Adapt select/frequency to new path struct (#686)

pica-matcher/src/subfield_matcher.rs

@@ -52,7 +52,7 @@ pub trait Matcher {
 /// This matcher can be used to determine if a single subfield or a list
 /// of subfields contains at least one subfield with a code, that is
 /// contained in the matcher's code list.
-#[derive(Debug, PartialEq, Eq)]
+#[derive(Clone, Debug, PartialEq, Eq)]
 pub struct ExistsMatcher {
     codes: Vec<char>,
 }
@@ -139,7 +139,7 @@ impl Matcher for ExistsMatcher {
 /// * StartsWith (`=^`)
 /// * EndsWith (`=$`)
 /// * Similar (`=*`)
-#[derive(Debug, PartialEq, Eq)]
+#[derive(Clone, Debug, PartialEq, Eq)]
 pub struct RelationMatcher {
     codes: Vec<char>,
     op: RelationalOp,
@@ -331,7 +331,7 @@ impl Matcher for RelationMatcher {
 }
 
 /// A matcher that checks a subfield value against a regex.
-#[derive(Debug, PartialEq, Eq)]
+#[derive(Clone, Debug, PartialEq, Eq)]
 pub struct RegexMatcher {
     codes: Vec<char>,
     pattern: String,
@@ -421,7 +421,7 @@ impl Matcher for RegexMatcher {
 }
 
 /// A matcher that checks if a subfield value is in a predefined list.
-#[derive(Debug, PartialEq, Eq)]
+#[derive(Clone, Debug, PartialEq, Eq)]
 pub struct InMatcher {
     codes: Vec<char>,
     values: Vec<BString>,
@@ -520,7 +520,7 @@ impl Matcher for InMatcher {
 }
 
 /// A matcher that checks the number of occurrences of a subfield.
-#[derive(Debug, PartialEq, Eq)]
+#[derive(Clone, Debug, PartialEq, Eq)]
 pub struct CardinalityMatcher {
     code: char,
     op: RelationalOp,
@@ -615,7 +615,7 @@ impl Matcher for CardinalityMatcher {
 ///
 /// This matcher combines all atomic, singleton matcher into a new
 /// matcher.
-#[derive(Debug, PartialEq, Eq)]
+#[derive(Clone, Debug, PartialEq, Eq)]
 pub enum SingletonMatcher {
     Cardinality(CardinalityMatcher),
     Exists(ExistsMatcher),
@@ -685,7 +685,7 @@ impl Matcher for SingletonMatcher {
 
 /// A matcher that allows grouping, negation and connecting of singleton
 /// matcher.
-#[derive(Debug, PartialEq, Eq)]
+#[derive(Clone, Debug, PartialEq, Eq)]
 pub enum SubfieldMatcher {
     Singleton(SingletonMatcher),
     Group(Box<SubfieldMatcher>),

pica-matcher/src/tag_matcher.rs

@@ -12,7 +12,7 @@ use pica_record::{Tag, TagMut};
 use crate::ParseMatcherError;
 
 /// A matcher that matches against PICA+ [Tags](`pica_record::Tag`).
-#[derive(Debug, PartialEq, Eq)]
+#[derive(Clone, Debug, PartialEq, Eq)]
 pub enum TagMatcher {
     Simple(TagMut),
     Pattern([Vec<char>; 4]),

pica-path/src/lib.rs

@@ -26,7 +26,7 @@ use thiserror::Error;
 #[error("invalid path expression, got `{0}`")]
 pub struct ParsePathError(String);
 
