Merge pull request #823 from RumovZ/rework-search-parser

Rework search parser
This commit is contained in:
Damien Elmes 2020-11-20 16:22:12 +10:00 committed by GitHub
commit e23d40e850
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
6 changed files with 365 additions and 187 deletions

View File

@ -48,8 +48,10 @@ def test_findCards():
# tag searches # tag searches
assert len(col.findCards("tag:*")) == 5 assert len(col.findCards("tag:*")) == 5
assert len(col.findCards("tag:\\*")) == 1 assert len(col.findCards("tag:\\*")) == 1
assert len(col.findCards("tag:%")) == 5 assert len(col.findCards("tag:%")) == 1
assert len(col.findCards("tag:\\%")) == 1 assert len(col.findCards("tag:sheep_goat")) == 0
assert len(col.findCards('"tag:sheep goat"')) == 0
assert len(col.findCards('"tag:* *"')) == 0
assert len(col.findCards("tag:animal_1")) == 2 assert len(col.findCards("tag:animal_1")) == 2
assert len(col.findCards("tag:animal\\_1")) == 1 assert len(col.findCards("tag:animal\\_1")) == 1
assert not col.findCards("tag:donkey") assert not col.findCards("tag:donkey")

View File

@ -1207,7 +1207,7 @@ QTableView {{ gridline-color: {grid} }}
if i % 2 == 0: if i % 2 == 0:
txt += a + ":" txt += a + ":"
else: else:
txt += re.sub("[*%_]", r"\\\g<0>", a) txt += re.sub(r"[*_\\]", r"\\\g<0>", a)
for c in '  ()"': for c in '  ()"':
if c in txt: if c in txt:
txt = '"{}"'.format(txt.replace('"', '\\"')) txt = '"{}"'.format(txt.replace('"', '\\"'))

View File

@ -9,17 +9,15 @@ use crate::{
use lazy_static::lazy_static; use lazy_static::lazy_static;
use nom::{ use nom::{
branch::alt, branch::alt,
bytes::complete::{escaped, is_not, tag, take_while1}, bytes::complete::{escaped, is_not, tag},
character::complete::{anychar, char, one_of}, character::complete::{anychar, char, none_of, one_of},
combinator::{all_consuming, map, map_res}, combinator::{all_consuming, map, map_res, verify},
sequence::{delimited, preceded, tuple}, sequence::{delimited, preceded, separated_pair},
{multi::many0, IResult}, {multi::many0, IResult},
}; };
use regex::Regex; use regex::{Captures, Regex};
use std::{borrow::Cow, num}; use std::{borrow::Cow, num};
// fixme: need to preserve \ when used twice in string
struct ParseError {} struct ParseError {}
impl From<num::ParseIntError> for ParseError { impl From<num::ParseIntError> for ParseError {
@ -63,7 +61,7 @@ pub(super) enum SearchNode<'a> {
}, },
AddedInDays(u32), AddedInDays(u32),
EditedInDays(u32), EditedInDays(u32),
CardTemplate(TemplateKind), CardTemplate(TemplateKind<'a>),
Deck(Cow<'a, str>), Deck(Cow<'a, str>),
DeckID(DeckID), DeckID(DeckID),
NoteTypeID(NoteTypeID), NoteTypeID(NoteTypeID),
@ -75,12 +73,12 @@ pub(super) enum SearchNode<'a> {
Tag(Cow<'a, str>), Tag(Cow<'a, str>),
Duplicates { Duplicates {
note_type_id: NoteTypeID, note_type_id: NoteTypeID,
text: String, text: Cow<'a, str>,
}, },
State(StateKind), State(StateKind),
Flag(u8), Flag(u8),
NoteIDs(Cow<'a, str>), NoteIDs(&'a str),
CardIDs(Cow<'a, str>), CardIDs(&'a str),
Property { Property {
operator: String, operator: String,
kind: PropertyKind, kind: PropertyKind,
@ -113,9 +111,9 @@ pub(super) enum StateKind {
} }
#[derive(Debug, PartialEq)] #[derive(Debug, PartialEq)]
pub(super) enum TemplateKind { pub(super) enum TemplateKind<'a> {
Ordinal(u16), Ordinal(u16),
Name(String), Name(Cow<'a, str>),
} }
/// Parse the input string into a list of nodes. /// Parse the input string into a list of nodes.
