aboutsummaryrefslogtreecommitdiffstats
path: root/src/xml/parsers.rs
diff options
context:
space:
mode:
Diffstat (limited to 'src/xml/parsers.rs')
-rw-r--r--src/xml/parsers.rs131
1 files changed, 77 insertions, 54 deletions
diff --git a/src/xml/parsers.rs b/src/xml/parsers.rs
index 6326593..93ff5b1 100644
--- a/src/xml/parsers.rs
+++ b/src/xml/parsers.rs
@@ -15,18 +15,18 @@ use nom::{
use crate::xml::NSAttName;
use super::{
- AttDef, AttDefName, AttType, AttValue, AttlistDecl, Attribute, CDEnd, CDSect, CDStart, CData,
- Char, CharData, CharRef, Children, ChildrenKind, Choice, Comment, ConditionalSect, Content,
- ContentItem, Contentspec, Cp, CpKind, DeclSep, DefaultAttName, DefaultDecl, DoctypeDecl,
- Document, ETag, Element, Elementdecl, EmptyElemTag, EncName, EncodingDecl, EntityDecl,
- EntityDef, EntityRef, EntityValue, EnumeratedType, Enumeration, Eq, ExtParsedEnt, ExtSubset,
- ExtSubsetDecl, ExtSubsetDeclaration, ExternalID, GEDecl, Ignore, IgnoreSect,
- IgnoreSectContents, IncludeSect, IntSubset, IntSubsetDeclaration, LiteralData, LocalPart,
- MarkupDecl, Misc, Mixed, NCName, NDataDecl, Name, NameChar, NameStartChar, Names, Nmtoken,
- Nmtokens, NotationDecl, NotationDeclID, NotationType, Occurence, PEDecl, PEDef, PEReference,
- PITarget, Prefix, PrefixedAttName, PrefixedName, Prolog, PubidChar, PubidLiteral, PublicID,
- QName, Reference, SDDecl, STag, Seq, StringType, SystemLiteral, TextDecl, TokenizedType,
- UnprefixedName, VersionInfo, VersionNum, XMLDecl, PI, S,
+ AttDef, AttDefName, AttType, AttValue, AttValueData, AttlistDecl, Attribute, CDEnd, CDSect,
+ CDStart, CData, Char, CharData, CharRef, Children, ChildrenKind, Choice, Comment,
+ ConditionalSect, Content, ContentItem, Contentspec, Cp, CpKind, DeclSep, DefaultAttName,
+ DefaultDecl, DoctypeDecl, Document, ETag, Element, Elementdecl, EmptyElemTag, EncName,
+ EncodingDecl, EntityDecl, EntityDef, EntityRef, EntityValue, EntityValueData, EnumeratedType,
+ Enumeration, Eq, ExtParsedEnt, ExtSubset, ExtSubsetDecl, ExtSubsetDeclaration, ExternalID,
+ GEDecl, Ignore, IgnoreSect, IgnoreSectContents, IncludeSect, IntSubset, IntSubsetDeclaration,
+ LocalPart, MarkupDecl, Misc, Mixed, NCName, NDataDecl, Name, NameChar, NameStartChar, Names,
+ Nmtoken, Nmtokens, NotationDecl, NotationDeclID, NotationType, Occurence, PEDecl, PEDef,
+ PEReference, PITarget, Prefix, PrefixedAttName, PrefixedName, Prolog, PubidChar, PubidLiteral,
+ PublicID, QName, Reference, SDDecl, STag, Seq, StringType, SystemLiteral, TextDecl,
+ TokenizedType, UnprefixedName, VersionInfo, VersionNum, XMLDecl, PI, S,
};
pub trait Parser<'s, T> {
@@ -141,9 +141,11 @@ impl Parser<'_, Char> for Char {
}
/// [3] S ::= (#x20 | #x9 | #xD | #xA)+
-impl<'s> Parser<'s, S<'s>> for S<'s> {
- fn parse(input: &'s str) -> IResult<&str, S<'s>> {
- map(is_a("\u{20}\u{9}\u{D}\u{A}"), |s| S(s))(input)
+impl Parser<'_, S> for S {
+ fn parse(input: &str) -> IResult<&str, S> {
+ // TODO?: whitespacing
+ // map(is_a("\u{20}\u{9}\u{D}\u{A}"), |s| S(s))(input)
+ value(S, is_a("\u{20}\u{9}\u{D}\u{A}"))(input)
}
}
@@ -221,43 +223,46 @@ impl<'s> Parser<'s, Nmtokens<'s>> for Nmtokens<'s> {
/// | "'" ([^%&'] | PEReference | Reference)* "'"
impl<'s> Parser<'s, EntityValue<'s>> for EntityValue<'s> {
fn parse(input: &'s str) -> IResult<&str, EntityValue<'s>> {
- map(
- alt((
+ alt((
+ map(
delimited(
char('"'),
many0(alt((
map(
recognize(many_till(take(1usize), peek(one_of("%&\"")))),
- |string| LiteralData::String(string),
+ |string| EntityValueData::String(string),
),
map(PEReference::parse, |pe_reference| {
- LiteralData::PEReference(pe_reference)
+ EntityValueData::PEReference(pe_reference)
}),
map(Reference::parse, |reference| {
- LiteralData::Reference(reference)
+ EntityValueData::Reference(reference)
}),
))),
char('"'),
),
+ |entity_value| EntityValue::DoubleQuoted(entity_value),
+ ),
+ map(
delimited(
char('\''),
many0(alt((
map(
recognize(many_till(take(1usize), peek(one_of("%&'")))),
- |string| LiteralData::String(string),
