diff --git a/src/parser/xml.rs b/src/parser/xml.rs index 877881f..32b798a 100644 --- a/src/parser/xml.rs +++ b/src/parser/xml.rs @@ -422,11 +422,14 @@ impl<'a> XmlParser<'a> { let _ = found_dup; // suppress unused warning } - // --- Apply #FIXED default attributes from DTD ATTLIST declarations --- - // Per XML 1.0 §3.3.2, if an attribute declared with #FIXED is not + // --- Apply DTD ATTLIST default attributes (#FIXED and #DEFAULT) --- + // Per XML 1.0 §3.3.2, when an attribute declared in an ATTLIST is not // present on the element, the parser must add it with the declared - // default value. #DEFAULT attributes are tracked for amplification - // factor checking but not inserted into the tree (matching libxml2). + // default value. This applies to both `#FIXED "v"` and bare `"v"` + // (so-called #DEFAULT) declarations. libxml2 applies both during + // normal parsing — verifiable via `xmllint --c14n` on a document + // with an ATTLIST default, which emits the default attribute in the + // canonical form. // Namespace declarations (xmlns, xmlns:prefix) are inserted before // other attributes to match libxml2's attribute ordering. if let Some(defaults) = if self.attr_defaults.is_empty() { @@ -453,27 +456,29 @@ impl<'a> XmlParser<'a> { .any(|a| a.name == decl_local && a.prefix.as_deref() == decl_pfx); if !already_present { // Track expansion for amplification factor check - // (both #FIXED and #DEFAULT contribute to expansion) + // (both #FIXED and #DEFAULT contribute to expansion). self.expansion_size += attr_name.len() + value.len(); - // Only insert #FIXED attributes into the tree - if is_fixed { - let (decl_prefix, decl_local) = split_name(attr_name); - let attr = Attribute { - name: decl_local.to_string(), - value, - prefix: decl_prefix.map(String::from), - namespace: None, - raw_value: None, - }; - let is_ns_decl = - attr_name == "xmlns" || attr_name.starts_with("xmlns:"); - if is_ns_decl { - attributes.insert(insert_pos, attr); - insert_pos += 1; - } else { - attributes.push(attr); - } + // Insert both #FIXED and #DEFAULT defaults into the + // tree. The `is_fixed` flag is no longer used to gate + // insertion; it would only matter if we additionally + // validated source attributes against #FIXED values, + // which is a separate validation step. + let _ = is_fixed; + let (decl_prefix, decl_local) = split_name(attr_name); + let attr = Attribute { + name: decl_local.to_string(), + value, + prefix: decl_prefix.map(String::from), + namespace: None, + raw_value: None, + }; + let is_ns_decl = attr_name == "xmlns" || attr_name.starts_with("xmlns:"); + if is_ns_decl { + attributes.insert(insert_pos, attr); + insert_pos += 1; + } else { + attributes.push(attr); } } } diff --git a/src/serial/c14n.rs b/src/serial/c14n.rs index 576112c..e87ce09 100644 --- a/src/serial/c14n.rs +++ b/src/serial/c14n.rs @@ -329,10 +329,26 @@ impl<'a> C14nContext<'a> { let mut ns_to_output: Vec<(String, String)> = Vec::new(); for (ns_prefix, ns_uri) in &ns_decls { - if current_rendered.get(ns_prefix) != Some(ns_uri) { - ns_to_output.push((ns_prefix.clone(), ns_uri.clone())); - current_rendered.insert(ns_prefix.clone(), ns_uri.clone()); + if current_rendered.get(ns_prefix) == Some(ns_uri) { + continue; } + + // Special case for `xmlns=""`: per Canonical XML 1.0 §3.7 and + // c14n11 §3.1, the empty default-namespace declaration is only + // emitted to undeclare a *non-empty* inherited default. If no + // non-empty default is currently in scope (parent has no default, + // or the inherited default is itself empty), the `xmlns=""` from + // the source must not appear in the canonical form. + if ns_prefix.is_empty() && ns_uri.is_empty() { + let has_nonempty_inherited_default = + current_rendered.get("").is_some_and(|s| !s.is_empty()); + if !has_nonempty_inherited_default { + continue; + } + } + + ns_to_output.push((ns_prefix.clone(), ns_uri.clone())); + current_rendered.insert(ns_prefix.clone(), ns_uri.clone()); } if !self.options.exclusive { diff --git a/tests/c14n_conformance.rs b/tests/c14n_conformance.rs new file mode 100644 index 0000000..190c268 --- /dev/null +++ b/tests/c14n_conformance.rs @@ -0,0 +1,366 @@ +//! W3C Canonical XML conformance test harness. +//! +//! Drives `canonicalize` / `canonicalize_subtree` against the worked examples +//! published in the C14N 1.0 and Exclusive C14N 1.0 specifications. Each +//! input is loaded from `tests/data/c14n_w3c/`, canonicalized in the +//! requested mode, and compared byte-for-byte to the expected output. +//! +//! Sources: +//! - C14N 1.0: §3 ("Examples") +//! - Exclusive C14N 1.0: §2 +//! +//! Some published examples require infrastructure outside the c14n module +//! (XPath-defined node-sets, DTD-driven attribute normalization, external +//! entity resolution). Those subsections are not represented here. +//! +//! On byte mismatch the harness reports the first differing byte plus a +//! 200-byte window from each side. Trailing newlines are normalized away +//! because file-save tooling routinely adds them; the spec's canonical +//! forms have no trailing newline. + +use xmloxide::serial::c14n::{canonicalize, canonicalize_subtree, C14nOptions}; +use xmloxide::tree::{Document, NodeId}; + +#[derive(Debug, Clone, Copy)] +enum Apex { + /// Whole-document canonicalization. + Document, + /// Subtree rooted at the first element with the given local name. + SubtreeAt(&'static str), +} + +#[derive(Debug)] +struct C14nCase { + name: &'static str, + input: &'static str, + expected: &'static str, + apex: Apex, + with_comments: bool, + exclusive: bool, + inclusive_prefixes: &'static [&'static str], + /// Provenance for failure reports. + source: &'static str, +} + +const CASES: &[C14nCase] = &[ + // -------- W3C Canonical XML 1.0 -------- + C14nCase { + name: "c14n10_3_1_no_comments", + input: "c14n10_3_1_input.xml", + expected: "c14n10_3_1_no_comments.xml", + apex: Apex::Document, + with_comments: false, + exclusive: false, + inclusive_prefixes: &[], + source: "C14N 1.0 §3.1 (without comments)", + }, + C14nCase { + name: "c14n10_3_1_with_comments", + input: "c14n10_3_1_input.xml", + expected: "c14n10_3_1_with_comments.xml", + apex: Apex::Document, + with_comments: true, + exclusive: false, + inclusive_prefixes: &[], + source: "C14N 1.0 §3.1 (with comments)", + }, + C14nCase { + name: "c14n10_3_2_whitespace", + input: "c14n10_3_2_input.xml", + expected: "c14n10_3_2_expected.xml", + apex: Apex::Document, + with_comments: false, + exclusive: false, + inclusive_prefixes: &[], + source: "C14N 1.0 §3.2 (whitespace in document content)", + }, + C14nCase { + // §3.3 exercises start/end tags, attribute sorting, and namespace + // handling. It also relies on a DTD `` + // which adds an attribute via DTD defaulting. xmloxide's parser may or + // may not apply that — if it doesn't, this test surfaces it as a + // missing `attr="default"`. + name: "c14n10_3_3_tags", + input: "c14n10_3_3_input.xml", + expected: "c14n10_3_3_expected.xml", + apex: Apex::Document, + with_comments: false, + exclusive: false, + inclusive_prefixes: &[], + source: "C14N 1.0 §3.3 (start/end tags)", + }, + C14nCase { + name: "c14n10_3_6_utf8", + input: "c14n10_3_6_input.xml", + expected: "c14n10_3_6_expected.xml", + apex: Apex::Document, + with_comments: false, + exclusive: false, + inclusive_prefixes: &[], + source: "C14N 1.0 §3.6 (UTF-8 encoding)", + }, + // -------- W3C Exclusive XML Canonicalization 1.0 -------- + C14nCase { + name: "excc14n_2_1_standalone", + input: "excc14n_2_1_standalone_input.xml", + expected: "excc14n_2_1_standalone_expected.xml", + apex: Apex::Document, + with_comments: false, + exclusive: true, + inclusive_prefixes: &[], + source: "Exc-C14N 1.0 §2.1 (standalone)", + }, + C14nCase { + // §2.1 enveloped: extracting elem1 from a pdu apex must yield the + // same canonical form as the standalone case. Subtree extraction. + name: "excc14n_2_1_enveloped", + input: "excc14n_2_1_enveloped_input.xml", + expected: "excc14n_2_1_enveloped_expected.xml", + apex: Apex::SubtreeAt("elem1"), + with_comments: false, + exclusive: true, + inclusive_prefixes: &[], + source: "Exc-C14N 1.0 §2.1 (enveloped, subtree at elem1)", + }, + C14nCase { + // §2.2 input1: elem2 subtree under different enveloping context. + name: "excc14n_2_2_input1", + input: "excc14n_2_2_input1.xml", + expected: "excc14n_2_2_expected.xml", + apex: Apex::SubtreeAt("elem2"), + with_comments: false, + exclusive: true, + inclusive_prefixes: &[], + source: "Exc-C14N 1.0 §2.2 (input1, subtree at elem2)", + }, + C14nCase { + // §2.2 input2: same elem2 subtree under a different enveloping + // context. Spec property: byte-equal output to input1's elem2. + name: "excc14n_2_2_input2", + input: "excc14n_2_2_input2.xml", + expected: "excc14n_2_2_expected.xml", + apex: Apex::SubtreeAt("elem2"), + with_comments: false, + exclusive: true, + inclusive_prefixes: &[], + source: "Exc-C14N 1.0 §2.2 (input2, subtree at elem2)", + }, +]; + +fn fixture_path(name: &str) -> String { + format!("{}/tests/data/c14n_w3c/{name}", env!("CARGO_MANIFEST_DIR")) +} + +fn find_first_local(doc: &Document, root: NodeId, local: &str) -> Option { + if doc.node_name(root) == Some(local) { + return Some(root); + } + for child in doc.children(root) { + if let Some(n) = find_first_local(doc, child, local) { + return Some(n); + } + } + None +} + +/// Strip a single trailing newline if present. C14N's spec output has no +/// trailing newline; file-save tooling routinely adds one. Normalizing both +/// sides means we can keep readable fixture files without false positives. +fn trim_trailing_newline(s: &str) -> &str { + s.strip_suffix('\n').unwrap_or(s) +} + +fn first_diff_window(expected: &str, actual: &str) -> String { + let exp = expected.as_bytes(); + let act = actual.as_bytes(); + let mut at = 0usize; + while at < exp.len() && at < act.len() && exp[at] == act[at] { + at += 1; + } + let pre = at.saturating_sub(80); + let exp_end = (at + 80).min(exp.len()); + let act_end = (at + 80).min(act.len()); + format!( + " first diff at byte {at}\n expected window: {:?}\n actual window: {:?}", + String::from_utf8_lossy(&exp[pre..exp_end]), + String::from_utf8_lossy(&act[pre..act_end]), + ) +} + +#[test] +fn w3c_c14n_conformance() { + let mut failures: Vec = Vec::new(); + + for case in CASES { + let input_path = fixture_path(case.input); + let expected_path = fixture_path(case.expected); + + let input = match std::fs::read_to_string(&input_path) { + Ok(s) => s, + Err(e) => { + failures.push(format!("[{}] read {input_path}: {e}", case.name)); + continue; + } + }; + let expected = match std::fs::read_to_string(&expected_path) { + Ok(s) => s, + Err(e) => { + failures.push(format!("[{}] read {expected_path}: {e}", case.name)); + continue; + } + }; + + let doc = match Document::parse_str(&input) { + Ok(d) => d, + Err(e) => { + failures.push(format!( + "[{}] parse failed: {e:?}\n source: {}", + case.name, case.source + )); + continue; + } + }; + + let opts = C14nOptions { + with_comments: case.with_comments, + exclusive: case.exclusive, + inclusive_prefixes: case + .inclusive_prefixes + .iter() + .map(|s| (*s).to_string()) + .collect(), + }; + + let actual = match case.apex { + Apex::Document => canonicalize(&doc, &opts), + Apex::SubtreeAt(local) => { + let Some(root) = doc.root_element() else { + failures.push(format!