Merge pull request #7 from popzxc/parser-functions

Manishearth · web-flow · commit b0cda4054fde · 2021-06-28T07:25:22.000-07:00
Add function to parse script from string
diff --git a/scripts/unicode.py b/scripts/unicode.py
@@ -34,7 +34,16 @@
 
 #![allow(missing_docs, non_upper_case_globals, non_snake_case)]
 
-use super::ScriptExtension;
+pub use tables_impl::*;
+
+#[rustfmt::skip]
+mod tables_impl {
+use crate::ScriptExtension;
+'''
+
+# Close `mod impl {`
+ending='''
+}
 '''
 
 UNICODE_VERSION = (13, 0, 0)
@@ -239,7 +248,21 @@ def emit_enums(f, script_list, extension_list, longforms):
         f.write("    /// %s\n    pub const %s: ScriptExtension = %s;\n" % (longform, name, expr))
     f.write("""}
 
-impl Script {
+""")
+
+    # Generate implementation for the `Script`
+    generate_script_impl(f)
+
+
+def generate_script_impl(f):
+    """Generates an `impl Script { ... }` section with all the required functions"""
+
+    # Open `impl Script` section.
+    f.write("""impl Script {
+""")
+
+    # Generate impl of `inner_full_name`.
+    f.write("""
     #[inline]
     pub(crate) fn inner_full_name(self) -> &'static str {
         match self {
@@ -251,7 +274,26 @@ def emit_enums(f, script_list, extension_list, longforms):
         f.write("            Script::%s => \"%s\",\n" % (longforms[script], longforms[script]))
     f.write("""        }
     }
+""")
+
+    # Generate impl of `inner_from_full_name`.
+    f.write("""
+    #[inline]
+    pub(crate) fn inner_from_full_name(input: &str) -> Option<Self> {
+        match input {
+            "Unknown" => Some(Script::Unknown),
+            "Common" => Some(Script::Common),
+            "Inherited" => Some(Script::Inherited),
+""")
+    for script in script_list:
+        f.write("            \"%s\" => Some(Script::%s),\n" % (longforms[script], longforms[script]))
+    f.write("            _ => None,\n" )
+    f.write("""        }
+    }
+""")
 
+    # Generate impl of `inner_short_name`
+    f.write("""
     #[inline]
     pub(crate) fn inner_short_name(self) -> &'static str {
         match self {
@@ -263,7 +305,25 @@ def emit_enums(f, script_list, extension_list, longforms):
         f.write("            Script::%s => \"%s\",\n" % (longforms[script], script))
     f.write("""        }
     }
+""")
 
+    # Generate impl of `inner_from_short_name`
+    f.write("""
+    #[inline]
+    pub(crate) fn inner_from_short_name(input: &str) -> Option<Self> {
+        match input {
+            "Zyyy" => Some(Script::Common),
+            "Zinh" => Some(Script::Inherited),
+""")
+    for script in script_list:
+        f.write("            \"%s\" => Some(Script::%s),\n" % (script, longforms[script]))
+    f.write("""        _ => None,\n""")
+    f.write("""        }
+    }
+""")
+
+    # Generate impl of `for_integer`
+    f.write("""
     #[inline]
     pub(crate) fn for_integer(value: u8) -> Self {
         match value {
@@ -273,6 +333,10 @@ def emit_enums(f, script_list, extension_list, longforms):
     f.write("""            _ => unreachable!(),
         }
     }
+""")
+
+    # Close `impl Script` section
+    f.write("""
 }
 """)
 
@@ -281,8 +345,6 @@ def extension_name(ext):
     return "script_extensions::%s" % "_".join([e.upper() for e in ext])
 
 
-
-
 if __name__ == "__main__":
     r = "tables.rs"
     if os.path.exists(r):
@@ -336,3 +398,5 @@ def extension_name(ext):
                    is_pub=False , pfun=lambda x: "(%s,%s,%s)" % (escape_char(x[0]), escape_char(x[1]), extension_name(x[2])))
 
         # emit_table(rf, "FOObar", properties)
+
+        rf.write(ending)
diff --git a/src/lib.rs b/src/lib.rs
@@ -4,7 +4,6 @@
 #![cfg_attr(not(test), no_std)]
 #![cfg_attr(feature = "bench", feature(test))]
 
-#[rustfmt::skip]
 mod tables;
 
 use core::convert::TryFrom;
@@ -15,16 +14,30 @@ use tables::{get_script, get_script_extension, NEXT_SCRIPT};
 pub use tables::{Script, UNICODE_VERSION};
 
 impl Script {
-    /// Get the full name of a script
+    /// Get the full name of a script.
     pub fn full_name(self) -> &'static str {
         self.inner_full_name()
     }
 
-    /// Get the four-character short name of a script
+    /// Attempts to parse script name from the provided string.
+    /// Returns `None` if the provided string does not represent a valid
+    /// script full name.
+    pub fn from_full_name(input: &str) -> Option<Self> {
+        Self::inner_from_full_name(input)
+    }
+
+    /// Get the four-character short name of a script.
     pub fn short_name(self) -> &'static str {
         self.inner_short_name()
     }
 
+    /// Attempts to parse script name from the provided string.
+    /// Returns `None` if the provided string does not represent a valid
+    /// script four-character short name.
+    pub fn from_short_name(input: &str) -> Option<Self> {
+        Self::inner_from_short_name(input)
+    }
+
     /// Is this script "Recommended" according to
     /// [UAX #31](www.unicode.org/reports/tr31/#Table_Recommended_Scripts)?
     pub fn is_recommended(self) -> bool {
diff --git a/src/tables.rs b/src/tables.rs