-#[derive(Debug, PartialEq, Eq)]
+#[derive(Clone, Debug, PartialEq, Eq)]
 pub struct Path {
     tag_matcher: TagMatcher,
     occurrence_matcher: OccurrenceMatcher,

pica-select/src/lib.rs

@@ -104,6 +104,12 @@ impl FromStr for Query {
     }
 }
 
+impl From<Path> for Query {
+    fn from(path: Path) -> Self {
+        Self(vec![path.into()])
+    }
+}
+
 #[derive(Debug, Copy, Clone)]
 enum Quotes {
     Single,
@@ -467,15 +473,16 @@ impl<T: AsRef<[u8]> + Debug + Display> QueryExt for Record<T> {
                             }
                         })
                         .map(|field| {
-                            // FIXME
-                            path.codes_flat()
+                            path.codes()
                                 .iter()
-                                .map(|code| {
+                                .map(|codes| {
                                     field
                                         .subfields()
                                         .iter()
                                         .filter(|subfield| {
-                                            subfield.code() == *code
+                                            codes.contains(
+                                                &subfield.code(),
+                                            )
                                         })
                                         .map(|subfield| {
                                             subfield.value()

pica-toolkit/src/commands/frequency.rs

@@ -5,11 +5,10 @@ use std::fs::File;
 use std::io::{self, Write};
 use std::str::FromStr;
 
-use bstr::BString;
 use clap::{value_parser, Parser};
-use pica_matcher::MatcherOptions;
-use pica_path::{Path, PathExt};
+use pica_path::Path;
 use pica_record::io::{ReaderBuilder, RecordsIterator};
+use pica_select::{Query, QueryExt, QueryOptions};
 use serde::{Deserialize, Serialize};
 
 use crate::config::Config;
@@ -123,8 +122,8 @@ impl Frequency {
             Path::from_str(&self.path)?
         };
 
-        let mut ftable: HashMap<BString, u64> = HashMap::new();
-        let options = MatcherOptions::new()
+        let mut ftable: HashMap<Vec<String>, u64> = HashMap::new();
+        let options = QueryOptions::new()
             .strsim_threshold(self.strsim_threshold as f64 / 100f64)
             .case_ignore(self.ignore_case);
 
@@ -151,10 +150,15 @@ impl Frequency {
                         }
                     }
                     Ok(record) => {
-                        for value in record.path(&path, &options) {
-                            *ftable
-                                .entry(BString::from(value.to_vec()))
-                                .or_insert(0) += 1;
+                        let outcome = record.query(
+                            &Query::from(path.clone()),
+                            &options,
+                        );
+
+                        for key in outcome.clone().into_iter() {
+                            if key.iter().any(|e| !e.is_empty()) {
+                                *ftable.entry(key).or_insert(0) += 1;
+                            }
                         }
                     }
                 }
@@ -165,8 +169,9 @@ impl Frequency {
             writer.write_record(header.split(',').map(|s| s.trim()))?;
         }
 
-        let mut ftable_sorted: Vec<(&BString, &u64)> =
+        let mut ftable_sorted: Vec<(&Vec<String>, &u64)> =
             ftable.iter().collect();
+
         if self.reverse {
             ftable_sorted.sort_by(|a, b| match a.1.cmp(b.1) {
                 Ordering::Equal => a.0.cmp(b.0),
@@ -179,7 +184,7 @@ impl Frequency {
             });
         }
 
-        for (i, (value, frequency)) in ftable_sorted.iter().enumerate()
+        for (i, (values, frequency)) in ftable_sorted.iter().enumerate()
         {
             if self.limit > 0 && i >= self.limit {
                 break;
@@ -189,12 +194,13 @@ impl Frequency {
                 break;
             }
 
-            let value = translit_maybe(
-                &value.to_string(),
-                self.translit.as_deref(),
-            );
+            let mut record = values
+                .iter()
+                .map(|s| translit_maybe(s, self.translit.as_deref()))
+                .collect::<Vec<_>>();
 
-            writer.write_record(&[value, frequency.to_string()])?;
+            record.push(frequency.to_string());
+            writer.write_record(record)?;
         }
 
         writer.flush()?;

pica-toolkit/tests/snapshot/frequency/027-frequency-multiple-cols.toml

@@ -0,0 +1,5 @@
+bin.name = "pica"
+args = "frequency -s \"065R{ (9, 4) | 9? && 4 =^ 'ort' }\" dump.dat.gz"
+status = "success"
+stdout = "040660095,ortg,1\n040787044,ortx,1\n041178548,orts,1\n041178548,ortw,1\n"
+stderr = ""