("[{}] no root element", case.name)); + continue; + }; + let Some(apex) = find_first_local(&doc, root, local) else { + failures.push(format!( + "[{}] subtree apex <{local}> not found in input", + case.name + )); + continue; + }; + canonicalize_subtree(&doc, apex, &opts) + } + }; + + let exp_trim = trim_trailing_newline(&expected); + let act_trim = trim_trailing_newline(&actual); + + if exp_trim != act_trim { + failures.push(format!( + "[{}] BYTE MISMATCH ({} bytes expected, {} bytes actual)\n source: {}\n{}", + case.name, + exp_trim.len(), + act_trim.len(), + case.source, + first_diff_window(exp_trim, act_trim), + )); + } + } + + if !failures.is_empty() { + let total = CASES.len(); + let failed = failures.len(); + panic!( + "W3C C14N conformance: {} / {} cases failed\n\n{}\n", + failed, + total, + failures.join("\n\n"), + ); + } +} + +/// Idempotency roundtrip: feeding canonical output back through the parser +/// + canonicalizer must produce byte-identical output. This is the formal +/// statement of "the canonical form is canonical" — a c14n implementation +/// that doesn't satisfy it is by definition non-canonical. +/// +/// Runs against every case in `CASES`. If a fixture's expected output +/// re-canonicalizes to anything other than itself, the test fails with +/// the same byte-window diff used by the conformance test above. +#[test] +fn w3c_c14n_idempotency() { + let mut failures: Vec = Vec::new(); + + for case in CASES { + // The "expected" file already represents canonical output for this + // case's mode; feeding it back through must reproduce it exactly. + let Ok(canonical_input) = std::fs::read_to_string(fixture_path(case.expected)) else { + continue; + }; + let canonical_input = trim_trailing_newline(&canonical_input).to_string(); + + let doc = match Document::parse_str(&canonical_input) { + Ok(d) => d, + Err(e) => { + failures.push(format!( + "[{}] re-parse of canonical output failed: {e:?}", + case.name, + )); + continue; + } + }; + + let opts = C14nOptions { + with_comments: case.with_comments, + exclusive: case.exclusive, + inclusive_prefixes: case + .inclusive_prefixes + .iter() + .map(|s| (*s).to_string()) + .collect(), + }; + + // Re-canonicalize using the same apex selection as the original + // case so subtree-apex tests stay well-defined. + let actual = match case.apex { + Apex::Document => canonicalize(&doc, &opts), + Apex::SubtreeAt(local) => { + let Some(root) = doc.root_element() else { + continue; + }; + let Some(apex) = find_first_local(&doc, root, local) else { + // The canonical-output file may have stripped enveloping + // ancestors, so the original apex element is the new + // root. Fall back to canonicalizing the whole doc. + let actual = canonicalize(&doc, &opts); + let act_trim = trim_trailing_newline(&actual); + if act_trim != canonical_input { + failures.push(format!( + "[{}] subtree-apex idempotency failed\n source: {}\n{}", + case.name, + case.source, + first_diff_window(&canonical_input, act_trim), + )); + } + continue; + }; + canonicalize_subtree(&doc, apex, &opts) + } + }; + + let act_trim = trim_trailing_newline(&actual); + if act_trim != canonical_input { + failures.push(format!( + "[{}] re-canonicalize differs from input\n source: {}\n{}", + case.name, + case.source, + first_diff_window(&canonical_input, act_trim), + )); + } + } + + assert!