Merge pull request 'implement basic citekey formatting' (#58) from format-citekeys-#55 into main

Reviewed-on: https://codeberg.org/lukeflo/bibiman/pulls/58

1 files changed, 125 insertions, 5 deletions

diff --git a/src/config.rs b/src/config.rs
index 00a35b7..47e145c 100644
--- a/src/config.rs
+++ b/src/config.rs
@@ -16,21 +16,65 @@
 /////
 
 use std::{
-    fs::{create_dir_all, File},
-    io::{stdin, Write},
+    fs::{File, create_dir_all},
+    io::{Write, stdin},
     path::PathBuf,
     str::FromStr,
+    sync::LazyLock,
 };
 
 use color_eyre::{eyre::Result, owo_colors::OwoColorize};
 use figment::{
-    providers::{Format, Serialized, Toml},
     Figment,
+    providers::{Format, Serialized, Toml},
 };
 use ratatui::style::Color;
 use serde::{Deserialize, Serialize};
 
-use crate::{bibiman::bibisetup::CustomField, cliargs::CLIArgs};
+use crate::{
+    bibiman::{bibisetup::CustomField, citekeys::CitekeyCase},
+    cliargs::CLIArgs,
+};
+
+pub const IGNORED_SPECIAL_CHARS: [char; 33] = [
+    '?', '!', '\\', '\'', '.', '-', '–', ':', ',', '[', ']', '(', ')', '{', '}', '§', '$', '%',
+    '&', '/', '`', '´', '#', '+', '*', '=', '|', '<', '>', '^', '°', '_', '"',
+];
+
+pub static IGNORED_WORDS: LazyLock<Vec<String>> = LazyLock::new(|| {
+    vec![
+        String::from("the"),
+        String::from("a"),
+        String::from("an"),
+        String::from("of"),
+        String::from("for"),
+        String::from("in"),
+        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"),
+        String::from("ein"),
+        String::from("eine"),
+        String::from("eines"),
+        String::from("des"),
+        String::from("auf"),
+        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"),
+    ]
+});
 
 const DEFAULT_CONFIG: &str = r##"
 # [general]
@@ -95,6 +139,55 @@ const DEFAULT_CONFIG: &str = r##"
 # author_color = "38"
 # title_color = "37"
 # year_color = "135"
+
+# [citekey_formatter]
+## Define the patterns for creating citekeys. Every item of the array consists of
+## five components separated by semicolons. Despite the field name every component
+## can be left blank:
+## - name of the biblatex field ("author", "title"...)
+## - number of max words from the given field
+## - number of chars used from each word
+## - delimiter to separate words of the same field
+## - trailing delimiter separating the current field from the following
+# fields = [ "author;2;;-;_", "title;3;6;_;_", "year" ]
+
+## Convert chars to specified case. Possible values:
+## "upper", "uppercase", "lower", "lowercase"
+# case = "lowercase"
+
+## Map all unicode chars to their pure ascii equivalent
+# ascii_only = true
+
+## List of special chars that'll be ignored when building citekeys.
+## A custom list will overwrite the default list
+# ignored_chars = [
+#     "?", "!", "\\", "\'", ".", "-", "–", ":", ",", "[", "]", "(", ")", "{", "}", "§", "$", "%", "&", "/", "`", "´", "#", "+", "*", "=", "|", "<", ">", "^", "°", "_", """,
+# ]
+
+## List of words that'll be ignored when building citekeys.
+## A custom list will overwrite the default list
+# ignored_words = [
+#     "the",
+#     "a",
+#     "an",
+#     "of",
+#     "for",
+#     "in",
+#     "at",
+#     "to",
+#     "and",
+#     "der",
+#     "die",
+#     "das",
+#     "ein",
+#     "eine",
+#     "eines",
+#     "des",
+#     "auf",
+#     "und",
+#     "für",
+#     "vor",
+# ]
 "##;
 
 /// Main struct of the config file. Contains substructs/headings in toml
@@ -102,6 +195,7 @@ const DEFAULT_CONFIG: &str = r##"
 pub struct BibiConfig {
     pub general: General,
     pub colors: Colors,
+    pub citekey_formatter: CitekeyFormatter,
 }
 
 /// Substruct [general] in config.toml
@@ -143,6 +237,15 @@ pub struct Colors {
     pub year_color: Color,
 }
 
+#[derive(Debug, Clone, Serialize, Deserialize, PartialEq)]
+pub struct CitekeyFormatter {
+    pub fields: Option<Vec<String>>,
+    pub case: Option<CitekeyCase>,
+    pub ascii_only: bool,
+    pub ignored_chars: Option<Vec<char>>,
+    pub ignored_words: Option<Vec<String>>,
+}
+
 impl Default for BibiConfig {
     fn default() -> Self {
         Self {
@@ -161,6 +264,13 @@ impl Default for BibiConfig {
                 custom_column: CustomField::Pubtype,
             },
             colors: Self::dark_colors(),
+            citekey_formatter: CitekeyFormatter {
+                fields: None,
+                case: None,
+                ascii_only: true,
+                ignored_chars: None,
+                ignored_words: None,
+            },
         }
     }
 }
@@ -187,6 +297,13 @@ impl BibiConfig {
             } else {
                 Self::dark_colors()
             },
+            citekey_formatter: CitekeyFormatter {
+                fields: None,
+                case: None,
+                ascii_only: true,
+                ignored_chars: None,
+                ignored_words: None,
+            },
         }
     }
 
@@ -344,8 +461,8 @@ fn select_opener() -> String {
 #[cfg(test)]
 mod tests {
     use figment::{
-        providers::{Format, Toml},
         Figment,
+        providers::{Format, Toml},
     };
 
     use super::BibiConfig;
@@ -382,6 +499,9 @@ mod tests {
                     author_color = "38"
                     title_color = "37"
                     year_color = "135"
+
+                    [citekey_formatter]
+                    ascii_only = true
                 "#,
             )?;