assert_eq!(err.level, XML_ERR_NONE as c_int);
973            assert_eq!(err.line, 0);
974        }
975    }
976
977    #[test]
978    fn test_copy_error() {
979        unsafe {
980            let from = _xmlError {
981                domain: XML_FROM_PARSER,
982                code: XML_ERR_NO_MEMORY,
983                message: ptr::null_mut(),
984                level: XML_ERR_FATAL as c_int,
985                file: ptr::null_mut(),
986                line: 100,
987                str1: ptr::null_mut(),
988                str2: ptr::null_mut(),
989                str3: ptr::null_mut(),
990                int1: 1,
991                int2: 2,
992                ctxt: ptr::null_mut(),
993                node: ptr::null_mut(),
994            };
995            let mut to = _xmlError {
996                domain: XML_FROM_NONE,
997                code: XML_ERR_OK as c_int,
998                message: ptr::null_mut(),
999                level: XML_ERR_NONE as c_int,
1000                file: ptr::null_mut(),
1001                line: 0,
1002                str1: ptr::null_mut(),
1003                str2: ptr::null_mut(),
1004                str3: ptr::null_mut(),
1005                int1: 0,
1006                int2: 0,
1007                ctxt: ptr::null_mut(),
1008                node: ptr::null_mut(),
1009            };
1010
1011            let result = copy_error(&from, &mut to);
1012            assert_eq!(result, 0);
1013            assert_eq!(to.domain, XML_FROM_PARSER);
1014            assert_eq!(to.code, XML_ERR_NO_MEMORY);
1015            assert_eq!(to.level, XML_ERR_FATAL as c_int);
1016            assert_eq!(to.line, 100);
1017            assert_eq!(to.int1, 1);
1018            assert_eq!(to.int2, 2);
1019        }
1020    }
1021
1022    #[test]
1023    fn test_raise_and_get_last_error() {
1024        unsafe {
1025            reset_last_error();
1026            assert!(get_last_error().is_null());
1027
1028            let file = b"test.xml\0" as *const u8 as *const c_char;
1029            let str1 = b"element\0" as *const u8 as *const c_char;
1030
1031            raise_error(
1032                ptr::null_mut(),
1033                ptr::null_mut(),
1034                ptr::null_mut(),
1035                ptr::null_mut(),
1036                ptr::null_mut(),
1037                XML_FROM_PARSER,
1038                XML_ERR_TAG_NAME_MISMATCH,
1039                XML_ERR_ERROR as c_int,
1040                file,
1041                10,
1042                str1,
1043                ptr::null(),
1044                ptr::null(),
1045                0,
1046                0,
1047                ptr::null(),
1048            );
1049
1050            let last = get_last_error();
1051            assert!(!last.is_null());
1052            assert_eq!((*last).domain, XML_FROM_PARSER);
1053            assert_eq!((*last).code, XML_ERR_TAG_NAME_MISMATCH);
1054            assert_eq!((*last).level, XML_ERR_ERROR as c_int);
1055            assert_eq!((*last).line, 10);
1056
1057            // Check file was stored
1058            let last_file = (*last).file;
1059            assert!(!last_file.is_null());
1060
1061            reset_last_error();
1062            assert!(get_last_error().is_null());
1063        }
1064    }
1065
1066    #[test]
1067    fn test_structured_error_callback() {
1068        unsafe {
1069            reset_last_error();
1070
1071            // Set up a structured error handler that captures the error
1072            let mut captured_domain: c_int = 0;
1073            let captured_ptr = &mut captured_domain as *mut c_int as *mut c_void;
1074
1075            // SAFETY: The callback writes to captured_ptr which lives on the stack
1076            // for the duration of this test.
1077            extern "C" fn test_handler(ctx: *mut c_void, _err: *const _xmlError) {
1078                // SAFETY: ctx is valid for the test duration.
1079                unsafe {
1080                    let captured = &mut *(ctx as *mut c_int);
1081                    *captured = 42;
1082                }
1083            }
1084
1085            set_structured_error_func(captured_ptr, Some(test_handler as xmlStructuredErrorFunc));
1086
1087            raise_error(
1088                ptr::null_mut(),
1089                ptr::null_mut(),
1090                ptr::null_mut(),
1091                ptr::null_mut(),
1092                ptr::null_mut(),
1093                XML_FROM_PARSER,
1094                XML_ERR_OK as c_int,
1095                XML_ERR_WARNING as c_int,
1096                ptr::null(),
1097                0,
1098                ptr::null(),
1099                ptr::null(),
1100                ptr::null(),
1101                0,
1102                0,
1103                ptr::null(),
1104            );
1105
1106            assert_eq!(captured_domain, 42);
1107
1108            // Reset
1109            set_structured_error_func(ptr::null_mut(), None);
1110            reset_last_error();
1111        }
1112    }
1113
1114    #[test]
1115    fn test_format_error_message() {
1116        unsafe {
1117            // Test with direct message
1118            let msg = b"test error\0" as *const u8 as *const c_char;
1119            let formatted = format_error_message(
1120                XML_FROM_NONE,
1121                XML_ERR_OK as c_int,
1122                msg,
1123                ptr::null(),
1124                ptr::null(),
1125                ptr::null(),
1126            );
1127            assert!(!formatted.is_null());
1128            let formatted_str = std::ffi::CStr::from_ptr(formatted);
1129            assert_eq!(formatted_str.to_bytes(), b"test error");
1130
1131            // Free the allocated message
1132            allocator::xmlFreeImpl(formatted as *mut c_void);
1133
1134            // Test with domain and str1
1135            let str1 = b"foo\0" as *const u8 as *const c_char;
1136            let formatted2 = format_error_message(
1137                XML_FROM_PARSER,
1138                XML_ERR_OK as c_int,
1139                ptr::null(),
1140                str1,
1141                ptr::null(),
1142                ptr::null(),
1143            );
1144            assert!(!formatted2.is_null());
1145            allocator::xmlFreeImpl(formatted2 as *mut c_void);
1146        }
1147    }
1148}
libxml_rs/xml/errors/mod.rs

libxml_rs/xml/errors/
mod.rs
