fix: process of records without a ppn (#908)

Signed-off-by: Nico Wagner <[email protected]>
deutsche-nationalbibliothek · Feb 13, 2025 · 003a405 · 003a405
1 parent 7465d04
commit 003a405
Show file tree

Hide file tree

Showing 7 changed files with 37 additions and 12 deletions.
diff --git a/crates/pica-cli/src/commands/concat.rs b/crates/pica-cli/src/commands/concat.rs
@@ -78,7 +78,7 @@ pub(crate) enum Strategy {
 #[inline]
 fn record_key(record: &ByteRecord, strategy: &Strategy) -> String {
     match strategy {
-        Strategy::Idn => record.ppn().to_string(),
+        Strategy::Idn => record.ppn().expect("ppn").to_string(),
         Strategy::Hash => {
             record.sha256().iter().fold(String::new(), |mut out, b| {
                 let _ = write!(out, "{b:02x}");

diff --git a/crates/pica-cli/src/commands/filter.rs b/crates/pica-cli/src/commands/filter.rs
@@ -196,8 +196,10 @@ impl Filter {
                     Ok(ref mut record) => {
                         progress.update(false);
 
-                        if !filter_set.check(record.ppn()) {
-                            continue;
+                        if let Some(ppn) = record.ppn() {
+                            if !filter_set.check(ppn) {
+                                continue;
+                            }
                         }
 
                         let mut is_match =

diff --git a/crates/pica-cli/src/commands/frequency.rs b/crates/pica-cli/src/commands/frequency.rs
@@ -202,8 +202,10 @@ impl Frequency {
                     Ok(ref record) => {
                         progress.update(false);
 
-                        if !filter_set.check(record.ppn()) {
-                            continue;
+                        if let Some(ppn) = record.ppn() {
+                            if !filter_set.check(ppn) {
+                                continue;
+                            }
                         }
 
                         if let Some(ref matcher) = matcher {

diff --git a/crates/pica-cli/src/commands/hash.rs b/crates/pica-cli/src/commands/hash.rs
@@ -80,7 +80,10 @@ impl Hash {
                         );
 
                         writer.write_record(&[
-                            record.ppn().to_string(),
+                            record
+                                .ppn()
+                                .unwrap_or_default()
+                                .to_string(),
                             hash,
                         ])?;
                     }

diff --git a/crates/pica-cli/src/commands/select.rs b/crates/pica-cli/src/commands/select.rs
@@ -221,8 +221,10 @@ impl Select {
                     Ok(ref record) => {
                         progress.update(false);
 
-                        if !filter_set.check(record.ppn()) {
-                            continue;
+                        if let Some(ppn) = record.ppn() {
+                            if !filter_set.check(ppn) {
+                                continue;
+                            }
                         }
 
                         if let Some(ref matcher) = matcher {

diff --git a/crates/pica-cli/tests/filter/mod.rs b/crates/pica-cli/tests/filter/mod.rs
@@ -741,3 +741,19 @@ fn filter_tee() -> TestResult {
     temp_dir.close().unwrap();
     Ok(())
 }
+
+/// https://github.com/deutsche-nationalbibliothek/pica-rs/issues/907
+#[test]
+fn filter_no_ppn() -> TestResult {
+    let mut cmd = Command::cargo_bin("pica")?;
+
+    let data = "036E/00 \x1faSpringer-Lehrbuch\x1e036E/01 \x1faSpringer-Link\x1fpBücher\x1e\n";
+    let assert =
+        cmd.args(["filter", "....?"]).write_stdin(data).assert();
+    assert
+        .success()
+        .code(0)
+        .stdout(predicates::ord::eq(data))
+        .stderr(predicates::str::is_empty());
+    Ok(())
+}
diff --git a/src/path.rs b/src/path.rs
@@ -250,15 +250,15 @@ pub trait PathExt {
     /// use pica_record::prelude::*;
     ///
     /// let record = ByteRecord::from_bytes(b"003@ \x1f0118540238\x1e\n")?;
-    /// assert_eq!(record.ppn(), "118540238");
+    /// assert_eq!(record.ppn().unwrap(), "118540238");
     ///
     /// # Ok::<(), Box<dyn std::error::Error>>(())
     /// ```
-    fn ppn(&self) -> &Self::Value {
+    fn ppn(&self) -> Option<&Self::Value> {
         static PATH: LazyLock<Path> =
             LazyLock::new(|| Path::new("[email protected]").unwrap());
 
-        self.first(&PATH, &Default::default()).unwrap()
+        self.first(&PATH, &Default::default())
     }
 }
 impl PathExt for RecordRef<'_> {
@@ -492,7 +492,7 @@ mod tests {
     fn test_path_ppn() -> TestResult {
         let data = ada_lovelace();
         let record = ByteRecord::from_bytes(&data)?;
-        assert_eq!(record.ppn(), "119232022");
+        assert_eq!(record.ppn().unwrap(), "119232022");
         Ok(())
     }
 }