@ -127,7 +125,6 @@ pub(super) fn parse(input: &str) -> Result<Vec<Node>> {
let (_, nodes) = let (_, nodes) =
all_consuming(group_inner)(input).map_err(|_e| AnkiError::SearchError(None))?; all_consuming(group_inner)(input).map_err(|_e| AnkiError::SearchError(None))?;
Ok(nodes) Ok(nodes)
} }
@ -205,32 +202,21 @@ fn text(s: &str) -> IResult<&str, Node> {
/// Determine if text is a qualified search, and handle escaped chars. /// Determine if text is a qualified search, and handle escaped chars.
fn search_node_for_text(s: &str) -> ParseResult<SearchNode> { fn search_node_for_text(s: &str) -> ParseResult<SearchNode> {
let mut it = s.splitn(2, ':'); let (tail, head) = escaped(is_not(r":\"), '\\', anychar)(s)?;
let (head, tail) = ( if tail.is_empty() {
unescape_quotes(it.next().unwrap()), Ok(SearchNode::UnqualifiedText(unescape(head)?))
it.next().map(unescape_quotes),
);
if let Some(tail) = tail {
search_node_for_text_with_argument(head, tail)
} else { } else {
Ok(SearchNode::UnqualifiedText(head)) search_node_for_text_with_argument(head, &tail[1..])
} }
} }
/// \" -> " /// Unquoted text, terminated by whitespace or unescaped ", ( or )
fn unescape_quotes(s: &str) -> Cow<str> {
if s.find(r#"\""#).is_some() {
s.replace(r#"\""#, "\"").into()
} else {
s.into()
}
}
/// Unquoted text, terminated by a space or )
fn unquoted_term(s: &str) -> IResult<&str, Node> { fn unquoted_term(s: &str) -> IResult<&str, Node> {
map_res( map_res(
take_while1(|c| c != ' ' && c != ')' && c != '"'), verify(
escaped(is_not("\"() \u{3000}\\"), '\\', none_of(" \u{3000}")),
|s: &str| !s.is_empty(),
),
|text: &str| -> ParseResult<Node> { |text: &str| -> ParseResult<Node> {
Ok(if text.eq_ignore_ascii_case("or") { Ok(if text.eq_ignore_ascii_case("or") {
Node::Or Node::Or
@ -256,57 +242,66 @@ fn quoted_term_str(s: &str) -> IResult<&str, &str> {
/// Quoted text, terminated by a non-escaped double quote /// Quoted text, terminated by a non-escaped double quote
fn quoted_term_inner(s: &str) -> IResult<&str, &str> { fn quoted_term_inner(s: &str) -> IResult<&str, &str> {
escaped(is_not(r#""\"#), '\\', anychar)(s) verify(escaped(is_not(r#""\"#), '\\', anychar), |s: &str| {
!s.is_empty()
})(s)
} }
/// eg deck:"foo bar" - quotes must come after the : /// eg deck:"foo bar" - quotes must come after the :
fn partially_quoted_term(s: &str) -> IResult<&str, Node> { fn partially_quoted_term(s: &str) -> IResult<&str, Node> {
let term = take_while1(|c| c != ' ' && c != ')' && c != ':'); map_res(
let (s, (term, _, quoted_val)) = tuple((term, char(':'), quoted_term_str))(s)?; separated_pair(
let quoted_val = unescape_quotes(quoted_val); verify(
escaped(is_not("\"(): \u{3000}\\"), '\\', none_of(": \u{3000}")),
match search_node_for_text_with_argument(term.into(), quoted_val) { |s: &str| !s.is_empty(),
Ok(search) => Ok((s, Node::Search(search))), ),
Err(_) => Err(nom::Err::Failure((s, nom::error::ErrorKind::NoneOf))), char(':'),
} quoted_term_str,
),
|p| match search_node_for_text_with_argument(p.0, p.1) {
Ok(search) => Ok(Node::Search(search)),
Err(e) => Err(e),
},
)(s)
} }
/// Convert a colon-separated key/val pair into the relevant search type. /// Convert a colon-separated key/val pair into the relevant search type.
