cli: add joining-type sub-command

PR #24

Author: wezm

src/app.rs

@@ -57,6 +57,11 @@ script-extension produces one table of Unicode codepoint ranges for each
 possible Script_Extension value.
 ";
 
+const ABOUT_JOINING_TYPE: &'static str = "\
+joining-type produces one table of Unicode codepoint ranges for each
+possible Joining_Type value.
+";
+
 const ABOUT_AGE: &'static str = "\
 age produces a table for each discrete Unicode age. Each table includes the
 codepoints that were added for that age. Tables can be emitted as a sorted
@@ -345,6 +350,25 @@ pub fn app() -> App<'static, 'static> {
             "List the properties that can be generated with this \
              command.",
         ));
+    let cmd_joining_type =
+        SubCommand::with_name("joining-type")
+            .author(crate_authors!())
+            .version(crate_version!())
+            .template(TEMPLATE_SUB)
+            .about("Create the Joining_Type property tables.")
+            .before_help(ABOUT_JOINING_TYPE)
+            .arg(ucd_dir.clone())
+            .arg(flag_fst_dir.clone())
+            .arg(flag_name("JOINING_TYPE"))
+            .arg(flag_chars.clone())
+            .arg(flag_trie_set.clone())
+            .arg(Arg::with_name("enum").long("enum").help(
+                "Emit a single table that maps codepoints to joining type.",
+            ))
+            .arg(Arg::with_name("rust-enum").long("rust-enum").help(
+                "Emit a Rust enum and a table that maps codepoints to \
+                 joining type.",
+            ));
     let cmd_prop_bool = SubCommand::with_name("property-bool")
         .author(crate_authors!())
         .version(crate_version!())
@@ -612,6 +636,7 @@ pub fn app() -> App<'static, 'static> {
         .subcommand(cmd_general_category)
         .subcommand(cmd_script)
         .subcommand(cmd_script_extension)
+        .subcommand(cmd_joining_type)
         .subcommand(cmd_age)
         .subcommand(cmd_bidi_mirroring_glyph)
         .subcommand(cmd_prop_bool)

src/general_category.rs

@@ -1,6 +1,6 @@
 use std::collections::{BTreeMap, BTreeSet};
 
-use ucd_parse::{self, UnicodeDataExpander};
+use ucd_parse::{self, UnicodeData, UnicodeDataExpander};
 
 use args::ArgMatches;
 use error::Result;
@@ -18,29 +18,8 @@ pub fn command(args: ArgMatches) -> Result<()> {
         return print_property_values(&propvals, "General_Category");
     }
 
-    // Expand all of our UnicodeData rows. This results in one big list of
-    // all assigned codepoints.
-    let rows: Vec<_> = UnicodeDataExpander::new(unexpanded).collect();
+    let mut bycat = expand_into_categories(unexpanded, &propvals)?;
 
-    // Collect each general category into an ordered set.
-    let mut bycat: BTreeMap<String, BTreeSet<u32>> = BTreeMap::new();
-    let mut assigned = BTreeSet::new();
-    for row in rows {
-        assigned.insert(row.codepoint.value());
-        let gc = propvals.canonical("gc", &row.general_category)?.to_string();
-        bycat
-            .entry(gc)
-            .or_insert(BTreeSet::new())
-            .insert(row.codepoint.value());
-    }
-    // As a special case, collect all unassigned codepoints.
-    let unassigned_name = propvals.canonical("gc", "unassigned")?.to_string();
-    bycat.insert(unassigned_name.clone(), BTreeSet::new());
-    for cp in 0..=0x10FFFF {
-        if !assigned.contains(&cp) {
-            bycat.get_mut(&unassigned_name).unwrap().insert(cp);
-        }
-    }
     // As another special case, collect all "related" groups of categories.
     // But don't do this when printing an enumeration, because in an
     // enumeration each codepoint should belong to exactly one category, which
@@ -74,6 +53,38 @@ pub fn command(args: ArgMatches) -> Result<()> {
     Ok(())
 }
 
+/// Expand a list of UnicodeData rows and group by category.
+pub fn expand_into_categories(
+    unexpanded: Vec<UnicodeData>,
+    propvals: &PropertyValues,
+) -> Result<BTreeMap<String, BTreeSet<u32>>> {
+    // Expand all of our UnicodeData rows. This results in one big list of
+    // all assigned codepoints.
+    let rows: Vec<_> = UnicodeDataExpander::new(unexpanded).collect();
+
+    // Collect each general category into an ordered set.
+    let mut bycat: BTreeMap<String, BTreeSet<u32>> = BTreeMap::new();
+    let mut assigned = BTreeSet::new();
+    for row in rows {
+        assigned.insert(row.codepoint.value());
+        let gc = propvals.canonical("gc", &row.general_category)?.to_string();
+        bycat
+            .entry(gc)
+            .or_insert(BTreeSet::new())
+            .insert(row.codepoint.value());
+    }
+    // As a special case, collect all unassigned codepoints.
+    let unassigned_name = propvals.canonical("gc", "unassigned")?.to_string();
+    bycat.insert(unassigned_name.clone(), BTreeSet::new());
+    for cp in 0..=0x10FFFF {
+        if !assigned.contains(&cp) {
+            bycat.get_mut(&unassigned_name).unwrap().insert(cp);
+        }
+    }
+
+    Ok(bycat)
+}
+
 /// Related returns a set of sets of codepoints corresponding to the "related"
 /// groups of categories defined by Table 12 in UAX#44 S5.7.1.
 ///

