diff options
| author | Philipp A | 2023-12-28 15:43:24 +0100 |
|---|---|---|
| committer | GitHub | 2023-12-28 15:43:24 +0100 |
| commit | 0f4f1a420cbcf263a9118ec9b288c95f1b59ade8 (patch) | |
| tree | 030abd05e8894e4e9c6775794f890180cf6c5e91 /parser/src/conversion/inline.rs | |
| parent | 38b1c488601cfc2479f02df555b135ef01aa5618 (diff) | |
| parent | 774dd4798aedc40b38c6480e9c47f34c482f12d0 (diff) | |
| download | rust-rst-0f4f1a420cbcf263a9118ec9b288c95f1b59ade8.tar.bz2 | |
Merge branch 'main' into allow-rst-to-read-from-stdin
Diffstat (limited to 'parser/src/conversion/inline.rs')
| -rw-r--r-- | parser/src/conversion/inline.rs | 278 |
1 files changed, 137 insertions, 141 deletions
diff --git a/parser/src/conversion/inline.rs b/parser/src/conversion/inline.rs index 82a74e7..a0dcb88 100644 --- a/parser/src/conversion/inline.rs +++ b/parser/src/conversion/inline.rs @@ -2,159 +2,155 @@ use failure::Error; use pest::iterators::Pair; use document_tree::{ - HasChildren, - elements as e, - url::Url, - element_categories as c, - extra_attributes as a, - attribute_types as at, + attribute_types as at, element_categories as c, elements as e, extra_attributes as a, url::Url, + HasChildren, }; -use crate::{ - pest_rst::Rule, -// pair_ext_parse::PairExt, -}; use super::whitespace_normalize_name; - +use crate::pest_rst::Rule; pub fn convert_inline(pair: Pair<Rule>) -> Result<c::TextOrInlineElement, Error> { - Ok(match pair.as_rule() { - Rule::str | Rule::str_nested => pair.as_str().into(), - Rule::ws_newline => " ".to_owned().into(), - Rule::reference => convert_reference(pair)?, - Rule::substitution_name => convert_substitution_ref(pair)?.into(), - Rule::emph => e::Emphasis::with_children(convert_inlines(pair)?).into(), - Rule::strong => e::Strong::with_children(convert_inlines(pair)?).into(), - Rule::literal => e::Literal::with_children(vec![pair.as_str().to_owned()]).into(), - rule => unimplemented!("unknown rule {:?}", rule), - }) + Ok(match pair.as_rule() { + Rule::str | Rule::str_nested => pair.as_str().into(), + Rule::ws_newline => " ".to_owned().into(), + Rule::reference => convert_reference(pair)?, + Rule::substitution_name => convert_substitution_ref(pair)?.into(), + Rule::emph => e::Emphasis::with_children(convert_inlines(pair)?).into(), + Rule::strong => e::Strong::with_children(convert_inlines(pair)?).into(), + Rule::literal => e::Literal::with_children(vec![pair.as_str().to_owned()]).into(), + rule => unimplemented!("unknown rule {:?}", rule), + }) } pub fn convert_inlines(pair: Pair<Rule>) -> Result<Vec<c::TextOrInlineElement>, Error> { - pair.into_inner().map(convert_inline).collect() + pair.into_inner().map(convert_inline).collect() } fn convert_reference(pair: Pair<Rule>) -> Result<c::TextOrInlineElement, Error> { - let name; - let refuri; - let refid; - let mut refname = vec![]; - let mut children: Vec<c::TextOrInlineElement> = vec![]; - let concrete = pair.into_inner().next().unwrap(); - match concrete.as_rule() { - Rule::reference_target => { - let rt_inner = concrete.into_inner().next().unwrap(); // reference_target_uq or target_name_qu - match rt_inner.as_rule() { - Rule::reference_target_uq => { - refid = None; - name = Some(rt_inner.as_str().into()); - refuri = None; - refname.push(rt_inner.as_str().into()); - children.push(rt_inner.as_str().into()); - }, - Rule::reference_target_qu => { - let (text, reference) = { - let mut text = None; - let mut reference = None; - for inner in rt_inner.clone().into_inner() { - match inner.as_rule() { - Rule::reference_text => text = Some(inner), - Rule::reference_bracketed => reference = Some(inner), - _ => unreachable!() - } - } - (text, reference) - }; - let trimmed_text = match (&text, &reference) { - (Some(text), None) => text.as_str(), - (_, Some(reference)) => { - text - .map(|text| text.as_str().trim_end_matches(|ch| " \n\r".contains(ch))) - .filter(|text| !text.is_empty()) - .unwrap_or_else(|| reference.clone().into_inner().next().unwrap().as_str()) - } - (None, None) => unreachable!() - }; - refid = None; - name = Some(trimmed_text.into()); - refuri = if let Some(reference) = reference { - let inner = reference.into_inner().next().unwrap(); - match inner.as_rule() { - // The URL rules in our parser accept a narrow superset of - // valid URLs, so we need to handle false positives. - Rule::url => if let Ok(target) = Url::parse_absolute(inner.as_str()) { - Some(target) - } else if inner.as_str().ends_with('_') { - // like target_name_qu (minus the final underscore) - let full_str = inner.as_str(); - refname.push(full_str[0..full_str.len() - 1].into()); - None - } else { - // like relative_reference - Some(Url::parse_relative(inner.as_str())?) - }, - Rule::target_name_qu => { - refname.push(inner.as_str().into()); - None - }, - Rule::relative_reference => { - Some(Url::parse_relative(inner.as_str())?) - }, - _ => unreachable!