fn search_node_for_text_with_argument<'a>( fn search_node_for_text_with_argument<'a>(
key: Cow<'a, str>, key: &'a str,
val: Cow<'a, str>, val: &'a str,
) -> ParseResult<SearchNode<'a>> { ) -> ParseResult<SearchNode<'a>> {
Ok(match key.to_ascii_lowercase().as_str() { Ok(match key.to_ascii_lowercase().as_str() {
"added" => SearchNode::AddedInDays(val.parse()?), "added" => SearchNode::AddedInDays(val.parse()?),
"edited" => SearchNode::EditedInDays(val.parse()?), "edited" => SearchNode::EditedInDays(val.parse()?),
"deck" => SearchNode::Deck(val), "deck" => SearchNode::Deck(unescape(val)?),
"note" => SearchNode::NoteType(val), "note" => SearchNode::NoteType(unescape(val)?),
"tag" => SearchNode::Tag(val), "tag" => SearchNode::Tag(unescape(val)?),
"mid" => SearchNode::NoteTypeID(val.parse()?), "mid" => SearchNode::NoteTypeID(val.parse()?),
"nid" => SearchNode::NoteIDs(check_id_list(val)?), "nid" => SearchNode::NoteIDs(check_id_list(val)?),
"cid" => SearchNode::CardIDs(check_id_list(val)?), "cid" => SearchNode::CardIDs(check_id_list(val)?),
"did" => SearchNode::DeckID(val.parse()?), "did" => SearchNode::DeckID(val.parse()?),
"card" => parse_template(val.as_ref()), "card" => parse_template(val)?,
"is" => parse_state(val.as_ref())?, "is" => parse_state(val)?,
"flag" => parse_flag(val.as_ref())?, "flag" => parse_flag(val)?,
"rated" => parse_rated(val.as_ref())?, "rated" => parse_rated(val)?,
"dupe" => parse_dupes(val.as_ref())?, "dupe" => parse_dupes(val)?,
"prop" => parse_prop(val.as_ref())?, "prop" => parse_prop(val)?,
"re" => SearchNode::Regex(val), "re" => SearchNode::Regex(unescape_quotes(val)),
"nc" => SearchNode::NoCombining(val), "r" => SearchNode::UnqualifiedText(unescape_raw(val)),
"w" => SearchNode::WordBoundary(val), "nc" => SearchNode::NoCombining(unescape(val)?),
"w" => SearchNode::WordBoundary(unescape(val)?),
// anything else is a field search // anything else is a field search
_ => parse_single_field(key.as_ref(), val.as_ref()), _ => parse_single_field(key, val)?,
}) })
} }
/// ensure a list of ids contains only numbers and commas, returning unchanged if true /// ensure a list of ids contains only numbers and commas, returning unchanged if true
/// used by nid: and cid: /// used by nid: and cid:
fn check_id_list(s: Cow<str>) -> ParseResult<Cow<str>> { fn check_id_list(s: &str) -> ParseResult<&str> {
lazy_static! { lazy_static! {
static ref RE: Regex = Regex::new(r"^(\d+,)*\d+$").unwrap(); static ref RE: Regex = Regex::new(r"^(\d+,)*\d+$").unwrap();
} }
if RE.is_match(s.as_ref()) { if RE.is_match(s) {
Ok(s) Ok(s)
} else { } else {
Err(ParseError {}) Err(ParseError {})
@ -360,13 +355,13 @@ fn parse_rated(val: &str) -> ParseResult<SearchNode<'static>> {
} }
/// eg dupes:1231,hello /// eg dupes:1231,hello
fn parse_dupes(val: &str) -> ParseResult<SearchNode<'static>> { fn parse_dupes(val: &str) -> ParseResult<SearchNode> {
let mut it = val.splitn(2, ','); let mut it = val.splitn(2, ',');
let mid: NoteTypeID = it.next().unwrap().parse()?; let mid: NoteTypeID = it.next().unwrap().parse()?;
let text = it.next().ok_or(ParseError {})?; let text = it.next().ok_or(ParseError {})?;
Ok(SearchNode::Duplicates { Ok(SearchNode::Duplicates {
note_type_id: mid, note_type_id: mid,
text: text.into(), text: unescape_quotes(text),
}) })
} }
@ -411,27 +406,116 @@ fn parse_prop(val: &str) -> ParseResult<SearchNode<'static>> {
}) })
} }
fn parse_template(val: &str) -> SearchNode<'static> { fn parse_template(val: &str) -> ParseResult<SearchNode> {
SearchNode::CardTemplate(match val.parse::<u16>() { Ok(SearchNode::CardTemplate(match val.parse::<u16>() {
Ok(n) => TemplateKind::Ordinal(n.max(1) - 1), Ok(n) => TemplateKind::Ordinal(n.max(1) - 1),
Err(_) => TemplateKind::Name(val.into()), Err(_) => TemplateKind::Name(unescape(val)?),
}))
}
fn parse_single_field<'a>(key: &'a str, val: &'a str) -> ParseResult<SearchNode<'a>> {
Ok(if val.starts_with("re:") {
SearchNode::SingleField {
field: unescape(key)?,
text: unescape_quotes(&val[3..]),
is_re: true,
}
} else if val.starts_with("r:") {
SearchNode::SingleField {
field: unescape(key)?,
text: unescape_raw(&val[2..]),
is_re: false,
}
} else {
SearchNode::SingleField {
field: unescape(key)?,
text: unescape(val)?,
is_re: false,
}
}) })
} }
fn parse_single_field(key: &str, mut val: &str) -> SearchNode<'static> { /// For strings without unescaped ", convert \" to "
let is_re = if val.starts_with("re:") { fn unescape_quotes(s: &str) -> Cow<str> {
val = val.trim_start_matches("re:"); if s.contains('"') {
true s.replace(r#"\""#, "\"").into()
} else { } else {
false s.into()
};
SearchNode::SingleField {
field: key.to_string().into(),
text: val.to_string().into(),
is_re,
} }
} }
/// Unescape quotes but escape wildcards and \s.