( + failures.is_empty(), + "C14N idempotency: {} / {} cases failed\n\n{}\n", + failures.len(), + CASES.len(), + failures.join("\n\n"), + ); +} diff --git a/tests/data/c14n_w3c/c14n10_3_1_input.xml b/tests/data/c14n_w3c/c14n10_3_1_input.xml new file mode 100644 index 0000000..ed450c7 --- /dev/null +++ b/tests/data/c14n_w3c/c14n10_3_1_input.xml @@ -0,0 +1,14 @@ + + + + + + +Hello, world! + + + + + + diff --git a/tests/data/c14n_w3c/c14n10_3_1_no_comments.xml b/tests/data/c14n_w3c/c14n10_3_1_no_comments.xml new file mode 100644 index 0000000..a5703c1 --- /dev/null +++ b/tests/data/c14n_w3c/c14n10_3_1_no_comments.xml @@ -0,0 +1,4 @@ + +Hello, world! + diff --git a/tests/data/c14n_w3c/c14n10_3_1_with_comments.xml b/tests/data/c14n_w3c/c14n10_3_1_with_comments.xml new file mode 100644 index 0000000..5193ede --- /dev/null +++ b/tests/data/c14n_w3c/c14n10_3_1_with_comments.xml @@ -0,0 +1,6 @@ + +Hello, world! + + + diff --git a/tests/data/c14n_w3c/c14n10_3_2_expected.xml b/tests/data/c14n_w3c/c14n10_3_2_expected.xml new file mode 100644 index 0000000..74eeea1 --- /dev/null +++ b/tests/data/c14n_w3c/c14n10_3_2_expected.xml @@ -0,0 +1,11 @@ + + + A B + + A + + B + A B + C + + diff --git a/tests/data/c14n_w3c/c14n10_3_2_input.xml b/tests/data/c14n_w3c/c14n10_3_2_input.xml new file mode 100644 index 0000000..74eeea1 --- /dev/null +++ b/tests/data/c14n_w3c/c14n10_3_2_input.xml @@ -0,0 +1,11 @@ + + + A B + + A + + B + A B + C + + diff --git a/tests/data/c14n_w3c/c14n10_3_3_expected.xml b/tests/data/c14n_w3c/c14n10_3_3_expected.xml new file mode 100644 index 0000000..c2b8631 --- /dev/null +++ b/tests/data/c14n_w3c/c14n10_3_3_expected.xml @@ -0,0 +1,14 @@ + + + + + + + + + + + + + + diff --git a/tests/data/c14n_w3c/c14n10_3_3_input.xml b/tests/data/c14n_w3c/c14n10_3_3_input.xml new file mode 100644 index 0000000..09567cc --- /dev/null +++ b/tests/data/c14n_w3c/c14n10_3_3_input.xml @@ -0,0 +1,18 @@ +]> + + + + + + + + + + + + + + diff --git a/tests/data/c14n_w3c/c14n10_3_6_expected.xml b/tests/data/c14n_w3c/c14n10_3_6_expected.xml new file mode 100644 index 0000000..a2addeb --- /dev/null +++ b/tests/data/c14n_w3c/c14n10_3_6_expected.xml @@ -0,0 +1 @@ +© diff --git a/tests/data/c14n_w3c/c14n10_3_6_input.xml b/tests/data/c14n_w3c/c14n10_3_6_input.xml new file mode 100644 index 0000000..31e2071 --- /dev/null +++ b/tests/data/c14n_w3c/c14n10_3_6_input.xml @@ -0,0 +1,2 @@ + +© diff --git a/tests/data/c14n_w3c/excc14n_2_1_enveloped_expected.xml b/tests/data/c14n_w3c/excc14n_2_1_enveloped_expected.xml new file mode 100644 index 0000000..e016aa5 --- /dev/null +++ b/tests/data/c14n_w3c/excc14n_2_1_enveloped_expected.xml @@ -0,0 +1,3 @@ + + content + diff --git a/tests/data/c14n_w3c/excc14n_2_1_enveloped_input.xml b/tests/data/c14n_w3c/excc14n_2_1_enveloped_input.xml new file mode 100644 index 0000000..db8aa7a --- /dev/null +++ b/tests/data/c14n_w3c/excc14n_2_1_enveloped_input.xml @@ -0,0 +1,5 @@ + + + content + + diff --git a/tests/data/c14n_w3c/excc14n_2_1_standalone_expected.xml b/tests/data/c14n_w3c/excc14n_2_1_standalone_expected.xml new file mode 100644 index 0000000..917afcd --- /dev/null +++ b/tests/data/c14n_w3c/excc14n_2_1_standalone_expected.xml @@ -0,0 +1,3 @@ + + content + diff --git a/tests/data/c14n_w3c/excc14n_2_1_standalone_input.xml b/tests/data/c14n_w3c/excc14n_2_1_standalone_input.xml new file mode 100644 index 0000000..917afcd --- /dev/null +++ b/tests/data/c14n_w3c/excc14n_2_1_standalone_input.xml @@ -0,0 +1,3 @@ + + content + diff --git a/tests/data/c14n_w3c/excc14n_2_2_expected.xml b/tests/data/c14n_w3c/excc14n_2_2_expected.xml new file mode 100644 index 0000000..a62a15e --- /dev/null +++ b/tests/data/c14n_w3c/excc14n_2_2_expected.xml @@ -0,0 +1,3 @@ + + + diff --git a/tests/data/c14n_w3c/excc14n_2_2_input1.xml b/tests/data/c14n_w3c/excc14n_2_2_input1.xml new file mode 100644 index 0000000..0648fa4 --- /dev/null +++ b/tests/data/c14n_w3c/excc14n_2_2_input1.xml @@ -0,0 +1,7 @@ + + + + + diff --git a/tests/data/c14n_w3c/excc14n_2_2_input2.xml b/tests/data/c14n_w3c/excc14n_2_2_input2.xml new file mode 100644 index 0000000..60ed873 --- /dev/null +++ b/tests/data/c14n_w3c/excc14n_2_2_input2.xml @@ -0,0 +1,9 @@ + + + + +