skip set and xdata entries by default

4 files changed, 22 insertions, 4 deletions

diff --git a/src/bibiman/citekeys.rs b/src/bibiman/citekeys.rs
index 0cec28e..999c6cb 100644
--- a/src/bibiman/citekeys.rs
+++ b/src/bibiman/citekeys.rs
@@ -28,7 +28,7 @@ use owo_colors::OwoColorize;
 use serde::{Deserialize, Serialize};
 
 use crate::{
-    bibiman::citekeys::citekey_utils::{build_citekey, formatting_help},
+    bibiman::citekeys::citekey_utils::{SKIPPED_ENTRIES, build_citekey, formatting_help},
     config::{BibiConfig, IGNORED_SPECIAL_CHARS, IGNORED_WORDS},
 };
 
@@ -167,6 +167,10 @@ impl CitekeyFormatting {
     pub fn do_formatting(&mut self, ignored_chars: &[char], ignored_words: &[String]) -> &mut Self {
         let mut old_new_keys: Vec<(String, String)> = Vec::new();
         for entry in self.bib_entries.iter() {
+            // Skip specific entries
+            if SKIPPED_ENTRIES.contains(&entry.entry_type.to_string().to_lowercase().as_str()) {
+                continue;
+            }
             old_new_keys.push((
                 entry.key.clone(),
                 build_citekey(
@@ -181,7 +185,6 @@ impl CitekeyFormatting {
         }
 
         self.old_new_keys_map = old_new_keys;
-        // dbg!(&self.old_new_keys_map);
 
         self
     }
diff --git a/src/bibiman/citekeys/citekey_utils.rs b/src/bibiman/citekeys/citekey_utils.rs
index 5f70224..58a8274 100644
--- a/src/bibiman/citekeys/citekey_utils.rs
+++ b/src/bibiman/citekeys/citekey_utils.rs
@@ -15,6 +15,8 @@
 // along with this program.  If not, see <https://www.gnu.org/licenses/>.
 /////
 
+use std::sync::LazyLock;
+
 use biblatex::{ChunksExt, Entry, Type};
 use deunicode::deunicode;
 use indoc::formatdoc;
@@ -25,6 +27,8 @@ use owo_colors::{
 
 use crate::bibiman::{citekeys::CitekeyCase, sanitize::sanitize_single_string_fully};
 
+pub(super) const SKIPPED_ENTRIES: [&str; 2] = ["set", "xdata"];
+
 pub(super) fn formatting_help() {
     let help = vec![
         formatdoc!(
@@ -200,6 +204,7 @@ pub(super) fn build_citekey(
     match case {
         Some(CitekeyCase::Lower) => new_citekey.to_lowercase(),
         Some(CitekeyCase::Upper) => new_citekey.to_uppercase(),
+        // otherwise skip, since camelcase is processed in char loop
         _ => new_citekey,
     }
 }
diff --git a/src/config.rs b/src/config.rs
index 7c1a0f8..b8d8b45 100644
--- a/src/config.rs
+++ b/src/config.rs
@@ -52,6 +52,10 @@ pub static IGNORED_WORDS: LazyLock<Vec<String>> = LazyLock::new(|| {
         String::from("at"),
         String::from("to"),
         String::from("and"),
+        String::from("him"),
+        String::from("her"),
+        String::from("his"),
+        String::from("hers"),
         String::from("der"),
         String::from("die"),
         String::from("das"),
@@ -63,6 +67,12 @@ pub static IGNORED_WORDS: LazyLock<Vec<String>> = LazyLock::new(|| {
         String::from("und"),
         String::from("für"),
         String::from("vor"),
+        String::from("er"),
+        String::from("sie"),
+        String::from("es"),
+        String::from("ihm"),
+        String::from("ihr"),
+        String::from("ihnen"),
     ]
 });
 
diff --git a/tests/biblatex-test-citekeys.bib b/tests/biblatex-test-citekeys.bib
index 9767f97..34c2f33 100644
--- a/tests/biblatex-test-citekeys.bib
+++ b/tests/biblatex-test-citekeys.bib
@@ -1,9 +1,9 @@
-@set{set,
+@set{SET,
     entryset = {article:herrmann-ofele_carboc=carben=as_2006,article:aksin-turkmen_effect=immobi=on_2006,article:yoon-ryu_pallad=pincer=comple_2006},
     annotation = {A \texttt{set} with three members.},
 }
 
-@set{set,
+@set{stdmodel,
     entryset = {article:glashow_partia=symmet=weak_1961,article:weinberg_model=lepton_1967,salam},
     annotation = {A \texttt{set} with three members discussing the standard
                   model of particle physics.},