fn unescape_raw(s: &str) -> Cow<str> {
lazy_static! {
static ref RE: Regex = Regex::new(r#"\\"?|\*|_"#).unwrap();
}
RE.replace_all(&s, |caps: &Captures| match &caps[0] {
r"\" => r"\\",
"\\\"" => "\"",
r"*" => r"\*",
r"_" => r"\_",
_ => unreachable!(),
})
}
/// Unescape chars with special meaning to the parser.
fn unescape(txt: &str) -> ParseResult<Cow<str>> {
if is_invalid_escape(txt) {
Err(ParseError {})
} else if is_parser_escape(txt) {
lazy_static! {
static ref RE: Regex = Regex::new(r#"\\[\\":()-]"#).unwrap();
}
Ok(RE.replace_all(&txt, |caps: &Captures| match &caps[0] {
r"\\" => r"\\",
"\\\"" => "\"",
r"\:" => ":",
r"\(" => "(",
r"\)" => ")",
r"\-" => "-",
_ => unreachable!(),
}))
} else {
Ok(txt.into())
}
}
/// Check string for invalid escape sequences.
fn is_invalid_escape(txt: &str) -> bool {
// odd number of \s not followed by an escapable character
lazy_static! {
static ref RE: Regex = Regex::new(
r#"(?x)
(?:^|[^\\]) # not a backslash
(?:\\\\)* # even number of backslashes
\\ # single backslash
(?:[^\\":*_()-]|$) # anything but an escapable char
"#
)
.unwrap();
}
RE.is_match(txt)
}
/// Check string for escape sequences handled by the parser: ":()
fn is_parser_escape(txt: &str) -> bool {
// odd number of \s followed by a char with special meaning to the parser
lazy_static! {
static ref RE: Regex = Regex::new(
r#"(?x)
(?:^|[^\\]) # not a backslash
(?:\\\\)* # even number of backslashes
\\ # single backslash
[":()-] # parser escape
"#
)
.unwrap();
}
RE.is_match(txt)
}
#[cfg(test)] #[cfg(test)]
mod test { mod test {
use super::*; use super::*;
@ -497,7 +581,7 @@ mod test {
})] })]
); );
// partially quoted text should handle escaping the same way // escaping is independent of quotation
assert_eq!( assert_eq!(
parse(r#""field:va\"lue""#)?, parse(r#""field:va\"lue""#)?,
vec![Search(SingleField { vec![Search(SingleField {
@ -507,13 +591,67 @@ mod test {
})] })]
); );
assert_eq!(parse(r#""field:va\"lue""#)?, parse(r#"field:"va\"lue""#)?,); assert_eq!(parse(r#""field:va\"lue""#)?, parse(r#"field:"va\"lue""#)?,);
assert_eq!(parse(r#""field:va\"lue""#)?, parse(r#"field:va\"lue"#)?,);
// any character should be escapable in quotes // only \":()-*_ are escapable
assert!(parse(r"\").is_err());
assert!(parse(r"\a").is_err());
assert!(parse(r"\%").is_err());
// parser unescapes ":()-
assert_eq!( assert_eq!(
parse(r#""re:\btest""#)?, parse(r#"\"\:\(\)\-"#)?,
vec![Search(Regex(r"\btest".into()))] vec![Search(UnqualifiedText(r#"":()-"#.into())),]
); );
// parser doesn't unescape unescape \*_
assert_eq!(
parse(r#"\\\*\_"#)?,
vec![Search(UnqualifiedText(r#"\\\*\_"#.into())),]
);
// escaping parentheses is optional (only) inside quotes
assert_eq!(parse(r#""\)\(""#), parse(r#"")(""#));
assert!(parse(")(").is_err());
// escaping : is optional if it is preceded by another :
assert!(parse(":test").is_err());
assert!(parse(":").is_err());
assert_eq!(parse("field:val:ue"), parse(r"field:val\:ue"));
assert_eq!(parse(r#""field:val:ue""#), parse(r"field:val\:ue"));
assert_eq!(parse(r#"field:"val:ue""#), parse(r"field:val\:ue"));
// escaping - is optional if it cannot be mistaken for a negator
assert_eq!(parse("-"), parse(r"\-"));
assert_eq!(parse("A-"), parse(r"A\-"));
assert_eq!(parse(r#""-A""#), parse(r"\-A"));
assert_ne!(parse("-A"), parse(r"\-A"));
// any character should be escapable on the right side of re:
assert_eq!(
parse(r#""re:\btest\%""#)?,
vec![Search(Regex(r"\btest\%".into()))]
);
// treat all chars as literals in raw searches
assert_eq!(parse(r"r:\*_"), parse(r"\\\*\_"));
assert_eq!(parse(r"field:r:\*_"), parse(r"field:\\\*\_"));
// no exceptions for escaping "
assert_eq!(
parse(r#"re:te\"st"#)?,
vec![Search(Regex(r#"te"st"#.into()))]
);
assert!(parse(r#"re:te"st"#).is_err());
assert_eq!(
parse(r#"r:te\"st"#)?,
vec![Search(UnqualifiedText(r#"te"st"#.into()))]
);
assert!(parse(r#"r:te"st"#).is_err());