src/joining_type.rs

@@ -0,0 +1,73 @@
+use std::collections::{BTreeMap, BTreeSet};
+
+use ucd_parse::{self, ArabicShaping};
+
+use args::ArgMatches;
+use error::Result;
+use general_category;
+use util::PropertyValues;
+
+pub fn command(args: ArgMatches) -> Result<()> {
+    let dir = args.ucd_dir()?;
+    let propvals = PropertyValues::from_ucd_dir(&dir)?;
+    let rows: Vec<ArabicShaping> = ucd_parse::parse(&dir)?;
+    let unexpanded_gc = ucd_parse::parse(&dir)?;
+    let gc =
+        general_category::expand_into_categories(unexpanded_gc, &propvals)?;
+
+    // Collect each joining type into an ordered set.
+    let mut by_type: BTreeMap<String, BTreeSet<u32>> = BTreeMap::new();
+    let mut assigned = BTreeSet::new();
+    for row in rows {
+        assigned.insert(row.codepoint.value());
+        let jt =
+            propvals.canonical("jt", row.joining_type.as_str())?.to_string();
+        by_type
+            .entry(jt)
+            .or_insert(BTreeSet::new())
+            .insert(row.codepoint.value());
+    }
+    // Process the codepoints that are not listed as per the note in
+    // ArabicShaping.txt:
+    //
+    // Note: Code points that are not explicitly listed in this file are either
+    // of joining type T or U:
+    //
+    // - Those that are not explicitly listed and that are of General Category
+    //   Mn, Me, or Cf have joining type T.
+    // - All others not explicitly listed have joining type U.
+    let transparent_name = propvals.canonical("jt", "transparent")?;
+    let non_joining_name = propvals.canonical("jt", "non_joining")?;
+    let transparent_categories = ["Mn", "Me", "Cf"]
+        .iter()
+        .map(|cat| propvals.canonical("gc", cat).map(|name| &gc[&name]))
+        .collect::<Result<Vec<_>>>()?;
+    for cp in 0..=0x10FFFF {
+        if assigned.contains(&cp) {
+            continue;
+        }
+        // See if the code point is in any of the general categories that
+        // map to the Transparent joining type. Otherwise add to the
+        // Non_Joining type.
+        if transparent_categories.iter().any(|cat| cat.contains(&cp)) {
+            by_type.get_mut(&transparent_name).unwrap().insert(cp);
+        } else {
+            by_type.get_mut(&non_joining_name).unwrap().insert(cp);
+        }
+    }
+
+    let mut wtr = args.writer("joining_type")?;
+    if args.is_present("enum") {
+        wtr.ranges_to_enum(args.name(), &by_type)?;
+    } else if args.is_present("rust-enum") {
+        let variants = by_type.keys().map(String::as_str).collect::<Vec<_>>();
+        wtr.ranges_to_rust_enum(args.name(), &variants, &by_type)?;
+    } else {
+        wtr.names(by_type.keys())?;
+        for (name, set) in by_type {
+            wtr.ranges(&name, &set)?;
+        }
+    }
+
+    Ok(())
+}

src/main.rs

@@ -35,6 +35,7 @@ mod case_folding;
 mod case_mapping;
 mod general_category;
 mod jamo_short_name;
+mod joining_type;
 mod names;
 mod property_bool;
 mod regex;
@@ -74,6 +75,7 @@ fn run() -> Result<()> {
         ("jamo-short-name", Some(m)) => {
             jamo_short_name::command(ArgMatches::new(m))
         }
+        ("joining-type", Some(m)) => joining_type::command(ArgMatches::new(m)),
         ("names", Some(m)) => names::command(ArgMatches::new(m)),
         ("property-names", Some(m)) => cmd_property_names(ArgMatches::new(m)),
         ("property-values", Some(m)) => {

ucd-parse/src/arabic_shaping.rs

@@ -37,6 +37,19 @@ pub enum JoiningType {
     Transparent,
 }
 
+impl JoiningType {
+    pub fn as_str(&self) -> &str {
+        match self {
+            JoiningType::RightJoining => "R",
+            JoiningType::LeftJoining => "L",
+            JoiningType::DualJoining => "D",
+            JoiningType::JoinCausing => "C",
+            JoiningType::NonJoining => "U",
+            JoiningType::Transparent => "T",
+        }
+    }
+}
+
 impl Default for JoiningType {
     fn default() -> JoiningType {
         JoiningType::NonJoining