10000 Merge pull request #7 from popzxc/parser-functions · unicode-rs/unicode-script@b0cda40 · GitHub
[go: up one dir, main page]
Skip to content

Commit b0cda40

Browse files
ManishearthManishearth
authored
Merge pull request #7 from popzxc/parser-functions
Add function to parse script from string
2 parents 95d0115 + d2c5a10 commit b0cda40

File tree

3 files changed

+498
-86
lines changed

3 files changed

+498
-86
lines changed

‎scripts/unicode.py

Lines changed: 68 additions & 4 deletions
10000
Original file line numberDiff line numberDiff line change
@@ -34,7 +34,16 @@
3434
3535
#![allow(missing_docs, non_upper_case_globals, non_snake_case)]
3636
37-
use super::ScriptExtension;
37+
pub use tables_impl::*;
38+
39+
#[rustfmt::skip]
40+
mod tables_impl {
41+
use crate::ScriptExtension;
42+
'''
43+
44+
# Close `mod impl {`
45+
ending='''
46+
}
3847
'''
3948

4049
UNICODE_VERSION = (13, 0, 0)
@@ -239,7 +248,21 @@ def emit_enums(f, script_list, extension_list, longforms):
239248
f.write(" /// %s\n pub const %s: ScriptExtension = %s;\n" % (longform, name, expr))
240249
f.write("""}
241250
242-
impl Script {
251+
""")
252+
253+
# Generate implementation for the `Script`
254+
generate_script_impl(f)
255+
256+
257+
def generate_script_impl(f):
258+
"""Generates an `impl Script { ... }` section with all the required functions"""
259+
260+
# Open `impl Script` section.
261+
f.write("""impl Script {
262+
""")
263+
264+
# Generate impl of `inner_full_name`.
265+
f.write("""
243266
#[inline]
244267
pub(crate) fn inner_full_name(self) -> &'static str {
245268
match self {
@@ -251,7 +274,26 @@ def emit_enums(f, script_list, extension_list, longforms):
251274
f.write(" Script::%s => \"%s\",\n" % (longforms[script], longforms[script]))
252275
f.write(""" }
253276
}
277+
""")
278+
279+
# Generate impl of `inner_from_full_name`.
280+
f.write("""
281+
#[inline]
282+
pub(crate) fn inner_from_full_name(input: &str) -> Option<Self> {
283+
match input {
284+
"Unknown" => Some(Script::Unknown),
285+
"Common" => Some(Script::Common),
286+
"Inherited" => Some(Script::Inherited),
287+
""")
288+
for script in script_list:
289+
f.write(" \"%s\" => Some(Script::%s),\n" % (longforms[script], longforms[script]))
290+
f.write(" _ => None,\n" )
291+
f.write(""" }
292+
}
293+
""")
254294

295+
# Generate impl of `inner_short_name`
296+
f.write("""
255297
#[inline]
256298
pub(crate) fn inner_short_name(self) -> &'static str {
257299
match self {
@@ -263,7 +305,25 @@ def emit_enums(f, script_list, extension_list, longforms):
263305
f.write(" Script::%s => \"%s\",\n" % (longforms[script], script))
264306
f.write(""" }
265307
}
308+
""")
266309

310+
# Generate impl of `inner_from_short_name`
311+
f.write("""
312+
#[inline]
313+
pub(crate) fn inner_from_short_name(input: &str) -> Option<Self> {
314+
match input {
315+
"Zyyy" => Some(Script::Common),
316+
"Zinh" => Some(Script::Inherited),
317+
""")
318+
for script in script_list:
319+
f.write(" \"%s\" => Some(Script::%s),\n" % (script, longforms[script]))
320+
f.write(""" _ => None,\n""")
321+
f.write(""" }
322+
}
323+
""")
324+
325+
# Generate impl of `for_integer`
326+
f.write("""
267327
#[inline]
268328
pub(crate) fn for_integer(value: u8) -> Self {
269329
match value {
@@ -273,6 +333,10 @@ def emit_enums(f, script_list, extension_list, longforms):
273333
f.write(""" _ => unreachable!(),
274334
}
275335
}
336+
""")
337+
338+
# Close `impl Script` section
339< 57AE /td>+
f.write("""
276340
}
277341
""")
278342

@@ -281,8 +345,6 @@ def extension_name(ext):
281345
return "script_extensions::%s" % "_".join([e.upper() for e in ext])
282346

283347

284-
285-
286348
if __name__ == "__main__":
287349
r = "tables.rs"
288350
if os.path.exists(r):
@@ -336,3 +398,5 @@ def extension_name(ext):
336398
is_pub=False , pfun=lambda x: "(%s,%s,%s)" % (escape_char(x[0]), escape_char(x[1]), extension_name(x[2])))
337399

338400
# emit_table(rf, "FOObar", properties)
401+
402+
rf.write(ending)

‎src/lib.rs

Lines changed: 16 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -4,7 +4,6 @@
44
#![cfg_attr(not(test), no_std)]
55
#![cfg_attr(feature = "bench", feature(test))]
66

7-
#[rustfmt::skip]
87
mod tables;
98

109
use core::convert::TryFrom;
@@ -15,16 +14,30 @@ use tables::{get_script, get_script_extension, NEXT_SCRIPT};
1514
pub use tables::{Script, UNICODE_VERSION};
1615

1716
impl Script {
18-
/// Get the full name of a script
17+
/// Get the full name of a script.
1918
pub fn full_name(self) -> &'static str {
2019
self.inner_full_name()
2120
}
2221

23-
/// Get the four-character short name of a script
22+
/// Attempts to parse script name from the provided string.
23+
/// Returns `None` if the provided string does not represent a valid
24+
/// script full name.
25+
pub fn from_full_name(input: &str) -> Option<Self> {
26+
Self::inner_from_full_name(input)
27+
}
28+
29+
/// Get the four-character short name of a script.
2430
pub fn short_name(self) -> &'static str {
2531
self.inner_short_name()
2632
}
2733

34+
/// Attempts to parse script name from the provided string.
35+
/// Returns `None` if the provided string does not represent a valid
36+
/// script four-character short name.
37+
pub fn from_short_name(input: &str) -> Option<Self> {
38+
Self::inner_from_short_name(input)
39+
}
40+
2841
/// Is this script "Recommended" according to
2942
/// [UAX #31](www.unicode.org/reports/tr31/#Table_Recommended_Scripts)?
3043
pub fn is_recommended(self) -> bool {

0 commit comments

Comments
 (0)
0