// spaces are optional if node separation is clear
assert_eq!(parse(r#"a"b"(c)"#)?, parse("a b (c)")?);
assert_eq!(parse("added:3")?, vec![Search(AddedInDays(3))]); assert_eq!(parse("added:3")?, vec![Search(AddedInDays(3))]);
assert_eq!( assert_eq!(
parse("card:front")?, parse("card:front")?,

View File

@ -9,12 +9,12 @@ use crate::{
err::Result, err::Result,
notes::field_checksum, notes::field_checksum,
notetype::NoteTypeID, notetype::NoteTypeID,
text::{matches_wildcard, text_to_re}, text::{
text::{normalize_to_nfc, strip_html_preserving_media_filenames, without_combining}, escape_sql, is_glob, matches_glob, normalize_to_nfc, strip_html_preserving_media_filenames,
to_custom_re, to_re, to_sql, to_text, without_combining,
},
timestamp::TimestampSecs, timestamp::TimestampSecs,
}; };
use lazy_static::lazy_static;
use regex::Regex;
use std::{borrow::Cow, fmt::Write}; use std::{borrow::Cow, fmt::Write};
pub(crate) struct SqlWriter<'a> { pub(crate) struct SqlWriter<'a> {
@ -119,7 +119,7 @@ impl SqlWriter<'_> {
// note fields related // note fields related
SearchNode::UnqualifiedText(text) => self.write_unqualified(&self.norm_note(text)), SearchNode::UnqualifiedText(text) => self.write_unqualified(&self.norm_note(text)),
SearchNode::SingleField { field, text, is_re } => { SearchNode::SingleField { field, text, is_re } => {
self.write_single_field(field.as_ref(), &self.norm_note(text), *is_re)? self.write_single_field(&norm(field), &self.norm_note(text), *is_re)?
} }
SearchNode::Duplicates { note_type_id, text } => { SearchNode::Duplicates { note_type_id, text } => {
self.write_dupes(*note_type_id, &self.norm_note(text)) self.write_dupes(*note_type_id, &self.norm_note(text))
@ -132,11 +132,9 @@ impl SqlWriter<'_> {
SearchNode::AddedInDays(days) => self.write_added(*days)?, SearchNode::AddedInDays(days) => self.write_added(*days)?,
SearchNode::EditedInDays(days) => self.write_edited(*days)?, SearchNode::EditedInDays(days) => self.write_edited(*days)?,
SearchNode::CardTemplate(template) => match template { SearchNode::CardTemplate(template) => match template {
TemplateKind::Ordinal(_) => { TemplateKind::Ordinal(_) => self.write_template(template)?,
self.write_template(template)?;
}
TemplateKind::Name(name) => { TemplateKind::Name(name) => {
self.write_template(&TemplateKind::Name(norm(name).into()))?; self.write_template(&TemplateKind::Name(norm(name).into()))?
} }
}, },
SearchNode::Deck(deck) => self.write_deck(&norm(deck))?, SearchNode::Deck(deck) => self.write_deck(&norm(deck))?,
@ -148,6 +146,7 @@ impl SqlWriter<'_> {
} }
SearchNode::NoteType(notetype) => self.write_note_type(&norm(notetype))?, SearchNode::NoteType(notetype) => self.write_note_type(&norm(notetype))?,
SearchNode::Rated { days, ease } => self.write_rated(*days, *ease)?, SearchNode::Rated { days, ease } => self.write_rated(*days, *ease)?,
SearchNode::Tag(tag) => self.write_tag(&norm(tag))?, SearchNode::Tag(tag) => self.write_tag(&norm(tag))?,
SearchNode::State(state) => self.write_state(state)?, SearchNode::State(state) => self.write_state(state)?,
SearchNode::Flag(flag) => { SearchNode::Flag(flag) => {
@ -167,7 +166,7 @@ impl SqlWriter<'_> {
fn write_unqualified(&mut self, text: &str) { fn write_unqualified(&mut self, text: &str) {
// implicitly wrap in % // implicitly wrap in %
let text = format!("%{}%", convert_glob_char(text)); let text = format!("%{}%", &to_sql(text));
self.args.push(text); self.args.push(text);
write!( write!(
self.sql, self.sql,
@ -178,7 +177,7 @@ impl SqlWriter<'_> {
} }
fn write_no_combining(&mut self, text: &str) { fn write_no_combining(&mut self, text: &str) {
let text = format!("%{}%", without_combining(text)); let text = format!("%{}%", without_combining(&to_sql(text)));
self.args.push(text); self.args.push(text);
write!( write!(
self.sql, self.sql,
@ -192,27 +191,27 @@ impl SqlWriter<'_> {