() - } - } else { - refname.push(trimmed_text.into()); - None - }; - children.push(trimmed_text.into()); - }, - _ => unreachable!() - } - }, - Rule::reference_explicit => unimplemented!("explicit reference"), - Rule::reference_auto => { - let rt_inner = concrete.into_inner().next().unwrap(); - match rt_inner.as_rule() { - Rule::url_auto => match Url::parse_absolute(rt_inner.as_str()) { - Ok(target) => { - refuri = Some(target); - name = None; - refid = None; - children.push(rt_inner.as_str().into()); - }, - // if our parser got a URL wrong, return it as a string - Err(_) => return Ok(rt_inner.as_str().into()) - }, - Rule::email => { - let mailto_url = String::from("mailto:") + rt_inner.as_str(); - match Url::parse_absolute(&mailto_url) { - Ok(target) => { - refuri = Some(target); - name = None; - refid = None; - children.push(rt_inner.as_str().into()); - }, - // if our parser got a URL wrong, return it as a string - Err(_) => return Ok(rt_inner.as_str().into()) - } - }, - _ => unreachable!() - } - }, - _ => unreachable!(), - }; - Ok(e::Reference::new( - Default::default(), - a::Reference { name, refuri, refid, refname }, - children - ).into()) + let name; + let refuri; + let refid; + let mut refname = vec![]; + let mut children: Vec<c::TextOrInlineElement> = vec![]; + let concrete = pair.into_inner().next().unwrap(); + match concrete.as_rule() { + Rule::reference_target => { + let rt_inner = concrete.into_inner().next().unwrap(); // reference_target_uq or target_name_qu + match rt_inner.as_rule() { + Rule::reference_target_uq => { + refid = None; + name = Some(rt_inner.as_str().into()); + refuri = None; + refname.push(rt_inner.as_str().into()); + children.push(rt_inner.as_str().into()); + } + Rule::reference_target_qu => { + let (text, reference) = { + let mut text = None; + let mut reference = None; + for inner in rt_inner.clone().into_inner() { + match inner.as_rule() { + Rule::reference_text => text = Some(inner), + Rule::reference_bracketed => reference = Some(inner), + _ => unreachable!(), + } + } + (text, reference) + }; + let trimmed_text = match (&text, &reference) { + (Some(text), None) => text.as_str(), + (_, Some(reference)) => text + .map(|text| text.as_str().trim_end_matches(|ch| " \n\r".contains(ch))) + .filter(|text| !text.is_empty()) + .unwrap_or_else(|| { + reference.clone().into_inner().next().unwrap().as_str() + }), + (None, None) => unreachable!(), + }; + refid = None; + name = Some(trimmed_text.into()); + refuri = if let Some(reference) = reference { + let inner = reference.into_inner().next().unwrap(); + match inner.as_rule() { + // The URL rules in our parser accept a narrow superset of + // valid URLs, so we need to handle false positives. + Rule::url => { + if let Ok(target) = Url::parse_absolute(inner.as_str()) { + Some(target) + } else if inner.as_str().ends_with('_') { + // like target_name_qu (minus the final underscore) + let full_str = inner.as_str(); + refname.push(full_str[0..full_str.len() - 1].into()); + None + } else { + // like relative_reference + Some(Url::parse_relative(inner.as_str())?) + } + } + Rule::target_name_qu => { + refname.push(inner.as_str().into()); + None + } + Rule::relative_reference => Some(Url::parse_relative(inner.as_str())?), + _ => unreachable!(), + } + } else { + refname.push(trimmed_text.into()); + None + }; + children.push(trimmed_text.into()); + } + _ => unreachable!(), + } + } + Rule::reference_explicit => unimplemented!("explicit reference"), + Rule::reference_auto => { + let rt_inner = concrete.into_inner().next().unwrap(); + match rt_inner.as_rule() { + Rule::url_auto => match Url::parse_absolute(rt_inner.as_str()) { + Ok(target) => { + refuri = Some(target); + name = None; + refid = None; + children.push(rt_inner.as_str().into()); + } + // if our parser got a URL wrong, return it as a string + Err(_) => return Ok(rt_inner.as_str().into()), + }, + Rule::email => { + let mailto_url = String::from("mailto:") + rt_inner.as_str(); + match Url::parse_absolute(&mailto_url) { + Ok(target) => { + refuri = Some(target); + name = None; + refid = None; + children.push(rt_inner.as_str().into()); + } + // if our parser got a URL wrong, return it as a string + Err(_) => return Ok(rt_inner.as_str().into()), + } + } + _ => unreachable!(), + } + } + _ => unreachable!(), + }; + Ok(e::Reference::new( + Default::default(), + a::Reference { + name, + refuri, + refid, + refname, + }, + children, + ) + .into()) } fn convert_substitution_ref(pair: Pair<Rule>) -> Result<e::SubstitutionReference, Error> { - let name = whitespace_normalize_name(pair.as_str()); - Ok(a::ExtraAttributes::with_extra( - a::SubstitutionReference { - refname: vec![at::NameToken(name)] - } - )) + let name = whitespace_normalize_name(pair.as_str()); + Ok(a::ExtraAttributes::with_extra(a::SubstitutionReference { + refname: vec![at::NameToken(name)], + })) } |