+ |string| EntityValueData::String(string),
),
map(PEReference::parse, |pe_reference| {
- LiteralData::PEReference(pe_reference)
+ EntityValueData::PEReference(pe_reference)
}),
map(Reference::parse, |reference| {
- LiteralData::Reference(reference)
+ EntityValueData::Reference(reference)
}),
))),
char('\''),
),
- )),
- |entity_value| EntityValue(entity_value),
- )(input)
+ |entity_value| EntityValue::SingleQuoted(entity_value),
+ ),
+ ))(input)
}
}
@@ -265,67 +270,76 @@ impl<'s> Parser<'s, EntityValue<'s>> for EntityValue<'s> {
/// | "'" ([^<&'] | Reference)* "'"
impl<'s> Parser<'s, AttValue<'s>> for AttValue<'s> {
fn parse(input: &'s str) -> IResult<&str, AttValue<'s>> {
- map(
- alt((
+ alt((
+ map(
delimited(
char('"'),
many0(alt((
map(
recognize(many_till(take(1usize), peek(one_of("%&\"")))),
- |string| LiteralData::String(string),
+ |string| AttValueData::String(string),
),
map(Reference::parse, |reference| {
- LiteralData::Reference(reference)
+ AttValueData::Reference(reference)
}),
))),
char('"'),
),
+ |att_value| AttValue::DoubleQuoted(att_value),
+ ),
+ map(
delimited(
char('\''),
many0(alt((
map(
recognize(many_till(take(1usize), peek(one_of("%&'")))),
- |string| LiteralData::String(string),
+ |string| AttValueData::String(string),
),
map(Reference::parse, |reference| {
- LiteralData::Reference(reference)
+ AttValueData::Reference(reference)
}),
))),
char('\''),
),
- )),
- |att_value| AttValue(att_value),
- )(input)
+ |att_value| AttValue::SingleQuoted(att_value),
+ ),
+ ))(input)
}
}
/// [11] SystemLiteral ::= ('"' [^"]* '"') | ("'" [^']* "'")
impl<'s> Parser<'s, SystemLiteral<'s>> for SystemLiteral<'s> {
fn parse(input: &'s str) -> IResult<&str, SystemLiteral<'s>> {
- map(
- alt((
+ alt((
+ map(
delimited(char('"'), recognize(many0(none_of("\""))), char('"')),
+ |system_literal| SystemLiteral::DoubleQuoted(system_literal),
+ ),
+ map(
delimited(char('\''), recognize(many0(none_of("'"))), char('\'')),
- )),
- |system_literal| SystemLiteral(system_literal),
- )(input)
+ |system_literal| SystemLiteral::SingleQuoted(system_literal),
+ ),
+ ))(input)
}
}
/// [12] PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'"
impl<'s> Parser<'s, PubidLiteral<'s>> for PubidLiteral<'s> {
fn parse(input: &'s str) -> IResult<&str, PubidLiteral<'s>> {
- map(
- alt((
+ alt((
+ map(
delimited(char('"'), recognize(many0(PubidChar::parse)), char('"')),
+ |pubid_literal| PubidLiteral::DoubleQuoted(pubid_literal),
+ ),
+ map(
delimited(
char('\''),
recognize(many0(recognize(not(char('\''))).and_then(PubidChar::parse))),
char('\''),
),
- )),
- |pubid_literal| PubidLiteral(pubid_literal),
- )(input)
+ |pubid_literal| PubidLiteral::SingleQuoted(pubid_literal),
+ ),
+ ))(input)
}
}
@@ -477,15 +491,18 @@ impl<'s> Parser<'s, XMLDecl<'s>> for XMLDecl<'s> {
/// [24] VersionInfo ::= S 'version' Eq ("'" VersionNum "'" | '"' VersionNum '"')
impl Parser<'_, VersionInfo> for VersionInfo {
fn parse(input: &'_ str) -> IResult<&str, VersionInfo> {
- map(
- preceded(
- tuple((S::parse, tag("version"), Eq::parse)),
- alt((
+ preceded(
+ tuple((S::parse, tag("version"), Eq::parse)),
+ alt((
+ map(
delimited(char('\''), VersionNum::parse, char('\'')),
+ |version_info| VersionInfo::SingleQuoted(version_info),
+ ),
+ map(
delimited(char('"'), VersionNum::parse, char('"')),
- )),
- ),
- |version_num| VersionInfo(version_num),
+ |version_info| VersionInfo::DoubleQuoted(version_info),
+ ),
+ )),
)(input)
}
}
@@ -639,12 +656,18 @@ impl Parser<'_, SDDecl> for SDDecl {
alt((
delimited(
char('\''),
- alt((value(true, tag("yes")), value(false, tag("no")))),
+ alt((
+ value(SDDecl::SingleQuoted(true), tag("yes")),
+ value(SDDecl::SingleQuoted(false), tag("no")),
+ )),
char('\''),
),
delimited(
char('"'),
- alt((value(true, tag("yes")), value(false, tag("no")))),
+ alt((
+ value(SDDecl::DoubleQuoted(true), tag("yes")),
+ value(SDDecl::DoubleQuoted(false), tag("no")),
+ )),
char('"'),
),
)),
generated by cgit (git 2.34.1) at 2025-04-04 16:49:03 +0000