} }
fn write_tag(&mut self, text: &str) -> Result<()> { fn write_tag(&mut self, text: &str) -> Result<()> {
match text { if text.contains(" ") {
"none" => { write!(self.sql, "false").unwrap();
write!(self.sql, "n.tags = ''").unwrap(); } else {
} match text {
"*" | "%" => { "none" => write!(self.sql, "n.tags = ''").unwrap(),
write!(self.sql, "true").unwrap(); "*" => write!(self.sql, "true").unwrap(),
} s => {
text => { if is_glob(s) {
if let Some(re_glob) = glob_to_re(text) { write!(self.sql, "n.tags regexp ?").unwrap();
// text contains a wildcard let re = &to_custom_re(s, r"\S");
let re_glob = format!("(?i).* {} .*", re_glob); self.args.push(format!("(?i).* {} .*", re));
write!(self.sql, "n.tags regexp ?").unwrap(); } else if let Some(tag) = self.col.storage.preferred_tag_case(&to_text(s))? {
self.args.push(re_glob); write!(self.sql, "n.tags like ? escape '\\'").unwrap();
} else if let Some(tag) = self.col.storage.preferred_tag_case(&text)? { self.args.push(format!("% {} %", escape_sql(&tag)));
write!(self.sql, "n.tags like ?").unwrap(); } else {
self.args.push(format!("% {} %", tag)); write!(self.sql, "false").unwrap();
} else { }
write!(self.sql, "false").unwrap();
} }
} }
} }
Ok(()) Ok(())
} }
@ -312,18 +311,20 @@ impl SqlWriter<'_> {
// rewrite "current" to the current deck name // rewrite "current" to the current deck name
let native_deck = if deck == "current" { let native_deck = if deck == "current" {
let current_did = self.col.get_current_deck_id(); let current_did = self.col.get_current_deck_id();
self.col regex::escape(
.storage self.col
.get_deck(current_did)? .storage
.map(|d| d.name) .get_deck(current_did)?
.unwrap_or_else(|| "Default".into()) .map(|d| d.name)
.unwrap_or_else(|| "Default".into())
.as_str(),
)
} else { } else {
human_deck_name_to_native(deck) human_deck_name_to_native(&to_re(deck))
}; };
// convert to a regex that includes child decks // convert to a regex that includes child decks
let re = text_to_re(&native_deck); self.args.push(format!("(?i)^{}($|\x1f)", native_deck));
self.args.push(format!("(?i)^{}($|\x1f)", re));
let arg_idx = self.args.len(); let arg_idx = self.args.len();
self.sql.push_str(&format!(concat!( self.sql.push_str(&format!(concat!(
"(c.did in (select id from decks where name regexp ?{n})", "(c.did in (select id from decks where name regexp ?{n})",
@ -341,8 +342,8 @@ impl SqlWriter<'_> {
write!(self.sql, "c.ord = {}", n).unwrap(); write!(self.sql, "c.ord = {}", n).unwrap();
} }
TemplateKind::Name(name) => { TemplateKind::Name(name) => {
if let Some(re) = glob_to_re(name) { if is_glob(name) {
let re = format!("(?i){}", re); let re = format!("(?i){}", to_re(name));
self.sql.push_str( self.sql.push_str(
"(n.mid,c.ord) in (select ntid,ord from templates where name regexp ?)", "(n.mid,c.ord) in (select ntid,ord from templates where name regexp ?)",
); );
@ -351,7 +352,7 @@ impl SqlWriter<'_> {
self.sql.push_str( self.sql.push_str(
"(n.mid,c.ord) in (select ntid,ord from templates where name = ?)", "(n.mid,c.ord) in (select ntid,ord from templates where name = ?)",
); );
self.args.push(name.to_string()); self.args.push(to_text(name).into());
} }
} }
}; };
@ -359,15 +360,15 @@ impl SqlWriter<'_> {
} }
fn write_note_type(&mut self, nt_name: &str) -> Result<()> { fn write_note_type(&mut self, nt_name: &str) -> Result<()> {
if let Some(re) = glob_to_re(nt_name) { if is_glob(nt_name) {
let re = format!("(?i){}", re); let re = format!("(?i){}", to_re(nt_name));
self.sql self.sql
.push_str("n.mid in (select id from notetypes where name regexp ?)"); .push_str("n.mid in (select id from notetypes where name regexp ?)");
self.args.push(re); self.args.push(re);
} else { } else {
self.sql self.sql
.push_str("n.mid in (select id from notetypes where name = ?)"); .push_str("n.mid in (select id from notetypes where name = ?)");
self.args.push(nt_name.to_string()); self.args.push(to_text(nt_name).into());
} }
Ok(()) Ok(())
} }
@ -378,7 +379,7 @@ impl SqlWriter<'_> {
let mut field_map = vec![]; let mut field_map = vec![];
for nt in note_types.values() { for nt in note_types.values() {
for field in &nt.fields { for field in &nt.fields {
if matches_wildcard(&field.name, field_name) { if matches_glob(&field.name, field_name) {
field_map.push((nt.id, field.ord)); field_map.push((nt.id, field.ord));
} }
} }
@ -401,7 +402,7 @@ impl SqlWriter<'_> {
} else { } else {
cmp = "like"; cmp = "like";
cmp_trailer = "escape '\\'"; cmp_trailer = "escape '\\'";
self.args.push(convert_glob_char(val).into()) self.args.push(to_sql(val).into())
} }
let arg_idx = self.args.len(); let arg_idx = self.args.len();
@ -455,29 +456,10 @@ impl SqlWriter<'_> {
} }
fn write_word_boundary(&mut self, word: &str) { fn write_word_boundary(&mut self, word: &str) {
// fixme: need to escape in the no-glob case as well self.write_regex(&format!(r"\b{}\b", to_re(word)));
let re = text_to_re(word);
self.write_regex(&format!(r"\b{}\b", re))
} }
} }
/// Replace * with %, leaving \* alone.
fn convert_glob_char(val: &str) -> Cow<str> {
lazy_static! {
static ref RE: Regex = Regex::new(r"(^|[^\\])\*").unwrap();
}
RE.replace_all(val, "${1}%")
}
/// Convert a string with _, % or * characters into a regex.
/// If string contains no globbing characters, return None.
fn glob_to_re(glob: &str) -> Option<String> {
if !glob.contains(|c| c == '_' || c == '*' || c == '%') {
return None;
}
Some(text_to_re(glob))
}
#[derive(Debug, PartialEq, Clone, Copy)] #[derive(Debug, PartialEq, Clone, Copy)]
pub enum RequiredTable { pub enum RequiredTable {
Notes, Notes,
@ -601,10 +583,9 @@ mod test {
vec!["%te%st%".into()] vec!["%te%st%".into()]
) )
); );
assert_eq!(s(ctx, "te%st").1, vec!["%te%st%".to_string()]); assert_eq!(s(ctx, "te%st").1, vec![r"%te\%st%".to_string()]);
// user should be able to escape sql wildcards // user should be able to escape wildcards
assert_eq!(s(ctx, r#"te\%s\_t"#).1, vec!["%te\\%s\\_t%".to_string()]); assert_eq!(s(ctx, r#"te\*s\_t"#).1, vec!["%te*s\\_t%".to_string()]);
assert_eq!(s(ctx, r#"te\*s\_t"#).1, vec!["%te\\*s\\_t%".to_string()]);
// qualified search // qualified search
assert_eq!( assert_eq!(
@ -673,23 +654,26 @@ mod test {
) )
); );
// unregistered tag short circuits
assert_eq!(s(ctx, r"tag:one"), ("(false)".into(), vec![]));
// if registered, searches with canonical // if registered, searches with canonical
ctx.transact(None, |col| col.register_tag("One", Usn(-1))) ctx.transact(None, |col| col.register_tag("One", Usn(-1)))
.unwrap(); .unwrap();
assert_eq!( assert_eq!(
s(ctx, r"tag:one"), s(ctx, r"tag:one"),
("(n.tags like ?)".into(), vec![r"% One %".into()]) (
"(n.tags like ? escape '\\')".into(),
vec![r"% One %".into()]
)
); );
// unregistered tags without wildcards won't match
assert_eq!(s(ctx, "tag:unknown"), ("(false)".into(), vec![]));
// wildcards force a regexp search // wildcards force a regexp search
assert_eq!( assert_eq!(
s(ctx, r"tag:o*n\*et%w\%oth_re\_e"), s(ctx, r"tag:o*n\*et%w%oth_re\_e"),
( (
"(n.tags regexp ?)".into(), "(n.tags regexp ?)".into(),
vec![r"(?i).* o.*n\*et.*w%oth.re_e .*".into()] vec![r"(?i).* o\S*n\*et%w%oth\Sre_e .*".into()]
) )
); );
assert_eq!(s(ctx, "tag:none"), ("(n.tags = '')".into(), vec![])); assert_eq!(s(ctx, "tag:none"), ("(n.tags = '')".into(), vec![]));
@ -803,12 +787,4 @@ mod test {
RequiredTable::Notes RequiredTable::Notes
); );
} }
#[test]
fn convert_glob() {
assert_eq!(&convert_glob_char("foo*bar"), "foo%bar");
assert_eq!(&convert_glob_char("*bar"), "%bar");
assert_eq!(&convert_glob_char("\n*bar"), "\n%bar");
assert_eq!(&convert_glob_char(r"\*bar"), r"\*bar");
}
} }

View File

@ -240,17 +240,6 @@ pub(crate) fn ensure_string_in_nfc(s: &mut String) {
} }
} }
/// True if search is equal to text, folding case.
/// Supports '*' to match 0 or more characters.
pub(crate) fn matches_wildcard(text: &str, search: &str) -> bool {
if search.contains('*') {
let search = format!("^(?i){}$", regex::escape(search).replace(r"\*", ".*"));
Regex::new(&search).unwrap().is_match(text)
} else {
uni_eq(text, search)
}
}
/// Convert provided string to NFKD form and strip combining characters. /// Convert provided string to NFKD form and strip combining characters.
pub(crate) fn without_combining(s: &str) -> Cow<str> { pub(crate) fn without_combining(s: &str) -> Cow<str> {
// if the string is already normalized // if the string is already normalized
@ -301,9 +290,91 @@ pub(crate) fn text_to_re(glob: &str) -> String {
text2.into() text2.into()
} }
/// Check if string contains an unescaped wildcard.
pub(crate) fn is_glob(txt: &str) -> bool {
// even number of \s followed by a wildcard
lazy_static! {
static ref RE: Regex = Regex::new(
r#"(?x)
(?:^|[^\\]) # not a backslash
(?:\\\\)* # even number of backslashes
[*_] # wildcard
"#
)
.unwrap();
}
RE.is_match(txt)
}
/// Convert to a RegEx respecting Anki wildcards.
pub(crate) fn to_re(txt: &str) -> Cow<str> {
to_custom_re(txt, ".")
}
/// Convert Anki style to RegEx using the provided wildcard.
pub(crate) fn to_custom_re<'a>(txt: &'a str, wildcard: &str) -> Cow<'a, str> {
lazy_static! {
static ref RE: Regex = Regex::new(r"\\?.").unwrap();
}
RE.replace_all(&txt, |caps: &Captures| {
let s = &caps[0];
match s {
r"\\" | r"\*" => s.to_string(),
r"\_" => "_".to_string(),
"*" => format!("{}*", wildcard),
"_" => wildcard.to_string(),
s => regex::escape(s),
}
})
}
/// Convert to SQL respecting Anki wildcards.
pub(crate) fn to_sql<'a>(txt: &'a str) -> Cow<'a, str> {
// escape sequences and unescaped special characters which need conversion
lazy_static! {
static ref RE: Regex = Regex::new(r"\\[\\*]|[*%]").unwrap();
}
RE.replace_all(&txt, |caps: &Captures| {
let s = &caps[0];
match s {
r"\\" => r"\\",
r"\*" => "*",
"*" => "%",
"%" => r"\%",
_ => unreachable!(),
}
})
}
/// Unescape everything.
pub(crate) fn to_text(txt: &str) -> Cow<str> {
lazy_static! {
static ref RE: Regex = Regex::new(r"\\(.)").unwrap();
}
RE.replace_all(&txt, "$1")
}
/// Escape characters special to SQL: \%_
pub(crate) fn escape_sql(txt: &str) -> Cow<str> {
lazy_static! {
static ref RE: Regex = Regex::new(r"[\\%_]").unwrap();
}
RE.replace_all(&txt, r"\$0")
}
/// Compare text with a possible glob, folding case.
pub(crate) fn matches_glob(text: &str, search: &str) -> bool {
if is_glob(search) {
let search = format!("^(?i){}$", to_re(search));
Regex::new(&search).unwrap().is_match(text)
} else {
uni_eq(text, &to_text(search))
}
}
#[cfg(test)] #[cfg(test)]
mod test { mod test {
use super::matches_wildcard;
use crate::text::without_combining; use crate::text::without_combining;
use crate::text::{ use crate::text::{
extract_av_tags, strip_av_tags, strip_html, strip_html_preserving_media_filenames, AVTag, extract_av_tags, strip_av_tags, strip_html, strip_html_preserving_media_filenames, AVTag,
@ -351,15 +422,6 @@ mod test {
); );
} }
#[test]
fn wildcard() {
assert_eq!(matches_wildcard("foo", "bar"), false);
assert_eq!(matches_wildcard("foo", "Foo"), true);
assert_eq!(matches_wildcard("foo", "F*"), true);
assert_eq!(matches_wildcard("foo", "F*oo"), true);
assert_eq!(matches_wildcard("foo", "b*"), false);
}
#[test] #[test]
fn combining() { fn combining() {
assert!(matches!(without_combining("test"), Cow::Borrowed(_))); assert!(matches!(without_combining("test"), Cow::Borrowed(_)));